Published ahead of print on August 31, 2006, doi:10.1164/rccm.200606-780OC
© 2006 American Thoracic Society doi: 10.1164/rccm.200606-780OC
Magnetic Resonance Imaging of Uneven Pulmonary Perfusion in Hypoxia in HumansSports Medicine, Internal Medicine VII; Cardiology, Internal Medicine III; Pediatric Radiology, University Hospital, Heidelberg; Departments of Radiology and Medical Physics in Radiology, German Cancer Research Center, Heidelberg; and Department of Clinical Radiology, University Medical Center Grosshadern, Ludwigs-Maximilians-University, Munich, Germany Correspondence and requests for reprints should be addressed to Christoph Dehnert, M.D., University Hospital Heidelberg, Internal Medicine VII, Sports Medicine, Im Neuenheimer Feld 410, D-69120 Heidelberg, Germany. E-mail: christoph.dehnert{at}med.uni-heidelberg.de
Rationale: Inhomogeneous hypoxic pulmonary vasoconstriction causing regional overperfusion and high capillary pressure is postulated for explaining how high pulmonary artery pressure leads to high-altitude pulmonary edema in susceptible (HAPE-S) individuals. Objective: Because different species of animals also show inhomogeneous hypoxic pulmonary vasoconstriction, we hypothesized that inhomogeneity of lung perfusion in general increases in hypoxia, but is more pronounced in HAPE-S. For best temporal and spatial resolution, regional pulmonary perfusion was assessed by dynamic contrast-enhanced magnetic resonance imaging. Methods: Dynamic contrast-enhanced magnetic resonance imaging and echocardiography were performed during normoxia and after 2 h of hypoxia (FiO2=0.12) in 11 HAPE-S individuals and 10 control subjects. As a measure for perfusion inhomogeneity, the coefficient of variation for two perfusion parameters (peak signal intensity, time-to-peak) was determined for the whole lung and isogravitational slices. Results: There were no differences in perfusion inhomogeneity between the groups in normoxia. In hypoxia, analysis of coefficients of variation indicated a greater inhomogeneity in all subjects, which was more pronounced in HAPE-S compared with control subjects. Discrimination between HAPE-S and control subjects was best in gravity-dependent lung areas. Pulmonary artery pressure during hypoxia increased from 22 ± 3 to 53 ± 9 mm Hg in HAPE-S and 24 ± 4 to 33 ± 6 mm Hg in control subjects (mean ± SD; p < 0.001), respectively. Conclusion: This study shows that hypoxic pulmonary vasoconstriction is inhomogeneous in hypoxia in humans, particularly in HAPE-S individuals where it is accompanied by a greater increase in pulmonary artery pressure compared with control subjects. These findings support the hypothesis of exaggerated and uneven hypoxic pulmonary vasoconstriction in HAPE-S individuals.
Key Words: high-altitude pulmonary edema • hypoxic pulmonary vasoconstriction • pulmonary artery pressure
It has been postulated that hypoxic pulmonary vasoconstriction in humans is inhomogeneous in individuals susceptible to high-altitude pulmonary edema (HAPE-S). HAPE is a noncardiogenic pulmonary edema that occurs in otherwise healthy mountaineers after rapid ascent to altitudes above 3,000 to 3,500 m. An abnormal increase of pulmonary artery pressure (PAP) is essential for the development of HAPE; this is proved by the evidence that HAPE-S individuals show exaggerated hypoxic pulmonary vasoconstriction (1–4), and that abnormally high PAP precedes edema formation (5, 6) and lowering PAP prevents HAPE (6, 7) and is effective for treatment (8–10). Because hypoxic pulmonary vasoconstriction appears upstream to the capillary bed in precapillary arterioles, a common explanation for edema formation in a hypoxic environment is the hypothesis of inhomogeneous or uneven hypoxic pulmonary vasoconstriction. This was originally postulated by Visscher (11) to explain edema formation in microembolization of pulmonary blood vessels and later adapted to HAPE by Hultgren and colleagues and Staub (12–15). Although strong hypoxic pulmonary vasoconstriction protects subsequent regions of the lung from high perfusion, other regions with less hypoxic pulmonary vasoconstriction are overperfused. This regional overperfusion leads to an increase of capillary pressure and subsequently to a hydrostatic pulmonary edema in these areas. Several attempts to prove this hypothesis in humans using imaging failed because of insufficient resolution of the methods used (16–18). In a recent study, Hopkins and coworkers (19) were able to demonstrate higher perfusion inhomogeneity during hypoxic conditions for the first time in humans in a small number of subjects with HAPE in their history. But, in this study, no change of perfusion distribution was found in nonsusceptible mountaineers or control subjects without previous high-altitude exposure. These measurements were performed by arterial spin labeling (ASL) magnetic resonance imaging (MRI). An alternative MRI technique for the assessment of regional pulmonary perfusion is dynamic contrast-enhanced (DCE) MRI (20–22). Compared with ASL-MRI, DCE-MRI allows for the assessment of perfusion of the entire lung with a higher spatial resolution and higher signal-to-noise ratio than ASL (22–24). There is direct evidence for inhomogeneous hypoxic pulmonary vasoconstriction in different species of animals (25–27), suggesting that acute hypoxic pulmonary vasoconstriction in general may be inhomogeneous in the mammalian lung. Therefore, the aim of this study was to examine in a larger group of mountaineers with well-documented altitude tolerance whether inhomogeneity of pulmonary perfusion in general increases in hypoxia and whether it is more pronounced in HAPE-S individuals. To obtain optimal spatial resolution and signal-to-noise ratio, DCE-MRI was used. In addition, PAP was measured to confirm susceptibility to HAPE, because increased hypoxic pulmonary vasoconstriction is a hallmark of susceptibility. Some of the results of this study have been previously reported in the form of abstracts (28–30).
