CPET is clinically useful in the evaluation of patients with chronic obstructive pulmonary disease (COPD); when an objective determination of exercise capacity (V̇o₂peak) is necessary (see below) in establishing exercise limitations (71–74) and assessing other factors that may be contributing to exercise limitation (occult myocardial ischemia); in relating symptoms to exercise limitation (4, 5, 75, 76), especially when exertional symptoms are disproportionate to resting PFTs (5); and when hypoxemia may contribute to exercise limitation and O₂ requirements may be directly quantified (1, 3, 28, 42, 72, 77, 78).

Furthermore, CPET permits evaluation of the impact of therapeutic interventions on overall exercise capacity and components of the exercise response, especially as it relates to breathing strategies (45, 75), relief of dyspnea, and improvement in exercise tolerance (45, 72, 79). The efficacy of CPET in monitoring a variety of treatment modalities (continuous positive airway pressure, bronchodilators, exercise training, lung volume reduction surgery [LVRS], etc.) directed at improving breathing strategy and/or reducing dynamic hyperinflation (resulting in improved breathlessness and exercise capacity) has been demonstrated (21, 45, 75, 79–87) (see below). An endurance constant work exercise protocol (see Section III,3.3: Constant Work Rate Protocol) was more sensitive than the 6MWT in detecting the effects of therapeutic interventions (inhaled anticholinergic agents) on exercise performance in patients with COPD (88).

Using CPET in patients with COPD, studies have demonstrated that early onset metabolic acidosis is associated with skeletal muscle dysfunction (89) and that exercise alters amino acid metabolism including alterations in muscle oxidative capacity and exercise-related substrate levels, especially glutamate (90, 91).

CPET is efficacious in the early detection of subtle pulmonary gas exchange abnormalities not revealed by routine testing. This is important in establishing a timely diagnosis and accurate physiologic severity assessment, as well as in permitting the monitoring of therapeutic interventions (29, 31, 92–99). Whether CPET, in particular exercise pulmonary gas exchange, has prognostic value for interstitial lung disease (IPF) is controversial and requires additional investigation (100, 101).

In chronic pulmonary vascular disease the maximal oxygen consumption (V̇o₂max) provides an index of severity, being lower in patients with high pulmonary vascular resistance and lower cardiac index as well as being significantly correlated with the amount of functional vascular bed (102–104). More recent work in patients with primary pulmonary hypertension has confirmed that reductions in V̇o₂peak reflect reduced cardiac output (105) and functional capacity and that CPET and 6-minute walk tests provide complementary information in the evaluation of these patients (106). The presence of a patent foramen ovale and right-to-left shunting can also be diagnosed with the assistance of CPET while the patient respires 100% O₂. Because of significant mortality risk, exercise testing should be approached cautiously, especially in patients with primary pulmonary hypertension; if syncope, arrhythmia, and/or acute right heart failure is evident, exercise testing should not be performed (107). However, CPET can be performed safely in patients with primary pulmonary hypertension (105). Resting hemodynamic data correlate well with exercise results and may suffice in monitoring response to treatment (103, 108). Indications for CPET in the individual patient must reflect careful risk–benefit analysis.

Work has provided convincing evidence that the measurement of V̇o₂peak is valuable for prognosis and management of patients with cystic fibrosis (109). More recent work suggests that CPET results and estimates of muscle size (cross-sectional area) may provide an optimized exercise prescription for patients with cystic fibrosis (110).

CPET is useful in diagnosing suspected exercise-induced bronchospasm, especially when standard protocols and practices are modified to optimize conditions conducive for eliciting symptoms and monitoring responses (serial spirometry before and after exercise) (24, 111–113). A negative test, however, does not preclude the diagnosis of exercise-induced bronchospasm. Functional evaluation (CPET) of patients with asthma may be a useful tool for encouraging increased physical activity and for optimizing exercise prescription (114).

Whereas routine pulmonary function tests (FEV₁, diffusing capacity of the lung for CO [Dl_CO]) have the greatest utility in documenting physiologic operability in low-risk patients, other diagnostic modalities, including CPET and/or split function assessment by quantitative lung scintigraphy, is/are often necessary to more accurately assess moderate- to high-risk patients (20, 115, 116). Although there are proponents for both approaches (117–120), work suggests that CPET and the measurement of V̇o₂peak (especially when expressed as a percentage of the predicted value) appear to be particularly useful in predicting postoperative pulmonary complications (121–123). A V̇o₂peak less than 50–60% predicted is associated with higher morbidity and mortality after lung resection (121–123).

Preoperative CPET and split function study results may be used in tandem to predict the risk of postoperative pulmonary complications and functional capacity; this may be most beneficial to “borderline” patients who might otherwise be precluded from surgery (124–127). In addition, CPET permits the detection of clinically occult heart disease and provides a more reliable estimate of functional capacity postoperatively compared with PFTs, which routinely overestimate functional loss after lung resection (122). Prospective validation of an algorithm for the functional assessment of lung resection patients has demonstrated that morbidity and mortality were reduced by one-half without unnecessarily excluding patients from surgery (128). These algorithms have proved highly sensitive for excluding morbidity from conventional surgery, although in the context of a potentially life-saving cancer operation patients and physicians may decide to incur a somewhat greater risk than the algorithm allows. Furthermore, different surgical approaches and advances in less invasive surgical techniques and postoperative care may result in interinstitutional differences in risk. Continued prospective validation of algorithms for preoperative functional assessment for lung resection surgery is necessary.

The potential utility of CPET is highlighted by its emergence as an important tool in the evaluation of emphysema patients being considered for LVRS. The range of application of CPET in this patient group includes the determination of cardiopulmonary functional status and assessment of potential operative risk before surgery, the determination of exercise training prescription before and after LVRS, the quantification and monitoring of the clinical response to surgery, and also the definition of underlying pathophysiologic mechanisms responsible for improvements in exercise performance resulting from LVRS.

As LVRS continues to be actively investigated, an expanding CPET database from several studies has yielded improvements in the quantification of many clinically relevant exercise performance variables after LVRS, which may not be revealed in standard pulmonary function studies (81, 129–135). Significant improvement in right ventricular performance, particularly during exercise, has been reported at 6 months after bilateral LVRS (136). LVRS remains controversial largely because of uncertainties with respect to patient selection and long-term outcome (137). The National Institutes of Health-sponsored multicenter National Emphysema Treatment Trial, evaluating the efficacy of LVRS, has chosen the maximal work rate derived from CPET as its primary physiologic outcome parameter, because it was considered to be the best objective measure of functional status.

With the emergence of lung and heart–lung transplantation for patients with end-stage pulmonary vascular and parenchymal lung disease as a viable therapeutic option, comprehensive CPET is increasingly being used to evaluate these complex patients before and after transplantation. As such, CPET is useful in assessing disease progression before transplantation, and in assessing functional capacity, quantitating causes of exercise limitation, and providing exercise prescription for pulmonary rehabilitation before and after transplantation (138, 139). However, there is presently no consensus on how indices of exercise performance may impact the clinical decision-making process for lung transplantation selection. As previously noted, selection guidelines for cardiac transplantation based on exercise performance (V̇o₂max) have been established (12, 44, 59).

From a clinical perspective, integrative CPET results in the transplantation arena have reinforced the importance of the multifactorial etiology of exercise limitation and that of skeletal muscle dysfunction in patients with heart disease (140, 141) and chronic lung disease (1, 138, 139, 142–147). Furthermore, although a 6MWT distance of less than 400 m is useful in listing for lung transplantation (148), the information provided is imprecise in answering important questions related to patient management. Pretransplantation CPET results demonstrate severe exercise intolerance related to ventilatory and/or circulatory limitation. Despite significant improvement in quality of life, PFTs, and hemodynamics with transplantation, considerable exercise limitation persists (V̇o₂ approximately 45–55% predicted) and is remarkably similar for single-lung, double-lung, or heart–lung transplantation (138, 139, 145, 147). Cardiac and ventilatory factors are usually not limiting, although abnormal breathing patterns have been observed (145, 149–151). In turn, peripheral factors (peripheral circulation and/or peripheral muscle), especially skeletal muscle dysfunction, are purportedly primarily responsible (144, 146, 147). However, as noted, exercise limitation is most probably multifactorial.

Work has shown that CPET is helpful in objectively assessing the adequacy of cardiovascular reserve and in predicting cardiovascular risk in an elderly population (152).

Two modes of exercise are commonly employed in cardiopulmonary exercise tests: treadmill and cycle ergometer (Table 2). The motor-driven treadmill imposes progressively increasing exercise stress through a combination of speed and grade (elevation) increases. Several incremental protocols are popular and the choice among them depends on the objectives of the test and the degree of the patient's debilitation (178–183). Treadmill exercise testing has several advantages over cycle ergometry. For most individuals, treadmill walking is a more familiar activity than cycling. Walking on the treadmill, however, is more complex than ordinary walking, as evidenced by differences in 6MWT distance results among subjects performing on a treadmill versus walking (156). That fact notwithstanding, walking/running permits a larger muscle mass to be brought to bear during maximal treadmill exercise and more work against gravity is done, leading to greater stress on the organ systems mediating the exercise response. Consequently, on average, maximal oxygen uptake is reported to be 5–10% higher on a treadmill than on a cycle ergometer (184–187). This may be important for athletes, in whom the determination of V̇o₂max is critical, and in some patients in whom abnormalities (e.g., cardiac ischemia) may occur only with the highest metabolic demand. If exercise testing is being used to provide a prescription for subsequent exercise training, then it may be advantageous to use the same exercise modality in testing as for training.

The main disadvantage of treadmill exercise testing is that it is difficult to accurately quantify the external work rate of the subject during treadmill exercise. The relationship between speed and grade of the treadmill and the metabolic cost (V̇o₂) of performing work is only an estimate, and therefore it is difficult to predict V̇o₂ from treadmill testing (12). The weight of the subject and pacing strategy are important determinants in this regard. Body weight has much less effect on bicycle ergometer performance. Holding onto the treadmill handrails can alter (usually decreases) the metabolic cost of treadmill walking and should be discouraged whenever possible.

The cycle ergometer is generally less expensive and requires less space than the treadmill. It is also less prone to introduce movement or noise artifacts into measurements (e.g., ECG and blood pressure auscultation are generally easier). The principal advantage, however, is that the rate at which external work is performed is easily quantitated. There is a modestly greater metabolic requirement for moving heavier legs in obese individuals (3, 188–190), about 5.8 ml/minute per kilogram body weight (190); but as long as the pedaling cadence is kept constant, this represents a constant offset. The predictability of the relationship between work rate and metabolic energy expenditure is important for diagnosis (see Section IV,1.1: V̇o₂–Work Rate Relationship).

There are two types of cycle ergometers. Mechanically braked cycles regulate external work by adjustable frictional bands. Electrically braked cycles increase resistance to pedaling electromagnetically. Friction-braked cycle ergometers (191) generally do not offer sufficiently precise work rate settings and also require the subject to pedal at a fixed cadence to keep the work rate constant for a given setting of the bands. In contrast, the electrically braked cycle ergometer (192) allows direct quantification of the work rate performed and can be computer controlled; this allows the work rate to be incremented automatically and even continuously (e.g., “ramp pattern”) (193–200). This type of ergometer is often constructed so that moderate changes in pedaling rate (40 to 70 rpm) do not influence the work rate performed. Cycle ergometers have become available that allow true unloaded pedaling (199, 200). The internal resistance of the ergometer is overcome by means of a motor “assist,” so that patients do not have to pedal against flywheel inertia as exercise begins; this is especially important for the most debilitated patients.

On occasion, some patients are unable to perform lower extremity exercise. For such patients, arm crank ergometers can be adapted for incremental exercise testing. However, the metabolic stress that can be induced during arm cranking is limited; in healthy subjects peak V̇o₂ averages roughly 70% of that achievable during lower extremity exercise (201–204) and lactic acidosis is often evident at low work rates (201). It should be noted that patients with lung disease tolerate arm cranking poorly; arm cranking interferes with the use of the accessory muscles of respiration.

Recommendation: In most clinical circumstances, cycle ergometry is the preferable mode of exercise; however, depending on the reason(s) for which CPET was requested and equipment availability, a treadmill may be an acceptable alternative.

A number of flow-transducing devices have been adapted for exercise testing. The choice of a flow versus a volume transducer is no longer crucial because numeric integration or differentiation can be employed to calculate one entity from knowledge of the other. Digital computer processing can accommodate a nonlinear relationship between a flow or volume and transducer output. Transducers used for exercise testing must be lightweight, have low dead space, and low resistance to flow in the range of flows encountered during exercise and be immune to effects of water vapor or pools of saliva that may accumulate during testing. A key consideration is whether the transducer can be positioned near the mouth. Such transducers are capable of sensing flow or volume bidirectionally. They also eliminate the need for a nonrebreathing valve and this reduces the system dead space. All transducers listed below, with the exception of the pneumotachograph, can be used in a bidirectional configuration. Accurate values for ventilation and metabolic parameters obtained from exercise testing are critically dependent on the accuracy of the flow-sensing device. For this reason, accurate calibration is essential, and software must provide easy methods for either changing calibration factors or verifying the accuracy of current calibration before each test.

The ATS has established standards for flow and volume measurement in the context of spirometry (19). The transducers used in exercise testing should also meet these standards (25).

Transducers currently employed for measuring flow or volume in cardiopulmonary exercise testing are listed in the following sections. A summary of the major advantages and disadvantages of each is given in Table 3.

Suggested standards for gas analyzer performance in different modes of metabolic rate measurement are listed in Table 4. Two types of gas analyzers can be used: a mass spectrometer (considered the “gold” standard), which is capable of measuring all the required respiratory gases (CO₂, O₂ and, for some purposes, N₂), and separate analyzers for O₂ and CO₂.

The dynamics of analyzer response have two separable components: transport delay (the time required for gas to traverse the distance from the sampling site to the analyzer) and analyzer response (the kinetics of response to a change in gas composition introduced into the analyzer). Computer software allows compensation for gas transport delay, generally on the order of 0.2 to 0.4 second, depending on the length of the gas-sampling tube and the gas-sampling rate. The analyzer response, often taking the form of a sigmoid or exponential response to a stepwise change in gas composition, is extremely difficult to fully compensate (213, 214); the time constant of this exponential response must be kept as short as possible. An additional concern is sensitivity of the analyzer to water vapor partial pressure in the sampled gas. Because water vapor partial pressure in the sampled gas can be difficult to predict (principally because the gas temperature is difficult to predict), this can introduce substantial errors in metabolic rate calculations (see Beaver [215] for discussion). Some analyzers require gas to be physically dried before it reaches the analyzer.

The mass spectrometer ionizes gas molecules in a high-vacuum environment and then separates them on the basis of mass-to-charge ratio. This enables the measurement of a number of gases. These analyzers are linear, often highly stable, and have rapid response characteristics (analyzer half-times of response of roughly 25–50 milliseconds). They are configured to “ignore” water vapor, yielding “dry gas fractions.” However, the high cost of mass spectrometers has inhibited their use in most commercial cardiopulmonary exercise systems.

