The relationship between hepatopulmonary syndrome and altitude

1Department of Psychology, Colorado State University; 2Division of Gastroenterology and Hepatology, Baylor Health Care System; 3Altitude Research Center, University of Colorado Denver; 4Colorado School of Public Health; 5Department of Geography and Environmental Sciences, University of Colorado Denver; 6Department of Emergency Medicine; University of Colorado School of Medicine, Denver, Colorado, USA Correspondence: Ms Morgan A Valley, Colorado State University, 1876 Campus Delivery, Fort Collins, Colorado 80523, USA. Telephone 303-242-4808, e-mail morgan.valley@colostate.edu Hepatopulmonary syndrome (HPS) is characterized by an oxygenation defect induced by pulmonary vascular dilation in patients with end-stage liver disease (1). Up to 20% of liver transplant candidates have HPS (2); once diagnosed, survival is significantly shorter compared with other liver transplant candidates (3). While several therapies, including pentoxyfylline, methylene blue and garlic have been evaluated, the only proven effective treatment for HPS is liver transplantation (4-7). The hallmark of HPS is a defect in arterial oxygenation. In most patients with HPS, the primary disturbance is intrapulmonary vascular dilation, which primarily leads to marked ventilation-perfusion mismatch, with some contribution from right-to-left shunting and subsequent hypoxemia (1). In animal models of cirrhosis and HPS, there is increased hepatic production of endothelin, which in turn causes increased production of nitric oxide by endothelial cells in the pulmonary vasculature. This leads to relaxation of pulmonary vascular smooth muscle cells and pulmonary vasodilation (8). Animals with experimentally induced HPS also demonstrate increased levels of vascular endothelial-derived growth factor (9). Levels of another pulmonary vasodilator, carboxyhemoglobin, are also elevated in animal models and in individuals with HPS (10). Finally, oxygen in the associated alveoli cannot diffuse into the centre of the dilated alveolar capillary, leading to diffusion-perfusion impairment and impaired oxygenation. Recently, it has been suggested that living at higher altitudes may protect against the development of HPS. In fact, in a small retrospective study, we found that, among end-stage liver disease patients awaiting transplantation, HPS was less prevalent in those treated at original arTicle

The present study sought to further explore the relationship between HPS and altitude by using geographical information systems (GIS).We collected information regarding duration and severity of liver disease, the presence of HPS and the altitude estimate of zip code residence for all patients with end-stage liver disease who were listed on the national Organ Procurement and Transplantation Network (OPTN) liver transplant wait list database.We hypothesized that higher altitude would be associated with a lower prevalence of HPS among American orthotopic liver transplant candidates.

METHODS
The present study was a retrospective analysis of the characteristics of 65,264 patients on the OPTN liver transplant wait list from 1988 to 2006.The diagnosis of HPS was determined by the treating physician and transplant centre.
Altitude, the primary predictor variable, was determined for each patient by assigning the mean altitude of the zip code of residence at the time of entry onto the OPTN wait list.Mean zip code elevation was calculated using the National Elevation Dataset produced by the United States Geological Survey, which provides elevation measures for 30 m × 30 m geographical units across the entire United States (http://ned.usgs.gov/).National Elevation Dataset data were combined with zip code boundaries in GIS software, ESRI ArcGIS version 9.2, and calculated mean elevation for each zip code by averaging all of the 30 m × 30 m units within that zip code.Altitude was successfully matched to 60,262 (92.3%) patients on the wait list; these patients comprised the study group.
Additional covariates obtained from the OPTN database included age, sex and the raw Model for End-stage Liver Disease (MELD) score, an accepted measure of liver disease severity.MELD includes international normalized ratio, bilirubin and creatinine levels, and whether the patient underwent dialysis at least twice in the past week.While the MELD model was not adopted in the United States until February 2002, raw MELD scores were calculated for all participants based on their test scores.
Data analysis proceeded in three steps.First, demographic and clinical characteristics of the participants were summarized using means and SDs for measurement variables and proportions and 95% CIs for categorical data.Second, t tests and χ 2 statistics were used to test for bivariate associations between the primary outcome (ie, HPS) and participant characteristics, including age, sex and MELD score.A binary logistic regression was conducted to test for an association between altitude (in metres) of residence and the prevalance of HPS.Third, multiple logistic regression was performed to test for an independent, statistically significant association between HPS and altitude (as a continuous variable), after adjusting for age, sex and MELD score.
Patients with more severe liver disease were suspected to live preferentially at lower altitudes (a 'migration effect').An association between altitude of residence and MELD score was tested for by calculating Pearson's correlation coefficient.
The present study was determined to be exempt from review by the Colorado Multiple Institutional Review Board.

