Mr-derived cerebral spinal fluid hydrodynamics as a marker and a risk factor for intracranial hypertension in astronauts exposed to microgravity
Abstract
Purpose
To quantify the change in cerebral spinal fluid (CSF) production rate and maximum systolic velocity in astronauts before and after exposure to microgravity and identify any physiologic trend and/or risk factor related to intracranial hypertension.
Materials and Methods
Following Institutional Review Board (IRB) approval, with waiver of informed consent, a retrospective review of 27 astronauts imaged at 3T was done. Qualitative analysis was performed on T2-weighted axial images through the orbits for degree of flattening of the posterior globe according to the following grades: 0 = none, 1 = mild, 2 = moderate, and 3 = severe. One grade level change postflight was considered significant for exposure to intracranial hypertension. CSF production rate and maximum systolic velocity was calculated from cine phase-contrast magnetic resonance imaging and compared to seven healthy controls.
Results
Fourteen astronauts were studied. The preflight CSF production rate in astronauts was similar to controls (P = 0.83). Six astronauts with significant posterior globe flattening demonstrated a 70% increase in CSF production rate postflight compared to baseline (P = 0.01). There was a significant increase in CSF maximum systolic velocity in the subgroup without posterior globe flattening (P = 0.01).
Conclusion
The increased postflight CSF production rate in astronauts with positive flattening is compatible with the hypothesis of microgravity-induced intracranial hypertension inferring downregulation in CSF production in microgravity that is upregulated upon return to normal gravity. Increased postflight CSF maximum systolic velocity in astronauts with negative flattening suggests increased craniospinal compliance and a potential negative risk factor to microgravity-induced intracranial hypertension. J. MAGN. RESON. IMAGING 2015;42:1560–1571.
Although exposure to microgravity has immediate physiologic effects,1 it is the chronic exposure to microgravity that poses a more insidious consequence of space flight. Loss of bone density and muscle mass are well-documented biological effects of microgravity.2 However, with progressively longer mission duration there is increasing evidence of visual acuity changes in astronauts associated with structural abnormalities of the retina and optic nerve that raise new concerns for crewmember health and the viability of interplanetary travel.3 In contrast to microgravity-induced changes of bone and muscle, the effects on astronaut vision are considerably more variable,3 suggesting underlying anatomic and/or physiologic factors that require identification to predict risk and develop effective mitigation strategies.
The etiology of the ocular changes remains speculative; however, magnetic resonance imaging (MRI) abnormalities identified in postflight astronauts has implicated intracranial hypertension.4 Intracranial hypertension can hypothetically affect the visual pathway through propagation of elevated cerebral spinal fluid (CSF) pressure from the intracranial compartment to the orbit with resultant expansion of the optic nerve sheath 5 and structural modification of the posterior globe 6 (Fig. 1). In support of the intracranial hypertension hypothesis, a microgravity analog study showed immediate increase in intracranial pressure (ICP) with vacuum drop testing in rats.7 Moreover, ICP was elevated within minutes in an instrumented rhesus monkey upon reaching orbit on a biosatellite.8 Regardless, proof of this hypothesis in humans has been elusive due to the reluctance in subjecting asymptomatic astronauts to invasive CSF pressure measurements. Although postflight lumbar punctures have verified elevated CSF pressures in five of six symptomatic astronauts with papilledema, baseline and in-flight measurements remain unknown to unequivocally establish temporal relationship to microgravity exposure.9, 10
Given the inelastic nature of the cranium and incompressibility of fluid, any variance of intracranial volume results in changes in ICP.11 The capability of the intracranial compartment to absorb changes in volume before pressure is pathologically elevated is referred to as the compensatory reserve.12 An important component is passive CSF and venous outflow counteracting intracranial volume spikes due to pulsatile cerebral arterial inflow, as shown in Fig. 2.13 In this example venous and CSF outflow mirrors arterial inflow with slightly delayed peaks as indicated by respective arrows (Fig. 2). The resulting CSF maximum systolic velocity generated at the cerebral aqueduct is related to the compensatory reserve capacity.14
Passive mechanisms cannot completely correct for chronic expansion of intracranial volume. This may occur in microgravity due to the absence of a hydrostatic gradient, hypothetically causing cephalad venous congestion with diminished CSF reabsorption via the sagittal sinus or cranial nerve lymphatics.10 According to Davson's equation, mean ICP is proportional to CSF outflow resistance × CSF production rate + venous sagittal sinus pressure.15 In this model, elevation in intracranial pressure could be counteracted by downregulation of CSF production, which is a mechanism confirmed experimentally with induced intracranial hypertension in dogs.16 The combination of hydrostatic gradient shifts, intracranial volume expansion, and compensatory active modulation of CSF production rate is the basis for our conceptual model of CSF hydrodynamics in astronauts exposed to microgravity (Fig. 3).
