Risk factors underlying high-altitude pulmonary edema in the Ecuadorian Andes

Ascent to high altitude (> 2500 m) exposes people to hypobaric atmospheric pressure and blood hypoxemia. It provokes a syndrome whose symptoms vary from the mild acute mountain sickness (AMS) to the life-threatening, high-altitude pulmonary edema (HAPE). This study analyzed the risk for developing high-altitude sickness in a group of HAPE patients (n = 59), which was contrasted against a group of AMS patients (n = 240) as the NO HAPE group, after sojourning above 4,000 m height. The objective of this retrospective was to analyse the factors contributing to the HAPE prevalence among travellers and dwellers of the Ecuadorian Andes. Methods


Introduction
The term high-altitude illness (HAI), or simply mountain sickness, is commonly used to describe a series of cerebral and pulmonary syndromes that develop shortly after rapid ascent above 2,500 m (HA), where the atmospheric and inspired partial O 2 pressures drop down to 75% [1]. The signi cant reduction of the partial O 2 atmospheric pressure leads to its poor diffusion from the alveolar air into the blood, while the percentage of O 2 carried out by hemoglobin drops down below 90% (hypoxemia). Under such hypobaric atmospheric conditions, the human body reacts by a series of mechanisms, known as acclimatization, to compensate the desaturation of blood O 2 . The acclimatization process takes from hours to days, and even weeks [2]. A failure in acclimatization causes HAIs [3,4], whose symptoms and severity vary across individuals from the benign, self-limiting acute mountain sickness or AMS, to more serious clinical manifestations like the high-altitude pulmonary edema or HAPE [1] and eventually the high-altitude cerebral edema or HACE [3].
We investigated risk factors for HAPE in an elevation of 4,000 m above sea level in the Ecuadorian Andes, where atmospheric pressure decreases to levels of 60%. Two groups of patients suffering from either HAPE or AMS (the NO HAPE group) after sojourning at 3,600-4,400 m height, were retrospectively compared using statistical association analyses of the demographic, environmental, health, and blood tests variables. A more advanced statistical analysis allowed us to explore the relative in uence of the risk factors in HAPE susceptibility. Finally, a model for the most representative variables was constructed to t their contribution to HAPE risk.

Study zone
This was a single-centre observational study conducted in the Claudio Benati Hospital in Zumbahua (Ecuador), a remote Andean town sited at 3,600 m height above sea level. Low-altitude and high-altitude residents frequently commute between the coast and the Andes ridge by car in this zone, through the only road connecting the Paci c coast with the Andes, which overcomes a 4,000 m slope in a 2-h ride. Given that the closest hospital was 64 Km away, chances were high that, among the visitors sojourning at very high altitude (100,000 per year), and local dewellers (approx. 20,000), those a icted by HAI had to rush to the emergency room of the Claudio Benatí hospital.

Ethics statement
Written informed consent was waived due to the retrospective nature of the study. We followed STROBE guidelines to report this study, which was approved by the Ethics Committee of the Claudio Benatí Hospital. The authors declare they had no access to identifying patient information when analysing the data.

Diagnosis
AMS and HAPE clinical evaluations were recorded by quali ed medical staff serving people permanently living at high altitude. AMS patients showed unequivocal symptoms of HAI (headache, lassitude, fatigue, dizziness, nausea, and vomiting) associated with a rapid ascent above 4,000 m height [25]. None of the AMS patients had conclusive symptoms of HACE. As to HAPE, the diagnosis relied on the examination of the cardinal symptoms including dyspnea at rest, tachypnea, crackles, wheezing; dry cough, weakness, chest tightness, and tachycardia [5], as well as on chest X-ray exams showing central interstitial edema with peribronchial cu ng, ill-de ned vessels, and a patchy pattern of airspace consolidation. In this respect, patchy opacities produced by HAPE disappeared within days after receiving treatment [26].

