Biomarker-driven and molecular targeted therapies for colorectal cancers

Abstract

Improved clinical selection and identification of new molecules and innovative strategies have widened treatment options and increased overall survival in metastatic colorectal cancer patients in recent years. Biomarker-driven therapies represent an emerging issue in this field and new targeted treatments are under investigation and probably will be soon adopted into daily clinical practice. In the present review, the role RAS, BRAF mutations, Her2 amplification, microsatellite instability, and CpG island methylator phenotype are discussed according to their possible roles as prognostic, predictive markers, as well as possible biomarker-driven treatment options.

1. Introduction

The identification of biomarker-driven and molecularly targeted therapies represents an issue of critical importance in clinical oncology. Colorectal cancer (CRC) is one of the leading causes of cancer-related death worldwide [1]. However, in recent years, significant progress has been made and median survival for patients treated in modern clinical trials has surpassed 30 months [2,3].

Such improvement in observed survival is related on one hand to a better clinical selection that takes into account patient and tumor characteristics, such as age, performance status, comorbidities, symptoms, extent of disease, sidedness, patient preferences and expectations, and quality of life. On the other hand, the introduction of new molecules such as regorafenib and trifluridine-tipiracil, together with therapy using induction, maintenance, and re-induction periods, has widened treatment options. Lastly, the biological characterization of metastatic CRC (mCRC) has been recently expanded from the evaluation of KRAS exon 2 only to a more comprehensive panel of biological alterations to include additional characterization of KRAS, NRAS, and BRAF V600E mutations, microsatellite instability (MSI) status and HER amplification [4].

In terms of biological characterization, more and more evidence supports the concept of spatial and temporal molecular heterogeneity during CRC progression. This phenomenon is deeply connected with the improving role of “liquid biopsy” to evaluate circulating tumor DNA as a possible tool to assess the dynamic nature of tumor growth and progression.

In the present review, we discuss the most recent findings on targeted therapies in mCRC. We will describe the function of most promising genetic alterations in CRC and their possible role as prognostic, predictive markers, as well as possible biomarker-driven treatment options.

2. RAS as biomarker

2.1. RAS in the primary resistance to anti-EGFRs

The constitutive activation of RAS-RAF-MEK-ERK-MAP kinase pathway plays a pivotal role in the regulation of cellular responses to growth signals and is one of the signaling pathways through which the epidermal growth factor (EGF) influences cell differentiation and proliferation [5,6]. Cetuximab and panitumumab, two anti-epidermal growth factor receptor (EGFR) monoclonal antibodies, have been approved by the US Food and Drug Administration (FDA) >10 years ago for the treatment of mCRC, based on the results of randomized trials proving their efficacy in advanced lines of treatment [7,8]. No predictive biomarkers for anti-EGFR efficacy were known at the time of these studies, but exploratory analysis of progression-free survival (PFS) results demonstrated a subgroup of patients that clearly did not derive benefit from such treatment.

The turning point on biomarker selection for anti-EGFR therapy was the identification that KRAS exon 2 (codons 12 and 13) mutations were negative predictive markers of response to anti-EGFR therapy. KRAS is a small GTPase protein member of the RAS family [9,10], and its constitutive activation, gained through somatic gene mutations, results in cell proliferation and survival independent of external cellular signals [11]. KRAS mutations can be found in approximately 40%–50% of mCRCs and more frequently involve exon 2 [12,13]. The evidence of the predictive role of KRAS exon 2 mutation came initially from retrospective series [14] and was afterwards confirmed in post hoc analyses of phase III randomized trial [7,13,15–17]. Following these results, FDA and European Medicines Agency (EMA) in 2008 restricted the use of anti-EGFRs to patients with KRAS exon 2 wild-type mCRC tumors.

Despite a high specificity, an independent meta-analysis [18] showed a very low sensitivity for KRAS exon 2 mutations in predicting acquired resistance to anti-EGFR treatment, pointing out the potential existence of additional negative predictive biomarkers. Rare RAS activating mutations, involving exon 3 (codons 59 and 61) and exon 4 (codons 117 and 146) of KRAS and exons 2, 3, and 4 of NRAS were then identified as possible negative predictive factors [19–21]. Definitive evidence came from the extended RAS analyses data from the large randomized phase III PRIME trial. In this study, comparing FOLFOX with or without panitumumab as first-line therapy in mCRC patients, a detrimental effect of the addition of panitumumab was observed in patients carrying any RAS mutations (hazard ratio [HR] for PFS: 1.31 [ P = .008, P for interaction <.002]; HR for overall survival [OS]: 1.21 [ P = .04, P for interaction = .001]) [22]. Based on these results, regulatory agencies restricted the use of anti-EFGRs to RAS wild-type (exons 2,3, and 4 of each KRAS and NRAS ) tumors [23].

