Objective: To improve preoperative risk stratification in lung cancer lobectomy by identifying and comparing optimal thresholds for peak oxygen uptake (VO_2peak) presented as weight-indexed and percent of predicted values, respectively.

Methods: This was a longitudinal cohort study including national registry data on patients scheduled for cancer lobectomy that used available data from preoperative cardiopulmonary exercise testing. The measured VO_2peak was indexed by body mass (mL/kg/min) and also compared with 2 established reference equations (Wasserman-Hansen and Study of Health in Pomerania, respectively). By receiver operating characteristic analysis, a lower 90% specificity and an upper 90% sensitivity threshold were determined for each measure, in relation to the outcome of any major complication or death. For each measure and based on these thresholds, patients were categorized as low risk, intermediate risk, or high risk. The frequency of complications was compared between groups using χ².

Results: The frequency of complications differed significantly between the proposed low-, intermediate-, and high-risk groups when using % predicted Study of Health in Pomerania (5%, 21%, 35%, P = .007) or % predicted Wasserman-Hansen (5%, 25%, 35%, P = .002) but not when using the weight-indexed VO_2peak groups (7%, 23%, 15%, P = .08). Nonsignificant differences were found using the threshold <15 mL/kg/min (P = .34).

Conclusions: This study showed that weight-indexed VO_2peak was of less use as a marker of risk at the lower range of exercise capacity, whereas % predicted VO_2peak was associated with a continuously increasing risk of major complications, also at the lower end of exercise capacity. As identifying subjects at high risk of complications is important, % predicted VO_2peak is therefore preferable.