Data collection and categorization
The database was extracted from patient les dated from the years 2007 to 2018 and kept in logbooks, conforming to a non-probabilistic sampling. Consistency between physical and digital records was meticulously checked to avoid transcription errors. The terms of the study were de ned as follows: 2.4.1 Demographic variables: ethnicity was categorized as indigenes or mestizo; sex as male or female; and age classi ed in children (0-14 years), young (15-25 years), adult (26-64 years), and elder (≥ 65 years) according to the Statistics from Canada and the American Cancer Society [27]. The terms indigene and mestizo de ned the Panzaleo pedigree and the mixed-race (like most of the Ecuadorians) respectively. 2.4.2 Environmental variables: residence at either high altitude (from 2,700 m to 4,400 m above sea level) or at low altitude (< 100 m above sea level); a recent stay at low altitude, categorized as either sea leveltrip or no sea level-trip antecedent within the last 7 days. The term sea level-trip was coined to identify reentry-HAPE cases.

Statistics
Analyses were computed using the SPSS software (Version 22) for Windows. Multiple a priori Student ttests were applied to compare variables between the HAPE and NO HAPE (AMS) groups. After categorization of the risk factors, ordinal and nominal variables of HAPE and NO HAPE (AMS) groups were compared by contingency tables and their associations computed by the Cramer´s V analysis [31]. Odds ratios (ORs) were calculated using a simple logistic regression analysis [32]. The multiple logistic regression analysis [33] was run to estimate the likelihood of HAPE on a multiple variable basis. Alpha value was set at 0.05.

Differences between HAPE and AMS groups
Multiple a priori t-tests were run to check if the split of the sample into HAPE and NO HAPE (AMS) groups was appropriate. Thus, 10 out of 12 clinical endpoints were statistically different between groups (Table  1). Breathing rate and heart rhythm in the HAPE group were higher compared to the AMS group (t (256) = 7.78, p < 0.001 and (t (295) = 3.12, p < 0.01 respectively), which provoked a profound hypoxemia. The blood O 2 saturation in the HAPE group dropped to 60%, far below the AMS group. The intolerance to high altitude made the hematocrit (49.8%) and hemoglobin (16.5 g/dL) levels in the HAPE group be higher compared to the AMS group (40.47% and 13.2 g/dL respectively). The difference in the hemoglobin levels (3.3 g/dL) was not due the unequal sex distribution, since women have mean levels of approximately 12% (1.9 g/dL) lower than men. Finally, leukocytosis (white blood cell counts > 11•10 9 /L) was a distinctive feature of HAPE. An information bias could de nitely be ruled out.

Analysis of HAPE risk factors
HAPE prevalence and associated descriptive (risk) variables were computed by the Cramer's V test. Demographic factors like ethnicity, sex, and age, environmental factors like the altitude of residence, and antecedents of a recent stay at sea level (Table 2), as well as health status variables like systemic blood pressures, heart rhythm, and blood O 2 saturation (Table 3) were signi cantly associated with HAPE susceptibility. Finally, HAPE risk was signi cantly associated with elevated white blood cell numbers and hemoglobin (Table 3).
Adjusted ORs for HAPE prevalence compared to AMS prevalence were obtained after a simple logistic regression. High prevalence of HAPE was characteristic of indigenes. In contrast, the HAPE risk was lower in females and signi cantly varied with age. The HAPE prevalence was signi cantly lower in individuals living at low altitude compared to high-altitude residents (0.014 times). Indeed, only 11 out of 59 HAPE patients permanently lived at low altitude. In indigenes permanently living at high altitude, a recent sea level-trip increased the odds of suffering from reentry HAPE (OR = 0.049) so that two out of three of these patients suffered from reentry HAPE. With 95 % of the individuals living at low altitude, the mestizo group had 76.2 more odds of having AMS (p < 0.001). As to vital signs, an elevation of systemic blood pressure may decrease HAPE risk. Interestingly, high hemoglobin increased the odds of having HAPE. Finally, leukocytosis was present in 33% of HAPE patients, but none of the AMS patients.
The objective of the multiple logistic regression analysis was to nd an equation that best predicted the occurrence of HAPE compared to AMS as a function of its independent variables. Such an equation may provide information about their attributable fraction or weight. The analysis was only conducted on those signi cant variables that had an annotated value in at least 80% of the cases in the Cramer's V test. Demographic and environmental variables were computed separately given the complex collinearity structure that sometimes disguised the statistical relevance of these variables. The environmental variables were statistically signi cant for the t model ( Table 4). The likelihood of having HAPE compared to NO HAPE (AMS) was reckoned by the following equation: Predicted logit of HAPE = 2.642 -(5.665) x Residence altitude -(2.15) x Sea level trip. The residence at low altitude and no sea level-trip antecedents were inversely related to the likelihood of having HAPE. Their relative in uence on lowering HAPE risk was as follows: low altitude residence > no sea level-trip antecedents. Table 4 shows the multiple logistic regression analysis for demographic variables. The likelihood of having HAPE was estimated by the following equation: Predicted logit HAPE = -0.942 + (3.094) x Ethnicity -(1.099) x Sex -(1.303) x Age (1 -14) -(1.321) x Age (15)(16)(17)(18)(19)(20)(21)(22)(23)(24) -(1.535) x Age  . Nevertheless, given the in uence of the other demographic variables, the multiple logistic regression analyses re ected the dependence of HAPE prevalence on age (Table 5). Female sex as well as age were negatively related to the likelihood of having HAPE, whereas the indigenous ethnicity was positively related. The relative in uences on HAPE prevalence were as follows: indigenous ethnicity > age (1−14) ≃ age (15−24) > age (25−65) > female.