Following these findings, the outcome s of all recent randomized trials with anti-EGFR agents were re-evaluated according to extended RAS mutational status [24–26] and additional meta-analyses were performed. Consistently among different anti-EGFR agents, lines of therapy, and chemotherapy backbones, no improvement in PFS or OS was observed with the addition of anti-EGFRs for tumors harboring any RAS mutation ( P >.05) [27]. A potential detrimental effect from the addition of anti-EGFRs to oxaliplatin-based chemotherapy in mutated patients was also confirmed. Of note, when evaluated in the selected extended RAS wild-type population, efficacy endpoints of the trials reached striking results. In the phase II randomized PEAK trial, evaluating a head-to-head comparison of panitumumab versus bevacizumab in association with FOLFOX in the first-line setting [28], no differences were seen in PFS (10.9 months in the panitumumab arm v 10.1 months in the bevacizumab arm) and overall response rate (ORR) (57.8% v 53.5%, respectively) when analyzing the KRAS exon 2 wild-type population. When the extended mutational panel was considered, however, there was a more notable difference among patients with RAS wild-type tumors with an increase in ORR to 64% versus 60% in the panitumumab and bevacizumab arms, respectively. Similarly, PFS reached 13.0 versus 9.5 months (HR 0.65, 95% confidence interval [CI] 0.44–0.96, P = .029), and OS overtook 40 months in the selected all wild-type population (41.3 v 28.9 months, HR 0.63, 95% CI 0.39–1.02, P = .058).

Even among patients with RAS wild-type tumors, it is well known that a portion of patients still do not benefit from anti-EGFR agents, underlining the existence of additional primary resistance mechanisms. Several genes involved in primary resistance to anti-EGFRs have been identified in RAS wild-type mCRC, based on preclinical data and retrospective evaluations, including MET amplification, HER2 amplification, phosphatidylinositide-3-kinase ( PIK3CA ) mutations (exon 9 and 20 hotspot mutations), FGFR1 and PDGFRA mutations, low EGFR copy number, and loss of PTEN function [29]. Nevertheless, further validation is needed prior to the application of these biomarkers to standard clinical practice. To overcome primary resistance to anti-EGFRs, different combined strategies and new targeted agents are currently under study, such as combination with mammalian target of rapamycin (mTOR) inhibitors or anti-HER2 therapies [30,31].

2.2. Secondary resistance to anti-EGFRs

Secondary resistance to anti-EGFR agents is often related to clonal selection induced by targeted treatment pressure [32]. Mutations in the RAS-RAF-MAPK pathway can be identified at progression in patients previously diagnosed with a KRAS wild-type tumor and multiple mutations can coexist at the same time [33,34]. Mutations involving the extracellular domain of EGFR represent another mechanism of resistance, which develops only in the acquired setting [35,36]. Retrospective analyses from the ASPECCT trial, a non-inferiority study comparing panitumumab to cetuximab in chemorefractory patients, revealed that EGFR S492R mutations occurred in 16% versus 1% of patients treated with cetuximab and panitumumab, respectively [37]. Interestingly, EGFR S492R mutations seem to confer resistance to cetuximab, but not panitumumab [33]. Several trials are investigating different approaches to multiple target blockade to overcome acquired resistance to anti-EGFRs, such as the combination of anti-EGFRs with MEK or MET inhibitors [31].

The use of liquid biopsies and the analysis of circulating tumor DNA (ctDNA) might have a pivotal role in the future in refining molecular selection and assisting treatment strategies by providing a means of early detection of primary and secondary resistance and allowing a dynamic molecular profiling of patients [38,39]. Serial ctDNA analyses in patients undergoing treatment with EGFR inhibitors have shown, in fact, the emergence of RAS and/or BRAF mutations during EGFR blockade in KRAS WT patients, with an increase during anti-EGFR administration and a rapid decline after treatment withdrawal, leading to a possible regain of drug sensitivity [38,40]. Based on these findings, rechallenge strategies after treatment breaks in patients with RAS wild-type tumors that demonstrated an initial response to anti-EGFR agents are currently under study. At this time, extensive investigation and prospective validation is still needed prior to implementation of serial ctDNA measurement in clinical practice.

