Although experimental evidence indicates that NO has both a neurotoxic and neuroprotective role, NO generation has recently been related to brain damage in acute ischemic stroke. High NO metabolites (NOx) in CSF within the first 24 hours of the onset of symptoms were associated with higher stroke severity at admission, early neurologic deterioration, larger infarct volumes, and poor outcome at 3 months in a series of 102 patients with a hemispheric ischemic stroke (
Castillo et al., 2000). These findings suggest predominant cytotoxic effects of NO in human stroke, and open new therapeutic avenues with specific nNOS and iNOS selective inhibitors.
Markers of NO-mediated brain injury could be useful to guide neuroprotective strategies. Because the taking of CSF samples is not feasible in clinical practice, molecular markers in blood that reflect the importance of NO generation are needed. NOx concentrations in plasma are subjected to high variability due to the influence of diet and renal function (
Moshage, 1997). However, the plasma value of the NO precursor, l-arg, in predicting ischemic cerebral lesion has not been studied. We have found a highly significant negative correlation between NOx and L-arg concentrations in CSF, suggesting a consumption of extracellular l-arg for the synthesis of NO in patients with greater cerebral damage (
Castillo et al., 2000). Therefore, plasma l-arg concentrations might decrease as a result of NO generation after acute ischemic stroke.
In this study, we investigated the relationship of concentrations of l-arg in CSF and plasma with stroke outcome and infarct volume in patients with an acute ischemic stroke. In addition, the blood profile of l-arg was analyzed in a group of patients admitted within the first 8 hours of the onset of symptoms, and in a rat stroke model of permanent middle cerebral artery occlusion (pMCAO) after the administration of a vehicle or 1400W, a specific iNOS inhibitor.
MATERIALS AND METHODS
Clinical study
We studied 268 patients (61% men; mean age, 68 ± 10 years) admitted consecutively with an acute hemispheric ischemic stroke within 24 hours of the onset of symptoms, and 50 control subjects. The purpose of this investigation was to analyze clinical, biochemical, and radiologic factors related to early neurologic deterioration. This series has been the core of recent investigations (
Dávalos et al., 2000;
Vila et al., 2000) and the characteristics of the control group and the inclusion criteria for patients have already been described (
Castillo et al., 1996,
1997). In summary, controls were subjects without neurologic disorders subjected to epidural anesthesia (60% men; mean age, 56 ± 17.5 years), and patients had a persistent focal neurologic deficit and absence of mass effect or cerebral hemorrhage on the cranial computed tomography (CT) performed before inclusion. The mean time from the onset of symptoms to the arrival at the hospital was 8.2 ± 5.9 hours (range, 1.5–23 hours). Blood samples were taken, and stroke severity was quantified using the Canadian Stroke Scale (CSS) on admission and 48 hours after hospitalization. As soon as informed consent was obtained and the cerebral CT was completed, a spinal tap was performed in 242 patients. The average time between taking the first blood sample and the CSF sample was 1.2 ± 0.8 hours. A second brain CT scan was performed between days 4 and 7 after clinical onset. Early signs of infarction were carefully evaluated in the first examination. In the control CT, infarct volume (0.5 × a × b × c, where a and b are largest perpendicular diameters measured on the computed tomography, and c is the slice thickness) was calculated. The same radiologist who was blind to clinical and biochemical results performed all CT evaluations. Patients received standard treatment according to published guidelines (
Adams et al., 1994). Two clinical outcome measures were evaluated: (1) early neurologic deterioration (END), as a potential sign of enlarging brain injury, and (2) functional capacity at 3 months. Following already published criteria (
Dávalos et al., 1990), END was diagnosed when the CSS score dropped 1 or more points within the first 48 hours of hospitalization. We used the Barthel index to assess patients' functional capacity at 3 months. Poor outcome was defined as death or Barthel index lower than 85 (
Sluter et al., 1999).
To analyze the plasma l-arg profile, we studied a separate group of 29 patients (mean age, 69 ± 8 years) with an acute ischemic stroke of less than 8 hours duration. Blood samples were collected at admission (mean time from stroke onset, 4.9 ± 1.4 hours) and at 12, 24, 36, 48, and 120 hours from the onset of symptoms. Cerebrovascular risk factors and stroke subtypes were similar to those of the total series. The median [quartile] CSS score at admission was 5.5 [3.0,8.0]. One patient died on the second day of hospitalization, so the complete profile was available in 28 patients.
