Prediction formula for predicted diffusion capacity of lung for carbon monoxide in pulmonary surgery

Diffusion capacity of the lung for carbon monoxide (D_LCO) is a useful value for perioperative risk assessment of non-small cell lung cancer (NSCLC). The percentage of the predicted D_LCO (%D_LCO: D_LCO/predicted D_LCO × 100) is often evaluated by setting cutoff values as in the clinical field, but several formulae are available for calculating the predicted D_LCO, and the %D_LCO thus varies depending on the formula used to predict D_LCO. We examined differences in %D_LCO calculated using several commonly used prediction formulae.

A total of 490 eligible patients who underwent completed video-assisted thoracoscopic surgery (c-VATS), especially radical pulmonary lobectomy, for NSCLC were analyzed retrospectively. Predicted D_LCO was calculated using the prediction formulae described by Burrows, Nishida, Cotes, and Kanagami, then the relationships with postoperative complications were evaluated.

The %D_LCO from Nishida's formula was two-thirds the value of that from Burrows' (p < 0.05). On logistic regression analysis, predicted postoperative %D_LCO (ppo-D_LCO) based on the formulae of Burrows, Cotes and Kanagami were independent factors related to postoperative pulmonary complications after c-VATS lobectomy for NSCLC (odds ratios 2.46, 1.79 and 2.33, p = 0.005, 0.043 and 0.009, respectively).

The %D_LCO is a useful index for surgical risk assessment of c-VATS lobectomy for NSCLC, while the results differ markedly between individual prediction formulae. Specification of the formula used is necessary in cases considering risk evaluations.