The exercise response in pulmonary hypertension (PH) has characteristic features, including decreased peak oxygen consumption (VO2-peak), increased ventilatory inefficiency (VE/VCO2 slope) and widened alveolar-arterial oxygen-gradient (AaG). We wished to evaluate if the AaG at peak exercise predicted those patients likely to have PH who would subsequently require catheter studies.

Methods We performed a retrospective analysis of patients referred to Hammersmith Hospital between Feb 2008 and Feb 2012 for investigation of Pulmonary Hypertension (PH) who underwent cardiopulmonary exercise testing (CPX) with testing of AaG using arterial blood gas analysis at peak exercise. Patients found to have alternative cardiac or respiratory diagnoses were excluded. Patients given diagnoses of Pulmonary Arterial Hypertension or Pulmonary Hypertension due to Left Heart Disease and with temporally coincident data from CPX and RHC (within 3 months) were included. Patients without cardiorespiratory diagnoses were healthy controls. The VE/VCO2 slope and AaG were compared to the diagnosis of PH and the trans-pulmonary pressure gradient (TPG), (the difference between mean pulmonary artery pressure (mPAP) and pulmonary capillary wedge pressure (PCWP) or left ventricular end diastolic pressure (LVEDP) where available).

Results Using logistic regression to predict a diagnosis of PH, AaG had an odds ratios of 2.98 (p < 0.01) and receiver operating characteristic curve for sensitivity and specificity had area under the curve (ROC-AUC) of 0.92. An AaG cut-off of 2.5kPA had 90% sensitivity and 80% specificity. Similarly, VE/VCO2 had an odds ratio of 1.21 (p < 0.01) and ROC-AUC 0.85 for predicting PH. Combining AaG and VE/VCO2 had ROC-AUC of 0.94 for diagnosing PH without significant interaction between AaG and VE/VCO2. For predicting a TPG >12mmHg, AaG had an odds ratios of 4.54 (p < 0.01) and ROC-AUC of 0.95. VE/VCO2 had an odds ratio of 1.10 (p < 0.01) and ROC-AUC 0.74 for predicting TPG>12mmHg.

Conclusion CPX has become part of the diagnostic workup of patients with PH. AaG measured at peak exercise has a high sensitivity and specificity in predicting patients with PH, which may help determining which patients will require invasive catheter studies. The AaG provides independent information than VE/VCO2 alone in predicting PH and may be useful in the investigation of PH.