Mortality in the randomized, controlled lung intergroup trial of isotretinoin

Abstract

In 2001, we reported that mortality may have been higher with isotretinoin (30 mg/d for 3 years) than with placebo in the subgroup of current smokers among the 1,166 patients with definitively resected early-stage non-small cell lung cancer who participated in the randomized, controlled Lung Intergroup Trial. We report the overall and cause (cancer, cardiovascular disease, or other)-specific mortality associated with long-term isotretinoin after an extended median follow-up of 6.2 years that included the capture of cause-of-death data from 428 deceased patients. Overall mortality was 36.7% in each of the two trial arms, about two thirds related to cancer and one third to other or unknown causes. Overall and cancer deaths increased in current smokers in the isotretinoin arm during the treatment and the extended follow-up period. No mortality end point increased among never smokers and former smokers taking isotretinoin, and cancer deaths decreased marginally in this combined subgroup. Isotretinoin also increased deaths from cardiovascular disease in current smokers. The present analysis supports the safety of protracted isotretinoin use in the combined group of never smokers and former smokers, which has important public health implications, for example, for treating acne in young people. The increased mortality in current smokers in this study is further evidence of the multifaceted danger of active smoking. The overall indications of this study have public health implications for treating acne in young people and other uses of retinoids in smokers.

Fig. 1

Cumulative incidence curves of cancer death (solid lines), cardiovascular death (dashed lines), or other death (dotted lines) for the isotretinoin group (red lines) and placebo group (black lines) for all patients (panel A), never smokers (panel B), former smokers (panel C), and current smokers (panel D). The x-axes show the number of years after randomization. The numbers of deaths, or events (E), from cancer, cardiovascular and other causes and the number of patients (N) that went into constructing each cause-of-death curve are shown. In panel A for example, isotretinoin has 149 cancer, 24 cardiovascular, and 43 non-cancer deaths with an N of 589, and placebo has 146 cancer, 23 cardiovascular, and 43 non-cancer deaths with an N of 577. Numbers of events (E) for each cause of death and total number of patients (N) are also shown in panels B, C, and D.

Fig. 2

Hazard ratios (HRs) of isotretinoin (versus placebo) over time for cancer death and cardiovascular-disease (CVD) death stratified by never smokers plus former smokers or current smokers. Results are shown for all patients. The hazard rates were estimated by the non-parametric Nelson and Aalen method and triquadratic kernel smoother with boundary corrections at different time points (14). HRs were plotted on the logarithmic scale to provide a symmetrical comparison of the potential protective effect (HR < 1) or harmful effect (HR > 1). Pointwise 95% confidence intervals (dashed blue lines) were calculated using the bootstrap method for each smoothed HR curve.