The purpose of this retrospective study was to quantify postflight differences in CSF production rate and maximum systolic velocity to further test the hypothesis of microgravity-induced intracranial hypertension and identify any associated physiologic risk factors.
Materials and Methods
The Institutional Review Board approved this Health Insurance Portability and Accountability Act (HIPAA)-compliant retrospective analysis of 45 consecutive astronaut studies from 2009 to 2012 and waived the requirement for informed consent. Data from seven healthy adult controls were analyzed from a HIPAA-compliant prospective study having informed consent and where the nature of the procedure was fully explained to the subject. MR studies were performed utilizing an eight-channel SENSE compatible head coil on a dedicated 3T Philips Intera scanner, having a maximum gradient amplitude of 80mT/m and a maximum slew rate of 200mT/ms/m (Philips Medical Systems, Best, Netherlands). Open-source image processing software (Osirix, v. 3.8.1) and freely available Segment v. 1.9 R3656 (http://segment.heiberg.se) software 17 were utilized to analyze imaging data.
Orbits
Axial 3D T2-weighted fast spin-echo high-resolution orbital images with spectral inversion recovery fat suppression were obtained using the following parameters: repetition time (TR) = 2000 msec, echo time (TE) = 100 msec, echo train length = 29, slice thickness = 1.56 mm, slice spacing = 0.78 mm, field of view (FOV) = 100 mm, matrix = 128 × 128. The astronaut dataset was intermixed with seven nonastronaut healthy controls. Raters were blinded to subject identifier, date, and other raters. Posterior globe flattening was used as an anatomical marker of intracranial hypertension, as it achieved 100% specificity in a double-blind study of patients with idiopathic intracranial hypertension.6 Posterior globe flattening of the right eye was graded using the following ordinal scale: 0 = none, 1 = mild, 2 = moderate, and 3 = severe (Fig. 4) individually by three radiologists (L.K., W.C., J.H.) with 24, 6, and 5 years of MRI experience, respectively. Interobserver variation in grade of posterior globe flattening was calculated for the three raters and the median score was tabulated for each astronaut.
The difference in grade of flattening between preflight and postflight datasets was used to establish a change due to microgravity exposure and eliminate any existing baseline abnormalities. A median score of one or more was defined as positive flattening and considered significant for anatomic evidence of intracranial hypertension. A median score of 0 was defined as negative flattening. Optic disc protrusion, which lags behind initial increases in intracranial pressure, is more characteristically associated with sustained high levels of intracranial hypertension causing prelaminar optic disc edema.18 It was coded as positive or negative by one radiologist (L.K.) as an independent indicator of intracranial hypertension.
CSF Hydrodynamics
The CSF production rate and CSF maximum systolic velocity in astronauts and healthy controls were calculated from velocity data obtained perpendicular to the mid cerebral aqueduct and presumed CSF flow direction (Fig. 5). The mid-aqueduct was selected due to a more uniform cylindrical morphology compared to the divergent margins of the aqueduct in an effort to minimize partial volume averaging artifacts and maintain slice orientation perpendicular to flow throughout the slice volume.
A 2D gradient-echo cine phase-contrast MRI (cine PC-MRI) sequence was utilized having bipolar velocity encoding gradients with the following parameters adapted from Huang et al 19: sequential cardiac phases per cardiac cycle = 51, maximum velocity encoding = 20 cm/sec, TR = 14 msec, TE = 9 msec, slice thickness = 4 mm, matrix = 256 × 256, repetitions = 2, FOV = 100 mm, flip angle = 15°. Using four precordial leads the sequence was retrospectively gated using vectorcardiogram triggering to minimize gating errors. Heart rate in beats per minute (bpm) was recorded for each study.
