Targeting the MAP kinase pathway in colorectal cancer: A journey in personalized medicine

Abstract

The anti-EGFR agents, cetuximab and panitumumab, were the first targeted agents to be licensed in colorectal cancer and marked a significant advancement in personalized care. Initial biomarkers provided poor discrimination between responders and non-responders. Through hypothesis-led translational studies, tumor genomic negative predictive markers were identified, and treatment is now limited to patients with RAS and BRAF wild-type disease. Guidelines further recommend treatment limitation to those with left-primary tumor location (PTL). Despite such progress, anti-EGFR response remains variable within the biomarker-selected population, indicating the presence of additional mechanisms of resistance and underscoring the need for novel positive predictive biomarkers, and novel targeted agents. This review explores established and emerging predictive biomarkers of anti-EGFR efficacy, including tumor genetic alterations beyond RAS and BRAF, as well as the EGFR ligands, amphiregulin (AREG) and epiregulin (EREG). To date, biomarker discovery and validation have largely been performed within post hoc analyses of existing clinical trial datasets. We highlight ongoing prospective clinical trials aiming to validate earlier findings and describe how novel biomarkers are being used to re-evaluate anti-EGFR agents in treatment settings where earlier trials, among non-biomarker selected populations, yielded negative results – including right-PTL, locally advanced disease, and anti-EGFR rechallenge strategies. Additionally, we discuss how our improved understanding of the molecular mechanisms underpinning anti-EGFR response and resistance is being leveraged to develop novel targeted agents.

1. Introduction

Colorectal cancer (CRC) is the third most diagnosed malignancy internationally and has the second highest cancer mortality rate. Alarmingly, the global burden of disease is predicted to increase by 60% to 2.2 million new cases annually by 2030.(1) Also of concern, prevalence in younger patients is increasing to the extent that early-onset disease now represents 10% of all diagnosed cases.(2)

Work over the past decade has demonstrated marked biological heterogeneity within CRC. This can be characterized by anatomy (primary tumor location – PTL), tumor genomic markers and RNA signatures.(3) This information is not just descriptive, but can be prognostic and even predictive of benefit from certain agents.

The epidermal growth factor receptor (EGFR), also known as ERBB1, is one of four transmembrane growth factor receptors belonging to the ERBB family of receptor tyrosine kinases. It is frequently overexpressed in CRC but, contrary to initial expectations, overexpression is not associated with anti-EGFR sensitivity.(4) Ligand binding to EGFR activates two main signaling cascades: the mitogen-activated protein kinase (MAPK) pathway and the PI3K/AKT/MTOR pathway. These in turn promote cell proliferation and survival (Figure 1).(5)

Figure 1. EGFR-mediated signaling pathway.

EGFR ligands bind to the extracellular domain of EGFR and activate downstream MAPK (RAS-RAF-MEK-ERK) and PI3K/AKT/MTOR pathways. Anti-EGFR monoclonal antibodies (mAbs) - cetuximab or panitumumab - compete to prevent signal transduction (represented by the ‘prohibited’ symbol). RAS/BRAF mutations, ERBB2 (HER2)/MET amplifications, ERBB3 (HER3) overexpression, PIK3CA mutations/PTEN loss – among others – have all been shown to contribute to anti-EGFR resistance. De novo RAS mutation and MET amplification are major mechanisms of acquired anti-EGFR resistance. HER2 generally forms heterodimers with other ERBB family members as it has no ligand binding domain (represented by the ‘X’ symbol), although homodimers can be formed when HER2 is expressed at very high levels. HER3 exists as an obligate heterodimer with other ERBB family members as it has no intrinsic tyrosine kinase activity (represented by the ‘X’ symbol).

