Predicting postrecurrence survival among completely resected nonsmall-cell lung cancer patients

Background. Survival after recurrence subsequent to complete resection of nonsmall-cell lung cancer (NSCLC) has been considered a multifactorial process dependent on demographic, clinical, biological, and treatment characteristics. This study sought to quantify the prognostic effects of these characteristics on postrecurrence survival. Methods. Three hundred ninety NSCLC patients who underwent complete resection and subsequently had recurrent cancer were studied. The associations between characteristics of both the initial and recurrent disease with postrecurrence survival were evaluated by Cox proportional hazards models. A multivariable Cox model determined those factors most strongly associated with postrecurrence survival. A simple algorithm based on this model facilitates estimating risk of postrecurrence mortality, as quantified by risk score points. Results. The factors most strongly associated with postrecurrence survival were performance status at recurrence (3 or 4, 4.2 points; 2, 2.8 points; and 1, 1.5 points), symptoms at recurrence (3.6 points), liver recurrence (2.3 points), initial lung cancer stage IIB or worse (1.8 points), and multiple recurrences (1.0 points). Based on these factors, patients were stratified as low risk (4.0 or fewer total points), moderate-low risk (4.1 to 6.1 points), moderate-high risk (6.1 to 8.0 points), and high risk (more than 8.0 points), with 12-month survival of 75%, 51%, 25%, and 9%, respectively. Postrecurrence survival was significantly different across groups (p < 0.01). Conclusions. The proposed prediction instrument offers clinicians a succinct tool for rapidly evaluating mortality risk after recurrence. The characteristics comprising this instrument can be easily ascertained and measured, making it of potential clinical value.