Discrete O₂ and CO₂ analyzers have been modified specifically for the demands of cardiopulmonary exercise testing. Carbon dioxide analyzers, based on absorption of infrared light by CO₂ (216), are common. Oxygen analyzers based on two principles have been employed. In paramagnetic analyzers, the effect of oxygen molecules on a magnetic field is utilized. In the electrochemical (“fuel cell”) analyzer, high-temperature reactions between O₂ and substrate are measured. These analyzers have potential disadvantages. The analyzer output is usually not a linear function of gas concentration; however, computerized correction can be made for these nonlinearities. The sensitivity of these analyzers to water vapor content has been circumvented by using sampling tubing (polymer Nafion) that absorbs water. The gas that reaches the analyzer therefore contains little water vapor. Failure of the drying process can be a source of errors in measurement of V̇o₂ and V̇co₂.

Conceptually, oxygen uptake (V̇o₂) and carbon dioxide output (V̇co₂) each represent the difference between the volume of gas (O₂ and CO₂, respectively) inhaled and the volume exhaled per unit of time. Under steady state conditions V̇o₂ and V̇co₂ will be equal to the rate of metabolic O₂ consumption and CO₂ production. The measurement of V̇o₂ is based on the mass balance equation:

Vi and V̇e represent the volumes of inhaled and exhaled gas, respectively, and t is the time period of the gas volume measurement. Fi_O2 and Fe_O2 represent the O₂ concentration in the inhaled and “mixed” exhaled gas, respectively. Vi, however, is not commonly measured; rather, it is calculated from V̇e on the assumption that the virtually insoluble gas N₂ is neither absorbed into nor discharged from the capillary blood:

The measurement of V̇co₂ is simpler, because Fi_CO2 in room air is practically zero and may be safely ignored in the calculation:

1.4.3. Breath-by-breath mode.

With the ready availability of online digital computer analysis of physiologic measurements (Figure 1), it has become practical to compute V̇co₂ and V̇o₂ on a breath-by-breath basis (223, 224). Utilizing algorithms first reported in 1973 (223), a signal proportional to expired airflow and signals proportional to fractional concentrations of CO₂ and O₂ measured near the mouth are typically sampled 50 or 100 times per second. In this way, each breath is broken down into a large number of parts and the O₂ uptake and CO₂ output are calculated for each interval (Figure 2). These measurements are summed over the entire expiration to compute the total volume of O₂ uptake and CO₂ output per breath. The values of each breath are extrapolated to the minute.

where V̇e is the instantaneous expired airflow rate; Δt is the sampling interval; and Fi_O2, Fe_O2, and Fe_CO2 are the fractional instantaneous concentrations of inspired O₂, expired O₂, and expired CO₂, respectively.

These calculations must accommodate water vapor, barometric pressure, and ambient temperature variations (see Beaver [215]). As importantly, compensation is necessary for the delay between the time at which gas is sampled at the mouth and the time at which the gas concentration is measured within the gas analyzers (usually on the order of 0.2–0.4 second). Breath-by-breath analysis, therefore, requires precise knowledge of gas analyzer delays and response kinetics (224).

Although breath-by-breath data collection/analysis is presently the most popular, it is important to recognize that the confidence with which these “metabolic” indicators can be calculated depends on the combined measurement errors for each of the determined variables. Unless great care is taken with the calibration of the sensors, these errors can be additive and large. A further concern is the assumptions of the algorithms (225–227). For example, breath-by-breath changes in end-expiratory lung volume (EELV) will violate the strict requirement that V̇o₂ be measured in expired-only gas, and substantial error can be introduced with each breath in which EELV changes; however, these effects are generally damped out by averaging over time (228–230). Several factors can also introduce additional errors in the measurements. In some patients, the concentrations recorded during each expiration may not represent the composition of the average alveolar gas (i.e., in patients with COPD and other patients with inhomogeneous distribution of ventilation). Furthermore, inaccurate integration of the three signals (flow, Fo₂, and Fco₂) can also occur in patients who do not have a uniform and smooth breathing pattern, due to circulatory oscillations, as has been reported in patients with heart failure (231).

To improve the reliability of breath-by-breath measurements, algorithms have been developed and implemented (225, 227) to enable breath-by-breath compensation for changes in lung gas stores. This allows approximation of the rate of gas transfer between the airspaces and the pulmonary capillaries.

Recommendations: Both breath-by-breath and mixing chamber data collection/analysis of CPET results can be used for clinical purposes. The convenience of the breath-by-breath methodology and the flexibility in the treatment of the data have made this technique especially attractive and widely utilized. However, there is considerable potential for erroneous data. The Douglas bag method should be used for quality control of the automated systems with minimal quality assurance standards developed for use by industry.

Heart rate usually is measured from the R-R interval of the electrocardiogram. A key requirement is that the electrodes and the detection electronics be specially designed for movement artifact rejection. For optimal detection of myocardial ischemia and cardiac arrhythmias during exercise, serial 12-lead electrocardiograms should be monitored in most clinical exercise tests (7, 191). However, in some exercise tests, three lead electrocardiograms may be used to monitor for rhythm disturbances and to screen for ischemia. Computerized systems enabling continuous CRT display contribute to test safety; averaging of ECG complexes is not recommended and it should be used only as a supplement to the raw data. Skin preparation and use of sweat-resistant adhesive electrodes are important to decrease the noise on the ECG tracing. The 12-lead ECG electrodes positioning proposed by Mason and Likar (233) best resembles the standard ECG.

Auscultation of blood pressure becomes more difficult during exercise because of the increase in ambient noise and movement artifacts. Yet detection of exercise-induced hypertension (or, less commonly, hypotension) is an important goal in many circumstances (191). Automated blood pressure measurement systems have been developed specifically for use during exercise. Many operate on the oscillometric method, in which, as the cuff is automatically deflated in stages, pressure oscillations induced within the cuff by pulsations in the arm are detected (234). Despite algorithms designed to decrease the effects of artifacts, blood pressure measurements may be inaccurate when, for example, the arm is flexed during the measurement cycle. Periodic checks against manual determinations are therefore important.

For studies in which an arterial catheter is inserted to facilitate blood sampling, it may be useful to measure blood pressure directly. It should be appreciated that intraarterial blood pressure measurements are modestly higher than those produced by auscultation, particularly in systolic pressure measurements. The diastolic blood pressure does not change or slightly decreases on exercise with auscultation but slightly rises with intraarterial measurement (235, 236). Miniature transducers that can be attached to the thorax or to the arm while the subject exercises are available. Meticulous attention to technique (e.g., exclusion of air bubbles) is necessary to assure good frequency response. The transducer should be placed and “zeroed” at the left atrium level (fourth intercostal space at the midclavicular line). Presently, most laboratories employ single-use disposable blood pressure transducers.

Pulse oximeters detect the variation in transmission of light of two different wavelengths that occur with arterial pulsations in an extremity (usually the finger or ear lobe). Because oxygenated and reduced hemoglobin transmit certain light wavelengths differently, this information is used to estimate arterial O₂ saturation (237). Although useful and convenient for continuous monitoring (238, 239), awareness of several important issues is necessary. In general, pulse oximeters have reasonable accuracy: 95% confidence limits of ± 4–5% as compared with directly measured arterial O₂ saturation (240), provided that a good pulse signal is obtained. The measurement is thought to be less accurate at saturations below about 88% (241), which is exacerbated in black patients (242).

Some authors have reported that pulse oximeters tend to overestimate true arterial O₂ saturation (77, 243, 244). On the other hand, poor perfusion of the extremity (yielding decreased pulsatility), which may occur in cardiovascular disease, may yield falsely low readings (245). Movement and stray light can yield artifacts. Dark skin color can interfere with signal detection (242, 246). These devices cannot detect carboxyhemoglobin or methemoglobin; the calculations approximate the oxygenated fraction of available hemoglobin. An additional disadvantage of pulse oximetry is that arterial O₂ saturation rather than Po₂ is measured. Arterial Po₂ is more relevant in assessing the effects of lung disease on pulmonary gas exchange. It should be noted that despite a fall in arterial Po₂ to 70 mm Hg, saturation would still remain above 93%, as the O₂ dissociation curve at this point is insensitive to changes in Po₂. In general, pulse oximeters are good for monitoring trending phenomenon but not reliable for determining absolute magnitude of change (247). Quality assurance for pulse oximeters should include validation with arterial oxygen saturation. Significant desaturation (i.e., a change in arterial oxygen saturation [ΔSp_O2] ⩾ 5%) should be confirmed with arterial blood gases (13, 25).

For cardiopulmonary exercise testing systems purchased as a unit, the manufacturer bears the responsibility for demonstrating that the system is accurate and precise. This should include description of the methods used in their validation. If possible, the ventilatory and gas exchange measurements of each unit should be validated by the bag collection method, with the gas concentrations analyzed by Scholander or mass spectrometry techniques. Minimal equipment requirements are presented in Table 5. Ideally, independent reference laboratories would validate CPET systems akin to the process available for spirometers.

However, it must be stressed that the user bears the responsibility for assuring that measurements remain accurate. Cardiopulmonary exercise testing, especially when it features breath-by-breath gas exchange analysis, requires meticulous attention to calibration procedures to assure accurate and reproducible measurements. A good practice is to calibrate the system daily and to maintain a calibration log book so that long-term trends can be monitored.

Daily calibration begins with the determination of ambient barometric pressure, temperature, and relative humidity. The exercise system must be calibrated daily (or ideally before each test if more than one test is done) to check the operation of key transducers. Verification of calibration of the air flow or volume transducer can be performed with a calibrated 3-L syringe. A range of flow rates should be performed to simulate the wide range of flow rates that occur in going from rest to heavy exercise; syringe strokes varying from less than 1 to 15 seconds in duration cover most of this range. Agreement in calculated volumes to within ± 3% signifies adequate performance.

Although the output of most CO₂ and O₂ gas analyzers is a nonlinear function of gas concentrations, electronic algorithms aim to create linear outputs over the desired operating range. For CO₂, this is 0–8%; for oxygen this is 13–21% (unless testing while breathing hyperoxic gas mixtures is intended). Daily two-point calibrations of each analyzer with two precision-analyzed gas mixtures should be performed. Typically this is done with one 6% CO₂ and 15% O₂ tank and one 0% CO₂ and 21% O₂ tank. On occasion, additional certified tanks with other relevant compositions should be used to verify linearity. A good practice is to maintain a single precision gas cylinder (grandfather) for occasional use; such a tank can last a number of years and provide long-term validation of calibration accuracy.

A third routine calibration must be performed in systems featuring breath-by-breath gas exchange measurements. The transport delays between the gas sampling point and each gas analyzer need to be known with precision so that the respiratory airflow and gas concentration signals can be properly time aligned. A solenoid allows an abrupt switch between two gas sources with different O₂ and CO₂ compositions, and the time delays between solenoid activation and the detection of change in gas analyzer output are measured. This should be an automated process provided by the manufacturer so that the delay time can be checked or changed on a daily basis.

Gas exchange simulators/calibrators have been developed (248–250) for quality control of gas exchange measurements on automated systems. They feature a reciprocating piston; injection of a precision gas mixture at a precisely metered rate yields simulation of known V̇e, V̇co₂, and V̇o₂. Day-to-day variation of these calculations and the variation with changes in pump rate should be roughly in the range of ± 3%. However, these simulators have not been well validated. It should be noted that these gas exchange simulators do not simulate the normal variation in breathing pattern waveforms nor do they simulate moist, room temperature exhalate. Thus, effectiveness of drying exhaled air before analysis, or application of temperature and humidity corrections, are not tested. Therefore, these calibrators are perhaps most useful for the detection of variations in the performance of the system (precision) rather than accuracy.

Other calibration procedures need to be performed, but on a less frequent basis. Blood pressure transducers can be calibrated periodically with a mercury manometer, but with modern disposable units this calibration is not meaningful unless performed daily, which is often not practical. Most electrically braked cycles have static calibration procedures that can be performed as a checkup procedure. However, some errors can be introduced by several factors, especially motion. That is why the electrically braked cycle ergometer requires dynamic calibration with the use of a dynamometer (torque meter) (251–253), which should be performed yearly or whenever the cycle ergometer is moved (jarring often disturbs the calibration). For most clinical testing purposes, the calibration should be linear in the range of 0 to 400 W. Because most laboratories do not have access to a dynamometer, cycle ergometer manufacturers should provide this service. Currently, this is also available from independent commercial vendors. For treadmills, belt speed should be verified by timing revolutions of the belt with a subject on the treadmill; accuracy of the grade indication should also be validated (191).

There are two other overall calibration checks that are advisable. The first is a physiologic/biologic validation (254), in which a healthy member of the laboratory staff, consuming a stable diet, performs a constant work rate test at varying workloads (50 and 100 or 150 W, etc.) at regular intervals depending on machine use (247). Subsequent steady state values for V̇e, V̇o₂, and V̇co₂ are then compared with the database and values outside the 95% CI for that individual should engender a thorough system-wide reassessment. If within tolerance, they are then added to the quality control database. Finally, timed expired gas collections (Douglas bag) made during steady state exercise can be used as a gold standard to validate ventilation and gas exchange measurements. Although laborious, when carefully done this method is generally accurate to within 2–3% for healthy subjects undergoing moderate to heavy exercise.

It is important to take into consideration the reproducibility of the variables measured during clinical exercise testing for appropriate interpretation of the results (Table 6). Factors that may contribute to variability in these measurements include the following:

1. Changes in the underlying disease process(es)

2. Changes in medication

3. Patient motivation

4. Patient instructions/inducement

5. The time of day

6. Testing procedures

7. Equipment/calibration errors

Care must be taken to ensure that these factors, which may contribute to alteration of measured exercise responses, are meticulously controlled.

A number of studies have closely examined the variability of measurements obtained during clinical exercise testing. The results of studies of normal subjects (255–257), of patients with chronic obstructive pulmonary disease (COPD) (258–263), of patients with interstitial lung disease (ILD) (264), and of patients with chronic heart failure (265–267) are outlined and summarized in Table 6.

In addition to specific guidelines for individual measurements, a number of other factors may influence the reproducibility of results. An important factor to consider is that of a potential learning effect (Table 6), and therefore the need for preliminary/familiarization testing. Various studies have provided conflicting results, with some reporting a significant change with repeated testing (257, 258, 261, 265), and others reporting no significant change (255, 256, 259, 260, 262–264, 266). A number of reasons may explain these discrepancies, and include whether the test was submaximal (i.e., not maximal and symptom limited), whether repeated testing was undertaken within a short period of time (i.e., four tests within 7–10 days), and whether there were changes in the underlying disease process (e.g., chronic heart failure). An additional factor that may influence the reproducibility of measurements is the time of testing. Preferably, repeated testing should be undertaken at the same time of day, as significant diurnal variation in results has been reported (255). Furthermore, the testing protocol, procedure, and instructions to the patient must be rigidly controlled, as these have been shown to significantly affect performance (258, 268). Finally, disease severity may also affect the variability of some measurements during exercise (266), and may affect the interpretation of results in some patients with more severe disease.

This protocol is widely used in clinical practice (Figure 3)

Figure 3. Flow chart of a maximal symptom-limited cardiopulmonary incremental protocol on a cycle ergometer. ECG = electrocardiogram; PFT = pulmonary function test.

