Acquired Resistance to Targeted Therapies against Oncogene-Driven Non-Small Cell Lung Cancer: Approach to Subtyping Progressive Disease and Clinical Implications

Abstract

In the emerging era of targeted therapy for advanced stage non-small cell lung cancer (NSCLC), it is becoming increasingly important to anticipate potential underlying driver oncogene alterations at the time of initial diagnosis and tumor tissue acquisition, so that patients can be selected in a timely fashion for first line tyrosine kinase inhibitor (TKI) therapy if their cancers are found to harbor activating mutations in the epidermal growth factor receptor (EGFR) gene or gain-of-function rearrangements of anaplastic lymphoma kinase (ALK). However, despite clear benefits for TKI therapy over chemotherapy in these settings, eventual emergence of acquired resistance and progressive disease (PD) is universal. How to best approach oncogene-driven NSCLC at the time of acquired resistance to initial TKI therapy is an increasingly complex question, due to variability in mechanisms of resistance, extent of PD and inter- and intra-patient tumor heterogeneity. Here we propose an approach to subtyping PD in the setting of acquired resistance as well as subsequent clinical implications.

Background

A transition from empiric to targeted and personalized therapy of NSCLC is well underway, largely as a result of extensive efforts in genomic characterization. (1, 2, 3) Over the last 10–15 years, NSCLC, previously viewed as a single disease, has been “ungrouped”, initially through histologic subtyping, and more recently through recognition of multiple clinically and biologically distinct molecular subsets, the magnitude of which has only been demonstrated through next generation sequencing of large numbers of cancers. (3, 4) These studies have delineated the complexity of NSCLC at the genomic level, both differentiating it from less complex cancers as well as pointing out a striking degree of inter- and intra-patient tumor heterogeneity. At present, EGFR activating mutations and ALK fusion genes represent the most actionable of these oncogene-driven and molecularly-defined subsets, based on availability of effective TKI therapy for each (5, 6). Undoubtedly others will join this actionable category in the not too distant future, as newer targeted therapies become available (7).

In EGFR-mutated NSCLC, randomized clinical trials comparing EGFR TKIs such as gefitinib, erlotinib or afatinib with chemotherapy have repeatedly demonstrated superior patient outcomes for the TKIs, as measured by response rate and progression-free survival (PFS). (8, 9, 10, 11, 12) More recently, the same has been shown for the ALK inhibitor crizotinib. (13) While no improvement in overall survival has been demonstrated in these trials, this finding has largely been attributed to cross-over from chemotherapy to targeted therapy. Regardless, TKI therapy for cancers harboring EGFR activating mutations or ALK fusions can be viewed as a positive step toward personalized therapy: selecting the right therapy for the right patient.

Yet despite the observed clinical benefits, the overall impact of these targeted therapies has been limited by almost universal development of acquired resistance. Even in these most TKI-sensitive subsets of NSCLC, progressive disease (PD) is typically observed within 10–14 months. Initial studies evaluating therapeutic decision-making at the time of acquired resistance and RECIST PD have tended to lump all patients together, regardless of the location or number of PD sites and/or magnitude of PD. Results from small pilot studies addressing the topic of TKI acquired resistance in NSCLC could thus be affected by great heterogeneity in patient prognosis, treatment options and likely outcomes, leading to confusion about the appropriate therapeutic approaches outside of the clinical trial arena. Here we describe an algorithm for subtyping PD which accounts in part for this variability. Further, we hypothesize that “best” management options at the time of PD differ dependent on the PD subtype. Lastly, we describe clinical trial designs suitable for addressing the complexity of acquired resistance to TKI therapy against oncogene-driven NSCLC.

Proposal for PD Subtyping in the setting of Acquired Resistance to EGFR- or ALK-directed TKI therapy

Conceptually, it is obvious that not all NSCLC patients who develop acquired resistance to targeted TKIs are created equal in terms of the extent and/or sites of progressive disease. Inter- and intra-patient tumor heterogeneity adds to the complexity. (14, 15, 16, 17, 18) Moreover, treatment options vary widely. (19, 20, 21, 22) Thus, as depicted in Figure 1, we propose that PD in the setting of acquired resistance to EGFR- or ALK-directed TKI therapy in NSCLC be broadly subtyped into: 1) CNS Sanctuary PD, 2) Oligo-PD and 3) Systemic PD, both for clinical considerations as well as clinical trial design. In this categorization, CNS Sanctuary PD represents isolated CNS failure, primarily parenchymal brain metastasis, in the absence of systemic PD. Due to the lack of good treatment options for long term control of leptomeningeal carcinomatosis, we propose excluding it from this category. Oligo PD refers to new sites or regrowth in a limited number of areas. For consistency, we propose a maximum of three PD sites for this category. Lastly, Systemic PD represents what Oncologists generally perceive of as PD following chemotherapy, that is, multi-site progression, which may include both new metastatic sites as well as regrowth in previously responsive sites of disease.