Two anti-EGFR agents have been licensed for the treatment of metastatic colorectal cancer (mCRC): cetuximab and panitumumab. Efficacy between the two monoclonal antibodies (mAbs) is similar but there are differences in toxicity profile and delivery.(6) Monotherapy in the chemotherapy-refractory setting has shown efficacy(7,8) but treatment is now more commonly delivered in the first-line setting in combination with doublet chemotherapy – folinic acid, fluorouracil and either irinotecan (FOLFIRI) or oxaliplatin (FOLFOX).(9,10) Data from the COIN trial suggest anti-EGFR agents have lesser efficacy in combination with capecitabine-based chemotherapy,(11) and the TRIPLETE trial has shown a lack of additional benefit from the combination of panitumumab with triplet chemotherapy (folinic acid, fluorouracil, oxaliplatin and irinotecan – FOLFOXIRI) in adequately selected patients with RAS/BRAF wild-type and mostly left-sided disease.(12)

Following initial trials showing limited impact in patients without prior biomarker selection,(7,8) a pivotal moment in the clinical development of these drugs was the identification of KRAS mutations as negative predictive biomarkers for efficacy, translating preclinical evidence into translational data using clinical samples.(13) Further retrospective analyses using clinical trial samples have resulted in current guidelines recommending treatment be limited to patients with extended RAS and BRAF wild-type disease with left PTL.(14,15) Despite restricting treatment to this population, heterogeneity in response and outcome is still observed – suggesting further precision through mechanistic biomarkers is possible. This is particularly pertinent as novel therapies, targets and approaches are currently emerging in mCRC: it has never been more important to prioritize targeted agents to those patients most likely to achieve meaningful response.

Here we describe the evidence base for predictive biomarkers to select patients with mCRC for anti-EGFR agents in current practice and review emerging biomarkers – both negative biomarkers of resistance and positive biomarkers indicating EGFR pathway dependence. We discuss how our improved understanding of MAPK signaling is being leveraged to develop new targeted therapies. Finally, we examine how novel biomarkers are being used to explore the use of anti-EGFR agents in new therapeutic settings – such as right PTL, locally advanced disease, and anti-EGFR rechallenge.

2. Mechanistic biomarkers of anti-EGFR response and resistance, and novel targets

a. RAS mutations

RAS mutations have a prognostic effect in CRC and a predictive effect on anti-EGFR efficacy. The RAS signaling molecule lies downstream of EGFR in the MAPK pathway (Figure 1). Activating KRAS and NRAS mutations are present in around half of CRCs and indicate an inferior prognosis.(16) In 2006, Lièvre et al. found no KRAS exon 2 mutations among 11 cetuximab responders, while KRAS mutations were seen in 13 of 19 (68.4%) tumors that failed to respond.(17) This apparent negative predictive effect was confirmed by others,(13,18–22) informing current guidelines.

Extension of RAS mutation testing to include additional hot-spot codons within KRAS exons 3 (codons 59,61) and 4 (codons 117,146), and in exons 2 (codons 12,13), 3 (codons 59,61) and 4 (codons 117,146) of the NRAS gene has produced incremental improvements in anti-EGFR patient selection.(23,24) However, the objective response rate (ORR) in the PRIME trial (first-line FOLFOX ± panitumumab) among patients with RAS wild-type tumors by the new broader definition remained less than 50%.(24) RAS wild-type status is therefore necessary for anti-EGFR benefit, but not sufficient, with PTL additionally used as a validated surrogate – but again still imperfect.

The RAS protein was at one time considered “undruggable” due to its relatively smooth protein structure. Sotorasib and adagrasib are two KRAS p.G12C inhibitors licensed for use in non-small cell lung cancer that have shown activity in KRAS p.G12C -mutant CRC (3-4% of CRC).(25,26) While RAS mutation is a negative predictor of anti-EGFR efficacy, sotorasib and adagrasib efficacy is enhanced through combination with an anti-EGFR agent due to rapid feedback activation through EGFR when RAS is inhibited in isolation. The CodeBreaK 300 trial demonstrated a significant improvement in the primary endpoint of progression-free survival (PFS) with both higher (960 mg, once daily) and lower (240 mg, once daily) dose sotorasib in combination with panitumumab vs standard care in the chemotherapy-refractory setting (median PFS, high dose sotorasib vs standard care: 5.6 vs 2.2 months; HR 0.49; 95%CI, 0.30-0.80; P=0.006),(26) with final overall survival (OS) data set at 50% of events and recently showing no significant benefit.(27) Results from the phase III KRYSTAL-10 trial of second-line adagrasib plus cetuximab vs chemotherapy are awaited (NCT04793958) and a next-generation of first-line trials with chemotherapy combination is planned. Sotorasib and panitumumab is also currently being tested in locally advanced disease in the ongoing UNICORN trial (NCT05845450). Encouragingly, pan-(K)RAS inhibitors are now in early phase trials.(28)