Download Figure | Download Powerpoint

[More] [Minimize]

Exercise tests in which the incremental phase lasts 8–12 minutes are efficient and provide useful diagnostic information (182). In healthy subjects (healthy in the sense that the predicted V̇o₂max is achieved) it is possible to estimate the rate of work rate increment that will yield an incremental phase of about 10 minutes in duration. Because the V̇o₂–work rate relationship is approximately linear for incremental exercise and has a slope of about 10 ml/minute per watt, the V̇o₂ after 10 minutes of incremental exercise will be

Prediction equations are available for V̇o₂max and V̇o₂unl (3). Deviation from the watts per minute calculated using the equation should be based on clinical judgment and knowledge of the subject's physical activity. For the subject who appears to be fit, work rate increments as high as 25–30 W/minute may be selected. For debilitated patients, lower work rate increments (e.g., 5 W/minute) may be selected.

New protocols have been introduced for clinical exercise testing; among them is the standardized exponential exercise protocol, which was designed to allow a single protocol to be used for subjects with a wide range of exercise capacities (279, 280). This protocol is suitable for use on either a cycle ergometer or a treadmill. In this protocol the work rate is increased exponentially by 15% of the previous workload every minute. Because bicycle exercise is relatively independent of body weight, the protocol is adjusted for different body weights. This protocol should elicit similar V̇o₂ in ml/kg per minute at each stage for different individuals; the most severely debilitated patients will exercise for fewer minutes and the more fit for about 15 minutes. No clear advantage has been shown in the use of this protocol over conventional protocols.

If a treadmill is used (281), an incremental protocol, similar to that for a cycle ergometer, may be used. The lowest treadmill speed (e.g., 0.6–1.0 mph) may be used for the exercise baseline. The work rate can than be incremented at regular intervals with a combination of speed and grade. The Bruce protocol, originally designed for testing and for the evaluation of coronary artery disease (282), may be used for more functional subjects. However, the work rate of its first stage (5 metabolic equivalents [METs]) and the subsequent increments (2 to 4 METs) are high (i.e., equivalent to increases of about 50 W per stage for an average person) and may not be achievable in patients with moderate to severe cardiac and/or pulmonary disease. For these patients, a modified Naughton protocol may be utilized, in which the initial work rate and subsequent increments are more than 1 but less than 2 METs (283). A drawback of both protocols, however, is the relatively long increment duration (3 minutes), which is likely to impair noninvasive assessment of the lactate threshold.

For cardiopulmonary measurements, the Balke protocol (284), in which the speed is kept constant at 3.3 mph, and elevation is increased by 1% every minute, and the modified Balke protocol, in which a fixed treadmill speed is chosen and the treadmill grade is increased by a constant amount each minute (3, 285), are the most appropriate. These protocols most closely approximate a constant rate of increase in work rate. Other incremental treadmill protocols that combine speed and grade changes have also been used for evaluating patients with pulmonary disease (286). As mentioned previously, the standardized exponential exercise protocol has also been designed for use with treadmills (279, 280). At each stage either the treadmill speed or the elevation is increased to elicit a 15% increase in work rate. The same protocol can be used for individuals of different body weights and obtain similar V̇o₂ in ml/kg per minute for each stage. Again, no clear advantage has been shown in the use of this protocol over more conventional and well-established exercise protocols for CPET.

In cardiology, and especially when treadmill protocols are used, it is common to express the metabolic requirement for external work as the metabolic equivalent (MET). “MET” is defined as the equivalent of the resting metabolic oxygen requirement. One metabolic equivalent equals 3.5 ml/kg per minute. Exercise capacity (in METS) is a powerful predictor of mortality among men referred for exercise testing (287)

This protocol is gaining popularity because of its clinical applicability, particularly for monitoring response(s) to a spectrum of therapeutic interventions including cardiopulmonary rehabilitation, bronchodilators LVRS, medical devices, and so on (88). It is also useful for the analysis of exercise tidal flow–volume loops and dynamic hyperinflation (45, 87, 247, 288), gas exchange kinetics (289–291), and validation of pulmonary gas exchange during incremental exercise testing (IET) (275).

If additional pulmonary gas exchange information (ABG data) is required after an IET has been performed, it may be appropriate to perform a constant work rate protocol (276). Treadmill or cycle ergometry exercise may be used at levels approximating the subject's usual daily activities (e.g., up to 3.0 mph on a treadmill, or up to 50 W on a cycle ergometer). A constant work rate may be performed about 1 hour after an IET. This test should involve at least 6 minutes of continuous exercise. Alternatively, using 50 to 70% of the maximal work rate achieved during an incremental exercise, a constant work rate test for 5 to 10 minutes often achieves about 70 to 90% of V̇o₂max achieved during IET. ABGs obtained at Minute 5 by single radial artery puncture approximate nearly maximal IET values and may provide an alternative to an arterial line (275, 276). Additional studies are required.

The clinical exercise test order should include reasons for the test and clinical diagnosis with a summary of the medical history (Table 7)

The following are required: (1) spirometry and maximal voluntary ventilation (MVV) should be measured, and lung volumes and Dl_CO can be included if clinically warranted; (2) if hypoxemia is clinically suspected, resting arterial blood gases should be obtained; (3) a recent hemogram and electrolytes should be determined, if warranted; (4) patients who smoke should be asked to abstain from smoking for at least 8 hours; (5) consultation with a cardiologist is recommended, when appropriate, for patients with a history of coronary artery disease; (6) for functional evaluation and disability, patients should be tested with their optimal medication regimen; (7) the morning of the test, patients should not exercise and should have a light breakfast no less than 2 hours before the test; and (8) patients should come to the laboratory in exercise clothing, including tennis shoes.

On the basis of the preliminary evaluation, a decision must be made concerning whether an invasive test is warranted (see Section III,2: Exercise Test with Arterial Blood Sampling). A consent form must be signed. With patient properly dressed, place the ECG electrodes and indwelling arterial catheter, if necessary. A supine resting 12-lead ECG should be obtained and used as the “standard” ECG tracing for determination of resting abnormalities before exercise testing (293).

Normally, V̇o₂ increases nearly linearly as external work (power output) increases. However, accurate determination of the external work rate in watts (or kilopond · meters [kpm] per minute) is required to determine this relationship. The external work rate is accurately measured by cycle ergometry but can only be estimated by treadmill exercise. The slope of V̇o₂ versus external work rate reflects the efficiency of the metabolic conversion of chemical potential energy to mechanical work and the mechanical efficiency of the musculoskeletal system. The slope determined from the rate of change in V̇o₂ divided by the rate of change in external work during incremental exercise testing on a cycle ergometer (ΔV̇o₂/ΔWR) is normally about 8.5–11 ml/minute per watt (3, 195) and is independent of sex, age, or height. Obese individuals may show an increase in V̇o₂ for a given external work rate, but the rate of rise in V̇o₂ with increasing external work rate (slope) is normal (301, 302). Because there are few processes that affect the metabolic efficiency of muscles, a reduction in the value of this relationship most often indicates inadequacies of O₂ transport, as may occur with diseases of the heart, lungs, or circulation. However, an abnormal O₂ utilization process—mitochondrial myopathy (54) and the muscle-related abnormality in oxygen metabolism reported in cystic fibrosis (110)—may also be associated with a reduced V̇o₂–work rate slope (see Section VIII,5.2: Is Metabolic Rate Appropriate during Exercise?); further studies are needed to assess this issue.

As V̇o₂ increases with increasing external work, one or more of the determinants of V̇o₂ approach limitations (e.g., SV, HR, or tissue extraction) and V̇o₂ versus work rate may begin to plateau. Achieving a clear plateau in V̇o₂ has traditionally been used as the best evidence of V̇o₂max. V̇o₂max is the best index of aerobic capacity and the gold standard for cardiorespiratory fitness. It represents the maximal achievable level of oxidative metabolism involving large muscle groups. However, in clinical testing situations, a clear plateau may not be achieved before symptom limitation of exercise (229, 303, 304). Consequently, V̇o₂peak is often used as an estimate for V̇o₂max. For practical purposes, V̇o₂max and V̇o₂peak are used interchangeably. Aerobic capacity should be directly measured because its estimation from resting indices, work rate, or submaximal exercise protocols is limited by physiologic mechanisms and methodologic inaccuracies and as such are unreliable (12). In turn, direct measurement of V̇o₂max is reliable and reproducible in normal subjects and patients (Table 6). The main determinants of normal V̇o₂max or V̇o₂peak are genetic factors and quantity of exercising muscle. V̇o₂max or V̇o₂peak is also dependent on age, sex, and body size, and it can be affected by training. V̇o₂peak should be expressed in absolute values (liters per minute) and as a percentage of the predicted value. The selection of predicted values is critical and should reflect the population being tested (see Section V: Reference Values).

V̇o₂max is often normalized by some index of body size. However, there is no consensus on the best method for adjusting for these indices. The most commonly used normalization is body weight in kilograms (American Heart Association, American College of Sports Medicine) and is the easiest to calculate. However, it may not be the most appropriate frame of reference for comparing or “normalizing” the metabolic rate across subjects of different sizes (305). Small but normal subjects have a higher V̇o₂ per kilogram than larger subjects (305). Because fat metabolism does not contribute significantly to V̇o₂max, normalization by body weight can produce deceptively low values in obese individuals. In obesity, normalization by height (V̇o₂/ht) may prove to be a better correlate of lean body mass and, in turn, a more reliable index of aerobic capacity (3). Additional studies are necessary.

Some have suggested that V̇o₂ referenced to lean body mass, also known as fat-free mass (FFM), would be a better index and has the further advantage of accounting for most of the sex differences in V̇o₂max; however, its routine measurement would be difficult to implement in the clinical exercise laboratory (306, 307). In addition, population variance may be reduced by including both weight and height (169, 308); this can be done by using as a reference body mass index (wt/ht²) or, preferably, by using fat-free mass index (FFM/ht²).

Recommendation: V̇o₂max or V̇o₂peak should be expressed as an absolute value and as a percentage of the predicted value; V̇o₂max should also be referenced to body weight (in kilograms) and/or height in the formatting of the report so that the impact of body size on exercise results is readily recognized (see Section VIII,4.6: Obesity). This is especially important in patients for whom actual weight is greater than ideal body weight.

V̇o₂ can increase from a resting value of about 3.5 ml/minute per kilogram (about 250 ml/minute in an average individual) to V̇o₂max values about 15 times the resting value (30–50 ml/minute per kilogram). Athletes may attain values over 20 times their resting values (up to 80 ml/minute per kilogram) (309). A reduced V̇o₂peak may reflect problems with oxygen transport (cardiac output, O₂-carrying capacity of the blood), pulmonary limitations (mechanical, control of breathing or gas exchange), oxygen extraction at the tissues (tissue perfusion, tissue diffusion), neuromuscular or musculoskeletal limitations, and, of course, effort. In addition, in patients, perceptual responses (symptoms) rather than a physiologic process as defined in the Fick equation may be responsible for a low V̇o₂max. The multifactorial etiology of reduced V̇o₂max in many clinical settings has been increasingly appreciated (1, 5, 142, 143, 276, 310). Decreases in V̇o₂max or V̇o₂peak are therefore general indicators of reduced exercise capacity. Underlying causes of exercise limitation are determined, in turn, by inspecting the pattern of responses in other variables. A reduced V̇o₂peak is the starting point in the evaluation of reduced exercise tolerance.

Recommendation: V̇o₂max/V̇o₂peak should be obtained from the maximal V̇o₂ value measured during an incremental exercise protocol taken to symptom limitation even if a plateau in V̇o₂ is not seen. Noteworthy symptoms should be appropriately recorded.

Energy for muscle contraction is provided by high-energy phosphate groups supplied in the form of adenosine triphosphate (ATP). ATP is supplied by the breakdown of glycogen to pyruvate, which enters the tricarboxylic acid (TCA) cycle via acetyl-coA, and breakdown of fats to produce acetyl-coA. Further processing of acetyl-coA in the TCA cycle and electron transport chain produces ATP needed for muscle contraction. If processing occurs only in the glycolytic pathway, a smaller amount of ATP is formed along with lactic acid. To produce ATP, the TCA and electron transport chains require oxygen, with the final by-products being water and CO₂. Cellular events defining oxidative and glycolytic processes during exercise are also impacted by muscle fiber type. Muscle fibers vary in the balance of oxidative versus glycolytic enzymes, that is, “aerobic” versus “anaerobic” metabolism. At low exercise intensities, fibers that are primarily oxidative are recruited, but as intensity increases, fibers that rely primarily on glycolytic pathways are recruited, thus increasing the output of lactic acid (323, 324). The extra acid produced causes an increase in V̇co₂ by buffering of CO₂ in the blood (see above). Controversy persists as to whether a deficiency of oxygen delivery versus oxidative capacity also contributes to the onset of lactic acid production, hence the term “anaerobic threshold.” It is possible that both processes, that is, the pattern of muscle fiber recruitment and a potential imbalance between oxygen supply and oxidative metabolism, contribute to the increase in lactic acid as exercise intensity increases.

Furthermore, there is no convincing evidence to substantiate that anaerobiosis at the cellular level is responsible for the increased arterial lactate above the AT; lactate accumulation may occur above and below a critical Po₂, which suggests that other factors (i.e., glycolytic enzymes) may also be involved (314). As such, the term AT should be used in a descriptive sense. The relative contribution of the different sources of lactic acid may also vary with disease. For example, in heart failure reduced oxygen delivery may be the predominant factor, so that as exercise intensity increases, the rate of rise in V̇o₂ starts to decline and the rate of rise in lactate increases earlier than in normal individuals (325, 326).

Regardless of mechanism, the increase in lactic acid that appears in the blood as exercise intensity increases has important physiologic consequences. First, the buildup in lactic acid reduces the pH of both blood and interstitial fluid, which in turn could ultimately compromise cellular function. Second, the reduced pH, or some event related to the change in pH, likely stimulates ventilation as the body attempts to buffer the increased acid by decreases in Pco₂. Because lactic acid buildup affects cellular function, the magnitude of the rise in lactate and the pattern of rise in lactate relative to change in V̇o₂ during exercise may be a useful diagnostic indicator in exercise testing. Also, the earlier lactate buildup occurs, the lower the long-term sustainable V̇o₂.

However, some authors have questioned whether the arterial [lactate] profile actually evidences threshold behavior (314, 321, 322). These findings (321, 322) were corroborated by other investigators (327) who, using venous blood lactate measurements, reported that in contrast to the “AT hypothesis,” their results were more consistent with a continuous development of acidosis, rather than a sudden onset of blood lactate accumulation during progressive exercise. Finally, Myers and coworkers (328) did not find a meaningful difference between the continuous and threshold models in the increase in blood lactate during ramp exercise.

In normal individuals, the AT occurs at about 50–60% V̇o₂max predicted in sedentary individuals, with a wide range of normal values extending from 35 to 80% (112, 113). The AT determination is age, modality, and protocol specific. The AT, when expressed as %V̇o₂max predicted, increases with age (3). The AT is highly modality specific, with arm exercise resulting in lower values versus leg exercise and with cycle ergometry resulting in lower (5–11%) values versus treadmill, a reflection of differences in exercising muscle mass and possibly differences in the dominant fiber type of exercising muscle (203, 204).