Figure 1.

Subtyping Progressive Disease. Abbreviations: CNS = central nervous system; PD = progressive disease.

Clinical Implications of PD Subtyping

PD subtyping in the setting of acquired resistance to TKI therapy for EGFR mutation- or ALK-positive NSCLC provides a rational approach to both clinical trial design and day-to-day patient management. Ensuring homogeneity in patient characteristics and prognostic factors is a hallmark of clinical trial design: comparing “apples to apples”, as discussed below. Just as important, clinical decision-making outside of a clinical trial that takes into account the individual patient situation is in fact another step toward the goal of personalized therapy. For example, in CNS Sanctuary PD as defined here, there appears to be a good rationale for a local-regional approach to PD by surgical resection, focused radiotherapy or whole brain radiotherapy, and for continuing TKI therapy as long as systemic remission is maintained. In this case, isolated CNS relapse may well represent pharmacodynamic failure due to poor TKI penetrance into the brain, rather than emergence of resistant tumor clones. Indeed, pilot studies employing this approach suggest that PFS can be extended in a clinically meaningful way. (19, 20, 21, 22) In the setting of Oligo PD, a similar approach may be possible dependent on the exact location and number of new or recurrent sites of disease. Although a minority of NSCLC patients with TKI acquired resistance fall within the Oligo-PD category, some studies suggest that in ALK-positive cases, this subset may represent up to 1/3 of all patients. (21) Treating the sites of Oligo-PD with SBRT or surgical resection where feasible, while continuing the same TKI therapy, may be a reasonable approach as long as systemic remission is otherwise maintained. In one such study in ALK-positive NSCLC, of 31 patients with PD after initial response to crizotinib, approximately 30% were deemed candidates for local ablative therapy (LAT). In this series, LAT extended the time to further disease progression by an additional 9 months. In a few cases, LAT was again utilized at the time of subsequent Oligo-PD. (21) Despite these favorable preliminary results, only randomized trials comparing this approach to “switch therapy” can fully address the standard of care issue for Oligo-PD, as described In greater detail below.

When TKI acquired resistance can be categorized as Systemic PD, a number of different therapeutic options are possible, as described in Figure 2. Assuming multiple new sites or regrowth in multiple areas, options include 1) Switch therapy, 2) Continued therapy with the same TKI alone, with the hopes of slowing further PD or 3) Addition of another agent(s) to the same TKI. The classic approach to acquired resistance, option 1) Switch therapy, most commonly represents chemotherapy in this setting, with the presumption that a majority of new growth/regrowth is due to TKI-resistant clones, and that the best therapeutic option is an entirely new strategy, such as chemotherapy. With acquired resistance to either EGFR-directed TKIs in EGFR-mutated NSCLC or the ALK-directed TKI crizotinib in ALK-positive NSCLC, there is good rationale for a switch to chemotherapy in the second line setting. In the EURTAC trial, in which erlotinib proved superior to chemotherapy with paclitaxel-carboplatin for response and PFS in EGFR-mutated NSCLC, the benefits of erlotinib appear similar whether the TKI is administered first line or second line following chemotherapy, suggesting that chemotherapy does not significantly alter EGFR TKI sensitivity (10) Moreover, EGFR-mutated NSCLC may be more sensitive to platinum-based chemotherapy than non-mutated cancers, based on deficient DNA repair, as exemplified by low ERCC1 levels. (23) Similarly, pemetrexed-based therapy seems to be particularly active in ALK-positive cancers, perhaps due to interrelationship with ALK signaling and/or target gene expression levels for pemetrexed activity. ( 24, 25, 26) Switch to a second generation TKI with activity in the acquired resistance setting is an attractive approach and an area of considerable pre-clinical and clinical research, based on presumed mechanisms of resistance (27, 28, 29) Somewhat surprisingly, in the setting of acquired resistance to EGFR TKIs, second generation agents directed against the T790M resistance mutation, while typically showing good activity in vitro, have to date largely failed in the clinic, raising the question of how often T790M actually represents a true second driver oncogene. . (29) In contrast, preliminary data suggest that ALK-directed second generation TKIs are highly active clinically (27, 30) While much has been made of the rationale for option 2) “post-PD”, continuation of the same TKI alone in order to prevent disease flair or slow progression due to residual sensitive tumor clones, in reality this approach is of limited appeal in the setting of systemic PD, especially in a symptomatic patient with treatment alternatives.