RESULTS
Of the 60,524 patients, 262 (0.43% [95% CI 0.39% to 0.49%]) had a reported diagnosis of HPS.The majority (60.5%) of the participants were male.The mean (± SD) age of participants was 44.59±17.94years (range <1 year to 84 years).Participants had a mean MELD score of 16.7±8.44(range 6 to 77) and mean residence zip code altitude of 241.39±352.76m (range 0 m to 3096 m).The mean MELD score of participants with HPS was 2.1 points lower than those without HPS (t=5.86;P<0.001).There was no correlation between MELD scores and altitude (Pearson's r=−0.005;P=0.27), suggesting that the distribution of severe illness was similar at lower and higher altitudes.
Table 1 summarizes the prevalence rates of HPS at each successively higher altitude.The unadjusted logistic regression analysis showed that HPS was significantly less common at higher resident altitudes (P=0.015).No patient living above 2000 m had a diagnosis of HPS (0.0% [95% CI 0% to 0.91%]).After adjusting for age, sex and MELD score, there was a 46% decrease in the odds of reporting HPS with every increase of 1000 m of resident elevation (OR 0.54 [95% CI 0.33 to 0.89]).

DISCUSSION
The present study revealed a strong, negative association between HPS and altitude.While the reasons for this observation are unknown, several possibilities exist.Although the pulmonary vascular dynamics are complex in cirrhotic patients, one plausible explanation is that the lower ambient oxygen concentration found at higher elevation leads to pulmonary vasoconstriction, which mitigates the primary physiological lesion of HPS (namely, pulmonary vasodilation).Evidence of increased pulmonary vasoconstriction exists among residents at moderate and high altitudes (12).In addition to residents living at altitude, travellers to altitude may also develop pulmonary hypertension due to pulmonary vasoconstriction.This is believed to occur when the alveolar partial pressure of oxygen is <70 mmHg, which corresponds to an altitude of approximately 2000 m, although lower altitudes can precipitate pulmonary vasoconstriction when anomalous pulmonary vessels are present (13).
An alternative explanation is a 'migration effect'; that is, patients with symptoms or signs of HPS may relocate from higher to lower altitudes to lessen the effects of hypoxia (14).However, we found that the distribution of severe illness in our study, as measured by MELD score, was similar at higher and lower elevations.Nevertheless, these data are cross-sectional, and cross-sectional data cannot establish a temporal sequence or causal relationship between altitude and prevention of HPS.
The prevalence of HPS in the present study (0.43%) was lower than expected and what has been reported in other publications (2).This may be due to patients on the wait list who have subclinical HPS.The reasons for this is that there is no therapy for this condition and the severity of hypoxemia in many affected patients is minimal.We are unsure as to how this may have influenced our results.
The present study had several limitations.Most importantly, the OPTN dataset does not require treating physicians to adhere to strict diagnostic criteria before assigning a diagnosis of HPS; consensus regarding the definition of HPS was not achieved until 2004 (15).As a result, the definitions of HPS in the study cohort may have varied across transplant centres over time.In addition, we did not have access to physiological variables to ensure the diagnosis of HPS in our patients.However, each of the patients in the present study underwent a comprehensive evaluation at a liver transplant centre by physicians who were experienced with the diagnosis and treatment of HPS.Additionally, we did not have access to other important clinical covariates (for example, smoking status, alcohol consumption) that could influence the severity of liver disease and the relationship between altitude and HPS.Some additional limitations are inherent in the GIS methods used to assign altitudes of residence.For example, we used the mean zip code altitude at time of entry into the liver transplant wait list for each patient; however, we were unable to determine patients' length of exposure at each altitude.In addition, the mean altitude of a zip code may not represent the true altitude of residence for every patient.