Investigators discuss the use of targeted therapies to treat metastatic colorectal cancer.
Metastatic colorectal cancer (mCRC) is the second most common cause of cancer-related death worldwide.1 In the mid-1980s, the median overall survival (OS) for patients with mCRC ranged from 10 to 12 months from the time of initial diagnosis.2 In more recent studies, this median has more than doubled and is commonly reported at more than 25 to 30 months.3 These improvements are due, in large part, to the introduction of multiple novel agents during the last 3 decades. Despite these improvements, however, nearly all patients treated with palliative chemotherapy will eventually develop resistance and ultimately succumb to progression of metastatic disease. Understanding the mechanisms by which malignant cells evade treatment could unlock novel therapeutic strategies that overcome resistance and improve survival. In this review, we will discuss some of the drivers of therapeutic resistance in patients with mCRC and present some novel strategies to overcome resistance.
Therapeutic resistance develops in nearly all patients with advanced cancer who are treated with palliative systemic therapy. Resistance to cancer-directed therapy is generally divided into primary and secondary (acquired) resistance. Primary resistance is defined as a lack of objective clinical or radiographic response to therapy. Secondary resistance is defined as therapeutic resistance that emerges after a period of disease stability or response.4 Although these terms will be used throughout this review, it is important to note that this simple division becomes less clear as the precision of surveillance imaging and/or blood-based tests improves. For example, biochemical or molecular assessment of response may be discordant with imaging studies, which may either be a function of disease natural history, imaging frequency, or—in the case of immunotherapy—pseudoprogression. Despite these caveats, it is helpful to understand therapeutic resistance fundamentally in terms of whether a therapy is “not effective” or “no longer effective” in order to better understand and target the underlying mechanisms driving resistance.
Until recently, there have been few attempts to study secondary therapeutic resistance in large-scale clinical trials. We will outline a few of the reasons why this research has been difficult to conduct but also why, in the age of targeted therapeutics, it is of utmost importance. The established clinical trial infrastructure is designed to assess the safety and efficacy of drugs largely based on objective response, progression-free survival (PFS), and OS. If a drug is effective at decreasing or controlling disease for an extended period of time, it is generally felt to be a positive outcome, which can lead to a regulatory approval. As clinical trials move more to response-based end points, it is possible that we will see more drugs that demonstrate excellent initial responses but lead to rapid development of secondary resistance. The need for rapid drug approval in this setting will have to be balanced against long-term outcomes of patients treated with novel therapeutics.
Historically, objective response rate (ORR) has been considered a surrogate for clinical benefit, even where OS benefit is unknown. However, therapies that are designed to target a single aberrant growth signal are susceptible to rapid outgrowth of resistant subclones, limiting the durability of response. The promise of targeted therapy—targeting only the gene or protein that is driving cellular proliferation—may become a liability when resistant subclones drive signaling through alternate pathways. A recurring challenge for drug development in mCRC is the rapid development of treatment resistance. For the vast majority of colon cancers, targeting a single mutation is either ineffective (primary resistance) or leads to relatively rapid disease progression after a period of response (secondary resistance). The future of clinical trials for mCRC will need to shift to dynamic assessment of not only primary resistance mechanisms but also those that develop in response to therapeutic inhibition.
Understanding mechanisms of secondary resistance is of utmost importance, but until recently such understanding has relied almost exclusively on postprogression tissue biopsies. Given the invasive nature of these biopsies, it is understandable that it has been difficult to conduct these ancillary studies in large-scale clinical trials. However, serial tissue biopsies are neither cost-effective nor practical. Furthermore, a single-site tumor biopsy is a relatively poor representation of the spectrum of intra- and intertumoral heterogeneity.5 If we are to successfully reach the ultimate goal of practicing precision oncology, we must understand the dynamic interplay between preexisting mutational profiles and those that are induced or unmasked by treatment. Fortunately, circulating tumor DNA (ctDNA) is a technology that can be harnessed to overcome most of these shortcomings and to provide a diagnostic tool for longitudinal, dynamic disease monitoring.