The AT demarcates the upper limit of a range of exercise intensities that can be accomplished almost entirely aerobically. Whereas work rates below the AT can be sustained essentially indefinitely, a progressive increase in work rate above AT is associated with a progressive decrease in exercise tolerance (329). Some have postulated that, in patients who become symptom limited with premature cessation of exercise, the AT as an effort-independent measurement may represent a submaximal variable that may assist in clinical decision making (3) Additional validation is necessary.

The AT is reduced in a wide spectrum of clinical conditions/diseases and, as such, has limited discriminatory ability in distinguishing between different clinical entities. A reduction in AT, as in V̇o₂peak, is somewhat nonspecific, often requiring inspection of other patterns of response to determine the underlying etiology (1, 112) (see Section VIII: Interpretation). Values below 40% of predicted V̇o₂max may indicate a cardiac, pulmonary (desaturation), or other limitation in O₂ supply to the tissues, or underlying mitochondrial abnormality (e.g., muscle dysfunction in cardiopulmonary diseases, mitochondrial myopathies, etc.).

AT determination is helpful as an indicator of level of fitness, for exercise prescription, and to monitor the effect of physical training (155, 158). However, if the AT is not reached, as in some patients with severe COPD (161, 330), or cannot be determined from the ventilatory response (160), an exercise prescription can still be established by using as a reference a percentage of peak WR, V̇o₂, or HR (22, 153). Similarly, physiologic improvement can be determined by monitoring changes in the variables mentioned above as well as in V̇e and lactate levels (160). Also, in patients undergoing cardiac rehabilitation, despite significant improvements in peak V̇o₂ and submaximal/maximal heart rate responses, no significant increase in noninvasively determined AT was noted (64).

Several methods are available for determination of the AT, and include the following: invasive determinations of AT (lactic acid and standard bicarbonate) and noninvasive determinations of AT (ventilatory equivalents method [V̇e/V̇o₂, V̇e/V̇co₂, Pet_O2, and Pet_CO2], V-slope method, and modified V-slope method).

Clinically, increasing lactic acidosis can be determined noninvasively by observing the pattern of change in V̇co₂ and V̇e relative to V̇o₂ as exercise intensity increases.

In healthy subjects, heart rate increases nearly linearly with increasing V̇o₂. Increases in HR are initially mediated by a decrease in parasympathetic activity (vagal withdrawal) and, subsequently, almost exclusively by increased sympathetic activity. Achievement of age-predicted values for maximal HR during exercise is often used as a reflection of maximal or near maximal effort and presumably signals the achievement of V̇o₂max. However, the use of this marker as a strict exercise end point is not recommended, in agreement with other consensus documents (12, 14). Considerable variability (10 to 15 beats/minute) within an age group is noted when available estimates of maximal HR are used, and as such, may complicate interpretation. The difference between the age-predicted maximal HR and the maximal HR achieved during exercise is referred to as the HR reserve (HRR). Normally, at maximal exercise, there is little or no HRR.

Significant differences in percent predicted maximal heart rate achieved can be observed, depending on the reference equation selected. There are several references value equations for maximal heart rate; the most widely used are 220 − age (3) and 210 − (age × 0.65) (345); both give similar values for people younger than 40 years. The first equation appears to underestimate the maximal heart rate in older people (346).

Peak heart rate is reduced in many, but not all, patients with different cardiorespiratory diseases (either because of the disease itself or because of medications used to treat the disease). If a patient reaches his/her predicted maximal HR, this suggests that the patient made a maximal or near maximal effort during exercise and that cardiovascular function may have contributed to exercise limitation (see Section VIII,4: Patterns of Exercise Response in Different Clinical Entities).

The HR–V̇o₂ relationship is often nonlinear at low work rates for upright exercise, becoming relatively linear as work rate increases to maximum. This relationship can be described by the slope and position of the regression line. The slope of the HR–V̇o₂ relationship is a function of the subject's SV: the higher the SV the lower the HR and, typically, its rate of change. HR at a given V̇o₂ is higher than normal in patients with lung disease, implying that SV must be lower, because cardiac output is similar to that of normal subjects (347). This may reflect deconditioning or relative unfitness, ventilatory limitation to exercise, and, possibly, the hemodynamic consequences of dynamic hyperinflation.

Patients with reduced O₂ delivery due to reduced O₂ content (hypoxemia, anemia, carboxyhemoglobin, etc.), patients with abnormal O₂ utilization (metabolic myopathy), as well as patients with deconditioning may also have an upward and steep HR–V̇o₂ relationship with (near) attainment of maximal heart rate (see Section VIII,4: Patterns of Exercise Response in Different Clinical Entities).

The heart rate response (ΔHR/ΔV̇o₂), defined as (peak HR–resting HR)/(peak V̇o₂ – resting V̇o₂), is another way to evaluate the relationship between HR and V̇o₂. This index has been suggested as an indicator of hyperdynamic cardiovascular response when its value is greater than 50 (348); however, additional validation is required.

Recommendation: Heart rate should be reported along with other variables during incremental testing protocols. A maximal HR that approaches the predicted maximal value suggests achievement of maximal or near maximal patient effort. A reduced maximal heart rate must be interpreted in light of a patient's disease and current medications.

The ratio of V̇o₂ to HR is conventionally termed the “oxygen pulse” and reflects the amount of O₂ extracted per heart beat. The O₂ pulse has been used by some as an estimator of stroke volume during exercise (3, 42). However this remains controversial, especially in patients who desaturate (see below). According to the modified Fick equation, the O₂ pulse is numerically equal to the product of SV and the arterial-to-mixed venous O₂ content difference, CO₂.

The profile of the O₂ pulse response during exercise therefore reflects the product of these two variables. The O₂ pulse normally increases with incremental exercise because of increases in both SV and O₂ extraction. At a near maximal/maximal work rate, in which C(a–v)O₂ is assumed to be maximal and relatively constant, the pattern of change of the O₂ pulse will represent the concomitant pattern of change of the SV as long as the previous assumption is correct. The basic profile of the O₂ pulse over the range in which V̇o₂ increases linearly with HR appears to be hyperbolic, with a rapid rise at low work rates followed by a slow approach to an asymptotic value. A low, unchanging, flat O₂ pulse with increasing work rate may therefore be interpreted as resulting from a reduced SV and/or as a failure for further skeletal muscle O₂ extraction. A low O₂ pulse therefore may reflect deconditioning, cardiovascular disease, and early exercise limitation due to ventilatory constraint or symptoms. Consequently, caution should be exercised in interpreting changes in O₂ pulse response solely as an index of cardiovascular dysfunction.

There are several mathematical models that have been developed in an attempt to more accurately estimate SV from the V̇o₂–HR relationship. Work with normal subjects has suggested that the “asymptotic O₂ pulse,” which is the product of 1/Ca_O2 and the slope of the V̇o₂–HR relationship rather than the maximal O₂ pulse achieved, may provide a more reliable estimator of the average SV during incremental exercise in normal subjects (349). Although theoretically attractive and applicable in normal subjects, validation in patients is necessary, especially in subjects in whom Ca_O2 might be expected to change during exercise (i.e., patients with lung disease) or for whom the assumption of linearity between cardiac output and V̇o₂ may not reasonably be made, as in patients with cardiovascular disease.

Stringer and coworkers (350), using a linear regression equation to calculate C(a–v)O₂ at a given V̇o₂ during incremental exercise in normal subjects, computed stroke volume according to the Fick equation. Calculated and measured stroke volume were comparable in normal subjects. Likewise, this method requires validation in patients. Agostoni and coworkers (351) calculated the stroke volume in patients with congestive heart failure, using a fixed value for C(a–v)O₂. Good agreement was noted between calculated and measured stroke volume; the best correlation occurred with measurements made at the AT. These data require corroboration. Another concept that has appeared in the literature, which requires validation, is the “extended O₂ pulse” (352), which corresponds to the O₂ pulse calculated from the extrapolation of the slope of the measured HR to the predicted V̇o₂.

The rise in V̇e with exercise is associated with an increase in both depth and frequency of breathing. In health, increases in tidal volume are primarily responsible for increases in ventilation during low levels of exercise (353, 354). As exercise progresses, both Vt and fr increase until 70 to 80% of peak exercise; thereafter fr predominates (309, 353). Vt usually plateaus at 50 to 60% of vital capacity (VC); however there is considerable variation (355). Younger adults typically increase Vt by three- to fivefold whereas in older adults Vt tends to increase less (two- to fourfold) (356). Breathing frequency typically increases one- to threefold in most subjects, but in fitter athletes it may be increased by six- to sevenfold at high levels of V̇e. In some subjects at extremely high ventilatory demands, Vt actually decreases as fr increases (357).

In health, the increase in Vt is due to both a decrease in end-expiratory lung volume (EELV) through encroachment on the expiratory reserve volume but predominantly to an increase in end-inspiratory lung volume (EILV) through a decrease in the inspiratory reserve volume (358). With progressive increases in exercise intensity and V̇e, EELV continues to fall to 0.5–1.0 L below the resting FRC (358). It has been hypothesized that the fall in EELV optimizes inspiratory muscle length (for force development) and helps prevent too large an increase in EILV, which would increase the inspiratory elastic load of breathing (359). In addition, the energy stored in the abdominal wall because of active expiration may provide some passive recoil at the initiation of the ensuing inspiration (360).

The increase in fr with exercise reflects a decrease in both inspiratory time (Ti) and expiratory time (Te). Typically, however, at the moderate to higher ventilatory demands, a greater fractional decrease is noted in Te so that Ti/Ttot increases from 0.4 at rest to 0.5 to 0.55 at maximal exercise (361, 362). Because of the greater decrease in Te, the increase in mean expiratory flow rate is greater than the increase in mean inspiratory flow rate.

Whether ventilatory limitation causes or contributes to exercise intolerance has traditionally been evaluated by the ventilatory reserve, which reflects the relationship of ventilatory demand to ventilatory capacity. In most healthy adults, peak exercise ventilation approaches 70% of the MVV (V̇e reserve typically greater than 15% of the MVV), although this percentile increases with increased fitness and with normal aging.

7.2.1. Ventilatory demand.

Ventilatory demand is dependent on multiple factors including metabolic requirement, degree of lactic acidosis, dead space ventilation, behavioral factors, deconditioning, body weight, mode of testing (e.g., arm versus leg exercise), and additional factors involved in ventilatory control.

Physiologic factors associated with the level of ventilation are linked through the following equation:

where 863 is the constant that corrects for the different conditions of reporting the gas volumes (for a body temperature of 37°C) and also the transformation of fractional concentration to partial pressure. V̇co₂ (stpd) and V̇e (btps) are in liters per minute. Pa_CO2 is in millimeters of mercury. Vd/Vt is expressed as a fraction. From the above equation, it is apparent that the ventilatory demand for a given work rate can be different depending on the current values of these defining variables. The ventilatory demand is usually increased at rest and at any given level of exercise in patients with COPD, ILD, and pulmonary vascular disease (PVD) due to ventilation–perfusion (V̇/Q̇) inequality with increased Vd/Vt, hypoxemia, and probably increased stimulation of lung receptors (which drive ventilation and reduce Pa_CO2).

Ventilatory reserve determination corresponds to how close the peak minute ventilation (ventilatory demand) achieved during exercise (V̇emax) approaches the MVV (ventilatory capacity) or some estimate of the MVV (FEV₁ × 35–40). Traditionally, ventilatory “reserve” (3) has been defined as the percentage of MVV achieved at peak exercise [(V̇epeak/MVV) × 100] or, alternatively, as the difference between MVV and the V̇e achieved at peak exercise. However, the “normal” values for this index standard have a high variance. And although the values can vary by up to 50% in normal subjects, a lower limit of a 15% difference between V̇e and MVV appears to be a reasonable reserve, based on 95% confidence limits of normative data. The value is not independent of fitness or aging, as the reserve is less in both fit and aged normal subjects. Patients with pulmonary diseases characteristically have reduced ventilatory capacity and increased ventilatory demand, resulting in reduced ventilatory reserve.

Analysis of the V̇e–V̇o₂ relationship during incremental exercise enables a graphic overview and appreciation of important ventilatory trending patterns related to the appropriateness of the ventilatory response to metabolic demands. Physiologically, the V̇e–V̇o₂ relationship would be the better option to analyze the ventilatory response to metabolic needs because, although V̇e varies with V̇o₂, V̇o₂ is nearly independent of V̇e. However, the relationship between V̇e and V̇o₂ is complex, commonly nonlinear, and difficult to standardize. Abnormal ventilatory response patterns including excessive ventilation due to hyperventilation or increased dead space, periodic breathing, or erratic or other psychogenic abnormalities may also be revealed (see Section VIII: Interpretation).

Others prefer plotting V̇e versus V̇co₂, as ventilation is so closely linked to CO₂ production. However, because V̇co₂ is dependent on V̇e, some have suggested that it would be inappropriate for it to be used as the reference variable. V̇e correlates closely with V̇co₂ during moderate exercise, whether incremental or constant-load (see Figure 11D). In healthy subjects the V̇e–V̇co₂ relationship is linear, with an intercept on the V̇e axis of some 4–5 L/minute. The slope of this relationship reflects that in normal subjects 23–25 L of V̇e is required to eliminate 1 L of CO₂. The slope can be appreciably higher when Vd/Vt is high (368) or when the “set point” for arterial Pco₂ is low. The incorporation of the MVV value into the graph permits visual inspection of the ventilatory reserve; approaching or exceeding this limit indicates that there is little or no ventilatory reserve (see Figure 9D). Currently, either a plot of V̇e versus V̇co₂ or a plot of V̇e versus V̇o₂ is acceptable for the graphic representation of ventilatory data.

The ratio of V̇e to V̇o₂ is called the ventilatory equivalent for O₂ and the ratio of V̇e to V̇co₂ is called the ventilatory equivalent for CO₂. The ventilatory equivalents for O₂ and CO₂ are both related to Vd/Vt, being higher as Vd/Vt increases. However, the ventilatory equivalents also increase with hyperventilation, and therefore interpretation must be made carefully. Because the relationships of V̇e versus V̇o₂ and of V̇e versus V̇co₂ have a positive intercept on the V̇e axis, a graph of the ventilatory equivalents versus work rate or V̇o₂ decreases hyperbolically at low work rates. For a more detailed explanation of the contour of the ventilatory equivalents versus V̇o₂ the interested reader is referred to other sources (113). The normal pattern of change in V̇e/V̇o₂ is a drop early in exercise to its nadir near the AT, and then an increase as maximal exercise capacity is approached. The ventilatory equivalent for CO₂ also decreases hyperbolically as work rate increases. For incremental tests, the increase in V̇e/V̇o₂ that typically occurs in concert with the development of metabolic acidemia occurs at a time when V̇e/V̇co₂ has not yet increased (see Figure 11F). It is this profile that separates this response from the onset of hyperventilation from other causes (e.g., anxiety, pain, or hypoxemia), in which case both V̇e/V̇o₂ and V̇e/V̇co₂ would increase in concert. The normal subsequent increase in V̇e/V̇co₂ reflects the onset of frank compensatory hyperventilation for metabolic acidosis, with concomitant reduction in Pa_CO2 and typically Pet_CO2. The mean of the minimal V̇e/V̇co₂ is about 25 in healthy young subjects, but may exceed 30 in older individuals. Usually V̇e/V̇co₂ is less than 32–34 at or near the AT and less than 36 (rarely 40) at peak exercise in normal subjects. V̇e/V̇o₂ also decreases to minimal values similar to or slightly greater than that for V̇e/V̇co₂. High V̇e/V̇co₂ at its nadir suggests a high Vd/Vt or a low Pa_CO2. Lack of a subsequent increase in V̇e/V̇o₂ or V̇e/V̇co₂ reflects either insensitivity to the stimuli associated with metabolic acidosis or the presence of high airway resistance or a general increase in respiratory muscle load. This may be observed in some patients with COPD who are clearly ventilatory limited but able to exert themselves so that Pa_CO2 begins to rise, indicating an inadequate ventilatory response (369). Similarly, in morbid obesity, some subjects may not ventilate adequately for the metabolic load and V̇e/V̇co₂ will remain low (370).