b. BRAF mutations

BRAF mutations have a prognostic effect in CRC and a probable predictive effect on anti-EGFR efficacy. BRAF mutations are present in approximately 10% of CRC with the class I, RAS activation-independent mutation, BRAF p.V600E accounting for 80% of these.(29) In addition to the relative scarcity of BRAF mutations, their strong negative prognostic effect (which exceeds that of RAS mutations(16)) has hampered attempts to define their predictive effect on anti-EGFR efficacy, despite a clear biological rationale.(30) The phase II FIRE-4.5 trial sought to identify the optimal first-line management for patients with BRAF p.V600E-mutant mCRC, randomizing patients to receive FOLFOXIRI with either cetuximab or bevacizumab.(31) PFS was significantly lower with cetuximab compared with bevacizumab (6.7 vs 10.7 months; HR=1.89; P=0.006), with a trend towards inferior OS – confirming BRAF-mutant status as a probable negative predictive biomarker.

The introduction of BRAF targeted agents has reduced the relevance of this debate. The BRAF inhibitor, encorafenib is licensed for use together with cetuximab (EC) in the second- or third-line management of BRAF-mutant disease,(32) and the combination has recently received FDA approval for first-line use together with FOLFOX following tumor response results from the phase III BREAKWATER trial.(33) EC is also currently being tested in locally advanced disease in the ongoing FOxTROT 4 trial (ISRCTN83842641). As with KRAS p.G12C inhibition, an anti-EGFR agent is required in the combination to prevent feedback activation through EGFR in the presence of isolated BRAF inhibition.(34)

c. ERBB2 (HER2) amplification and ERBB3 (HER3) overexpression

ERBB2 (HER2) amplification has a putative predictive effect on anti-EGFR efficacy. HER2 amplification and/or overexpression is present in around 4% of CRCs, 95% of these occurring in the left colon and rectum.(29) It is more frequent in RAS/BRAF wild-type tumors and is mutually exclusive with microsatellite instability (MSI).(35) HER2 lacks a ligand binding domain and, under ordinary conditions, forms heterodimers with other members of the ERBB receptor family. When overexpressed, spontaneous homodimerization can occur.(36) In the presence of EGFR blockade, this process appears to enable bypass activation, with HER2 amplification associated with inferior PFS with anti-EGFR therapy in a meta-analysis of five retrospective cohort studies (N=594; HR=2.84; 95%CI, 1.44-5.60).(37)

No formal analysis of the predictive effect of HER2 amplification on anti-EGFR efficacy has yet been performed, and HER2 testing is not universally conducted in CRC. There have been encouraging results from several phase II trials of HER2-directed therapy. Based on non-randomized data, trastuzumab/tucatinib has received FDA approval in patients with HER2-positive and RAS wild-type disease, while trastuzumab-deruxtecan at the dose of 5.4 mg/kg every 3 weeks has received tumor agnostic, accelerated FDA approval for patients with treatment-refractory, unresectable/metastatic HER2 immunohistochemistry (IHC) 3+ disease based on the results of trials across a number of tumor types. These include the DESTINY-CRC02 trial in HER2-positive CRC where the ORR was 37.8% (31/82 patients; 95%CI, 27.3-49.2) in the 5.4 mg/kg dose group (irrespective of RAS mutational status).(38) With the phase III MOUNTAINEER-03 trial of first-line FOLFOX with trastuzumab and tucatinib vs FOLFOX-based standard care also ongoing in RAS/BRAF wild-type, HER2-positive tumors (NCT05253651), upfront HER2 testing in CRC may soon become more widespread, potentially influencing anti-EGFR prescribing. The SAGITTARIUS trial is also examining the potential benefit of HER2-targeted therapy in the adjuvant setting (NCT06490536).

