Distinct evolutionary paths to TKI resistance in NSCLC

Despite initially responding, EGFR-mutant non-small cell lung cancer (NSCLC) patients treated with EGFR tyrosine kinase inhibitors (TKIs) invariably acquire drug resistance. A better understanding of how resistance develops will be valuable to efforts to design strategies to overcome, delay and prevent acquired resistance. A number of resistance mechanisms to EGFR TKIs have been identified and characterized, with the most common being a secondary mutation in EGFR, T790M, that prevents the EGFR TKI erlotinib or gefitinib from effectively inhibiting activity.¹ The mechanistic details underlying the evolution of resistance, however, are much less clear. Our recent study provides key insights into the different evolutionary paths that EGFR-mutant cancers can take to acquire resistance to TKI.² Similar to prior studies,^3,4 we observed that cells with a bona fide genetic resistance mechanism capable of restoring growth in the presence of TKI (e.g., MET amplification, or in this case, EGFR^T790M) can pre-exist prior to treatment and be rapidly selected for following the onset of the selective pressure applied by TKI treatment. However, we also found that the EGFR^T790M mutation can also be acquired de novo by drug-tolerant “persister” cells during months of drug treatment, thereby rendering these cells more fully resistant and conferring a selective advantage in the presence of TKI (Fig. 1).

Figure 1.

Acquired resistance mechanisms can emerge via 2 distinct evolutionary pathways Top – Cells harboring EGFR^T790M can be present prior to treatment and rapidly elected upon treatment with EGFR inhibitor (EGFRi). Bottom - EGFR^T790M and other bona fide resistance mechanisms evolve de novo during treatment in drug tolerant cells that did not undergo apoptosis after EGFR inhibitor treatment. The fully resistant cells retain epigenetic features of the drug tolerant cells and are less responsive to third generation EGFR inhibitors that target EGFR^T790M.

The evolutionary path that is followed to acquire resistance may have an important impact on the response to subsequent therapy. For resistant cancers harboring EGFR^T790M, the next therapy is a third-generation EGFR TKI. This class of drugs can suppress EGFR with activating mutations as well as EGFR with activating mutations plus T790M, all while sparing wild-type EGFR. This unique specificity profile of third-generation EGFR inhibitors is responsible for the exciting results being observed in the clinic. In our studies, third-generation EGFR inhibitors were less effective at inducing apoptosis in late-developing EGFR^T790M cells that emerged from the drug-tolerant cells compared to those derived from pre-existing EGFR^T790M cells. This is likely a consequence of the former having already undergone selection for cells that did not undergo apoptosis following EGFR inhibition. RNA sequencing revealed that these cells retained some of the epigenetic features of drug-tolerant cells, such as upregulation of mesenchymal genes that likely contribute to their lack of apoptotic response to third-generation TKIs. EGFR^T790M cells that emerged from pre-existing cells, on the other hand, behaved and had a gene expression pattern more consistent with parental cells that had never been exposed to TKI. Pre-existing EGFR^T790M cells may be functionally TKI naïve, in that EGFR was never inhibited and therefore they never underwent selection for a drug-tolerant state. An examination of EGFR^T790M cell lines derived directly from erlotinib-resistant patients revealed varying degrees of apoptosis following treatment with a third-generation EGFR inhibitor. Whether the resistant cells with a robust apoptotic response came from patients in which EGFR^T790M pre-existed erlotinib treatment and those with low apoptosis response came from drug tolerant cells that acquired EGFR^T790M during therapy is an intriguing possibility, but challenging to assess given current technical limitations in detecting very rare genetic alterations in clinical samples. Combining the BCL-2/BCL-XL inhibitor ABT-263 with third-generation EGFR TKIs boosted the apoptotic response of EGFR^T790M cells with low apoptotic response to TKI alone and converted tumor stasis to regression in vivo.

One important implication of our study is that drug tolerant cells that are present at the point of minimal residual disease may serve as a cellular reservoir from which fully resistant cells with bona fide genetic resistance mechanisms can emerge. This is further supported by another recent study that demonstrated that multiple distinct resistance mechanisms can emerge from drug-tolerant EGFR-mutant NSCLC cells during EGFR inhibitor therapy.⁵ We speculate that evolution of resistance mechanisms within drug tolerant cell populations may contribute to heterogeneity of resistant tumors in some cases. In addition, a greater number of drug tolerant cells may be linked to greater potential for heterogeneity of resistance to evolve, which presents a considerable challenge for treating resistant patients in the clinic. Altogether, these data implicate drug-tolerant cells in the development of acquired resistance and underscore the need to identify more effective therapeutic strategies for eliminating them. Previous studies have implicated epigenetic mechanisms in maintaining the drug tolerant state.⁶ Undoubtedly, the tumor microenvironment also plays a role in supporting these cells during TKI treatment.⁷ For example, factors such as TGFβ known to alter the epigenetic state and protect cancer cells from EGFR TKIs, as well as growth factors that can sustain activation of key signaling pathways may be provided by cancer-associated stromal cells. A more complete characterization of drug-tolerant cells both in vitro and in vivo will help to determine the extent to which these factors and others are responsible for their survival. Efforts to identify synthetic lethalities and unique vulnerabilities of drug tolerant cells may also yield insights. Treatment strategies targeting epigenetic modifiers and stromal-mediated protection, as well as other orthogonal approaches designed to undermine the drug tolerant state warrant further investigation. In addition, drug combinations such as third-generation EGFR inhibitors + BCL-2 family inhibitors that induce a greater apoptotic response could be employed in the first line setting to effectively reduce both the number of surviving drug tolerant cells and pre-existing EGFR^T790M cells. In summary, it is becoming increasingly clear that eliminating drug-tolerant cells is a requirement in order to dramatically improve patient outcomes.

Disclosure of potential conflicts of interest

M.J.N. has provided consulting services for Boehringer Ingelheim Pharmaceuticals. J.A.E. is a consultant for Novartis, Sanofi, Genentech, Clovis and Astra Zeneca; owns equity in Gatekeeper Pharmaceuticals, which has interest in EGFR inhibitors; and has research agreements with Novartis and AstraZeneca. A.N.H. has provided consulting services for Amgen.