Pet_CO2 increases and Pet_O2 decreases throughout the moderate work rate range. Pet_O2 typically begins to increase in concert with the increase in V̇e/V̇o₂ during incremental tests (see Figure 11I). With the increasing work rate, Pet_CO2 typically increases (although Pa_O2 does not) until the subsequent increase in V̇e/V̇o₂ occurs, at which time Pet_CO2 stabilizes until the subsequent increase in V̇e/V̇co₂ occurs; then Pet_CO2 starts to fall concomitantly with the increase in Pet_O2. The period of increasing Pet_O2 with relatively stable Pet_CO2 has been termed “isocapnic buffering” (371). A fall in Pet_CO2 when V̇e/V̇co₂ is high suggests hyperventilation whereas high V̇e/V̇co₂ without a fall in Pet_CO2 suggests increased dead space ventilation.

Additional methods have evolved in an attempt to better identify the “true” ventilatory capacity available for producing ventilation during exercise. Analysis of the exercise tidal flow–volume loop during exercise plotted within the maximal flow–volume envelope enables a theoretical estimation of this “true” ventilatory capacity (V̇ecap) (377, 378). By taking into consideration the patient's breathing strategy (tidal volume), operational lung volume (EELV), and exercise-induced changes in airway tone (MFVL performed during exercise), maximal available ventilation can be determined. It has been reported that V̇ecap was similar to measured MVV in normal individuals (377).

A more direct and, in principle, simple means for testing whether expiratory flow limitation has, in fact, occurred has been described by Kolouris and coworkers (379, 380). In this test, a small negative pressure is applied at the mouth at the initiation of tidal expiration. A lack of increase in expiratory flow over that seen during spontaneous breathing suggests expiratory flow limitation. This type of technique offers theoretical advantages over the use of the exercise tidal flow–volume loop method in that it is not dependent on an accurate assessment of the MFVL or on accurate placement of the tidal flow–volume loops by assessment of IC. Clearly, however, flow limitation is volume dependent, typically beginning at low lung volume near EELV and gradually increasing along with the tidal breath until either hyperinflation occurs or ventilation is fully constrained. A limitation to the use of negative expiratory pressure is that the method may demonstrate an increase in expiratory flow early in expiration, suggesting no flow limitation, when in fact flow may be limited at the lower lung volumes. This technique only allows for evaluation of the presence or absence of flow limitation, and flow limitation is not an all-or-none phenomenon (372). In addition, no information about operational lung volumes and, in turn, breathing strategy is provided. However, the negative expiratory pressure technique combined with an assessment of the extFVL may yield the most accurate assessment of the degree of expiratory flow limitation as well as other indices of constraint. Further study and validation of this technique are required.

Other techniques that have also been applied to assess the degree of ventilatory constraint (limitation) during exercise include the following: breathing through an increased dead space volume (378), hypercapnic stimulation (309, 381, 382), and heliox administration (377, 383–386). However, these techniques also provide little information about breathing strategy and, apart from the negative pressure technique, tell little about the specific source of ventilatory constraint and still remain in the domain of research laboratories.

An important index of abnormality in pulmonary gas exchange is the alveolar-to-arterial oxygen tension difference. The alveolar–arterial Po₂ difference measures the difference between the “ideal” alveolar Po₂ and arterial Po₂. Alveolar gas composition varies from one breath to another and in different alveoli. These normal biologic variations may be significantly larger in patients with cardiopulmonary disease.

The equation used for calculation of the ideal Pa_O2 by indirect means is based on the assumption that the Pa_CO2 is representative of the mean Pco₂ in all the perfused alveoli and that the respiratory exchange ratio (RER, or R) for these alveoli is equal to that of the lungs as a whole. The ideal Pa_O2 represents what the alveolar Po₂ of the whole lung would be if the composition of gases were homogeneous throughout the lungs. When CO₂ is absent in the inspired gas, alveolar Po₂ is computed from the alveolar gas equation (391–393):

A simplified version of this equation, in which Pa_CO2 is replaced with Pa_CO2, is the most commonly used:

In these equations R is the respiratory exchange ratio measured from expired gas, and Pa_CO2 is the ideal alveolar Pco₂ (defined as the Pa_CO2 of a mythical, perfectly homogeneous lung with the same R value as the lung under study) and is assumed to equal Pa_CO2. Pi_O2 is determined by the barometric pressure and fraction of inspired O₂, (Pb − 47) × Fi_O2 (25). Because the term in square brackets in these equations normally contributes only 2 mm Hg or less to the estimated Pa_O2 and becomes inconsequential when R = 1.0, in clinical practice it is commonly neglected.

P(a–a)O₂ is then simply (Pi_O2 − Pa_CO2/R) − Pa_O2, where Pa_O2 is the arterial Po₂ sampled concurrently with the measurement of R. It is common in clinical practice to assign a fixed value of 0.8 to R (when not actually measured) to calculate P(a–a)O₂ at rest; however, this value needs to be evaluated cautiously and used only as a rough estimate because the impact of R in the equation is significant. If the real R value for the patient is 1.0 the error in the estimate would be about 10 mm Hg. Real data contain random errors that, collectively, can have a large effect on P(a–a)O₂. It is important to recognize that, even when the true P(a–a)O₂ is normal (about 6 mm Hg at rest), the measured value may be calculated as negative because of the additive effects of acceptable error levels of the primary variables (394). This, of course, will be less likely during exercise, as P(a–a)O₂ increases (190).

The P(a–a)O₂ is helpful in determining (although only partially) the cause of a drop in Pa_O2 with exercise. Thus, if alveolar Po₂ falls as much as arterial Po₂ [normal P(a–a)O₂], the hypoxemia is essentially due to an inadequate ventilatory response to exercise (relative alveolar hypoventilation for the work rate) and, as a consequence, a concomitant increase in Pa_CO2 should occur. This could be due to mechanical derangement of the lungs and/or chest wall or to respiratory muscle fatigue or inadequate respiratory control mechanisms. The other cause of a drop in Pa_O2 without a change in P(a–a)O₂ is in a hypoxic environment (altitude, with reduced Pb, or reduced inspired O₂ fraction) (395).

If Pa_O2 falls without a drop in Pa_O2 [increased P(a–a)O₂], other mechanisms must be present. These could include (1) worsening V̇/Q̇ inequalities, (2) increasing right-to-left shunt, (3) alveolar–capillary diffusion limitation, and (4) a compounding of the first three factors by the inevitable fall in mixed venous Po₂ that occurs between rest and exercise (396). Mixed venous Po₂ falls simply because the relative increase in O₂ uptake exceeds that of cardiac output, with an obligatory rise in arterial–venous [O₂] difference and thus a fall in mixed venous [O₂]. Other things being equal, a lower mixed venous [O₂] will (1) reduce end-capillary Po₂ in exchanging units of all V̇/Q̇ ratios, (2) worsen the arterial hypoxemia due to shunting, and (3) prolong the time required for full oxygenation of capillary blood by diffusion (397).

Another index of gas exchange efficiency is Vd/Vt, the ratio of physiologic dead space to tidal volume. Vd/Vt can also be described as the fraction of each breath that would hypothetically be “wasted” on ventilating both the anatomic dead space and the alveoli that are unperfused by blood (physiologic dead space). An increase in Vd/Vt reflects an increased inefficiency of ventilation (ventilation–perfusion [V̇/Q̇] mismatching or right-to-left shunt) and, as a consequence, requires an increase in V̇e to maintain Pa_CO2. However, the increase in V̇e may be inadequate to compensate for the increased Vd/Vt and Pa_CO2 levels may rise depending on factors related to control of breathing and the degree of ventilatory constraint (404). Vd/Vt measurement is also highly dependent on the breathing pattern, that is, rapid shallow breathing increases Vd/Vt even without abnormalities of V̇/Q̇ (405, 406).

Vd/Vt is measured from arterial Pco₂, and Pe̅_CO2, using the following equation (407):

Pe̅_CO2 represents the mixed expired value of alveolar and dead space gas, in millimeters of mercury. Pe̅_CO2 can be measured directly by collecting expired gas in a large gas-impermeable bag over time and measuring the Pco₂ of the gas in the bag. Modern commercial exercise systems provide this value either by sampling gas from a mixing chamber, or, for breath-by-breath systems, by determining the V̇co₂/V̇e ratio, where both gas volumes are expressed in liters (stpd).

Because most exercise is accomplished with subjects connected to a mouthpiece and frequently to a nonrebreathing valve, the equation needs to be modified to incorporate the mouthpiece and valve dead space (30–110 ml):

There are no valid procedures that allow Pa_CO2 to be adequately estimated noninvasively, especially in patients with lung disease. End-tidal Pco₂ (Pet_CO2) should not be used as an index of Pa_CO2; indeed, this can be misleading, as it can exceed Pa_CO2 (272, 273). Normally, the Pet_CO2 is slightly less than Pa_CO2 at rest, but Pet_CO2 becomes greater than Pa_CO2 with exercise in normal individuals. This would lead to an overestimation of Vd/Vt. In lung disease, however, because of effects of lung regions with increased V̇a/Q̇, Pet_CO2 may remain below Pa_CO2 even during exercise, possibly leading to an underestimation of Vd/Vt (3).

Subjects should preferably be community based rather than hospital based. The subjects studied should possess characteristics similar to those of the patient population to which the reference values will be applied. Minimally, this should include age, sex, and anthropomorphic considerations. Also important is level of physical activity, racial composition, dwelling altitude, occupational exposure, and knowledge of any significant coexisting medical condition and/or medications. This is best achieved by a standardized questionnaire.

The number of subjects tested should be sufficiently large to permit an appropriately powered sample size with a uniform distribution of subjects for sex and age groups. Specific attention should be given to the need to include women and older individuals, given the changing demographics and paucity of reliable population-based CPET data for these groups.

The study design of future reference value initiatives should include a randomization process to avoid the potential bias seen when more physically active subjects volunteer for the study.

Appropriate quality was achieved using recommendations contained in the present document.

CPET protocols should be in accordance with recommendations specified in the present document.

CPET results obtained by either breath-by-breath analysis or mixing chamber should be treated in accordance with recommendations contained in the present document. CPET results should be interval averaged, preferably every 30–60 seconds (to avoid the noise of shorter intervals), and the peak value reported should represent the mean of the last completed stage or of all the data collected during the final stage, but preferably for not less than 30 seconds.

Reference equations must be validated in populations other than those used to generate the existing data.

The function that most accurately describes the distribution of the data should be used. For example, curvilinear (power) functions may more accurately describe the distribution of the data. Furthermore, the precision of the individual and population predicted values should be reported.

Reason(s) for which CPET was requested should be precisely determined. The reason(s) might be a specific exercise-related symptom appearing during or limiting daily or special activities. Alternatively, the exercise test may be obtained for preoperative evaluation, to assess the consequences of occupational exposure, for impairment/disability evaluation, or for the determination of exercise capacity of an asymptomatic subject before enrollment in an exercise program (see Section II: Indications for Cardiopulmonary Exercise Testing).

Interpretation of CPET results requires knowledge of a subject's clinical information including clinical diagnosis, results of the medical history and physical examination, PFTs, chest X-ray, ECG, and other testing as appropriate (echocardiography, bronchial provocation challenge, etc.). A health questionnaire may also be especially helpful for cardiopulmonary risk factor analysis, symptoms, occupational exposure, and determination of level of physical activity. Especially critical is knowledge of medications (i.e., β-blockers may affect heart rate response during exercise). Skeletal (i.e., joint, bone) abnormalities, especially of the lower limbs, should be noted. Laboratory data including hemoglobin and carboxyhemoglobin determinations may also be helpful. A more meaningful physiologic–clinical correlation and, in turn, accurate interpretation of CPET results is possible when a thorough clinical subject profile is available (1, 276). The above-described information should be part of the exercise “work sheet” and succinctly included in the final exercise report.

The major portion of this section focuses on CPET results generated during maximal, symptom-limited incremental cycle ergometry, which is currently the most popular, albeit not the exclusive, protocol. The impact of exercise modality (cycle exercise versus treadmill) as a function of muscle mass involved and protocol (constant work versus incremental, continuous versus discontinuous, etc.) as a reflection of differences in temporal patterns of work rate changes on interpretation is well appreciated and has been addressed previously (see Section V: Reference Values; Section III: Methodology; and Section IV: Conceptual and Physiologic Basis of Cardiopulmonary Exercise Testing Measurements). Accurate interpretation therefore requires that reference values for comparisons of patient data reflect patient factors, protocol, and equipment. The increasing use of constant work rate (submaximal) exercise testing (usually based on results generated during maximal IET) on clinical decision making was discussed in Section III,3.3 (Constant Work Rate Protocol).

As previously noted, the selection of an appropriate set of reference values is a function of the patient population, age, height, weight, sex, and physical activity and may vary from laboratory to laboratory (3, 43, 269) (see Section V: Reference Values). Prediction equations usually do not take physical activity into consideration; accurate interpretation certainly requires such knowledge. For example, very fit individuals and athletes may experience significant reductions in their peak V̇o₂ as a result of pulmonary, cardiac, or (peripheral) muscle illness and still be within the normal confidence interval for sedentary subjects. Because most exercise variables may vary with work rate/metabolic rate, observed data should be compared with these values predicted at the same work rate or metabolic rate if available. Limited submaximal reference value data are available (see Section V: Reference Values). Peak or maximal exercise values are most often used for comparison, because the greatest number of reference values is available for maximal exercise variables. However, exclusive reliance on peak exercise results should be avoided. Different sets of maximal (or peak) reference values can have significant impact on interpretation of CPET results (276).

An impressive number of primary and secondary (derived) variables can be measured during CPET. The number of variables to be measured in any one situation will depend on the reasons for which CPET was requested. Key measurements obtained during CPET and the plots of important graphic interrelationships between these measurements appear in Tables 10 and 11, respectively (see Section IV: Conceptual and Physiologic Basis of Cardiopulmonary Exercise Testing Measurements). Although the conceptual basis and clinical use/limitation for these measurements have been discussed previously, additional salient comments relevant to interpretation will be noted. Although the measurements in Table 10 are grouped into categories, abnormality of a variable does not necessarily define exercise limitation in that category. Also, in Table 10, the measurements are represented as noninvasive or invasive (i.e., arterial sampling intervention); decision analysis for obtaining invasive determinations is discussed in Section III (Methodology).