ERBB3 (HER3) overexpression has a weak positive prognostic effect in mCRC(39) and a possible predictive effect on anti-EGFR efficacy. Having no intrinsic tyrosine kinase activity, HER3 is an obligate ERBB family heterodimer. As with HER2, it has been suggested that HER3 overexpression might provide an escape pathway under EGFR blockade via PI3K/AKT activation.(40) However, in a retrospective analysis of the PICCOLO trial (second-line irinotecan ± panitumumab), increased HER3 mRNA expression was predictive of panitumumab PFS and OS benefit among RAS wild-type patients (adjusted P_interaction= 0.003 and 0.01 respectively), so HER3 positivity may rather indicate EGFR pathway dependence.(41) Complicating matters, tumor-derived heregulin (a HER3 ligand) may be associated with de novo and acquired resistance to cetuximab.(42) In a human CRC cell line exhibiting high heregulin production, cetuximab monotherapy inhibited EGFR and ERK phosphorylation, but not that of HER2, HER3 or AKT.(43) In contrast, cetuximab in combination with patritumab (a HER3 mAb) resulted in marked inhibition of both pathways. Such an approach has not yet proved successful in practice, with the FOCUS 4-D trial of a pan-ERBB inhibitor terminated prior to completion of recruitment due to futility.(44) A number of novel HER3-directed therapies are, however, currently in early phase trials.(45)

d. MET amplification

MET amplification and high MET protein expression are associated with early tumor invasion and metastasis, conferring an inferior prognosis in CRC.(46) The MET tyrosine kinase is activated by hepatocyte growth factor and shares downstream pathways with EGFR. De novo MET amplification is uncommon in CRC with a rate of 1.7% (10 of 590 patients) in one cohort.(47) Instead, it appears to have a role in acquired anti-EGFR resistance, as an alternative mechanism to acquired RAS mutation, with a frequency of 22.6% in the circulating tumor DNA (ctDNA) of 53 patients who had experienced progression of disease during anti-EGFR therapy.(47) Agents targeting MET are now becoming available and the antibody-drug conjugate ABBV-400 (a MET-targeting antibody, telisotuzumab with a topoisomerase 1 inhibitor payload) has shown promising antitumor activity in a phase I trial.(48) Preclinical data suggest that MET amplification-associated anti-EGFR acquired resistance can be overcome by MET inhibition.(49,50) Indeed, a case report has described efficacy with the class I MET inhibitor, capmatinib in combination with encorafenib in a patient with BRAF p.V600E-mutant, MET amplified mCRC that had progressed on EC.(51)

The EGFR-MET bispecific antibody, amivantamab (given intravenously) has shown promising activity among patients with RAS/BRAF/EGFR extracellular domain (ECD) wild-type and HER2 non-amplified, chemotherapy-refractory mCRC in a phase Ib/2 trial, including in patients with prior anti-EGFR exposure and right PTL disease.(52,53) Based on the encouraging activity and safety in the chemotherapy combination cohorts of the study, the phase III OrigAMI-2 and OrigAMI-3 trials will test amivantamab (administered subcutaneously) in the first- and second-line in combination with chemotherapy. The relevance to amivantamab of predictive biomarkers of anti-EGFR efficacy requires further investigation, but the use of next-generation anti-EGFR mAbs may overcome several resistance pathways and enable a stronger immune-redirecting effect.

e. PIK3CA mutations and PTEN loss

PIK3CA mutations and PTEN loss do not appear to have an independent prognostic effect in CRC,(54,55) and data have been conflicting regarding their predictive effect on anti-EGFR efficacy.(56) Activating mutations in exons 9 or 20 of PIK3CA occur in 10-20% of CRC but usually co-occur with KRAS and BRAF mutations,(57) making it difficult to discern their precise predictive effect on anti-EGFR efficacy.

30% of CRCs contain mutations or epigenetic changes that cause inactivation of the PTEN tumor suppressor gene. However, concordance between primary tumors and metastases is variable and so again a definitive predictive effect has not been demonstrated.(58) Consequently, neither US nor European guidelines recommend PIK3CA or PTEN mutation assessment in routine clinical practice.

The PI3Kα inhibitor, alpelisib is used in the palliative management of PIK3CA-mutant, hormone receptor-positive, HER2-negative breast cancer.(59) A phase II trial of third-line alpelisib with capecitabine in patients with PIK3CA-mutant CRC showed a modest ORR of 7.7% (2 of 26 patients).(60) Interestingly, of 10 patients who had a partial response or stable disease for at least 15 weeks, none had KRAS-mutant tumors or liver metastases.

