Genetic Admixture and Survival in Diverse Populations with Pulmonary Arterial Hypertension

Abstract

Rationale: Limited information is available on racial/ethnic differences in pulmonary arterial hypertension (PAH).Objectives: Determine effects of race/ethnicity and ancestry on mortality and disease outcomes in diverse patients with PAH.Methods: Patients with Group 1 PAH were included from two national registries with genome-wide data and two local cohorts, and further incorporated in a global meta-analysis. Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated for transplant-free, all-cause mortality in Hispanic patients with non-Hispanic white (NHW) patients as the reference group. Odds ratios (ORs) for inpatient-specific mortality in patients with PAH were also calculated for race/ethnic groups from an additional National Inpatient Sample dataset not included in the meta-analysis.Measurements and Main Results: After covariate adjustment, self-reported Hispanic patients (n = 290) exhibited significantly reduced mortality versus NHW patients (n = 1,970) after global meta-analysis (HR, 0.60 [95% CI, 0.41-0.87]; P = 0.008). Although not significant, increasing Native American genetic ancestry appeared to account for part of the observed mortality benefit (HR, 0.48 [95% CI, 0.23-1.01]; P = 0.053) in the two national registries. Finally, in the National Inpatient Sample, an inpatient mortality benefit was also observed for Hispanic patients (n = 1,524) versus NHW patients (n = 8,829; OR, 0.65 [95% CI, 0.50-0.84]; P = 0.001). An inpatient mortality benefit was observed for Native American patients (n = 185; OR, 0.38 [95% CI, 0.15-0.93]; P = 0.034).Conclusions: This study demonstrates a reproducible survival benefit for Hispanic patients with Group 1 PAH in multiple clinical settings. Our results implicate contributions of genetic ancestry to differential survival in PAH.

Keywords: Hispanic American; Native American; health disparities; pulmonary arterial hypertension; survival.

Figure 1.

Proportions of African, European, and Native American ancestry were generated in the (A) Pulmonary Arterial Hypertension Biobank and the (B) Allegheny Health Network cohort using ancestry informative markers input into STRUCTURE with three subpopulations (k = 3) using 1000 Genomes reference populations. Each observation is positioned within the triangle, reflecting the combined estimate for percentage of African, European, and Native American ancestry. Color for each dot indicates self-reported race/ethnicity. Confidence intervals (50%, 95%, and 99%) for percentage of ancestry by self-reported race/ethnicity were calculated using the Mahalanobis distance in R. (C) Combined Pulmonary Arterial Hypertension Biobank and Allegheny Health Network cohorts are overlaid with magnitude of pulmonary vascular resistance (PVR). In adjusted regressions, increased PVR was associated with Native American ancestry and decreased PVR was associated with African ancestry. AAs = African American patients.

Figure 2.

Kaplan-Meier curves for all-cause mortality by race/ethnicity in the (A) Pulmonary Arterial Hypertension Biobank and the (B) Allegheny Health Network cohort. Survival was calculated from enrollment up to date of death or censoring. In the Pulmonary Arterial Hypertension Biobank, January 1, 2018, was used as the date of censoring. In the Allegheny Health Network cohort, if a specific date for last contact was not recorded, the date December 19, 2014, was used as the censoring date. Logrank tests were performed for differential survival between race/ethnic groups. Kaplan-Meier curves were stopped when 10% of the original number at risk remained. AAs = African American patients; no. = number.

Figure 3.

All-cause mortality hazard ratios for self-reported race/ethnicity in the (A) Pulmonary Arterial Hypertension (PAH) Biobank and in (B) validation meta-analysis of the AHN, UA, and Stanford cohorts. Cox proportional hazards regression for all-cause mortality was performed in idiopathic PAH (PAH Biobank) and patients with Group 1 PAH (AHN, UA, and Stanford cohorts), adjusted for age, sex, prostacyclin use, and log (pulmonary vascular resistance). The Stanford cohort analyses did not adjust for prostacyclin use because these newly diagnosed patients were treatment naive. Patients with Hispanic ethnicity were compared with non-Hispanic white patients. Meta-analysis of Cox proportional hazards regression used a fixed effects model, and inverse SE was used as a weight for each study. AHN = Allegheny Health Network; CI = confidence interval; HR = hazard ratio; PAHB = Pulmonary Arterial Hypertension Biobank; UA = University of Arizona.

Figure 4.

Odds ratios for inpatient mortality in race/ethnic groups versus non-Hispanic white patients. Logistic regression for all-cause inpatient mortality was performed in patients with pulmonary arterial hypertension (adjusted for age, sex, and Elixhauser comorbidity index). Non-Hispanic white patients served as the reference group and were compared with non-Hispanic African American, Hispanic, non-Hispanic Asian, and Native American patients. CI = confidence interval; OR = odds ratio.