The study was approved by the institutional ethics committee. Twenty-four low-altitude natives who had not resided above 2,000 m during the 2 wk before the study commenced were enrolled after they had given informed consent. Their altitude tolerance was well documented from previous studies with rapid ascent (20–24 h) from sea level to 4,559 m. Subjects were classified as susceptible (HAPE-S) if they developed at least two episodes of HAPE, or as controls if they had not developed HAPE despite this rapid ascent. Measurements of echocardiography and MRI were performed in normoxia and after 2 h of normobaric hypoxia (oxygen content [FIO2]), 12%; corresponding to an altitude of 4,500 m), while still in hypoxia. Heart rate and oxygen saturation were continuously monitored by pulse oxymetry; blood pressure was measured every 30 min.
Determination of PAP
MRI
Data Analysis Data analysis was performed by one author who was unaware of the clinical and experimental data. For the assessment of perfusion inhomogeneity, the entire lungs were manually segmented in each partition of the three-dimensional dataset using an in-house–developed software. Due to physiologic temporal differences in maximum contrast enhancement between gravity-dependent and non–gravity-dependent lung regions, time-independent parametric maps were calculated after baseline correction (peak signal intensity [PSI]). To account for physiologic differences in the perfusion kinetics, parametric maps of the time-to-peak (TTP) were also computed (Figures 2 and 3).
Perfusion inhomogeneity was quantified by calculating the coefficient of variation (CV) (32–34) of these two parameters for the whole lung. Because the physiologic variation of pulmonary blood flow between nondependent and dependent lung regions could neutralize effects in the evaluation of the entire lung, three isogravitational slices defined by anatomic "landmarks" (i.e., dorsal edge of the descending aorta, left atrium, and right pulmonary artery) were also analyzed.
Statistics Additional detail and discussion on methods are provided in an online supplement.
Study Population The results of the study are based on data for 21 of the 24 subjects (11 HAPE-S and 10 control subjects). MR perfusion measurements could not be analyzed in three subjects (two female, one male) due to insufficient breath-holding or inconsistent breath-hold levels between the examinations performed during normoxia and hypoxia. According to the inclusion criteria, HAPE-S subjects had experienced two to five episodes of radiographically documented HAPE, whereas, by definition, no HAPE or HAPE-like episode was reported from any of the control subjects. There were no significant differences in anthropometric data between the two groups. Oxygen saturation was significantly reduced by the hypoxic exposure. This decrease was a little more pronounced in the HAPE-S subjects, but the difference was not significant. Resting heart rate in normoxia was the same in both groups and increased by 9 beats/min in both groups during hypoxia. There was no difference in blood pressure in normoxia or during hypoxic exposure. Data are presented in detail in Table 1.
Response of Pulmonary Artery Systolic Pressure Echocardiographic determination of systolic PAP showed similar values for both groups at normoxia. There were no signs of left or right ventricular dysfunction and there was no difference in pulmonary acceleration time. After 2 h of hypoxic exposure, systolic PAP increased from 24 ± 4 to 33 ± 6 mm Hg in control subjects, whereas it increased from 22 ± 3 to 53 ± 9 mm Hg in HAPE-S subjects. The difference between groups was highly significant (p < 0.001). Data are presented in Table 1.