Upregulation of PI3K in PIK3CA-mutant CRC enhances cyclo-oxygenase-2 (COX-2) activity and hence prostaglandin E₂ synthesis, preventing tumor cell apoptosis.(61) Aspirin may suppress cancer-cell growth and induce apoptosis by inhibiting COX-2(62) and was seen to prolong cancer-specific survival in a retrospective observational study of patients with PIK3CA-mutant CRC.(63) More recently, the phase III SAKK 41/13 trial randomized 112 patients with PIK3CA exon 9/20-mutant stage II/III CRC 2:1 between standard care with aspirin (100 mg daily for 3 years) and standard care with placebo, demonstrating a clinically meaningful but statistically non-significant improvement in 3-year DFS (88.3% vs 82.4%; HR=0.57; 90%CI, 0.27-1.22; P=0.11).(64) The larger ALASCCA trial in stage I-III rectal and stage II/III colon cancer tested adjuvant aspirin (160 mg daily for 3 years) against placebo in 314 patients with PIK3CA exon 9/20-mutant disease and 312 patients with tumors exhibiting less common PIK3CA/PTEN alterations.(65) Here, there was a significant improvement in DFS at 3 years in both biomarker-defined groups (HR=0.49; 95%CI, 0.24-0.98; P=0.044; and HR=0.42; 95%CI, 0.21-0.83; P=0.013, respectively). Given the ready availability of aspirin and the increasing availability of data regarding PIK3CA/PTEN tumor status in routine practice, these data could result in immediate patient benefit.

3. Primary tumor location

The distribution of different molecular subtypes of colorectal cancer along the colon is non-random.(66) Right PTL tumours are associated with Fusobacterium nucleatum, carcinogenic bile acids, deficient mismatch repair status, CpG island methylator phenotype, and KRAS, BRAF and PIK3CA mutations; while left PTL tumours are associated with younger age of onset, male sex, chromosomal instability, and NRAS and TP53 mutations.(67) Right PTL has been shown to predict lack of benefit from anti-EGFR therapy,(68,69) which is reflected in current clinical guidelines.(14,15) However, the dichotomous left vs right approach has been challenged and rather a continuum has been suggested, with no one colorectal cancer subtype exclusive to any particular location of origin.(70) Consequently, while right PTL patients appear not to benefit from anti-EGFR therapy as an overall group, there may well be subgroups within this population who do.

4. Novel strategies for defining anti-EGFR sensitivity

a. EGFR ligands: Amphiregulin and epiregulin

Autocrine and paracrine stimulation of EGFR by its ligands, amphiregulin (AREG) and epiregulin (EREG), are mechanisms of EGFR pathway dependence,(71) and the two molecules are commonly overexpressed in CRC.(22) A predictive effect of AREG/EREG mRNA on benefit from anti-EGFR therapy has now been examined in five retrospective analyses of randomized clinical trials.(72–75)

As AREG and EREG are highly co-expressed at the transcriptional and protein levels, and can each bind EGFR independently,(22,71) a model where high expression of one or both ligands is regarded as denoting biomarker positivity has been proposed.(76) A predictive effect on PFS using this model has been demonstrated in a post hoc analysis of the PICCOLO trial (irinotecan ± panitumumab) (P_interaction=0.01),(76) with benefit also observed among those with right PTL (HR, 0.20; P=0.04).(77)

IHC – rather than quantitative reverse transcription polymerase chain reaction (qRT-PCR) – might be a more convenient assay for quantifying AREG and EREG in routine practice and its utility in this regard has been demonstrated in both the PICCOLO dataset(78) and in a cohort of patients who received anti-EGFR therapy as part of standard care.(79) Interestingly, patients in PICCOLO with very low AREG/EREG expression by IHC (<20% tumor cells positive) appeared to be harmed by the addition of panitumumab to irinotecan, with significantly worse PFS (HR 1.73; 95%CI, 1.02-2.95; P=0.04),(78) again highlighting the importance of mechanistic biomarkers in this setting.