Although the presence of cell-free DNA (cfDNA) was initially described in 1948 in healthy individuals, it was not until the past decade that commercial ctDNA assays have become incorporated into routine clinical practice.6 The most obvious benefit of ctDNA-based molecular testing is its ability to provide a noninvasive mechanism for monitoring dynamic mutational profiles. The fact that ctDNA is noninvasive allows for repeated, longitudinal testing without the associated morbidity that tissue biopsies entail. As tumors progress, metastasize, or are exposed to stressors such as targeted inhibitors or chemotherapy, they develop acquired mutations; these are not uniformly distributed throughout single tumors or within multiple tumors in the same patient.7 This intra- and intertumoral heterogeneity cannot be accounted for in single-site tissue biopsies. In this situation, ctDNA can actually provide a better understanding of the fluid molecular landscape of cancers, because it is exposed to therapy in real time as resistance mechanisms are developing. Indeed, this approach has allowed for the design of multiple novel clinical trial platforms that are tailored not only to a specific disease state or line of therapy but to underpinnings of molecular resistance. Two such efforts are the Colorectal and Liquid Biopsy Molecularly Assigned Therapy (COLOMATE) umbrella trial (NCT03765736) and the Guardant Originates in Zipangu Liquid biopsy Arrival (GOZILA) trial (UMIN000016343). In these trials, patients with refractory mCRC undergo ctDNA-based screening and are assigned to one of several molecularly selected treatment arms. Patients are also offered molecular screening at the time of trial discontinuation. Although complete discussion of the COLOMATE and GOZILA trials is outside of the scope of this article, their design and implementation serve as a road map for future trial platforms. Herein, we will outline some of the most common mechanisms of resistance to targeted therapies in the treatment of mCRC.
EGFR is commonly overexpressed in CRC.8 This knowledge led to the initial development, and ultimately approval, of the monoclonal antibodies cetuximab (Erbitux) and panitumumab (Vectibix) to treat mCRC.9,10 Despite these approvals, a large proportion of patients did not gain significant benefit from treatment with cetuximab or panitumumab. Multiple reports later identified KRAS mutations as driving primary resistance to EGFR blockade.11-13 While initial reports investigated only KRAS exon 2 mutations, the negative predictive effect of KRAS exon 3 and 4 mutations as well as of NRAS exon 2, 3, and 4 mutations were later found to have a similar negative predictive value.14,15 These findings eventually led to revised approvals of these drugs, limiting treatment to patients with KRAS and NRAS (RAS) wild-type disease. RAS mutations remain the strongest negative predictor of response to anti-EGFR therapy.
BRAF V600E mutations, occurring downstream of KRAS, appear to have a similar negative predictive effect to EGFR inhibitor (EGFRi) therapy. Di Nicolantonio and colleagues initially described a cohort of 79 patients with mCRC treated with panitumumab or cetuximab. The response rate in 11 patients with BRAF V600E–mutant mCRC was 0% compared with 32% in those patients with BRAF V600E and KRAS wild-type mCRC.16 This finding has been subsequently bolstered by multiple retrospective reviews; however, given the relative rarity of BRAF V600E mutations in mCRC, there was insufficient statistical evidence to preclude benefit from anti-EGFR therapies in this patient subset.17,18
HER2 amplification has also been implicated as a driver of primary resistance to EGFRi therapy.18,19 Similar to BRAF, the rarity of HER2 amplification in mCRC has limited the ability to make definitive conclusions regarding its role in EGFR resistance. Nonetheless, HER2 amplification is associated with significantly worse outcomes in patients treated with EGFR antibodies with or without cytotoxic chemotherapy.20,21 In one retrospective analysis, 74 patients with HER2-amplified mCRC had significantly worse ORRs (31.2% vs 46.9%) and median PFS times (5.7 vs 7.0 months).21
