A primary objective of this phase IIa study was to assess safety at recommended phase II dose. As anticipated from the phase I study, the most common grade III–IV toxicity was thrombocytopenia (17
). However, this was not associated with clinically significant bleeding events and was reversible with interruption of study drug. Thrombocytopenia represents an expected and on-target toxicity of this drug, due to potent inhibition of Bcl-xL
, a close functional homolog of Bcl-2 that is expressed in platelets (15
inhibition has been shown to result in platelet apoptosis, with a bias toward apoptosis of older platelets in circulation (16
The level of single-agent activity of navitoclax in recurrent SCLC was disappointing, given its remarkable preclinical in vivo
activity as a single agent in multiple SCLC cell line xenograft models (12
). This activity was in keeping with our earlier report using the closely related parental drug, ABT-737, in both in vitro
and in vivo
testing of multiple cell lines (13
). In sharp contrast to these results, evaluation of ABT-737 activity in primary xenograft SCLC tumor models, derived from direct transfer of human tumors into recipient mice, showed low-level single-agent activity, varying from no effect relative to control to inhibition of growth trajectory without evident tumor reduction (34
). Gene expression patterns in primary SCLC xenografts more closely mimic the human tumors of origin than do related cancer cell lines (35
). In retrospect, the clinical data presented here support the predictive value of the primary xenograft model as a tool in cancer drug development and, more broadly, underline the importance of evaluating potential therapeutics across a spectrum of relevant preclinical disease models.
Multiple preclinical models, including primary xenografts, in fact do support that ABT-737 and navitoclax, through lowering of the tumor apoptotic threshold, enhance the efficacy of standard cytotoxic agents against SCLC and other solid tumors (36
). Thus, combinatorial regimens involving navitoclax with chemotherapy or radiation may be of particular interest. This strategy is being actively tested in a variety of solid tumors.
The correlative biomarkers evaluated in this study are currently exploratory in nature. The relationship between tumor size and these markers has not been explored for first-line treatments in SCLC, and the independent prognostic role of the markers is not clear. We hypothesize that some of these endpoints, including CTC enumeration and tumor marker expression, may represent prognostic biomarkers, associated with poor outcome across a variety of therapeutic strategies and in a variety of clinical contexts (reviewed in ref. 43
). NSE, CYRFA 21-1, CTCs, and M30 have been reported as poor prognostic factors in newly diagnosed SCLC (18
). Plasma pro-GRP, in contrast, we hypothesize may reflect tumor dependence on Bcl-2 and may be more closely linked to activity of potent and specific Bcl-2 inhibitors. If pro-GRP is confirmed as a predictive factor for sensitivity to Bcl-2–targeted inhibition, this could represent a strategy for selective treatment of patients for future clinical trials of navitoclax. In contrast, hypothesized prognostic markers (CYFRA 21-1, NSE, and CTCs), if confirmed in other settings, could be considered as stratification factors for future clinical research in SCLC. Evaluation of these putative biomarkers in future clinical trials in SCLC will better define their use as prognostic markers for SCLC or predictive markers for treatment with navitoclax.