Given the potential AREG and EREG have shown for identifying patients with right PTL who might benefit from anti-EGFR therapy, prospective validation is now being undertaken in the randomized ARIEL trial (ISRCTN11061442) of first-line chemotherapy ± anti-EGFR therapy in patients with RAS wild-type, AREG/EREG-high disease and right PTL. Additional prospective data is being acquired in the single-arm BIOMARCER-2 trial (ACTRN12623000874617) of second- (or later) line cetuximab with irinotecan-based chemotherapy, again in patients with right PTL.

b. Negative hyperselection for anti-EGFR therapy

The PRESSING (PRimary rESiStance IN RAS and BRAF wild-type mCRC patients treated with anti-eGfr monoclonal antibodies) panel was developed in an effort to create an assay with clinical utility, incorporating less common biomarkers of anti-EGFR primary resistance: HER2/MET amplifications, ALK/ROS1/NTRK1-3/RET fusions, PIK3CA/AKT1/MAP2K1 mutations and PTEN alterations.(80) In a case-control study including 94 patients treated with anti-EGFR agents in routine practice, PFS was significantly longer in those with PRESSING-negative than PRESSING-positive tumors (6.3 vs 2.7 months; HR, 0.18; P<0.001) – a process the authors termed “negative hyperselection”.(80) HER2 amplification was the most frequent alteration in resistant patients (7 of 47 [14.9%] patients), followed by MET amplification (4 [8.5%] patients), and all genetic alterations were mutually exclusive. This finding has since been corroborated in post hoc analyses of trials involving upfront(81) and maintenance panitumumab.(82)

The phase III PARADIGM trial of first-line FOLFOX with panitumumab vs FOLFOX with bevacizumab in RAS wild-type disease demonstrated superior OS with panitumumab among patients with left PTL (median OS: 37.9 vs 34.3 months; HR, 0.82; P=0.03), while outcomes were similar in the two groups for patients with right PTL (median OS: 20.2 vs 23.2).(83) The trial incorporated ctDNA collection at baseline. In a pre-specified post hoc analysis, negative hyperselection was performed using a broad gene panel (similar to the PRESSING panel) incorporating KRAS, NRAS, BRAF p.V600E, PTEN and EGFR ECD mutations, HER2 and MET amplifications, and ALK, RET and NTRK1 fusions.(84) Inclusive of patients with both left and right PTL, OS was longer with panitumumab in the ctDNA negative hyperselection population (median OS: 40.7 vs 34.4 months; HR, 0.76), with the predictive effect of the panel confirmed in this randomized dataset involving a non-anti-EGFR control arm (P_interaction=0.037). Interestingly, RAS mutations were the second most frequent alteration detected (53 of 733 [7.2%] patients) after BRAF p.V600E mutation – despite known tumor RAS wild-type status (but not BRAF wild-type status) at baseline being an eligibility criterion. Patients with RAS alterations detected in ctDNA had poorer survival with panitumumab (median OS: 20.9 months vs 25.7 months) than those with RAS wild-type ctDNA (36.3 vs 32.4 months), demonstrating the potential significance of assays with greater sensitivity – with ctDNA being a potentially useful complement in cases where turnaround time or tissue quality constraints are present. Importantly, following negative hyperselection, there was also a strong trend towards benefit from panitumumab in the smaller right PTL subgroup (n=85; median OS: 38.9 vs 30.9 months; HR, 0.82; 95%CI, 0.50-1.35).(84) Moreover, the lack of benefit from panitumumab in the overall right PTL subgroup appeared to be driven by disbenefit among those with mutations (median OS 14.1 vs 18.5 months; HR, 1.33; 95%CI, 0.84-2.11), highlighting the importance of mechanistic biomarkers in this setting.

5. Emergence of biomarkers of acquired anti-EGFR resistance in ctDNA: Implications for rechallenge

The emergence of RAS or BRAF mutations in previously wild-type tumors, and alternatively of EGFR ECD mutations that impair antibody binding to its target, are major mechanisms of acquired anti-EGFR resistance.(50,85) Indeed, in the FIRE-4 trial, 27 of 139 (19.4%) patients with RAS wild-type baseline ctDNA and a paired post-progression sample converted to RAS-mutant under continuous cetuximab exposure, while 7 of 155 (4.5%) converted from BRAF wild-type to BRAF-mutant.(86) Interestingly, in the absence of continued selective pressure from EGFR inhibition, the prevalence of RAS- and EGFR ECD-mutant clones in ctDNA declines with half-lives of 3.7 months and 4.7 months respectively(87) – suggesting anti-EGFR rechallenge may be effective in patients who have previously experienced a response to treatment and have subsequently gone on to experience disease progression.