Due to partial volume errors in CSF production rate calculations associated with manually drawn margins of the aqueduct,19 a method based on pixel velocity maps was developed to improve identification of the true aqueduct and exclude surrounding neural tissue (Fig. 5). In the expected region of the cerebral aqueduct, pixel elements were individually analyzed by two raters each having 24 years of experience (L.K., S.R.). Only pixels having a characteristic CSF velocity waveform were used to determine stroke volume and CSF maximum systolic velocity (Fig. 5). The CSF production rate was calculated by multiplying stroke volume by heart rate. Interobserver variation of CSF production rate and maximum systolic velocity was calculated on the resulting dataset. The aqueduct was manually traced on the magnitude image of the first cardiac phase to determine cross-sectional area.
Clinical Data and Exclusion/Inclusion Criteria
Duration of exposure to microgravity in days and time from reentry to imaging was tabulated for each astronaut using biographical data available via the National Aeronautics and Space Administration, Johnson Space Center (http://www.jsc.nasa.gov/Bios/). To minimize microgravity effects on preflight CSF data, only astronauts with imaging obtained ≥30 days after the previous mission were included. To maximize microgravity effects on postflight data, only astronauts with imaging obtained ≤20 days from reentry were included. Since there is no long-term longitudinal human or animal studies on the duration of microgravity effects on CSF hydrodynamics, the exclusion criteria were set arbitrarily. However, a rat microgravity study showed that a protein mediator of CSF production in the choroid plexus is significantly overexpressed when readapted 2 days postflight compared to 5.7–8 hours.20 This suggests that upregulation of CSF production is a slow process with a tendency to overcompensate and that it is reasonable to measure at postflight day 2 or later. Mission duration ≤30 was defined as short duration and >30 days was considered long duration. CSF pressure was measured by direct lumbar puncture using column manometry with the subject placed in full flexion in the lateral decubitus position.
Statistics
Paired t-tests were used to compare correlated or repeated measurements on the same subjects. The Mann–Whitney nonparametric test was used for independent group comparisons. Reproducibility of measurements was assessed using the intraclass correlation coefficient (ICC). Reproducibility was considered excellent for ICC ≥0.9. Agreement between more than two observers was assessed using the information-based measure of disagreement (Henriques et al 21). Linear regression and the Pearson correlation coefficient were used to relate different anatomical measurements. Statistical significance was considered at values of P ≤ 0.05. All statistical analyses were conducted using MatLab R12.1 Statistical Toolbox v. 3.0 (MathWorks, Natick, MA).
Results
Astronauts
Studies from 29 astronauts were retrospectively analyzed. Fourteen of these astronauts had complete pre- and postflight datasets and met all inclusion criteria. Thirteen out of 14 of the astronauts included in the dataset had a prior mission in space. The astronaut cohort had an average age = 48 ± 3 years and preflight imaging data that averaged 1116 ± 896 days after a previous mission. The time interval between pre- and postflight imaging was on average 208 ± 171 days. The average time to imaging from reentry was 8.1 ± 6.1 days (Table 1).
Flight duration | Astronaut | Globe flattening grade change | Optic disc protrusion | Postflight days-to-MRI |
---|---|---|---|---|
Long (n = 7) | 1 | 0 | no | 14 |
2 | 0 | no | 7 | |
3 | 0 | no | 7 | |
4 | 1 | no | 9 | |
5 | 2 | no | 6 | |
6 | 2 | no | 8 | |
7* | 3 | yes | 6 | |
Short (n = 7) | 8 | 0 | no | 20 |
9 | 0 | no | 20 | |
10 | 0 | no | 8 | |
11 | 0 | no | 2 | |
12 | 0 | no | 2 | |
13 | 1 | no | 2 | |
14 | 2 | yes | 2 |
- Note that the incidence of positive flattening and severity is greater in the long duration astronauts. The asterisk indicates the astronaut who underwent a CSF pressure measurement postflight.
The disagreement between raters of grade of posterior globe flattening as measured by the information-based measure of disagreement was 0.38 preflight and 0.27 postflight. The medians of the ratings were compared using a nonparametric sign test and a significant change was found postflight (P = 0.002), confirming that 6 out of 14 astronauts (43%) had positive flattening postflight consistent with evidence of intracranial hypertension (Table 1). Two of these subjects had concomitant optic disc protrusion, further supporting intracranial hypertension in this subgroup (Table 1). The remaining eight astronauts (57%) had negative flattening without optic disc protrusion (Table 1). Long- compared to short-duration astronauts had an increased incidence and severity of positive flattening (Table 1).