2.3. Targeting RAS mutant tumors

A new frontier in its early development is represented by treatment strategies targeting RAS-mutated tumors. Recent preclinical evidence demonstrated that the combination of cetuximab with BKM120, a PIK3CA inhibitor, was able to reduce cell proliferation in a concentration-dependent manner in human CRC cell lines with KRAS mutations. Additionally, combination treatment with cetuximab and BKM120 significantly reduced the growth of xenograft tumors originating from KRAS-mutant cells compared with cetuximab alone (P = .034) [41]. No data are available on the safety and efficacy of the combination in clinical studies, but the use of PIK3CA inhibitors is a promising new arena to overcome anti-EGFR resistance in KRAS-mutant CRC. In a different setting, Verissimo et al evaluated the efficacy of dual inhibition of the EGFR-MEK-ERK pathway in RAS-mutant CRC organoids. The dual target blockade was found to be able to induce a transient cell-cycle arrest rather than cell death; moreover, in vivo drug response of xenotransplanted RAS mutant organoids confirmed a similar growth arrest upon pan-HER/MEK combination therapy. These results showed preliminary evidence of preclinical activity of multiple target combined inhibition in RAS-mutated CRC, and demonstrate the potential of patient-derived CRC organoid libraries in evaluating inhibitors and drug combinations in a preclinical setting [42]. The combined inhibition of pan-HER and MEK is currently tested in patients with RAS mutant cancers, including mCRC, in several clinical trials, as well as the combined inhibition of MEK and ERK [43–46].

In preclinical models, high levels of vitamin C were able to kill cultured human KRAS mutant CRC cells due to an alteration of metabolic pathways related to reactive oxygen species [47]. A phase III trial evaluating the efficacy of the addition of ascorbic acid to standard first-line chemotherapy has been planned [48]. Another strategy showing promising preliminary results is the combination of cobimetinib (a MEK inhibitor) with atezolizumab (MPDL3280A), an engineered antibody that inhibits binding of PDL1 to its receptors, PD-1 and B7.1. This combination in a cohort of 23 chemorefractory mCRC (22 KRAS mutant, 1 WT) showed an ORR of 17% (four partial responses [PRs], five stable disease [SD]) and a 6-months OS of 72% with a safe toxicity profile. Of note, three responders were mismatch repair-proficient, and one was unknown [49]. A phase III study evaluating this combination in chemorefractory mCRC is currently open and recruiting, with a target of 50% enrollment of RAS-mutant tumors [50]. Results of these studies are highly anticipated.

3. BRAF as biomarker

3.1. Clinical and prognostic features of BRAF-mutant tumors

RAF is another key player in the EGFR-mediated downstream signaling pathway. Alongside MEK and ERK, RAF forms a three-tiered cytosolic tyrosine kinase signaling cascade activated by RAS, which affects cell growth, proliferation, and differentiation [51]. RAF activation is also able to regulate apoptosis, cell migration and survival (through the regulation of other pathways such as BCL-2, RHO small GTPases, and HIPPO) [52]. BRAF signaling is known to be implicated as an oncogenic driver in approximately 8%–10% of CRCs51. The majority of BRAF mutations involve the substitution of glutamic acid for valine at residue 600 (V600E) within the protein kinase domain, which accounts for approximately 80% of BRAF gene mutations in CRC. BRAF V600E mutation leads to constitutive activation of the MEK-ERK-MAP kinase downstream pathway even in absence of functional RAS. Working through the same signaling cascade RAS and BRAF V600E mutations are considered mutually exclusive drivers of oncogene addiction, and their concomitant detection is extremely rare (<.001%) [53].

The negative prognostic value of BRAF mutation in mCRC is currently well established. When retrospectively evaluated, in fact, BRAF-mutant patients were observed to have a median OS of less than 12 months across multiple series [54–56]. Moreover, BRAF-mutant tumors share distinct clinical and pathological characteristics: they are more frequent in women than men; are often right-sided; mucinous histology and microsatellite instability (MSI)-high; more advanced disease with preferential spread to lymph nodes and peritoneum [57]. A distinct gene signature [58] and a specific carcinogenesis pathway [59] have also been associated with the presence of BRAF V600E mutation. When liver metastases are radically resected, BRAF-mutated tumors often relapse early, with the occurrence of extra-hepatic lesions [60,61]. Notably, when BRAF status was assessed in a prospective cohort of unselected mCRC patients, a higher BRAF mutation incidence (21%) was reported, and the presence of BRAF mutation was found to be associated with a high rate of poor performance status (39% performance status 2–4) and advanced age (37% age >75 years), thus underlining that BRAF-mutant patients are probably underrepresented in clinical trials [62].

In contrast with BRAF V600E mutation, rare mutations of BRAF codons 594 and 596, occurring in less than 1% of CRCs, have also recently been shown to have different prognostic and clinical implications. These mutations were found to be associated with a rectal primary tumor location, non-mucinous histology, microsatellite stability, and lack of peritoneal disease. Moreover, although the number of patients was small, no negative prognostic impact was observed (BRAF 594 or 596 mutant v BRAF V600E median OS 62.0 v 12.6 months; HR 0.36; 95% CI 0.20–0.64; P = .002) [63].