All the causes of primary resistance mentioned above can also lead to secondary resistance in patients who are treated with EGFR antibodies. One novel class of mutations that develops as a mechanism of secondary resistance to EGFRi are the so-called EGFR ectodomain mutations, which arise within the receptor region of EGFR and alter cetuximab or panitumumab binding.22,23 Based on mathematical modeling, one can assume that there are numerous subclonal mutations in all patients with mCRC.24 The fact that these mutations generally remain subclonal suggests that they have some inherent growth disadvantage compared with the dominant (wild-type) clone in most cases. This dynamic is changed in the setting of targeted therapy, in which the dominant clone is selectively inhibited; this allows for growth of subclonal populations that are resistant to the inhibitor. When this selective pressure is removed, subclonal resistance mutations rapidly decay. This interplay between dominant and subclonal populations has been shown in a number of elegant preclinical and clinical studies.25-27 In one study that included 135 patients with RAS/BRAF wild-type mCRC who had progressed on anti-EGFR therapy, researchers were able to quantify the half-life of exponential decay of acquired mutations after withdrawal of an EGFRi. The half-life of clonal decay of acquired RAS mutations was 3.4 months, and the half-life of EGFR mutations was 6.9 months. Further, the investigators showed a nearly 20% improvement in ORR between those patients who were retreated with anti-EGFR therapy less than 1 half-life vs those who were retreated after more than 2 half-lives.25
The knowledge that acquired mutations decay after withdrawal of anti-EGFR therapy has spurred multiple trials aimed at EGFR rechallenge in patients who have developed secondary resistance to EGFRi. Two recent single-arm trials have provided evidence that EGFR rechallenge is a feasible and effective strategy for treatment of refractory mCRC. The 2 trials had similar designs and included patients who had an initial response to cetuximab-based chemotherapy followed by development of secondary resistance and exposure to a cetuximab-free regimen. In the CRICKET trial (NCT02296203), 28 patients were treated with cetuximab and irinotecan rechallenge. Although ctDNA was collected at the time of enrollment, patients were not excluded based on the presence of RAS, BRAF, or EGFR ectodomain mutations. The ORR and disease control rate (DCR) were 21% and 54%, respectively. In the CHRONOS trial (NCT03227926), patients with RAS, BRAF, or EGFR ectodomain mutations were excluded from EGFR rechallenge. In this trial, 27 patients were treated with single-agent panitumumab. The ORR and DCR were 30% and 63%, respectively. The COLOMATE companion trial PULSE (NCT03992456) will assess panitumumab rechallenge in patients with mCRC who have progressed on prior anti-EGFR therapy and who meet rigorous molecular eligibility based on cfDNA profiling. Enrollment is ongoing.
HER2 amplification occurs in approximately 3% of patients with mCRC.28 Multiple trials have evaluated different anti–HER2-based therapies in patients with mCRC with varying degrees of success.29-31 Despite the impressive activity of this therapeutic strategy, nearly all patients develop primary or secondary resistance. The HERACLES trial (NCT03225937) evaluated trastuzumab (Herceptin) and lapatinib (Tykerb) in patients with refractory, HER2-positive mCRC. Utilizing ctDNA, these patients were evaluated pre- and post treatment to determine the molecular alterations associated with primary and secondary resistance. Baseline RAS or BRAF mutations were detected in 6 of 7 patients (86%) with primary resistance to anti-HER2 therapy, as compared with 3 of 22 patients (14%) who had clinical benefit from therapy.32 At the time of progression, alterations in a known resistance pathway were identified in all but 1 patient. These acquired alterations included PIK3CA mutations (4 patients), HER2 mutations (3 patients), KRAS mutations (2 patients), EGFR mutation (1 patient), and MET amplification (3 patients). Similar results were reported in the MyPathway trial (NCT02091141), in which patients were treated with trastuzumab and pertuzumab (Perjeta). The ORR was significantly lower in patients with a baseline KRAS mutation vs those with wild-type disease (8% vs 40%).30 In addition to KRAS mutations, those patients with PIK3CA mutations also had significantly worse ORR than those without (13% vs 43%, respectively).
Because patients with HER2-amplified mCRC with concomitant RAS, BRAF, and/or PIK3CA mutations experience limited benefit from selective anti-HER2 strategies, a novel therapeutic strategy is needed. The current standard of care for these patients—regorafenib (Stivarga) and TAS-102 (trifluridine–tipiracil)—offers only limited survival benefit.33,34 The combination of tucatinib (Tukysa) and trastuzumab is active against RAS wild-type and HER2-amplified mCRC,35 but this selective anti-HER2 combination is unlikely to provide benefit in patients with concomitant resistance mutations. To target resistance mutations and HER2 amplification simultaneously, the 3T study will evaluate the safety, tolerability, and preliminary efficacy of tucatinib, trastuzumab, and TAS-102 in patients with PIK3CA-, RAS-, or BRAF-mutated and HER2-amplified mCRC. This will also be a COLOMATE companion trial.