The average CSF production rate postflight for all astronauts was 335 μL/min ± 164 and was not statistically different compared to 278 μL/min ± 127 preflight (P = 0.15). However, analyzed in subgroups, astronauts with positive flattening had a significant increase in CSF production rate postflight (P = 0.01), whereas the negative flattening subgroup did not have a significant increase postflight (P = 0.46) (Fig. 6). The positive flattening subgroup had a significantly lower preflight CSF production rate compared to the negative flattening subgroup (P = 0.02) (Fig. 6). CSF production rate measurements for two observers were highly reproducible (ICC = 0.95, P < 0.00001).
The average CSF maximum systolic velocity for all astronauts was 10.5 cm/s ± 3.0 preflight and increased significantly to 11.5 cm/s ± 3.6 postflight (P = 0.04). However, analyzed in subgroups, astronauts with positive flattening had no significant increase in CSF maximum systolic velocity postflight, whereas the negative flattening subgroup did have a significant increase postflight (P = 0.01) (Fig. 7). CSF maximum systolic velocity measurements for two observers were highly reproducible (ICC = 0.99, P = 1.0).
The area of the mid-cerebral aqueduct was not significantly different in preflight (2.7 mm2 ± 0.8) compared to postflight (2.7 mm2 ± 0.8) (P = 0.71) astronauts. Heart rate was not significantly different in preflight (59.5 bpm ± 5.5) compared to postflight (62 bpm ± 5.8) (P = 0.27) astronauts.
There was only one astronaut with symptoms warranting a lumbar puncture (Table 1). This was performed 12 days postflight and yielded an elevated opening pressure of 28 cm of H20. This astronaut was notable for grade 3 positive flattening (Fig. 8).
Healthy Controls
Seven healthy controls average age 47 years ± 12 had no globe flattening or evidence of optic disc protrusion. Mean CSF production rate and CSF maximum systolic velocity measured 291 μL/min ± 98 and 8.5 cm/sec ± 2.4, respectively. The cross-sectional area of the aqueduct measured 2.6 mm2 ± 0.5 and was not significantly different compared to preflight astronauts (P = 0.97).
There was no significant difference in preflight CSF production rate in astronauts compared to controls (P = 0.83). Similarly, astronaut-based subgroups with either positive or negative flattening had no significant difference in CSF production rate compared to controls (P = 0.11 and P = 0.36, respectively).
Mean CSF maximum systolic velocity in preflight astronauts was not significantly different compared to controls (P = 0.17). Similarly, astronaut-based subgroups with positive or negative flattening had no significant difference in CSF maximum systolic velocity compared to controls (P = 0.14 and P = 0.35, respectively).
Discussion
Increased CSF production rate in postflight astronauts with positive posterior globe flattening is compatible with the hypothesis of microgravity-induced intracranial hypertension. We propose that CSF production rate is significantly downregulated during space flight in this subgroup due to greater susceptibility to microgravity-induced cephalad fluid shifts. A new steady state of intracranial pressure and CSF production rate is then established, but remains above the preflight baseline, resulting in pathologic modification of the orbital structures. Upon return to normal gravity the cephalad fluid shift reverses, causing an abrupt decrease in ICP stimulus with respect to the in-flight level. This sudden drop in ICP triggers a compensatory CSF upregulation to reestablish homeostasis.
Hypothetically, a detectable residual upregulation in CSF production is not observed postflight in astronauts with negative flattening due to a more extensive compensatory reserve capacity maintaining relatively stable ICP levels through the transition of fluid shifts between normal gravity, microgravity, and the return to normal gravity. The larger compensatory reserve capacity hypothesized for this group most likely requires less CSF modulation to maintain ICP in the normal operating range.