Experimental study
Experiments were performed on male Fischer rats weighing 225 to 275 g. Rats were anesthetized with 2.5% halothane in a mixture of 70% nitrogen/30% oxygen, and permanent focal cerebral ischemia was induced by ligature of the left common carotid artery and occlusion of the ipsilateral distal middle cerebral artery as described previously (
Puig et al., 2000). During the process, body temperature was maintained at 37.5°C ± 0.5°C. All procedures conformed to the Committee of Animal Care at the Universidad Complutense of Madrid according with European Union rules (DC86/609/CEE).
Three groups were used for determinations of l-arg levels and infarct volume. Rats in which the middle cerebral artery was exposed but not occluded (sham-operated controls, n = 8), rats with pMCAO (n = 8), and rats which had received 20-mg/kg N-(3-(aminomethyl)benzyl) acetamidine (1400W; GlaxoSmithKline, UK) at onset of ischemia and at 8-hour intervals for 3 days after pMCAO by a subcutaneous injection volume of 1 mL/100 g body weight (MCAO + 1400W; n = 8). The dose and time of administration of 1400W was chosen according to our previous data (
Cárdenas et al., 1998;
Menchén et al., 2001). Blood samples were obtained from the tail before pMCAO (time 0) and 1, 2, 4, 6, 24, 48, and 72 hours after pMCAO in the presence or absence of 1400W.
The brains were removed 72 hours after pMCAO, and series of 2-mm coronal brain slices were obtained (Brain Matrix, WPI, UK) and stained in 1% TTC (2,3,5-triphenyl-tetrazolium chloride, Merck) in 0.1-mol/L phosphate buffer. The infarcted area, which is not stained, was quantified by image analysis (Scion Image for Windows 2000, Scion Corporation, Frederick, MD, U.S.A.).
Laboratory determinations
Blood and CSF samples were centrifuged and immediately stored at −80°C until l-arg determination. Quantification of l-arg was performed by high-performance liquid chromatography following the method described elsewhere (
Castillo et al., 1996). Amino acid determinations from patients and animals were done in the same laboratory and were blinded to the experimental group, clinical and neuroimaging findings, and to stroke outcome.
Statistical analyses
l-Arg concentrations are expressed as median [quartiles], because they were not normally distributed. The comparison of l-arg concentrations between two groups was performed with the Mann-Whitney test, and comparisons were made among more than two groups with the Kruskal-Wallis test. Comparisons of repeated measures were done with the Friedman test. Spearman analysis was used for bivariate correlations between l-arg and CSS score, time from stroke onset to inclusion, and infarct volume.
We used logistic regression analysis to study the importance of l-arg concentration on END and stroke outcome. The effect on infarct volume was analyzed by multiple linear regression analysis, after a log-transformation of infarct volume to achieve a normal distribution. The models were adjusted for the time from onset of symptoms to hospitalization, and for the variables associated with END and poor outcome in our previous investigations conducted in the same series of patients: age, body temperature, serum glucose, CSS score on admission, and early CT signs of cerebral infarct (
Castillo et al., 1996,
1997;
Dávalos et al., 2000;
Vila et al., 2000). One logistic model was built for each l-arg parameter (CSF levels on admission and plasma concentrations at 48 hours), so the odds ratios are given after adjusting for the same six covariates. In the same way, the effect of each l-arg parameter on infarct volume was studied in a separate linear regression model.
DISCUSSION
Our findings show that patients with acute ischemic stroke have lower levels of l-arg in plasma and CSF than control subjects, and that low concentrations of l-arg are associated with greater cerebral damage. l-Arg concentrations in CSF within 24 hours of the onset of symptoms, and in plasma 48 hours after acute stroke, had the highest correlation to the clinical signs of brain injury. In this study, two facts suggest that low l-arg levels resulted from the consumption of this amino acid in the molecular events triggered by cerebral ischemia. Firstly, we have observed that the longer the time from symptoms onset, the lower the CSF and plasma values of l-arg. Secondly, serial determinations in blood in a group of patients with acute stroke and in rats after pMCAO, showed a decrease in l-arg concentrations, with a peak value between 6 and 24 hours after the onset of ischemia.