BRAFV600E mutations occur in approximately 7% of patients with mCRC, and they are associated with significantly worse response to chemotherapy and poor prognosis.36,37 Several studies have sought to target BRAF V600E–mutated mCRC. Unlike patients with BRAFV600E–mutated melanoma, patients with mCRC appear to derive modest benefit, if any, from single-agent BRAF inhibitors.38,39 The mechanism of this primary resistance is due to EGFR-mediated reactivation of MAPK signaling through RAS, CRAF, and BRAF heterodimerization.40,41 Indeed, the combination of EGFR and BRAF inhibitors have led to significant improvements in responses, yet these responses tend to be short-lived due to relatively rapid development of secondary resistance. In the phase 3 BEACON trial, patients were treated with a combination of the BRAF inhibitor encorafenib (Braftovi) and cetuximab with or without the MEK inhibitor binimetinib (Mektovi) vs a control arm of irinotecan-based chemotherapy. ORRs (27% vs 20% vs 2%), median PFS (4.5 vs 4.3 vs 1.5 months) and median OS (9.3 vs 9.3 vs 5.9 months) were all improved with the targeted inhibitor combinations.42 Somewhat surprisingly, there was not a clinically significant difference in outcomes between patients who received triplet (BRAF, MEK, EGFR) vs doublet inhibition (BRAF, EGFR) despite a numerically increased response rate with the addition of MEK inhibition.
As mentioned above, the development of secondary resistance in patients treated with targeted BRAF inhibitors is inevitable and generally rapid when compared with other targeted therapy approaches. Multiple pathways of secondary resistance have been identified in both in vitro and patient samples. The most common of these is the development of acquired RAS mutations or amplifications, which accounts for approximately 70% of patients who develop secondary resistance.27,43 Other documented causes of resistance include BRAF and MET amplifications, MAP2K1 mutations, and ARAF mutations as well as BRAF exon 2-8 deletions and EGFR ectodomain mutations, both of which decrease the affinity of BRAF and EGFRi binding.43 The common theme of all these alterations is reactivation of signaling through the MAPK pathway, which bypasses inhibition.
Since resistance mutations decay upon withdrawal of anti-EGFR therapies, it is hypothesized that similar changes would occur in BRAFV600E–mutant mCRC as well. Using this rationale, we have designed the BRAFV600Erechallenge arm of the COLOMATE trial, which will enroll patients with BRAF V600E–mutant mCRC who develop secondary resistance to targeted inhibition, are subsequently treated with chemotherapy, and are then rechallenged with encorafenib, cetuximab, and binimetinib. The addition of binimetinib in this setting allows for negative selective pressure downstream of BRAF, which will hopefully decrease the reemergence of secondary resistance mutations that bypass BRAF signaling.
Non-V600 BRAF (BRAFnon-V600) mutations occur in approximately 2% of all patients with mCRC, and these patients have a significantly better prognosis than those with the more common BRAF V600E mutations.44 BRAF non-V600 mutations are functionally divided into class II or class III, based upon their mechanism of signaling and RAS activation dependence. Whereas BRAF V600E mutations signal as RAS independent, activated monomer, class II BRAF non-V600 mutants signal as constitutively active mutant BRAF dimers that are relatively insensitive to negative RAS feedback inhibition.45 Class III BRAF non-V600mutants have impaired or even absent kinase activity, but they can still amplify MAPK signaling in a RAS-dependent fashion through increased mutant/wild-type RAF heterodimerization.46
As mentioned above, class II BRAF non-V600 variants signal as constitutively activated dimers that are largely independent of RAS activation at baseline. Xenografts derived from patients with class II BRAF non-V600 variants who were exposed to dual BRAF/MEK inhibition showed significant shrinkage in 17 of 19 (89%) of tumors.47 Despite the preclinical evidence of benefit in this patient population, there has been limited success thus far in clinical trials evaluating dual BRAF-MEK inhibition in patients with class II mutations. One potential cause of resistance in patients with mCRC is EGFR upregulation/reactivation. Class II mutants are relatively insensitive to upstream activation (EGFR, RAS) at baseline due to ERK-mediated feedback suppression. However, in situations where this feedback suppression is eased, such as in the case of effective MEK/BRAF inhibition, one would expect rapid feedback reactivation as is seen in BRAF V600E mutations. Current RAF inhibitors selectively inhibit BRAF V600E–mutant monomers, but this can lead to modest ERK inhibition in patients with class II mutations. This is due to binding of the RAF inhibitor to one site in the dimer pair, leading to inhibition of that protomer while the other protomer remains activated. By itself, this modest inhibition is unlikely to provide any level of disease control; however, we hypothesize that the combination of modest BRAF inhibition combined with MEK and EGFR inhibition to prevent feedback reactivation will lead to improved disease control compared with other approved therapies.
