Incidence of pheochromocytoma and paraganglioma varies according to altitude: meta-regression analysis
Alexander A. C. Leung, Martin Hyrcza, Janice L. Pasieka, Gregory Kline
Abstract
We thank Drs Patel and Mihai for their interest in our article where we described the incidence of pheochromocytoma and paraganglioma in Alberta, Canada (1). In their letter (2), they raised important points related to the role of hypoxia in tumorigenesis, highlighting the supporting evidence to date. Exposure to high altitude may indeed be a modifiable risk factor for pheochromocytoma and paraganglioma. However, it is admittedly difficult to examine the epidemiology of this association on a large population scale because of the inherent challenges in accurately quantifying the cumulative risk and the intensity of exposure over the period of a person's lifetime (e.g. for those who live at different altitudes throughout their lives) (3). Accordingly, as Drs Patel and Mihai point out, the best available evidence may be derived from comparisons of existing studies that were conducted in populations of different altitudes. Addressing this, we examined the global incidence of pheochromocytoma and paraganglioma across available studies, accounting for differences in altitude and barometric pressure (Table 1). To facilitate comparisons, the annual incidence proportions per 100 000 people along with corresponding 95% CIs were extracted, if possible (or otherwise manually calculated) (4). The altitude of the population was based on the average elevation of the geographic location of the study (https://en-ca.topographic-map.com; accessed March 2, 2021) and the corresponding barometric pressure (PB) in Torr estimated using the Model Atmosphere equation (5): PB = e(6.63268 − 0.1112 × (altitude in km) − 0.00149 × (altitude in km) × (altitude in km)). We then conducted a meta-analysis using restricted maximum likelihood estimates with a random-effects model. Statistical heterogeneity was assessed and quantified using the Cochran Q test and I2 statistic, respectively (6). We explored potential explanations for between-study heterogeneity using meta-regression.