Discussion
This analysis depicted risk variables that may explain HAPE and specially the reentry HAPE, which has almost exclusively been documented in the South American Andean ridge. In the HAPE group, indigenes living at very high altitude outnumbered low-altitude residents. Mestizos residing at low altitude were largely stricken with AMS at high altitude, where indigenes from the highland suffer from reentry HAPE when returned to their homes after a brief (days) sojourn at lower altitude. From a logistic analysis, some factors may either increase (high hemoglobin) or reduce (female sex and high blood pressure) the risk for developing HAPE. Although the Andean-origin indigenes are supposedly well adapted to living at high altitude, latent individual factors for developing HAPE remain that may shed some light on its etiopathogenesis. Some hypotheses should be put forward to explain the in uence of the above risk factors in HAPE. The nding that high-altitude indigenous residents had a higher risk of HAPE compared to those living in low altitudes was interesting because it does nothing but con rm that the Andeans present a worse tolerance to high altitude than the Ethiopian and Himalayan populations [19,20]. Dwellers of this region of the Ecuadorian Andes were Panzaleo indigenes, an ethnicity of the Kichwa pedigree originally from the Amazonia, who presented a high prevalence of high altitude-induced erythrocytosis. The high prevalence of HAPE was largely due to a previous sojourn at sea level, which increased the odds of having HAPE, a phenomenon called reentry HAPE [34]. To this respect, our results are in agreement with the fact that reentry HAPE is more commonly seen in the South Americans because of their poorer adaptation to high altitude [24,35]. A rapid ascent (in this case in a single day) is known to be an important risk factor in this disease [5,34]. In addition, the reentry HAPE may also be due to change in blood volume because prolonged exposure to high altitude results in CMS [36]. In this vein, hematocrit and hemoglobin levels in native Andeans increase to physiologically adapt to high-altitude-induced hypoxemia [20,22,37,38]. A chronic hypoxia of high altitude causes increased muscularization of the pulmonary arterioles, thus generating excessive pulmonary arteriole pressure on the reascent to high altitude [39]. Enhancement of blood pressure (Table 4) may compensate pulmonary vasoconstriction, while a blunted reactivity of the blood pressure to high altitude may lead to HAPE [40,41].
Although the role of sex in the susceptibility to HAPE has been controversial [5], our results are in agreement with the fact that women were more tolerant to HAPE than men [42,43]. The reason is that, under the in uence of progesterone, women may manage alveolar ventilation and oxygenation better [44] and have less erythropoietin and hemoglobin than men [45,46]. Multiple linear regression analyses con rmed that age has an impact in HAPE [24,47,48]. Vulnerability to HAPE is likely to be high in childhood, and it then decreases with aging as a result of age-dependent decline in hemoglobin levels [49]. If that was the case, the role of hemoglobin should be investigated [50].

Conclusion
In summary, although this study was a retrospective and single-center observational analysis. it posited the hypothesis that susceptible Andean communities like the Panzaleo indigenes may dramatically underestimate the adverse effects of shear forces in pulmonary artery pressure induced by the return to high altitude, which along with their probability of having a high hematocrit would increase the rate of intravascular hemolysis leading to HAPE. What causes a high haemoglobin mass in this speci c Andean community may be an interesting avenue to elucidate the etiopathogenesis of HAPE.