MRI
Evaluation of single slices. For isogravitational lung slices during hypoxia, higher CV of both parameters was observed in HAPE-S subjects. In the slice at the level of the dorsal edge of the descending aorta, a significant increase of the CV of TTP and PSI from normoxia to hypoxia was observed only in HAPE-S subjects. During hypoxic exposure, this increase resulted in a significantly greater inhomogeneity in HAPE-S compared with control subjects for both parameters (Figures 4 and 5). In the medial slice at the level of the dorsal edge of left atrium, we could also demonstrate a significant increase of the CV of both parameters from normoxia to hypoxia in HAPE-S subjects. In this slice, there was also a significant increase of the CV of TTP in the control group. Nevertheless, group differences again showed significantly greater inhomogeneity among HAPE-S subjects during hypoxia for both parameters (Figures 4 and 5). In the ventral slice at the level of the right pulmonary artery, the CV of TTP again showed a significant increase from normoxia to hypoxia only in HAPE-S subjects, with a significantly greater inhomogeneity among the HAPE-S subjects in hypoxia. Here, the increase in the CV of PSI from normoxia to hypoxia was only significant in HAPE-S subjects, but there was no difference during hypoxia between the groups (Figures 4 and 5).
In this study, regional pulmonary perfusion was assessed by a contrast-enhanced MRI technique in HAPE-S subjects and HAPE-resistant control subjects in normoxia as well as after 2 h of exposure to hypoxia. The main finding was a greater dispersion in hypoxia of two perfusion parameters—that is, the peak signal intensity and the time to reach this peak (TTP)—in all subjects, indicating a greater perfusion inhomogeneity. The inhomogeneity was more pronounced in HAPE-S compared with control subjects. This finding is compatible with Hultgren's hypothesis of uneven hypoxic pulmonary vasoconstriction resulting in regional overperfusion. In our study, both groups showed a comparable degree of lung tissue perfusion inhomogeneity during normoxia. Parameters indicating perfusion inhomogeneity increased during hypoxia, particularly in HAPE-S subjects where they increased significantly for almost all evaluations. This finding is in part concordant with a previous study by Hopkins and colleagues (19), in which pulmonary perfusion distribution was assessed by ASL. Hopkins and colleagues showed increased pulmonary blood flow inhomogeneity only in mountaineers with previous episodes of HAPE, whereas pulmonary perfusion distribution was unaltered in the "HAPE-resistant" subjects and their control subjects without altitude experience. In contrast to Hopkins, we also observed an increased perfusion inhomogeneity in the control group. It is unlikely that a misclassification of some subjects led to this result, because HAPE susceptibility of all subjects had been documented in previous studies and, in addition, PAP response to hypoxia confirmed this classification. It is much more likely that acute hypoxic pulmonary vasoconstriction in general is inhomogeneous in the mammalian lung, because there is direct evidence for inhomogeneous hypoxic pulmonary vasoconstriction in dogs (26) and sheep (27). Furthermore, it has been shown in pigs, by direct measurements of pulmonary blood flow distribution with fluorescent microspheres (25), that hypoxic pulmonary vasoconstriction is inhomogeneously distributed over the lung. Together, our findings demonstrate that hypoxic pulmonary vasoconstriction is inhomogeneous in humans as in other mammals. Furthermore, exaggerated hypoxic pulmonary vasoconstriction and enhanced inhomogeneity of pulmonary perfusion can explain an increase of the capillary pressure to a level that is associated with edema formation in animals (35), as measurements by single occlusion technique in HAPE at 4,559 m have shown (5). Because perfusion inhomogeneity is assessed as the dispersion of signal intensity values, it is obvious that a large number of pixels with high signal intensity (i.e., resulting from vessels), as opposed to pixels with lower signal intensity (i.e., resulting from lung parenchyma), may cause a greater dispersion. We therefore excluded all image information resulting from large intrapulmonary vascular structures. Furthermore, the different result compared with Hopkins and coworkers (19), in that we also found an increase in perfusion inhomogeneity in control subjects, may be attributed to the substantially different imaging techniques for the assessment of lung perfusion. In contrast to ASL used by Hopkins and coworkers, which uses the protons of arterial blood as an endogenous tracer for perfusion imaging, DCE-MRI uses rapid dynamic imaging of the tissue of interest during the first pass of an injected contrast agent bolus as a perfusion tracer (20–22). This method allows for the assessment of the temporal dynamics of perfusion, whereas ASL images the amount of perfusion at a predefined time point. To address potential differences in the amount of perfusion as well as temporal dynamics of perfusion, we evaluated the inhomogeneity of two perfusion parameters extracted from the signal intensity time curves (Figure 2). In a direct comparison of both methods, it has been shown that DCE-MRI offers a substantially higher signal of the perfused lung, and is thus much more sensitive for perfusion variations of lung tissue (23). To compensate for this limitation, signal values of ASL perfusion are usually averaged over repetitive measurements, thus potentially changing the level of