Long-term findings from COMFORT-II, a phase 3 study of ruxolitinib vs best available therapy for myelofibrosis.

Ruxolitinib is a Janus kinase (JAK) (JAK1/JAK2) inhibitor that has demonstrated superiority over placebo and best available therapy (BAT) in the Controlled Myelofibrosis Study with Oral JAK Inhibitor Treatment (COMFORT) studies. COMFORT-II was a randomized (2:1), open-label phase 3 study in patients with myelofibrosis; patients randomized to BAT could crossover to ruxolitinib upon protocol-defined disease progression or after the primary end point, confounding long-term comparisons. At week 48, 28% (41/146) of patients randomized to ruxolitinib achieved ⩾35% decrease in spleen volume (primary end point) compared with no patients on BAT (P<0.001). Among the 78 patients (53.4%) in the ruxolitinib arm who achieved ⩾35% reductions in spleen volume at any time, the probability of maintaining response was 0.48 (95% confidence interval (CI), 0.35-0.60) at 5 years (median, 3.2 years). Median overall survival was not reached in the ruxolitinib arm and was 4.1 years in the BAT arm. There was a 33% reduction in risk of death with ruxolitinib compared with BAT by intent-to-treat analysis (hazard ratio (HR)=0.67; 95% CI, 0.44-1.02; P=0.06); the crossover-corrected HR was 0.44 (95% CI, 0.18-1.04; P=0.06). There was no unexpected increased incidence of adverse events with longer exposure. This final analysis showed that spleen volume reductions with ruxolitinib were maintained with continued therapy and may be associated with survival benefits.

Double-blind evaluation and benchmarking of survival models in a multi-centre study.

Accurate modelling of time-to-event data is of particular importance for both exploratory and predictive analysis in cancer, and can have a direct impact on clinical care. This study presents a detailed double-blind evaluation of the accuracy in out-of-sample prediction of mortality from two generic non-linear models, using artificial neural networks benchmarked against a partial logistic spline, log-normal and COX regression models. A data set containing 2880 samples was shared over the Internet using a purpose-built secure environment called GEOCONDA (www.geoconda.com). The evaluation was carried out in three parts. The first was a comparison between the predicted survival estimates for each of the four survival groups defined by the TNM staging system, against the empirical estimates derived by the Kaplan-Meier method. The second approach focused on the accurate prediction of survival over time, quantified with the time dependent C index (C(td)). Finally, calibration plots were obtained over the range of follow-up and tested using a generalization of the Hosmer-Lemeshow test. All models showed satisfactory performance, with values of C(td) of about 0.7. None of the models showed a systematic tendency towards over/under estimation of the observed survival at tau=3 and 5 years. At tau=10 years, all models underestimated the observed survival, except for COX regression which returned an overestimate. The study presents a robust and unbiased benchmarking methodology using a bespoke web facility. It was concluded that powerful, recent flexible modelling algorithms show a comparative predictive performance to that of more established methods from the medical and biological literature, for the reference data set.