Class III BRAF non-V600 mutations activate ERK signaling through amplification of mutant/wild-type RAF heterodimers. Class III mutants bind to RAS more effectively and activate wild-type CRAF, leading to downstream ERK activation.46 ERK activation in these mutants requires upstream activation by RAS,either by activating mutation or increased EGFR activity. This reliance on upstream activation can be exploited by effective EFGR inhibition in patients with RAS wild-type disease.48,49 EGFR inhibition significantly decreases RAS signaling, while MEK inhibition has the potential to not only decrease downstream signaling but also to reduce the competitive advantage of emergent RAS or other upstream mutations. BRAF inhibitor treatment in this setting offsets toxicity from the MEK and EGFRi as it has opposing effects in normal tissues. These patients with class II and class III BRAF non-V600 mutations will be eligible for the COLOMATE Umbrella trial to Evaluate Anti-BRAF Treatment Strategies, which will assess the safety and effectiveness of the combination of encorafenib, cetuximab, and binimetinib.
Although the mechanisms listed above are some of the most common pathways of resistance, they do not account for all the therapeutic targets driving secondary resistance. Noninvasive, longitudinal monitoring with ctDNA will continue to revolutionize clinical trial design. These advances have the opportunity to usher in new trials that allow for dynamic and adaptive treatment arms. In these studies, patients can be started on one trial and seamlessly switched to another at the first signs of resistance. This clinical trial approach will require rethinking clinical trial and statistical design, but it could offer a novel approach to overcoming resistance. We look forward to innovative clinical trial designs that will allow for this level of precision care for patients with mCRC.
Author Affiliations: ¹Mayo Clinic, Jacksonville, FL, USA; ²Vanderbilt University Medical Center, Nashville, TN, USA; ³Emory University School of Medicine, Atlanta, GA, USA; 4Mayo Clinic, Scottsdale, AZ, USA; 5Duke University Medical Center, Durham, North Carolina, USA.
JJ receives consulting fees and education grants from AstraZeneca, Bayer, E.R. Squibb & Sons, and OncLive. KC has research grants to her institution from Array, Bristol Myers Squibb, Daiichi Sankyo, Incyte, Merck, Nucana, Pfizer/Calithera. She receives consulting fees from Array, Foundation Medicine, Merck, Natera, Taiho, Pfizer. CW receives research grants to her institution from Boston Biomedical Inc, INHBRX, Lycera, Rapt Therapeutics, Seattle Genetics, Symphogen, and Vaccinex. She has consulting fees and education grants from Nova Research, Oncology Learning Network, and PrecisCA. JS is a consultant for AbbVie, Amgen, AstraZeneca, Bayer, Biopharma, Inivata, Mereo, Natera, Pfizer, Seagen, Silverback Therapeutics, and Viatris. He has received research grants to his institution from AbbVie, Amgen, AStar D3, Bayer, Curegenix, Daiichi Sankyo, Gossamer Bio, Nektar, Roche/Genentech, Sanofi, and Seagen. TBS receives research funding to his instiution from Abgenomics, Agios, Amgen, Arcus, Arys, Atreca, Bayer, Boston Biomedical, Bristol Myers Squibb, Celgene, Clovis, Incyte, Ipsen, Genentech, Lilly, Merck, Merus, Mirati, Novartis, Pfizer, and Seattle Genetics,. Hereceives consulting fees to his institution from Arcus, Array Biopharma, AstraZeneca, Bayer, Daiichi Sankyo, Eisai, Foundation Medicine Genentech, Incyte, Ipsen, Pfizer, Seattle Genetics, and Merck. Hereceives consulting fees (to self) from AbbVie, Beigene, Boehringer Ingelheim, Celularity, Exact Science, Janssen, Kanaph, Natera, Sobi, TreosBio, and Xilis. He serves on an Independent Data Monitoring Committee/Data and Safety Monitoring Board (IDMC/DSMB) (to self) for AstraZeneca, Exelixis, Lilly, PanCan, and 1Globe. He is on the Scientific Advisory Board for Imugene, Immuneering, and Sun Biopharma. He holds inventions/patents of WO/2018/183488; WO/2019/055687.