The measurements in Table 10 were selected because of their clinical utility and because of the way they impact the clinical decision-making process by (1) determining whether exercise capacity was reduced and establishing the most likely source(s) of exercise limitation, (2) aiding in detecting clinically occult disease, (3) helping to distinguish between clinical entities, and (4) assessing response to treatment. Finally, the measurements included in Table 10 are not intended to be exhaustive and will likely change in the future as experience with new methodologies (i.e., exercise tidal flow–volume loops) and variables (EELV) evolve (288, 372, 374, 378, 450, 451).

The quantification and characterization of symptoms during exercise are increasingly being utilized for clinical decision making (82, 411, 452). This is because measurement tools are available, are more rigorously applied, and relationships (of symptoms, especially dyspnea) to physiologic variables are more firmly established (4, 75, 83, 97, 416). Furthermore, as many patients with cardiopulmonary disorders are symptom limited rather than physiologically limited, quantitating and interrelating a symptom to physiologic variables are practical and helpful (5, 310). The most commonly evaluated symptoms are chest pain, exertional breathlessness, general fatigue, and leg fatigue. The modified Borg Scale is the most widely used in providing a rating of perceived exertion (414); its reproducibility for the measurement of dyspnea during exercise is good (85, 256, 453) (see Section IV,11: Perceptual Assessment).

Graphic representations of these variable interrelationships are routinely provided by automated exercise systems and are helpful. However, selection of the most appropriate format for data display is important for discriminating patterns of abnormality in the exercise response. Suggested forms for tabular and graphic reports of the results of incremental CPET in a normal middle-aged male appear in Table 19 and Figure 11, respectively. The trending phenomena as work rate progresses from submaximal to peak exercise provided in these graphs enhance interpretation so that the entire exercise response may be appreciated in addition to the maximal values, which are also usually presented in tabular form (see Table 19). Although underutilized, these submaximal graphic data are revealing and can often be diagnostic (49). Optimal interpretation requires that CPET results be evaluated on the basis of both tabular and graphic expressions of the data, with careful attention given to trending phenomena. In addition, the use of the appropriate corresponding reference values for each plot greatly enhances interpretation and is recommended (1, 276, 454, 455). Interpretation is facilitated if both the tabular and graphic data reflect interval-averaged data (30–60 seconds) rather than breath-by-breath data so that “noise” is minimized (see Section III: Methodology). The appropriateness or inappropriateness of the response of a variable, compared with the corresponding reference value, is the basis for determining normal or abnormal exercise responses and exercise limitation. Table 17

Knowledge of patient effort and motivation is necessary for the accurate interpretation of a maximal CPET. This is especially true when maximal exercise capacity is reduced. The reproducibility of maximal exercise capacity in normal subjects and patients with cardiac or pulmonary disease is well established (263, 264, 266, 267). Some studies have shown that measured V̇o₂ is more reproducible on repeat testing when compared with peak work rate (see Table 6). Consequently, this suggests that patient effort is usually maximal or that it does not vary significantly with time. Patient effort can be optimized by familiarization before and encouragement during CPET (see Section III,3: Exercise Testing Protocols). Submaximal efforts may interfere with the interpretation of CPET results and, in turn, patient management. If a subject's effort appears submaximal, it should be stated in the report. Patient effort becomes an important consideration when V̇o₂peak is reduced and physiologic limitation is not achieved. Questions then include the following: was the patient symptom limited? Was it poor effort, or possibly some other factor(s)? Another very important question: what is the impact on clinical decision making? Work has underscored this issue in patients with heart failure being evaluated for cardiac transplantation, as V̇o₂peak is so critical in clinical decision analysis (456).

There is considerable variability in the objective markers used for the determination of a “maximal” exercise test. These markers include heart rate achieved, lactate level, plateau in oxygen uptake, bicarbonate or pH drop, and RER (457). Currently, however, there is no gold standard for the assessment of maximal effort. Patient effort is usually considered to be maximal if one or more of the following occur:

1. The patient achieves predicted peak oxygen uptake and/or a plateau is observed.

2. Predicted maximal work rate is achieved.

3. Predicted maximal heart rate is achieved.

4. There is evidence of ventilatory limitation, that is, peak exercise ventilation approaches or exceeds maximal ventilatory capacity.

5. Although no one RER value defines maximal effort, values greater than 1.15 are more likely to be associated with near maximal or maximal effort (277).

6. Patient exhaustion/Borg Scale rating of 9–10 on a 0-to-10 scale.

It should be emphasized that a fall in Pa_O2 or Sa_O2/Sp_O2 (less than 55 mm Hg or less than 88%, respectively) does not necessarily indicate that a patient's effort was maximal, as such falls can be observed during submaximal exercise in several clinical entities including ILD and PVD. Although chest pain, ischemic ECG changes, and heart rate and BP drops may be indications for stopping an exercise test, these events may occur before maximal exercise and, therefore, are not good measures of maximal effort.

Quantification of intensity of symptoms during exercise, using either the Borg Scale or Visual Analog Scale, for ratings of perceived exertion may assist in the assessment of motivation or effort; at present they are useful adjuncts to physiologic data in determining adequacy of effort (see Section VIII,3.5: Symptoms). Only a minority of (apparently) well-motivated normal subjects or patients with chronic respiratory disease rate dyspnea or sensation of leg effort as maximal (10 on the modified Borg Scale) when they stop exercise (458). Work has noted that the Borg Scale dyspnea target intensity scores correlate with V̇o₂peak in patients with COPD (452). In the final analysis, the assessment of subject motivation and effort requires attention to several factors including physiologic variables, observation of the patient, and quantification of symptom(s).

Reason(s) for stopping the exercise test, using the modified Borg Scale or VAS and the subject's own words, should be included and reported with the variable responses. In addition, the observations of the individual monitoring the exercise test, including other occurrences during the test and condition of the subject at the point of discontinuing exercise, should also be noted as they may impact interpretation. Relevant comments would be, for instance, patient appeared exhausted, not obviously stressed, too breathless to talk, ataxic, sweating, pale, and so on.

Multiple factors contribute to exercise intolerance in patients with cardiovascular disease (459). These include inadequate O₂ delivery due to variably severe abnormalities in heart rate response, systolic and diastolic dysfunction, and abnormalities in the distribution and impedance of blood flow in the peripheral circulation; abnormal pulmonary vascular responses, as well as skeletal muscle abnormalities including deconditioning (i.e., O₂ utilization and muscle metabolism); and abnormal ventilatory responses (12, 36, 62, 141, 310, 459–467). In consequence, a reduced peak work rate and peak V̇o₂ compared with normal subjects, owing to one or several of these factors, is usually observed. Functional classification of severity of heart failure, prognosis, and selection for transplantation (see Section II: Indications for Cardiopulmonary Exercise Testing) is currently based on V̇o₂max (36, 44, 59–61).

The ratio of increase in V̇o₂ to increase in external work (ΔV̇o₂/ΔWR) has been reported to be decreased in patients with heart disease (195, 468), although this has not been consistently observed. Early onset of metabolic acidosis is manifested by a reduced anaerobic threshold and early fall in Pa_CO2. The O₂ pulse is usually low, with an abnormally “flattened” appearance compared with the normally “flattened” appearance (see Section IV: Conceptual and Physiologic Basis of Cardiopulmonary Exercise Testing Measurements, for comments concerning the constellation of factors that can impact O₂ pulse and for a discussion of the relationship of O₂ pulse to stroke volume).

Higher HR responses at submaximal levels of V̇o₂ are observed; the steepness (rate of rise of HR relative to V̇o₂) and deviation from linearity of the normal HR–V̇o₂ relationship is highly variable. Patients with mitral valvular disease (469) and angina, for instance, may have impressive increases in slope. However, so do some patients with metabolic myopathy (54); a (mildly) abnormal HR–V̇o₂ relationship is usually also seen in deconditioning. In general, patients with heart failure manifest impaired heart rate responses to exercise with greater HR response to any given V̇o₂ and with peak HR reduced as disease severity progresses (36, 470). Consequently, there is usually little or no heart rate reserve (HRR) in patients with mild heart failure, mitral valvular disease, or angina; there is usually greater HRR observed in patients with more severe heart failure. Patients with ischemic heart disease may also have an impaired HR response to exercise due to chronotropic incompetence. Abnormal heart rate recovery is an independent predictor of mortality in patients referred for exercise electrocardiography (471). The predictor value of heart rate recovery after cycle ergometry requires further evaluation. ECG and blood pressure abnormalities may also be observed (9, 10, 12).

Patients with cardiovascular diseases will often have abnormal exercise ventilatory responses, including increased V̇e at submaximal V̇o₂ due to early and usually more severe onset of metabolic acidosis, abnormal ventilation–perfusion relationships due to low cardiac output for the metabolic rate, subclinical interstitial pulmonary edema (leading to decreased lung compliance), diastolic dysfunction, increased airway resistance (see below), and stimulation of chest wall and/or lung mechanoreceptors. Patients with chronic heart failure are more likely to experience abnormal ventilatory responses compared with patients with coronary artery disease, reflecting that chronicity rather than acute flow changes may be responsible (472).

Increased respiratory frequency and reduced Vt are frequently seen. Early exercise cessation and reduced V̇o₂peak are associated with a reduced peak V̇e but usually with considerable ventilatory reserve (V̇epeak/MVV). Patients with heart failure have increased airway resistance (383), breathe at rest near residual volume, and are flow limited at low work rates (Figure 8)

Figure 8. Example of a patient with stable congestive heart failure (New York Heart Association Class III). Shown are rest, mild, moderate, and peak exercise tidal flow–volume loops plotted within the maximal flow–volume loop. EELV is reduced at rest and remains near RV throughout exercise despite significant expiratory flow limitation and apparent room to increase EELV to avoid the flow limitation (288). EELV = end-expiratory lung volume; EILV = end-inspiratory lung volume; MFVL = maximal flow–volume loop; TLC = total lung capacity.

Download Figure | Download Powerpoint

[More] [Minimize]

Importantly, there is usually no arterial desaturation or drop in Pa_O2 and normal P(a–a)O₂ in response to exercise (463). However, patients with combined cardiovascular and pulmonary disease may demonstrate abnormal P(a–a)O₂ and Pa_O2 responses (276, 443). Increases in Vd/Vt and V̇e/V̇co₂ due to reduced cardiac output for the metabolic rate and consequent V̇/Q̇ abnormalities are frequently observed, especially in moderate to severe heart failure (62, 460, 463, 474). An abnormal V̇e/V̇co₂ slope (greater than 34) during exercise has been suggested as an independent predictor of mortality in patients with heart failure and often as a correlate for advanced disease (62). Interestingly, abnormal V̇e/V̇co₂ responses return to normal after cardiac transplantation (475). The ratio of tidal volume at peak exercise to vital capacity is usually normal. Periodic breathing during exercise has been reported in some patients (476). In general, for patients with heart failure, respiratory abnormalities, although variably contributory to exercise intolerance, are usually not exercise limiting (477). The presence of V̇e/MVV approaching or exceeding 100% of predicted (reduced ventilatory reserve) suggests concurrent pulmonary disease and may signal the presence of combined cardiovascular and respiratory limitation (276, 443). However, this requires additional study.

Patients with pulmonary vascular disease (primary pulmonary hypertension, pulmonary embolism, chronic thromboembolic disease, pulmonary vasculitis, etc.) will often have a reduced peak work rate and peak V̇o₂ and are usually (but not invariably) cardiovascular limited (102, 104–106, 478, 479). The development of pathologic cardiovascular limitation may depend on several factors including extent of pulmonary vascular involvement, underlying pulmonary vascular pathology, the time course of disease progression, and, ultimately, the heart's inability to maintain adequate cardiac output in the face of increased pulmonary vascular resistance and consequent RV afterload (36, 102, 300). There is usually an early-onset metabolic acidosis. The HR–V̇o₂ relationship is left shifted with submaximal HR values trending higher than normal and peak HR responses usually (near) normal, although low peak HR responses are seen (little/no heart rate reserve). Peak O₂ pulse is usually reduced.

The V̇e–V̇o₂ relationship usually demonstrates increased levels of submaximal ventilation, although there is usually ventilatory reserve at peak exercise. The slope of the V̇e–V̇co₂ relationship is clearly abnormal. An abnormal breathing pattern of reduced Vt and increased respiratory frequency is usually observed throughout exercise in response to a spectrum of pulmonary gas exchange abnormalities including evidence of inefficient ventilation (increased V̇e/V̇co₂) due to increased dead space ventilation (abnormal Vd/Vt) and arterial hypoxemia with reductions in Pa_O2 and abnormal widening of the P(a–a)O₂. Respiratory frequency increases as a result of mechanical stimulation of J receptors (169). Patients with pulmonary vascular disease and elevated right-sided heart pressures may also experience hypoxemia due to right-to-left shunting across a patent foramen ovale, which occurs in about 20% of patients.

Reference values for normal subjects should be obtained from relatively sedentary persons who walk a modest amount in the course of their daily life, but who do not participate in regular aerobic exercise. Thus, fit athletes as well as those who are regular joggers or cyclists may achieve peak V̇o₂ and anaerobic threshold values that are well above predicted. Peak V̇o₂ is often low or at the lower limit of normal in subjects who are very deconditioned but otherwise normal. There is early-onset metabolic acidosis (low AT). There is often a left-shifted HR–V̇o₂ relationship (increased submaximal HR responses) with a normal slope and a normal peak HR and, consequently, little or no heart rate reserve (480, 481). Peak O₂ pulse is reduced. The ventilatory response to low levels of exercise is usually normal. However, increased submaximal V̇e is observed at any level of V̇o₂ above AT as a reflection of increased metabolic acidosis compared with normal subjects. There is usually significant ventilatory reserve. Pa_O2 and Vd/Vt responses are usually normal. Although V̇e/V̇co₂ responses may be elevated at a given V̇o₂ above the AT because of alveolar hyperventilation, peak V̇e/V̇co₂ is usually normal, as is the slope of the V̇e–V̇co₂ relationship.

Deconditioning (relative unfitness) is often difficult to distinguish from early or mild heart disease (28, 42, 48). Clinical history is extremely helpful in making this distinction, as are the changes in CPET responses to an aerobic training program with monitoring of responses (V̇o₂, O₂ pulse, AT, HR), which may help to distinguish between these clinical entities. Deconditioning is often associated with chronic illness and should be considered as a contributing factor to exercise intolerance (138, 139, 145, 310). Deconditioning is also an important factor contributing to a reduced V̇o₂peak and exercise intolerance in patients with mitochondrial myopathy (55, 56). Work has demonstrated that training significantly improves many exercise variables and overall exercise performance in these patients, although not to normal levels. It appears that they are deconditioned most probably because of the mitochondrial myopathy (56, 482, 483).

