Lung cancer remains one of the tumor types with the poorest survival, and although there have been gains in survival time with the identification of targetable driver mutations, the emergence of drug resistance remains a significant bottleneck to further therapeutic gains. A particularly salient instance of this occurs in the case of ALK-mutated non-small cell lung cancer (NSCLC), driven by the ALK-EML4 fusion gene. The ALK fusion product is targetable with a number of different small-molecule tyrosine kinase inhibitors (TKIs) and, while these, beginning with crizotinib, have shown significant benefits in regards to PFS (NEJM 2014;371(23):2167-2177), and possibly even OS (NEJM 2017; doi:10.1056/NEJMoa1704795) in the case of alectinib, they all share the commonality of the emergence of resistance that foils our attempts at cure. Once found to be resistant, these tumors are seldom chemo- or radioresponsive, and the decision regarding second-line therapies remains clinically uncertain.
In our recent study published in Scientific Reports (2017; doi:10.1038/s41598-017-00791-8), we were inspired by lessons from evolutionary biology and antibiotic resistance (PLoS Comput Biol 2015;11(9):e1004493, Sci Transl Med 2013;5(204):204ra132-204ra132), as well as a recent study of leukemia (Cell 2016;165(1):234-246) to study a phenomenon called collateral sensitivity. In these previous works, it has been shown that there can exist predictable patterns of induction of sensitivity to second-line drugs by given first-line drugs. As this had not been tested in solid tumors, we designed our study to do so in ALKmut NSCLC. Further, the clinical reality of drug holidays-that is time between first-line therapy and second-line therapy-has never been considered in these studies, and we endeavored to explore that here as well.
That is, just as opponents are lured into traps of weak positions in chess by the promise of short-term gains, the aim of collateral sensitivity is to use sequences or patterns of drug administration to sensitize the tumor to future drugs, and ultimately use this to "checkmate" the tumor.
Study Design & Outcomes
Our study design aimed to recapitulate the phenomenon of collateral sensitivity in an in vitro experiment using the ALK-mutant lung cancer cell line H3122. Over a 16-week period, resistance was evolved to four targeted ALK TKIs (crizotinib, lorlatinib, alectinib, and ceritinib), and also to six alternative chemotherapeutic agents including standard NSCLC drugs (pemetrexed, paclitaxel, and etoposide), as well as several heat shock protein-90 inhibitors (ganetespib, IPI-504, and AUY922), to study whether a different first-line therapy might have a more favorable collateral sensitivity profile further on-that is, may induce increased sensitivity to second line therapies.
Once the 10 distinct resistant cell lines were obtained, these were assayed for their sensitivities towards the other chemotherapeutic agents tested in a search for vulnerabilities in the form of collateral sensitivity. Importantly, these results show there may be exploitable sensitivities exposed in the resistant cell lines influenced greatly by the choice of first-line therapy. In essence, these are weaknesses induced in the cell line as a cost to their resistance. We also computationally derived drug cycling regimens, alternating sequences of drugs that all sequentially induce sensitivity to the next. While these sequences were not directly assayed, they provide a large number (84) of specific hypotheses for further testing.
A further outcome looking at the effects of drug holidays on the cell lines made resistant to ALK-targeting chemotherapy was studied to determine whether such holidays may have any effect on the resistant state. We tested to see how/if the reported collateral sensitivities changed after 1 day, 3 days, 7 days, 14 days, and 21 days of drug holiday (i.e., stopping administration of the drug, once in the resistant state). Notably, we found that collateral sensitivities changed significantly between resistant cell lines, even over these short, clinically-relevant timescales.
Implications for Future Studies
Based on these results, the first study of its kind in a solid tumor, it appears clear that the evolution into and out of the chemotherapeutic resistant state is a process necessitating much future work. The evolution of drug resistance is considered one of the greatest issues in cancer therapeutics in the current era, and will continue to be unless the resistant state can be better understood and taken advantage of. The study of the induction of collateral sensitivity, as we suggest here, provides one way forward. While this study is limited by its in vitro methods, we feel these results provide strong encouragement for future studies involving the study of drug cycling regimens, and the reproducibility of phenomena that might be clinically actionable, such as collateral sensitivity.
Future work in this arena will involve the study of other tumor types for this phenomenon in other tumors with targetable mutations, as well as those without specific mutations at fault. Further, understanding commonalities in the induced states of sensitivities at the molecular scale will help us determine the underpinnings of the resistant state and gain a greater understanding of the mechanisms of collateral sensitivity.
Clinical Implications
The clinical implications of this work rely significantly on its ability to be reproduced in vivo across samples. The reproducibility of evolutionary processes is not yet understood to a full extent, especially in the in vivo realm, and we caution that this may limit the generalizability of such a strategy. However, this study opens the door to further work to quantify and qualify the effects of evolution against our therapies and its consequences. This on its own warrants future study, and our additional findings of the instability of these sensitivities in the face of drug holidays and the existence of drug cycling regimens appear to be promising avenues that future clinical trials should consider.
Ultimately, a treatment based on evolutionary paradigms has the potential to change the management of diseases such as NSCLC, previously undefeatable due to resistance, into chronic diseases, which we may be able to keep at bay with cycling regimens by taking the long view of treatment and constantly staying one step ahead of the cancer.
JACOB SCOTT, MD, is a physician-scientist in Translational Hematology and Oncology Research and Radiation Oncology at Cleveland Clinic. ANDREW DHAWAN, MD, is a PhD student jointly appointed to Cleveland Clinic and University of Oxford.