CSF is produced in the choroid plexus of the lateral, third, and fourth ventricles controlled by a complex of mechanisms including protein expression and regulatory hormones.22 These endogenous signaling pathways ultimately influence membrane transport of water and solutes.23 One potential target of these pathways is aquaporin-1 (AQP1). AQP1 is a membrane protein that functions as a water channel within the apical membrane of the choroid plexus.24 In reference to this pathway, AQP1 null mice are associated with reduced ICP and CSF production rate.25 In relationship to microgravity, rats flown on space shuttle missions have demonstrated a decreased expression of AQP1, thus indirectly indicating a downregulation in CSF production rate.20 In the same experiment, AQP1 was subsequently overexpressed upon readaptation to normal gravity for 2 days, indicating an aggressive upregulation in CSF production rate consistent with a rebound phenomenon.20 Other rodent studies performed in microgravity have also demonstrated evidence of downregulation of CSF production in the form of histologic and biochemical alteration of the choroid plexus.26, 27 These animal studies are all consistent with an active physiologic response to elevated ICP consisting of downregulation of CSF production. Overexpression of AQ1 in postflight rats suggests a potential mechanism for observed CSF upregulation in postflight astronauts.
The reported CSF production rate measured by direct ventriculolumbar perfusion technique is 370 μL/min ± 100 28 and quantified at the level of the cerebral aqueduct using cine PC-MRI technique with semiautomated segmentation analysis is 305 μL/min ± 146.19 These reported measurements are in excellent agreement with calculated values for our control group and preflight astronauts, suggesting validation of our methodology.
In contrast, preflight mean CSF production rate in astronauts with positive flattening is more than a standard deviation lower than controls and significantly lower compared to the negative flattening subgroup. We hypothesize that a lower baseline CSF production rate in the positive flattening subgroup of astronauts may indicate a higher baseline CSF outflow resistance and/or sagittal sinus pressure as predicted per Davson's equation and resulting in a greater intracranial pressure sensitivity to intracranial volume change. Lower preflight CSF production rate may therefore represent a physiologic parameter useful in predicting which astronauts may be susceptible to microgravity-induced intracranial hypertension. Age was not considered a relevant factor in this baseline variance, as a prior study showed no significant differences in CSF production rate across a wide age range of healthy adults and the age range between astronaut subgroups was relatively narrow.29, 30 Additionally, there are no reported gender differences in CSF production rate 30 and it is therefore not considered contributory to subgroup differences.
CSF maximum systolic velocity measured at the cerebral aqueduct is a marker of the compensatory reserve capacity.14 CSF outflow at the aqueduct is closely coupled with a larger quantity of CSF outflow at the cervical spinal canal that is dependent on craniospinal compliance.31 The spinal canal provides the largest component of passive craniospinal compliance, having inherently more capacity to absorb transient increases in volume compared to the cranial compartment through expansion of the thecal sac, compression of the epidural venous plexus, and the large capacity of the spinal subarachnoid space.31, 32 A study using positive airway pressure to raise ICP demonstrated a decrease in CSF maximum systolic velocity.14 A possible explanation for this effect is compression of the thecal sac through transmission of intrathoracic pressure through the valveless epidural and neuroforaminal veins 14 reducing the capacity of the thecal sac to expand and/or epidural venous plexus to collapse, and thus a reduction in the overall craniospinal compliance. Additionally, a reduced spinal compliance found in patients with idiopathic intracranial hypertension 33 is also thought to be related to compression of the neural foramina by increased abdominal pressure associated with obesity characteristically found in this population.34 Both these studies indicate that spinal canal compression negatively impacts craniospinal compliance, and thus the compensatory reserve capacity.
In contrast, identification of increased CSF maximum systolic velocity postflight could represent evidence of increased craniospinal compliance. We hypothesize that spinal lengthening known to develop in microgravity 35 expands the capacity of the thecal sac and epidural venous plexus, resulting in increased craniospinal compliance. This in addition to a known 10% reduction in total blood volume 36 and predicted reduction in CSF volume due to downregulation would result in an improvement in overall compensatory reserve capacity. This hypothesis is compatible with our observation that astronauts with the least anatomic evidence of intracranial hypertension (negative flattening) had the largest increase in CSF maximum systolic velocity postflight indicative of the greatest expansion of the compensatory reserve capacity. Therefore, larger increases in CSF maximum systolic velocity in astronauts postflight could indicate intrinsic anatomic and physiologic factors that confer less susceptibility to microgravity-induced intracranial hypertension.
Normal pressure hydrocephalus (NPH) is another well-established cause of increased CSF maximum systolic velocity measured at the cerebral aqueduct. NPH is hypothesized to be due to decreased cerebrovascular compliance with increased venous resistance at the level of the superior sagittal sinus, resulting in decreased arterial perfusion.37 The lack of reported evidence of hydrocephalus or diminished cerebral blood flow in postflight astronauts 4, 38 does not, however, support this mechanism in astronauts.
