Modeling longer-term efficacy of neratinib in the extended adjuvant setting for early-stage breast cancer.

e12011 Background: Long-term efficacy in the extended adjuvant setting requires a long follow-up. In the ExteNET trial, neratinib was given for a year and invasive disease-free survival (iDFS) assessed at 2 and 5 years. In women with HER2+ hormone receptor positive (HR+) cancer who initiated neratinib, the observed difference in iDFS between neratinib and placebo at 5 years (4.5%) was greater than at 2 years (3.7%). The objective of these analyses was to extrapolate the effect of neratinib on iDFS beyond 5 years in patients with HER2+/HR+ early stage breast cancer who initiated treatment after receiving adjuvant trastuzumab. Methods: We analyzed the 5-year follow-up of the ExteNET trial using flexible spline-based parametric survival models to project iDFS risk at 10 years. Analyses included only HR+ patients. Several model specifications were explored including various combinations of the following variables: treatment as a time varying covariate, nodal status (0, 1-3, ≥4 nodes), and months from last trastuzumab treatment to randomization. The regression models included different combinations of interaction terms, either a proportional hazards or a proportional odds (PO) link function, and either 2 or 6 knots. The model fit was compared using Akaike Information Criterion (AIC). Results: In general, the fit statistics among the 16 models were comparable; however, the model with the lowest AIC included no interaction terms, 2 knots, and the PO link function. The mean iDFS event rate at 10 years was estimated to be 12.2% for neratinib and 18.9% for placebo for a difference of 6.8% (range across 16 models: 6.8%-7.8%). The recurrence rate and the difference between treatment groups were largest in women with ≥4 positive nodes. Conclusions: Based on these analyses, the projected 10-year difference in iDFS between placebo and neratinib patients seems likely to persist. However, because the data was limited to 5 years, we could not include the effect of hormonal therapy cessation. Also, low event rates in the node negative group may require longer follow-up to evaluate. In absence of longer-term data for neratinib, flexible parametric survival modeling of the iDFS suggests that the effect of neratinib therapy may remain stable.