The phase II CRICKET trial tested irinotecan with cetuximab rechallenge in patients with chemotherapy-refractory mCRC.(88) Among 28 patients, there were 6 (21%) partial responses, and 9 (32%) patients had stable disease. ctDNA was collected at rechallenge baseline. No RAS mutations were detected in patients with partial response, and patients with RAS wild-type ctDNA had significantly longer PFS than those with RAS-mutant ctDNA (median PFS: 4.0 vs 1.9 months; HR, 0.44; 95%CI, 0.18/0.98; P=0.03) – highlighting the potential role of ctDNA in patient selection for anti-EGFR rechallenge.

Building upon this, in the single-arm, phase II CHRONOS trial, participants were eligible for panitumumab rechallenge if they were RAS, BRAF and EGFR ECD wild-type by ctDNA at progression on a subsequent regimen that did not include an anti-EGFR agent.(89) Of 27 patients, eight (30%) experienced a partial response and 17 (63%) disease control, suggesting a potential survival advantage from repeat use of anti-EGFR agents in selected patients. The randomized phase II CITRIC trial had similar eligibility criteria.(90) 114 patients were screened, with 30 (26%) patients excluded due to the presence of RAS/BRAF/EGFR ECD mutations. 58 patients met all eligibility criteria and were randomized between cetuximab and irinotecan vs investigator’s choice third-line therapy without an anti-EGFR agent. The trial failed to meet its ambitious primary objective of a difference of 27% in the ORR between the intervention and control arms (12.9% vs 0%; P=0.116). There was however a significant improvement in the disease control rate (74.2% vs 44.4%; P=0.031) and a trend towards improved PFS (median PFS 4.4 vs 2.2 months; HR, 0.717; 95%CI, 0.403-1.276; P=0.255), providing a strong signal that some patients stand to benefit from this approach but suggesting greater precision in patient selection may be required – perhaps with inclusion of other markers of acquired resistance such as MET amplification.(47)

6. Locally advanced disease

Early in their development, attempts were made to demonstrate benefit from anti-EGFR agents in the adjuvant setting.(91,92) However, with the limited patient selection possible at the time, results were disappointing. More recently, in a post hoc analysis of the PETACC-8 trial (adjuvant FOLFOX ± cetuximab), a trend towards improved survival outcomes was seen among patients with RAS and BRAF wild-type disease (time to recurrence: HR, 0.77; 95%CI, 0.55–1.08; P=0.12).(93) The FOxTROT trial tested peri-operative vs adjuvant chemotherapy in locally advanced colon cancer, demonstrating a significant improvement in 3-year disease-free survival (DFS) (80.7% vs 75.5%; HR 0.73, 95%CI 0.55-0.97; P=0.030).(94) FOxTROT included an embedded randomized phase II trial of peri-operative FOLFOX ± panitumumab. Similar to PETACC-8, while the trial was negative overall,(94) there was a trend towards improved recurrence-free survival (RFS) among RAS and BRAF wild-type patients in the experimental arm (HR 0.51; 95%CI, 0.24-1.10; P=0.09).(95) Moreover, the AREG/EREG-high subset gained significant 3-year RFS benefit from panitumumab (5/42 [11.9%] vs 8/32 [25.0%]; HR, 0.29; 95%CI, 0.09-0.99; P=0.047), with no benefit seen in the AREG/EREG-low subgroup (6/43 [14.0%] vs 7/40 [17.5%]; HR, 0.75; 95%CI, 0.25-2.22; P=0.60; P_interaction=0.33).(95) Prospective validation of this finding is now being planned.

7. Conclusion

It has long been understood that RAS wild-type status is necessary but not sufficient for benefit from anti-EGFR therapy. Improved understanding of the mechanisms underlying anti-EGFR sensitivity and primary and acquired resistance have led to advances in patient selection and have highlighted additional therapeutic targets. Novel mechanistic biomarkers are now allowing use of anti-EGFR therapy to be re-evaluated in settings where they have proved ineffective in non-biomarker selected populations, such as right PTL and locally advanced disease. Similarly, improved understanding of tumor dynamics is enabling rechallenge strategies to be assessed. With additional information from ctDNA, alternating use of a variety of targeted agents in response to such dynamic changes may lead to improved survival outcomes in future practice.