Blood carbon dioxide tension and risk in pulmonary arterial hypertension.

ABSTRACT

BACKGROUND: Low partial pressure of blood carbon dioxide (PaCO2) is common in patients with pulmonary arterial hypertension (PAH) and may inform on clinical outcomes. We investigated whether PaCO2 measurements could provide prognostic information in addition to standard risk assessment in this group of patients. METHODS: We conducted a retrospective observational cohort study on patients with newly diagnosed idiopathic, heritable or drug/toxin-induced PAH recruited from two European centres. Arterialised capillary blood gas analyses at diagnosis and follow-up were incorporated into standard risk assessment strategies and related to outcomes, defined as lung transplant or death. C statistics from receiver-operated characteristics and Cox regression models were used to assess the predictive value of models with and without PaCO2 measurements. Unsupervised clustering was applied to assess the relation of PaCO2 to haemodynamic and pulmonary function variables. RESULTS: Low PaCO2 measured at diagnosis and follow-up was significantly associated with inferior outcomes in 204 patients with PAH. PaCO2 provided prognostic information independent of established non-invasive variables. Integrating PaCO2 in risk strata improved C statistics of non-invasive and mixed invasive/non-invasive models, and revealed more accurate outcome estimates in regression models. Pairwise correlation and unsupervised cluster analyses supported a link between PaCO2 and haemodynamic variables, particularly with cardiac output, in PAH. CONCLUSIONS: Measuring PaCO2 at diagnosis and during follow-up in patients with PAH provided independent prognostic information and has the potential to improve current risk assessment strategies.