Figure 2.

Approaches to Systemic PD in Oncogene-Driven NSCLC With TKI Acquired Resistance. Abbreviations: ALK = anaplastic lymphoma kinase gene; EGFR = epidermal growth factor receptor gene; NSCLC = non–small-cell lung cancer; PD = progressive disease; RECIST = Response Evaluation Criteria In Solid Tumors; TKI = tyrosine kinase inhibitor.

Clearly, the way forward is to direct attention toward the mechanisms of acquired resistance and to develop means of overcoming them. To this end, option 3) addition of another agent(s) to the original TKI offers considerable potential to both address emergent resistant tumor clones and to maintain suppression of suppressed but residual sensitive clones. (15) As an example of this approach in ALK-positive NSCLC with acquired TKI resistance, S1300 is a developing North American Intergroup trial which randomizes patients with acquired crizotinib resistance to either switch therapy (pemetrexed) versus continuing crizotinib with the addition of pemetrexed. (Figure 3). Based on pre-clinical modeling and clinical data from re-biopsy of ALK-positive NSCLC at the time of acquired resistance to crizotinib, multiple potential secondary drivers and/or bypass pathways have been identified (18, 31) The hypothesis to be explored in S1300 is that differential activity will be observed in individual patients, dependent on the mechanism(s) of resistance: ALK-dominant versus ALK-non-dominant. (18)

Figure 3.

Randomized Phase II Trial (S1300) in ALK-positive NSCLC Progressive After Crizotinib Therapy. Abbreviations: ALK+ = anaplastic lymphoma kinase positive; CR = complete response; DCR = disease control rate; FISH = fluorescence in situ hybridization; NSCLC = non–small-cell lung cancer; ORR = overall response rate; OS = overall survival; PD = progressive disease; PFS = progression-free survival; PR = partial response; SD = standard deviation; TKI = tyrosine kinase inhibitor.

Prevention or Circumvention of Acquired Resistance in Oncogene-driven NSCLC

Considering the difficulties in designing therapies to address acquired resistance once it is established, an attractive alternative is to design strategies to prevent or circumvent resistance before it develops, i.e. during first line therapy. As described in Figure 4, potential clinical trial approaches, in addition to the current approach on targeted monotherapy following at PD by a second line agent(s) directed at mechanisms of resistance, include sequential targeted monotherapies, or perhaps more appealing, utilization of multi-drug targeted therapy aimed at the most likely mechanisms of acquired resistance to the TKI in question. Multi-drug therapy could be employed concurrently, or in an intercalated fashion designed to achieve pharmacodynamics separation as a means of avoiding potential antagonism when given concurrently. (32) An example of the intercalated tactic is the recently published FASTACT-2, which combined intermittent dosing of the EGFR TKI erlotinib together with chemotherapy to achieve pharamcodynamic separation. (33) Multi-drug therapy employing two or more targeted agents (a “targeted therapy cocktail”) is another approach now being investigated in oncogene-driven NSCLC. An example of this approach to circumventing acquired resistance is S1403, a developing Phase II/III trial combining the second generation EGFR TKI afatinib with the EGFR-directed monoclonal antibody cetuximab in first line therapy, mimicking the strategy employed successfully against human immunodeficiency virus (HIV) disease. (34)

Figure 4.

Clinical Trial Design to Address Circumvention of Acquired Resistance in Oncogene-Driven NSCLC. Abbreviations: NSCLC = non–small-cell lung cancer; TKI = tyrosine kinase inhibitor.

Summary

In summary, acquired or adaptive resistance to targeted therapies in oncogene-driven NSCLC is an expected and almost universal phenomenon. However, individual cases of acquired resistance are highly variable from both a mechanistic and clinical standpoint. Therefore, therapeutic approaches to patients in the acquired resistance setting should be individualized as well. Subgrouping progressive disease into categories such as CNS-PD, Oligo-PD and Systemic –PD provides a basic foundation for integrating mechanistic data from tumor re-biopsy into therapeutic decision-making. Innovative clinical trial designs will be required to address the multifactorial nature of acquired resistance in order to reverse it once established, and even more importantly, to anticipate the most likely mechanisms and circumvent them.

Figure 5.

Phase II/III Trial of Afatinib With or Without Cetuximab in EGFR-Mutated NSCLC (S1403). Abbreviations: EGFR = epidermal growth factor receptor; NSCLC = non–small-cell lung cancer; PDX = patient-derived xenograft; TKI = tyrosine kinase inhibitor.