Despite all of the encouraging clinical results with molecularly targeted therapies against non-small cell lung cancer, there are still important and significant questions that continue to vex. The discovery of activating mutations in EGFR radically changed the treatment options for these patients, and, indeed, treatment of advanced stage patients with EGFR-mutated tumors with erlotinib (in the US) is now front-line therapy. However, even patients who initially show dramatic responses invariably relapse due to the development of drug resistance. Although such resistance can be explained in about 50% of these patients, what are the mechanisms for the other 50%? Conversely, why do some, albeit rare, patients who are wild-type for EGFR respond to EGFR-TKIs? Also, why don't conventional chemotherapy and molecularly targeted therapy produce a synergistic effect? In fact, patients treated with combined therapy seem to actually do worse than if they get just one or the other--why?
Numerous research groups are trying to identify more effective biomarkers to identify predictive markers for response or resistance to EGFR-TKIs in order to optimize therapy for individual patients. Our group at Rush University Medical Center has been focused on the role of epithelial-mesenchymal transition as a prognostic marker for early-stage NSCLC but also as a predictive marker for response to EGFR-TKIs. So I read with great interest the article by Byers et al. recently published online in Clinical Cancer Research on December 20, 2012 (abstract here).
Some important notes:
- The EMT first principal component correlated better with E-cadherin protein level than did even the best CDH1 mRNA probe set.
- Mesenchymal cell lines were highly resistant to erlotinib as compared to epithelial cell lines, which agrees with previous work. It is interesting, however, that while cell lines with EGFR-activating mutations were the most sensitive to erlotinib, in a subset of 40 cell lines WT for both EGFR and KRAS, the correlation between EMT signature and erlotinib resistance was still observed with significantly greater resistance in mesenchymal-like cell lines. An important corollary is that the EMT signature was a better predictor of erlotinib response than mRNA probe sets for either CDH1 (E-cadherin) or VIM genes individually--a finding that agrees with the work we've done in our lab.
- EMT does not appear to be a general marker of resistance. Indeed, mesenchymal cells were not more resistant to other targeted agents, such as sorafenib or to commonly used cytotoxic chemotherapies, including pemetrexed, docetaxel, paclitaxel, and platinum-doublets. Instead, a trend toward greater relative sensitivity was seen in mesenchymal cells as compared with epithelial for cisplatin.
- An EMT signature may be a predictive marker in EGFR wild-type/KRAS wild-type patients with advanced stage NSCLC. The authors analyzed the association between EMT signature and clinical outcome in patients enrolled in the BATTLE-1 trial but limited their analysis to patients whose tumors were WT for EGFR and KRAS. Although the N is only 20, patients with disease control at 8 weeks (the primary study endpoint) showed a more epithelial-like signature as compared with those without disease control, although the difference is of borderline significance (P=0.05, by t test). However, among the full group of 101 of 139 clinically evaluable patients (all treatment arms), expression of EMT signature genes was not prognostic of 8-week disease control or progression-free survival (PFS) in the overall group (all treatment arms).
This is a dense article but well-worth reading through in detail if you are interested in either EMT in NSCLC or mechanisms of EGFR-TKI drug resistance. I offer a few further comments.
First, note that this is a study of NSCLC cell lines, so the findings have to be taken as hypothesis-generating for confirmation in clinical tumor specimens. The limitations of studying cell lines ought to be well-known to you all. But one point that has to be appreciated is that the phenomenon of EMT in vivo is context-dependent and critically depends on interaction with the tumor microenvironment--and this crucial fact is not and cannot be studied in this type of model. Second, although it seems obvious to state this, histological subtype matters greatly with regard to EGFR mutations, EMT, and response to erlotinib. Why do so many recently published studies still lump together non-small-cell lung cancer? We recently focused exclusively on a group of patients with resected lung adenocarcinomas who received erlotinib upon disease progression in a study before the advent of EGFR mutation testing. We identified two separate "epithelial" (high E-cadherin/low vimentin) and "mesenchymal" (low E-cad/high VIM) as well as intermediate/transitional tumors and found a trend toward worse DFS that did not quite reach statistical significance. Unfortunately, our study was small (N=45) and I think that much larger numbers will be needed to show any importance difference. In this study, out of 35 cell lines with adenocarcinoma histology, 29 (83%) had epithelial signatures, while 8 (23%) had mesenchymal signatures. Of 4 cell lines with squamous histology, there was an even distribution between epithelial and mesenchymal subsets. Thus, the adenocarcinomas more commonly expressed an epithelial signature (P<0.0016, Chi-square test). Finally, I have observed in our studies that EMT in vivo is a heterogenous process throughout a tumor. Indeed, even if you focus on the invasive edge of a tumor, where one expects EMT to be occurring (as we did in a recently published abstract at AACR 2012), there is frequently significant variation in the expression and distribution of E-cadherin and vimentin. What is the significance of this variation? Can it be quantified and does the amount of variation in EMT have any prognostic or predictive significance? Don't know--image analysis and quantitative immunohistochemical studies could possibly answer these questions.