Background Low blood oxygen concentrations (hypoxaemia) usually result from alveolar hypoxia which also stimulates hypoxic pulmonary vasoconstriction, thereby increasing right ventricular work. In contrast, anatomic right-to-left shunts such as pulmonary arteriovenous malformations (PAVMs), cause hypoxaemia without alveolar hypoxia. Our aim was to test whether isolated arterial hypoxaemia is associated with dyspnoea or impaired exercise capacity.

Methods Two prospective studies were performed in patients with radiologically-proven PAVMs. Exercise tolerance was graded by a modified MRC dyspnoea scale. Other patient variables were quantified at rest and on exercise.

Results In 165 consecutive patients, aged 17–87 (median 49)ys, resting SaO₂ varied widely (78.5–99%, median 95%). Five patients were athletes, despite severe resting hypoxaemia (SaO₂<85%). SaO₂ displayed no clear relationship with dyspnoea grade. Higher grade dyspnoea was significantly more common in patients with coexisting cardiorespiratory disease who were more likely to report symptomatic improvement post embolisation. During cardiopulmonary exercise testing, 21 PAVM patients (SaO₂ 80–96%) were no more dyspnoeic than 12 age-matched volunteers (SaO₂ 96–99%). Within the 21 PAVM patients, the majority achieved their predicted workload and peak VO₂, and there was no difference in maximum workload, or peak VO₂ according to the severity of hypoxemia. PAVM patients demonstrated similar relationships betweenpeak heart rate and peak VO₂ as controls. For five patients retested post embolisation, when SaO₂ rose from 88–94% to 94–96% (p = 0.009), there was no difference in perceived dyspnoea, maximum workload (medians 119/113W) or peak VO₂ (medians 1.69/1.72Lmin^-1). Treated patients reset to virtually identical peak oxygen pulse following embolisation. Overall, workload and peak VO₂ were associated not with oxygenation parameters, but with body mass index, anaerobic threshold and ventilatory efficiency: Expired end-tidal PCO₂ at rest and on peak exercise was lower in PAVM patients compared to controls, and in the more hypoxemic patients. During exercise, PAVM patients increased minute ventilation (V’E) more than controls for a given increase in CO₂ output (V’CO₂), and post embolisation, end-tidal PCO₂ and V’E/V’CO₂ slopes normalised.

Conclusions Despite severe hypoxaemia, normal oxygen delivery can be maintained during peak exercise by harnessing integrated adaptive responses that maintain oxygen delivery and uptake with each heart beat (the “O₂ pulse”).