A spectrum of exercise response patterns can be seen in patients with COPD. Patients with moderate to severe COPD will usually experience exercise intolerance and have a reduced peak work rate and peak V̇o₂, whereas patients with mild COPD will often have an exercise response pattern and peak V̇o₂ similar to normal. However, altered CPET responses have been noted in some patients with mild COPD (290, 484, 485). Abnormal exercise tidal flow–volume loops with expiratory flow limitation despite otherwise “normal” CPET responses have also been reported in mild COPD (450). The slope of theV̇o₂–WR relationship is usually normal (486, 487).

One of the distinguishing features of many patients with moderate to severe COPD is a reduced ventilatory reserve (V̇e/MVV approaching or exceeding 100%), signaling a significant ventilatory contribution to exercise limitation (see Section IV,7: Ventilation). The breathing strategy adopted by patients with moderate to severe COPD during exercise includes a respiratory frequency that is higher and a Vt that is lower compared with normal subjects at the same V̇e, an increase in EELV due to dynamic hyperinflation, and a resultant reduction in inspiratory capacity (TLC − EELV = IC) (Figure 7) (45, 82, 87, 288, 488, 489). Consequently, inspiratory muscle dysfunction due to the increased elastic load resulting from dynamic hyperinflation may further compromise ventilatory capacity and neuroventilatory coupling (45, 82). Blunted tidal volume and inspiratory capacity responses due to dynamic hyperinflation are associated with increased work of breathing and dyspnea (45, 75, 76, 81, 85, 87, 489). Hyperoxia (Fi_O2 = 60%) has been shown to improve exercise endurance in patients with severe COPD (FEV₁ = 31%), purportedly by reducing ventilatory demand, reducing end-expiratory lung volume, and relieving dyspnea (490). Importantly, exercise cessation due to leg fatigue is reported in a notable number of patients with severe COPD, along with evidence of significant ventilatory constraint (4, 5). Exercise limitation is usually multifactorial.

Although patients with moderate to severe COPD will have increased submaximal HR responses (441), peak HR is usually reduced compared with normal subjects. There is usually significant heart rate reserve, a reflection that the cardiovascular system has been relatively unstressed. In a retrospective study of patients with COPD categorized as mild, moderate, or severe, it was demonstrated that as disease severity progressed, V̇o₂max and ventilatory reserve decreased and heart rate reserve increased (491). O₂ pulse is usually (but not invariably) proportionately reduced to V̇o₂peak and may be due to a combination of factors—ventilatory limitation, deconditioning, and, possibly, hypoxemia. A reduced O₂ pulse has also been suggested as possibly reflecting the hemodynamic consequences of dynamic hyperinflation (492).

Other respiratory abnormalities include a trending of increased submaximal V̇e and inefficiency of ventilation (increased V̇e/V̇co₂) due to the increased dead space ventilation with abnormal Vd/Vt responses. A reduced level of alveolar ventilation due to a blunted ventilatory response to metabolic acidosis, manifested by no change or increase in Pa_CO2, may be seen with moderate to severe COPD (404). It has been suggested that in advanced COPD, the likelihood of developing hypercapnia during exercise with marked V̇a/Q̇ abnormalities primarily reflects severe mechanical constraints on ventilation due to dynamic hyperinflation (493, 494).

In COPD, the AT response may be normal, low, or indeterminate (cannot be identified but does not mean that a metabolic acidosis is precluded) by noninvasive gas exchange criteria (see Section IV,4: Anaerobic Threshold). Blood samples for standard bicarbonate or lactate help avoid false-positive noninvasive AT determinations that have been reported in COPD (330) Most patients with mild to moderate COPD can achieve a metabolic acidosis (316, 319) whereas others with severe COPD cannot (38, 161). A low AT may reflect deconditioning due to physical inactivity and/or skeletal muscle dysfunction (73, 159, 445) including alterations in exercise-related glutamate levels (89, 90), concomitant pulmonary vascular disease, cor pulmonale, or occult left ventricular dysfunction. Consequently, a low AT may have limited discriminatory value for interpretative purposes (i.e., distinguishing heart versus lung) in patients with COPD.

Patients with COPD usually have low Pa_O2 values at rest. During exercise Pa_O2 may increase, decrease, or remain the same but is more likely to be reduced in patients with moderate to severe COPD (42) (see Section IV,10: Pulmonary Gas Exchange); P(a–a)O₂ usually increases abnormally, especially when Pa_O2 decreases (405). Exercise desaturation occurs more frequently in patients with predominantly emphysema (495, 496) than in those with predominantly chronic bronchitis, in whom increases in Pa_O2 may occur (405). Several mechanisms may be responsible for the arterial desaturation and increases in P(a–a)O₂ seen during exercise in patients with COPD. These include the impact of a fall in mixed venous O₂ on low V̇/Q̇ lung units and shunts and hypoventilation in some patients with COPD (497) as well as diffusion limitation in some severe cases (FEV₁, 36% predicted) (498). Arterial desaturation has been shown to be correlated to reduced diffusing capacity (271, 405). In turn, COPD patients with Pa_O2 values that either improve or remain the same with exercise have demonstrated improvement in V̇/Q̇ relationships (498, 499).

In interstitial lung disease a spectrum of respiratory (mechanical and pulmonary gas exchange) and cardiovascular abnormalities may be observed, which may reflect differences in disease severity and/or differences in ILD types (see below). Peak V̇o₂ and peak WR are usually reduced in ILD (94). The V̇e–V̇o₂ relationship reveals an often impressive increase in submaximal V̇e resulting from many stimuli but largely caused by increased levels of dead space ventilation. The slope of the V̇e–V̇co₂ relationship is also increased. A reduced ventilatory reserve (high V̇e/MVV) and ventilatory limitation to exercise are often seen, primarily reflecting deranged pulmonary mechanics (see Figure 14). In addition, many factors contribute to the increased ventilatory demand in patients with ILD including increased dead space ventilation (Vd/Vt), inefficient ventilation (V̇e/V̇co₂), hypoxemia, early onset metabolic acidosis, pulmonary hypertension, lung receptors, and increased central drive (98, 99). Because of mechanical constraints on tidal volume expansion, an abnormal breathing pattern consisting of increased respiratory frequency, reduced Vt, and minimal change in EELV is seen (94, 97–99, 378, 500, 501). In severe ILD, Vt approaches IC early in exercise and thereafter V̇e increases, mostly due to fr increases alone (3, 97–99).

Exercise tidal flow–volume loop analysis in those ILD patients who stop exercise because of dyspnea reveals significant expiratory flow limitation, increased EELV, abnormal inspiratory flow loops, and less arterial desaturation compared with those who stop because of leg fatigue (Figure 9)

Figure 9. Maximal and extFVL in patients with ILD. Left: Patients who stopped secondary to dyspnea. Right: Patients who stopped due to leg fatigue. Minimal change was observed in EELV in either group, with the group complaining of dyspnea demonstrated significant expiratory flow limitation (modified from Marciniuk and coworkers [378]). EELV = end-expiratory lung volume; ExtFVL = exercise tidal flow–volume loop; ILD = interstitial lung disease.

Download Figure | Download Powerpoint

[More] [Minimize]

At rest, patients with ILD will usually have either normal or various levels of reduced Pa_O2. During exercise, in most patients with significant ILD, impressive arterial desaturation and abnormal increases in P(a–a)O₂ are observed (29, 31, 94, 95, 504, 505). Multiple mechanisms, including V̇a/Q̇ mismatching, O₂ diffusion limitation, and low mixed venous Po₂, have been shown to contribute to the abnormally increased P(a–a)O₂ (497, 504). Patients with ILD may also experience hypoxemia as a consequence of right-to-left intracardiac shunting. Arterial desaturation has been shown to be correlated with resting Dl_CO measurements in patients with ILD (506). Although the prediction of exercise desaturation from resting Dl_CO measurements in patients with ILD may not be consistently reliable (95, 507), patients with Dl_CO less than 70% are more likely to desaturate (95).

Work has suggested that hypoxemia, not respiratory mechanics, appears to be the predominant factor contributing to exercise limitation during incremental exercise in patients with ILD (93, 508). In these studies, despite improvement of mean V̇o₂peak by about 20% with supplemental O₂, V̇o₂peak was still only 67% of predicted, suggesting that exercise limitation was indeed multifactorial (93).

Inefficient ventilation (increased V̇e/V̇co₂ responses) due primarily to increased Vd/Vt and also hyperventilation due to hypoxemia and mechanoreceptor stimulation are usually observed throughout exercise. Hypoxemia not only contributes to the exaggerated ventilatory response, but may also contribute to reduce O₂ delivery primarily through a reduced O₂ content and also via hypoxic vasoconstriction (and the attendant increase in right ventricular afterload). Pa_CO2 may increase, decrease, or remain the same (94). Pulmonary gas exchange abnormalities including increased Vd/Vt, reduced Pa_O2, and widened P(a–a)O₂ occur even in mild disease and may be a useful diagnostic tool for detection of early disease (48, 49).

Cardiovascular abnormalities are common and reflect pulmonary vascular and right ventricular dysfunction. The HR–V̇o₂ relationship reveals higher HR at submaximal levels of V̇o₂ compared with normal subjects. Low peak HR responses are usually observed, especially in more severe disease. Consequently, heart rate reserve may be increased or normal. The O₂ pulse is reduced. The AT response can be normal, although a low AT commonly occurs and may be due to pulmonary circulatory and/or RV dysfunction (O₂ delivery) and/or skeletal muscle dysfunction and deconditioning (O₂ utilization) (352).

Differences in CPET response(s) are found between different specific ILD types (31, 509). Patients with interstitial pulmonary fibrosis (IPF) experience a greater degree of arterial desaturation and abnormal pulmonary gas exchange during exercise than do patients with sarcoidosis (510) and patients with cryptogenic fibrosing alveolitis-scleroderma type (511, 512). Some investigators have suggested that patients with sarcoidosis may have disproportionate cardiovascular abnormalities due to associated left- and right-sided myocardial involvement (513–515). Additional investigation is necessary. Although patients with ILD have been traditionally thought to be ventilatory limited, a retrospective analysis has suggested that circulatory factors may be primarily exercise limiting (352). It would appear that patients with IPF (516, 517) are more likely to be circulatory limited than are those with scleroderma (478). However, this provocative concept requires additional work for IPF and for other ILD entities. The coexistence of a high V̇e/MVV (no ventilatory reserve) and cardiovascular abnormalities and limitations may signal the presence of combined exercise limitation (1).

A spectrum of exercise responses can be seen, depending mostly on the severity of the obesity. V̇o₂peak may be reduced when expressed per kilogram of actual body weight or normal when expressed per kilogram of ideal body weight (see Section IV,1: Oxygen Uptake). There is an excessive metabolic requirement (i.e., V̇o₂ may be increased for a given work rate) but with a normal slope (ΔV̇o₂/ΔWR) (301, 302). Resting V̇o₂ is not appreciably different in obese subjects compared with normal subjects of the same ideal or lean weight, reflecting the relatively low metabolic rate of the adipose tissue. During cycle exercise at unloaded (0 W) pedaling V̇o₂ increases excessively (190). The excessive metabolic requirement reflects the high-energy cost of moving the weight of the legs. Obese subjects generally cannot attain the same work rates compared with normal subjects. In the moderately obese subject, although work capacity is impaired, V̇o₂peak, peak O₂ pulse, and the AT are usually normal, reflecting the “training effect” induced consequent to the demands of performing their habitual activities while “loaded” with a greater body mass. Increased HR at submaximal work rates with attainment of a (near) normal peak HR usually results in little/no heart rate reserve. Work (518) comparing obese patients with normal subjects demonstrated that the AT was reduced and that noninvasively determined cardiac index (CO₂-rebreathing method) during exercise below the AT was lower in obese patients as compared with normal subjects. The authors postulated that obese patients might have relatively less efficient cardiac performance, relying on greater tissue extraction during exercise compared with normal subjects. In another interesting study (519), indices of left ventricular diastolic filling were measured by pulse Doppler–echocardiography in asymptomatic, morbidly obese patients and in a matched lean control group. Fifty percent of the obese patients had left ventricular diastolic filling abnormalities. The authors concluded that abnormalities of diastolic function occur frequently in asymptomatic, morbidly obese patients and that this may represent a subclinical form of cardiomyopathy.

As a result of the increased metabolic requirement, V̇e at a given external work rate is higher for obese subjects. V̇e/MVV (measured) is usually normal but may be increased in extreme obesity. However, if MVV is calculated as FEV₁ × 40, the V̇e/MVV ratio may be low. This may be misleading and underestimate true ventilatory limitation. Exercise tidal flow–volume loop analysis suggests that there is (variable) ventilatory constraint due to breathing at low lung volumes and inability to increase EELV sufficiently during exercise (presumably secondary to the increased inspiratory load), which results in expiratory flow limitation (288, 520). A “pseudo-asthma” syndrome in obese patients has been described (521).

In obesity, a breathing pattern during exercise characterized by increased respiratory rate and reduced Vt compared with normal subjects has been reported (522). This may represent an attempt to minimize the work of breathing due to the increased elastic load of excess adipose tissue. Abnormal resting Pa_O2 and P(a–a)O₂ resulting from the decrease in chest wall and lung compliances and basilar atelectasis occur (523); these abnormalities may improve as Vt increases with exercise and overall V̇/Q̇ relationships improve. The increased V̇e and normal Vd/Vt responses may result in alveolar hyperventilation and a reduced Pa_CO2 with exercise. Obesity often coexists with other conditions in impacting exercise capacity. For interpretative purposes, there is value in having peak V̇o₂ referenced to both ideal and actual body weight (see Section IV,1: Oxygen Uptake); a significant difference would suggest that obesity may be an important contributing factor to exercise intolerance.

Subjects with psychogenic disorders (anxiety reactions, hysteria, panic disorders, obsessional behavior, hyperventilation syndrome, etc.) may complain of several symptoms including exertional dyspnea, chest pain, and light-headedness. These symptoms may represent unrecognized hyperventilation due to anxiety and stress (524). Often, however, there may be no apparent psychopathology (525).

Subjects with hyperventilation syndrome will often have a normal or near normal V̇o₂peak and WR. Occasionally, reduced peak V̇o₂ and WR are observed and may reflect some other, concurrent etiology. Most importantly, abnormal breathing patterns at rest and during exercise may be revealing and, in some circumstances, almost diagnostic. CPET in these subjects often reveals impressive hyperventilation as evidenced by abnormal increases in V̇e, V̇e/V̇co₂, and respiratory frequency, and respiratory alkalosis (decreased Pet_CO2 and Pa_CO2) (526, 527). As an increased V̇e/V̇co₂ may reflect inappropriate ventilation due to either hyperventilation or excessive dead space, simultaneous arterial Pco₂ sampling is recommended. In contrast to the usual gradual increase in respiratory frequency seen during progressive exercise, subjects with psychogenic disorders may have an abrupt “turned on” onset of regular, rapid, shallow breathing disproportionate to the metabolic stress (49). Chronic respiratory alkalosis with a downregulated Pa_CO2 set point may also be observed at rest before exercise. Other symptoms (see above) due to respiratory alkalosis may also be elicited. Hyperventilation during exercise has been associated with ECG changes resembling ischemia in subjects with normal coronary arteries (528).