The association between l-arg consumption and poor stroke prognosis may be explained by the neurotoxic effects of NO generation, because l-arg is the only known precursor in the synthesis of NO (
Moncada and Higgs, 1993;
Sessa, 1994). NO is an endogenous free radical, which has a double role in cerebral ischemia (
Iadecola et al., 1994). In early stages, small quantities of NO generated by eNOS cause vasodilation and hence an increase in the collateral blood flow limiting the extent of brain injury, whereas in a second phase, a greater production of NO by nNOS and iNOS isoforms worsens cerebral damage (
Huang et al., 1994;
Iadecola et al., 1995b,
1997b). This explanation fits in with the time-dependent opposite effects of l-arg. In animal models, l-arg has a neuroprotective role when it is administered up to 2 hours after onset of cerebral ischemia, but increases infarct volume when its administration is delayed by 24 hours. The deleterious consequence of a delayed l-arg administration is thought to result from NO generation by iNOS expression after a time lag of 6 to 12 hours (
Iadecola et al., 1995a;
Zhang et al., 1995).
Some clinical findings support the idea that low l-arg levels in acute stroke are due to NO generation. In a previous study, we found a negative correlation between NO metabolites (NOx) and l-arg concentrations in CSF within 24 hours of stroke onset (
Castillo et al., 2000). CSF NOx were particularly high in patients with END and in those with l-arg levels below 6.0 μmol/L. The present findings replicate, in a larger series of patients, the results of our earlier investigation, because the median value of l-arg in CSF of patients with END was 6.4 μmol/L in contrast with 15 μmol/L in patients with good early and late outcome.
Interestingly, what we have observed in plasma may be a reflex of what happened to the CNS within the first 24 hours of stroke, given that there was a lineal correlation between levels of l-arg in the CSF at admission and the levels of l-arg at 48 hours in plasma. These findings, together with the high correlation between blood and CSF values in control subjects, suggest that NO generation after cerebral infarction by
de novo expression of the inducible NOS isoform might increase the demand of L-arg, and presumably reduce the extracellular concentrations, firstly in CSF and subsequently in plasma, due to good diffusion of this amino acid through the blood–brain barrier. l-Arg has been found to cross blood–brain barrier through a transporter with specificity for amino acid analogues possessing cationic terminal guanidine groups, such as those contained in l-arg (
Mahar et al., 2000), but its diffusion may be even easier as a result of stroke (
del Zoppo, 1994;
Sage et al., 1984). We can reasonably rule out a plasma l-arg decline due to a low l-arg intake in the acute phase of stroke because the concentrations of tryptophan, an amino acid not involved in the pathophysiology of cerebral ischemia, were stable during the same period. One of the major questions raised by our study is whether low l-arg levels in blood are the expression of brain ischemia or originate as a result of the acute-phase reaction or systemic causes like concomitant infections, these factors being responsible for the neurological worsening. Because recent infections have been associated with an increased risk of impending stroke and the release of proinflammatory molecules, NO generation and l-arg consumption could have been influenced by a recent infection (
Grau et al., 1995). To partially control for such effects, we included in the analysis indirect markers of infections, such as body temperature, and other prognostic factors of brain damage.
Although this clinical study does not allow us to rule out the possibility of a reduction of plasma l-arg after stroke onset due to metabolic reactions other than the l-arg–NO pathway (
Moncada et al., 1991;
Moncada and Higgs, 1993;
Guayao and Morris, 1998), the experimental study confirms our clinical hypothesis. The decrease in plasma l-arg concentrations after pMCAO was inhibited by the administration of one of the most selective inhibitors of iNOS isoform described to date, an effect that correlated with a significant reduction in infarct volume. In contrast with aminoguanidine (an iNOS inhibitor used to produce neuroprotection after focal ischemia in rats that also inactivates the constitutive isoforms of NO synthase in the simultaneous presence of Ca
2+, calmodulin, and other cofactors;
Iadecola et al., 1995b;
Wolff and Lubeskie, 1995), 1400W is an irreversible inhibitor or an extremely slowly reversible inhibitor of iNOS. Furthermore, physiologic concentrations of the substrate, l-arg, reverse the weak inhibition that 1400W exerts on the constitutive isoforms (nNOS and eNOS), while not affecting inhibition on iNOS activity. Although 1400W is closely related to bisisothioureas, compounds that are acutely toxic, it has been reported that toxicity is only observed at doses higher than the therapeutic ones (
Garvey et al., 1997). Therefore, we can attribute the reduction of plasma l-arg concentration after stroke onset to the NO generation by the expression of the iNOS isoform.
In conclusion, this study has demonstrated l-arg consumption after acute ischemic stroke, particularly in patients with greater ischemic brain damage and worse stroke outcome. Taken together with the experimental results, our data indicate that determination of l-arg levels in blood might be useful to evaluate the neurotoxic effects of NO generation. These findings might be helpful to guide future neuroprotective strategies in patients with ischemic stroke.