perfusion inhomogeneity. In addition, artificial signal variation may occur as a result of misregistration among the different images (e.g., caused by different breath-hold levels). Furthermore, DCE-MRI allows one to assess pulmonary perfusion with a higher spatial resolution than ASL (e.g., slice thickness < 5 mm compared with 15 mm) and the assessment of the entire lung, which is not practicable with ASL. On the other hand, the analysis of the whole lung may not necessarily be advantageous, because physiologic gravitational differences of the regional lung perfusion underlie some homogenization during hypoxic exposure (36, 37). A perfusion shift from gravity-dependent to nondependent lung areas may in part counterbalance regional perfusion differences caused by uneven hypoxic pulmonary vasoconstriction. Therefore, we analyzed the perfusion inhomogeneity in three individual isogravitational slices localized in a ventral, middle, and dorsal aspect of the lung. In general, the analysis of the single slices confirmed the results obtained from the whole lung, but in ventral, non–gravity-dependent lung areas, group differences were only observed for TTP. This can be attributed to the methodologic limitation of DCE-MRI in that the signal-to-noise ratio in nondependent lung is poor. In addition, the signal-to-noise ratio in these areas is further diminished by pulsation artifacts from the heart. Thus, measurements could be regarded as less valid. However, in the dorsal and medial slices with a better signal-to-noise-ratio and lower overall CV values, the discrimination between hypoxia and normoxia as well as between HAPE-S and control subjects was substantially better. There are some limitations of our study. DCE-MRI, a rather new method (the first report of this technique was published in 1996 [20]), has not yet been validated in depth. In animal experiments, DCE-MRI correlated well with microsphere measurements of lung perfusion (38). In clinical studies, an excellent agreement of DCE-MRI with conventional perfusion scintigraphy has been demonstrated (21, 24, 39, 40). Perfusion scintigraphy is still considered as the clinical gold standard for the assessment of lung perfusion. However, compared with perfusion MRI, scintigraphy has a significantly inferior spatial resolution (e.g., in-plane spatial resolution: < 4 vs. > 10 mm) (20, 41). Furthermore, the data acquisition is much longer, and can only be achieved while breathing freely. Thus, scintigraphy per se might not be sensitive enough to demonstrate regional perfusion inhomogeneity (36). On the basis of the results of previous studies (21, 24, 39, 40), we consider DCE-MRI to be a promising technique to assess perfusion inhomogeneity in HAPE-S individuals. Several studies have indicated that direct quantitative perfusion values may be obtained from DCE-MRI by applying the indicator dilution theory. Although MRI has delivered comparable perfusion values to reference data from H2O15 positron emission tomography (22, 42–44), valid results can only be expected for small contrast agent doses (see also the online supplement). If the contrast agent dose is reduced, however, valid measurements of perfusion inhomogeneity are questionable due to the limited signal-to-noise ratio in the lungs. This is already indicated by some findings in this study. Although we used a high dose of contrast agent, different results were observed in nondependent lung areas with a lower signal-to-noise ratio. Therefore, similar to previous studies (45–47), we decided to use descriptive parameters from the signal intensity time curves of the DCE-MRI data as surrogates for perfusion. In summary, we observed a greater inhomogeneity of pulmonary perfusion in hypoxia in healthy subjects, which was more pronounced in HAPE-S than in control subjects. As long as PAP is normal or only mildly elevated, this may not be of clinical relevance. But if hypoxic pulmonary vasoconstriction is abnormally increased, and therefore PAP considerably elevated, this can account for regional overperfusion due to an increased shift of blood to nonconstricted areas. Thus, our findings indicate a more uneven hypoxic pulmonary vasoconstriction in HAPE-S subjects, resulting in regional overperfusion during hypoxic exposure, as hypothesized by Hultgren 40 yr ago.
The authors thank Dr. Julia Zaporozhan for her assistance in the evaluation of data.
Supported by a research grant from GE Healthcare Buchler GmbH and Co. KG, Ismaning, Germany, and by a research grant from the German National Research Agency (DFG): "Image-based V/Q analysis" (FOR 474). This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org Originally Published in Press as DOI: 10.1164/rccm.200606-780OC on August 31, 2006
Conflict of Interest Statement: C.D. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. F.R. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. S.L. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. T.A.K. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. R.B. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. M.P. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. E.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. D.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. H.-U.K. received together with C.F. a research grant of Received in original form June 13, 2006; accepted in final form August 31, 2006
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
0.05; in case of repeated testing, a Bonferroni correction was used. 
p < 0.05, 
20,000 for 3 yr from GE Healthcare Buchler GmbH & Co. KG, Ismaning, Germany, and a research grant of