Irregular exercise breathing patterns punctuated by breath holding (noted by changes in Pet_CO2) and sighing have been noted in hysteria (see comments on nonphysiologic respiratory patterns in Section VIII,4.8: Poor Effort and Malingering). A complete, careful history and review of systems are essential for accurate interpretation. Identification of hyperventilation syndrome is important because appropriate treatment is usually successful (524).

Poor effort may be suspected with early cessation of exercise and a reduced V̇o₂peak, a normal or unattained AT, a low R value at exercise cessation, and substantial heart rate reserve and ventilatory reserve with no readily apparent peripheral abnormality. Malingering may be suspected when irregular, often erratic breathing patterns with intermittent hyperventilation and hypoventilation and irregular respiratory rates are noted with fluctuation in Pet_CO2 and Pa_CO2, which are unrelated to work rate. Other abnormal respiratory pattern responses characterized by frequent sighing, panting, erratic timing, and changing FRC may be associated with intentional malingering, anxiety, and swallowing or inappropriate attempts at vocalization. Such events are common but poorly characterized in the literature. The nonphysiologic respiratory patterns are important to recognize, because they may result in inaccurate characterization of exercise variables including V̇o₂, AT, and Vd/Vt (see discussion in Section IV: Conceptual and Physiologic Basis of Cardiopulmonary Exercise Testing Measurements). Patient symptom scores may be totally disproportionate to the level of effort and exhaustion noted by the person monitoring the test. Knowledge of secondary gain is critical. In both conditions, repeat testing may be helpful in demonstrating the lack of consistent response during exercise.

A reduced V̇o₂peak has many possible causes (see Section IV,1: Oxygen Uptake and Table 18), which are evaluated through responses to subsequent questioning. In turn, a normal V̇o₂peak reflects a normal aerobic capacity and provides reassurance that no significant functional impairment exists. However, in athletes and very fit subjects, a normal V̇o₂peak may be noted despite an actual significant reduction, which would not be appreciated without a previous V̇o₂peak for comparison. Clinical history including baseline physical activity determination would be helpful. A normal V̇o₂peak when referenced to ideal body weight but at a lower peak WR may also be reported in obesity (see Section VIII,4.6: Obesity). Also, patients with early or mild heart disease and lung disease may have a normal or borderline normal V̇o₂peak. Patients with gastroesophageal reflux disease may also have a normal CPET. Furthermore, although patients with psychogenic dysfunction will often have a normal or near normal V̇o₂peak, CPET may reveal other abnormalities (i.e., abnormal breathing patterns) that may be of diagnostic value (49).

The V̇o₂–WR relationship may reveal a metabolic requirement during exercise that is excessive (i.e., a greater amount of V̇o₂ per given WR). A high or left-shifted V̇o₂–WR relationship may be seen in obesity, poor exercise technique, and hyperthyroidism (301). The resulting “exercise inefficiency,” especially with coexisting cardiopulmonary disease, frequently contributes to exercise intolerance. However, a left-shifted V̇o₂–WR relationship may not be readily apparent without the appropriate corresponding reference values plotted and these are therefore strongly recommended. Although a reduced ratio of increase in V̇o₂ to increase in work rate (ΔV̇o₂/ΔWR) is reported to occur in cardiovascular disease, as noted previously, this has not been consistently observed in patients with known cardiovascular disease (195) and may also be seen in some patients with mitochondrial myopathy (54) and in patients with cystic fibrosis (110). Therefore, a low ΔV̇o₂/ΔWR may be useful in identifying an abnormal relationship but relatively nonspecific in establishing etiology (i.e., O₂ delivery versus O₂ utilization dysfunction). This is important for interpretative purposes in appropriate clinical scenarios (i.e., unexplained exertional dyspnea) (51). Additional work is required to establish the sensitivity and specificity of ΔV̇o₂/ΔWR and its discriminatory ability.

Achievement of age-predicted values for maximal HR during exercise is often used as a reflection of maximal or near maximal effort, presumably signals attainment of V̇o₂max and maximal cardiac output, and, furthermore, suggests that cardiovascular function contributes to exercise limitation. This is the usual situation in normal subjects when V̇o₂max predicted is achieved (see Section VI: Normal Integrative Exercise Response). However, when exercise stops at a low V̇o₂ but with a normal predicted HR, abnormal cardiovascular function and/or O₂ content factor (anemia, hypoxemia, or carboxyhemoglobin) and/or O₂ utilization (skeletal muscle dysfunction) is contributing to exercise limitation. This low peak V̇o₂–normal peak heart rate pattern may be seen in several different clinical entities including early or mild heart disease, mitral valve disease, pulmonary vascular disease, coronary artery disease, deconditioning, obesity, early/mild lung disease with or without hypoxemia, and mitochondrial myopathy.

Clearly, determining which of these entities is most likely requires the integration of additional exercise responses, both normal and abnormal, so that patterns of responses can be established and, when considered within a clinical setting framework, linked to appropriate clinical entities. For instance, an accompanying abnormal ECG, chest pain, and abnormal BP responses would favor coronary artery disease.

In a physically active individual with relatively recent onset exertional dyspnea, normal PFTs including a negative methacholine challenge, an unremarkable ECG, normal baseline laboratory studies, and reduced exercise tolerance, a low peak V̇o₂–normal peak HR pattern when accompanied by a low AT, low O₂ pulse, and a normal V̇e/MVV might be more suggestive of early/mild cardiovascular disease rather than deconditioning. However, without the clinical history, it would be more difficult to distinguish between these two entities. The differential diagnosis would most probably also include consideration of much less commonly encountered entities, that is, some mitochondrial myopathy in which a hyperdynamic, hyperventilatory pattern with a low O₂ pulse and low AT may also be seen (52, 54, 55). Although the AT is often low in patients with cardiovascular disease, it has limited discriminatory ability, as it is also low in deconditioning, mitochondrial myopathy, as well as other conditions. The AT is useful, however, in providing internal consistency to the integrated exercise responses.

A low peak V̇o₂–normal peak HR pattern would also include pulmonary vascular disease (PVD). Compared with early/mild cardiovascular disease and deconditioning, pulmonary vascular disease is more likely associated with abnormal V̇e/V̇co₂, Vd/Vt, Pa_O2, and P(a–a)O₂ responses. Although abnormal V̇e/V̇co₂ and Vd/Vt responses without desaturation are seen in patients with moderate to severe heart failure, the peak HR response in heart failure is highly variable, usually decreasing as disease severity increases (36). In turn, early or mild heart disease usually is associated with near normal peak HR.

If predicted peak HR is not achieved and the V̇o₂ is low, cardiac limitation due to coronary artery disease, peripheral vascular disease, chronotropic dysfunction, or noncardiovascular limitation to exercise is possible. The noncardiovascular causes include poor effort, lung disease, drug-induced (β-blockers) and peripheral factors related to muscles, and neuromuscular or arthritic abnormalities.

One of the most challenging tasks in the interpretation of CPET result relates to the assessment and influence of ventilatory limitation (or constraint) on exercise limitation. Assessing the degree of ventilatory constraint (limitation) has traditionally been based on the “ventilatory reserve,” defined as peak V̇e/MVV or peak V̇e − MVV, or as the relation between ventilatory demand represented by V̇epeak and ventilatory capacity estimated by MVV. Significant controversy, however, surrounds this practice in part because of a lack of definitive measurement of ventilatory capacity (see extensive discussion in Section IV,7: Ventilation). Although emerging technologies, in particular exercise tidal flow–volume loops referenced to maximal flow–volume loops, have provided additional valuable insight into how mechanical constraints limit exercise (Figures 5–9), clinical validation in different clinical settings is required (288). For the future, continued use of V̇e/MVV and the unique visual assessment provided by extFVL/MFVL may be useful together. Alternatively, the role of negative expiratory pressure applied while monitoring the exercise tidal flow–volume loops may provide additional insight (380). Additional investigation is required.

Consequently, peak V̇e/MVV, as an indirect index of ventilatory constraint, is practical and empirically useful. A high peak V̇e/MVV (or low/no ventilatory reserve) typically, but not invariably (see below), signals the presence of a pulmonary problem (485). In severe COPD, the peak V̇e/MVV usually approaches or exceeds 100%, whereas in mild COPD, it may be normal with significant ventilatory constraint apparent only with extFVL/MFVL analysis (450, 491). In moderate COPD, the peak V̇e/MVV may approximate the upper normal 95% CI (about 85%), but may not approach 100%. Ventilatory constraint contributing to exercise limitation may be clinically suspected if other respiratory abnormalities known to increase ventilatory requirements during exercise are present (see Section VIII,4.4: COPD). In this situation, extFVL/MFVL analysis might be helpful (Figure 7). In patients with ILD, the peak V̇e/MVV is usually higher than normal but usually not as high as in COPD (378). Ventilatory constraint to exercise limitation due primarily to mechanical derangements and other factors increasing ventilatory requirements may be a function of type and severity of ILD (see Section VIII,4.5: ILD). Active investigation of this topic is underway presently.

Importantly, the peak V̇e/MVV is usually normal in cardiovascular disease and, as such, is helpful in distinguishing heart from lung disease in patients with both abnormal cardiovascular and respiratory exercise responses. In patients with a low V̇o₂, low O₂ pulse, low AT, and either a normal or reduced heart rate reserve, the associated findings of abnormal ECG changes, no drop in Pa_O2 or Sa_O2, and a normal peak V̇e/MVV reinforce the likelihood of a cardiovascular etiology (see Section VIII,4.1: Heart Failure). In turn, the presence of a reduced or absent ventilatory reserve (peak V̇e/MMV > 100%) in patients with a clinical profile and exercise response pattern containing cardiovascular and respiratory abnormalities may signal the presence of combined cardiovascular and ventilatory limitation (276, 443). Likewise, patients with ILD can have exercise response patterns reflecting combined pulmonary vascular and ventilatory limitation to exercise (V̇e/MVV, about 90–100%). Interval evaluation for disease progression may establish the prevailing predominant exercise-limiting factor (1).

The peak V̇e/MVV is usually normal in deconditioning but may exceed or approach 100% in athletes, a reflection that V̇o₂max predicted has been reached or exceeded (529). In contrast, when peak V̇e/MVV is greater or equal to 100% in patients, the V̇o₂peak is usually reduced. Measured ventilatory reserve is usually normal or increased (peak V̇e/MVV is normal or low) in patients with neuromuscular disorders. Often, however, the calculated MVV (FEV₁ × 40) is significantly greater than the actual measured MVV. The disparity in ventilatory reserve should raise suspicion for neuromuscular disease. The spectrum of neuromuscular abnormalities, however, may be wide, reflecting different pathophysiologies (42, 530). Patients with apparent predominantly peripheral exercise limitation (e.g., status post lung and heart–lung transplantation) will have a low V̇o₂, low O₂ pulse, low AT, low R, and significant ventilatory reserve (low peak V̇e/MVV) and HRR at end exercise (138, 139, 145). Finally, poor effort is suspected when reduced V̇o₂peak is associated with high ventilatory reserve (low V̇e/MVV) and HRR and a low R without apparent peripheral problems (see Section VIII,4.8: Poor Effort and Malingering).

Exercise hypoxemia most often reflects intrinsic lung disease, pulmonary vascular disease, and, less frequently, intracardiac left-to-right shunts (see Section IV,10: Pulmonary Gas Exchange; Section VIII,4.2, Pulmonary Vascular Disease; Section VIII,4.4, COPD; and Section VIII,4.5: ILD). A fall in Sa_O2 (ΔSa_O2) of ⩾ 4%, Sa_O2 ⩽ 88%, or Pa_O2 ⩽ 55 mm Hg during CPET is usually considered clinically significant (13, 25). As such, exercise desaturation occurs in most patients with ILD, pulmonary vascular disease, in many patients with COPD, and in elite athletes (396). Furthermore, supplemental O₂ administered during exercise has been shown to significantly improve exercise performance in these patients (74, 93). Although the exact mechanism of action of O₂ in improving exercise performance in these populations is not yet fully understood (i.e., increased transport of O₂ to the exercising muscles, reduced hypoxic drive to breathe, reduced dyspnea, improved cardiovascular function), it is apparent that hypoxemia significantly contributes to limit exercise performance in those patients who become hypoxemic during exercise.

Hypoxemia is a significant contributor to exercise limitation in both pulmonary vascular disease and ILD. However, the distinction between ILD and pulmonary vascular disease is often difficult, as there may be significant overlap in several cardiovascular, metabolic, ventilatory, and pulmonary gas exchange exercise responses (see Table 18). The clinical profile (history, physical examination, chest X-ray, high-resolution and/or spiral computed tomography scan, etc.) is often critical. Importantly, hypoxemia is usually not seen in patients with anemia, carboxyhemoglobin, deconditioning, obesity, heart failure, or mitochondrial myopathy.

Inefficiency of ventilation as reflected in abnormal V̇e/V̇co₂ responses during exercise results from either increased dead space ventilation (Vd/Vt) due to ventilation–perfusion imbalance and/or relative alveolar hyperventilation. The latter can be caused by increased central drive, increased mechanoreceptor activity, or hypoxemia and is evidenced by a reduced Pa_CO2. Patients may be subject to each of the two major causes, with the predominant effect usually defined by the Pa_CO2. Abnormal exercise V̇e/V̇co₂ responses are observed in patients with cardiovascular disease, respiratory disease, and mitochondrial myopathy, and in subjects with psychogenic dysfunction (hyperventilation syndrome, anxiety, etc.).

Importantly, an abnormal V̇e/V̇co₂ response may signal the need to obtain arterial blood gases during exercise, so that Pa_CO2 and Vd/Vt can be determined directly (see Section III,3: Exercise Test with Arterial Blood Sampling). This may be helpful in more definitively establishing hyperventilation without intrinsic lung disease versus abnormal dead space ventilation.

Although abnormal V̇e/V̇co₂ responses are found in several clinical entities, it is likely that in COPD and ILD its impact on exercise limitation is most directly realized. In patients with COPD, abnormal V̇e/V̇co₂ responses are usually due to increased Vd/Vt, although other causes—hypoxemia, mechanoreceptor stimulation, cor pulmonale, and so on—and other abnormal respiratory exercise responses may coexist. These factors all contribute to an increased ventilatory requirement for exercise, which adversely affects the ventilatory demand/capacity relationship and may result in a reduced ventilatory reserve. The slope of the V̇e–V̇co₂ relationship is an independent prognostic indicator of poor outcome in patients with heart failure (62).

Early onset metabolic acidosis manifested by a low AT is commonly seen in patients with COPD, ILD, pulmonary vascular disease, heart failure, mitochondrial myopathy, and deconditioning. Because finding a low AT is not specific to any one disease or group of diseases, its role as a discriminatory variable is limited. However, the AT may be useful in interpretative schemes when used in conjunction with other variables to corroborate patterns of response with different clinical entities and to establish consistency among the measurements. Finding a low AT is evidence of abnormal oxygen delivery to working muscles and/or intrinsic muscle abnormalities. In situations in which the AT measured noninvasively is undetermined or questionable (see Section IV,4: Anaerobic Threshold) and the detection or quantification of metabolic acidosis is needed, blood samples for lactate or standard bicarbonate are recommended (269, 330).