Volumetric flow rate equals flow velocity multiplied by cross-sectional area. Therefore, increased CSF maximum systolic velocity postflight was not attributed to alteration in aqueductal cross-sectional area postflight, since this parameter remained unchanged from preflight measurements.
Cerebral arterial inflow, which is the driving component of CSF outflow,31 was not directly measured in our study. However, a previous study measuring middle cerebral blood flow velocity with transcranial Doppler in astronauts with short-duration microgravity exposure showed no significant change between pre- and postflight values.38 This study suggests that cerebral arterial inflow is not likely a factor in the observed postflight CSF maximum systolic velocity increase in short-duration astronauts.38
The effect of blood pressure and heart rate on CSF maximum systolic velocity has not been fully evaluated. In our study heart rates were within the low normal range and there was no significant change between pre- and postflight measurements, as indicated by the cine PC-MRI data. A study of blood pressure in long-duration astronauts also showed no significant change in the postflight compared to preflight measurements.39 Therefore, any potential effect on CSF maximum systolic velocity relative to heart or blood pressure change in postflight astronauts is not likely to be significant.
The increased incidence and severity of positive flattening with longer mission duration suggests a dose–response relationship of microgravity exposure to the development of intracranial hypertension.
This retrospective study has several limitations. First, all but one astronaut had previous exposure to microgravity, and thus true baseline levels of CSF production rate and maximum systolic velocity could not be established. Second, time to imaging was not constant postflight, therefore CSF production rate was not measured at a similar phase of readaptation leading to bias between subgroups. Third, although medications such as furosemide 40 and acetazolamide 41 can alter CSF production rate, common dietary components containing caffeine 42 can also influence CSF production rate and were not documented or controlled, potentially contributing to bias. Fourth, time of imaging was not controlled, but its effects relative to variability of CSF production rate due to circadium rhythm are not considered significant.43 Fifth, healthy control subjects were not reevaluated at a similar time interval to astronaut studies to test for drift in CSF production rate or CSF maximum systolic velocity. Sixth, quantification of cerebral arterial inflow was not obtained in this study and thus its contribution to any change on CSF maximum systolic velocity postflight could not be determined. Finally, since the duration of CSF hydrodynamic changes due to prior microgravity exposure is not yet established, the net effect of arbitrarily chosen exclusion criteria based on timing of initial reentry to baseline imaging study is unknown. Because CSF production rate and maximum systolic velocity between healthy controls and preflight astronauts were not significantly different suggests this exclusion criteria did not impact the results of this study.
Both invasive and noninvasive techniques to measure ICP that can be applied before, during, and after a mission would be definitive proof of the development of intracranial hypertension in microgravity and are currently under investigation. Preflight followed by postflight serial measurements of CSF maximum systolic velocity, cerebral volumetric blood flow, and spinal CSF volumes could determine if a change in CSF maximum systolic velocity postflight is related to expanded spinal canal volume predicted to increase craniospinal compliance. A larger prospective study of astronauts without prior exposure to microgravity and a consistent shorter reentry to imaging time is needed to confirm results of this study.
Although the etiology of visual change in astronauts remains unresolved, microgravity-induced intracranial hypertension continues to be implicated in the pathogenesis. Increased CSF production rate in postflight astronauts indicates that gravitational forces affect ICP homeostasis in a manner that is consistent with the primary hypothesis of microgravity-induced intracranial hypertension. CSF hydrodynamics using noninvasive MR phase-contrast techniques suggests a novel approach to define risk and measure physiologic responses to microgravity that could be utilized to further develop mitigation strategies for this unique clinical entity. Ultimately, there may be a plethora of anatomic and physiology risk factors that will determine an astronaut's sensitivity to developing visual abnormalities either through modulation of the ICP pathway or in combination with other factors.
Acknowledgments
Contract grant sponsor: National Multiple Sclerosis Society; Contract grant number: RC 1019-A-5; Contract grant sponsor: limited personnel support provided through our local Center for Clinical and Translational Sciences funded by the National Center for Research Resources; Contract grant number: UL1 RR024148; Contract grant sponsor: Dunn Research Foundation (to K.M.H.).