3.2. BRAF mutation as predictive factor to anti-EGFRs

Though still under debate, in addition to its prognostic value growing evidence is accumulating on the role of BRAF mutations as biomarker of resistance to anti-EGFR monoclonal antibodies [64]. Retrospective analyses from a large data set of 733 chemorefractory mCRC patients included in the European Consortium analyses showed that the response rate to cetuximab with or without chemotherapy was 8.3% (2/24) versus 38.0% (124/326) in BRAF-mutant versus wild-type patients, respectively (OR 0.15; 95% CI 0.02–0.51; P = .0012) [21]. On the other hand, however, in several subgroup analyses of phase III trials, the presence of the BRAF V600E mutation failed to demonstrate a predictive value, thought to be because of the small number of BRAF-mutant patients and lack of statistical power [22,65]. Nevertheless, two recent meta-analyses showed no improvement in the outcomes (both in terms of PFS and OS) of patients with BRAF-mutated mCRC when treated with either panitumumab or cetuximab combination treatment versus chemotherapy alone [66,67]. Based on these observations, anti-EGFRs do not demonstrate a clear outcome benefit in BRAF-mutant patients, and should be restricted to patients with no other alternative therapeutic options.

3.3. Treatment options for BRAF-mutant tumors

To date, in the first-line setting, FOLFOXIRI plus bevacizumab represents the most promising chemotherapeutic treatment option in clinically selected BRAF-mutated patients [2,68]. Data from the molecular subgroup analyses of the Italian phase III TRIBE study, investigating FOLFOXIRI plus bevacizumab versus FOLFIRI plus bevacizumab as first-line treatment, have shown, in fact, an unprecedented median OS of 19.0 months (HR 0.54, 95% CI 0.24–1.20) in the BRAF-mutated population in the FOLFOXIRI plus bevacizumab arm, with a treatment effect not significantly different across molecular subgroups (P interaction = .52).

Outcomes of BRAF-mutated patients, however, are largely unsatisfactory, underlining the need for further improvement and optimization of treatment strategies, alongside the development of biological agents and targeted therapies. Many studies have been conducted in the last few years aiming at the identification of possible effective BRAF inhibitors for mCRC patients. Vemurafenib and dabrafenib are two targeted agents that have been FDA-approved for the treatment of patients with BRAF-mutant melanoma. However, in marked contrast to the results seen in melanoma, single-agent vemurafenib did not show significant activity in patients with BRAF-mutant mCRC [69]. One possible explanation is that survival of BRAF-mutant mCRCs may not be as dependent on the MAPK signaling pathway as BRAF-mutant melanoma, relying on alternative or parallel signaling pathways that can maintain proliferation or survival such as the WNT/β-catenin or the PI3K pathway, as well as a possible incomplete MAPK pathway inhibition by BRAF inhibitors alone in BRAF-mutant mCRC or the development of EGFR-mediated feedback reactivation pathways following BRAF inhibition [70,71]. Thus, dual blockade of BRAF and alternative survival pathways have been tested in clinical trials.

The combination of a BRAF inhibitor (dabrafenib) and a MEK inhibitor (trametinib) [72] was firstly tested in a phase I/II clinical trial. Overall 43 patients with BRAF V600 mutant mCRC were enrolled. Five (12%) achieved a response, including one complete response with response duration >36 months, and 24 patients (56%) achieved stable disease. Another strategy investigated in several trials involve the combination of BRAF and EGFRs inhibitors, including vemurafenib with either cetuximab [73] or panitumumab [74], encorafenib plus cetuximab [75], and dabrafenib plus panitumumab [76]. Thus far, these combinations appear to be well tolerated, and many of these approaches are showing promising initial results, with a response rate ranging from 4%–22%. A substantial percentage of patients, however, still fail to respond. A possible resistance mechanism to combined inhibition in BRAF-mutant mCRC cell lines has been shown to derive from MET overexpression or amplification [77,78], thus targeting both EGFR-dependent and EGFR-independent resistance mechanisms in BRAF-mutant tumors seems to be a rational approach to optimize targeted treatment. Since MEK inhibitors act downstream of BRAF, adding a MEK inhibitor to a BRAF/EGFR inhibitor combination is expected to improve treatment efficacy through a better inhibition of MAPK pathway. Promising preliminary results from 83 BRAF-mutated mCRC patients treated with the triple combination of dabrafenib, panitumumab, and trametinib showed 18% and 67% rates of confirmed complete response (CR)/PR and SD, respectively, and a median OS at the time of data presentation of 9.1 months (95% CI 7.6–20.0) with acceptable tolerability and treatment toxicity profile [79]. A second triple-targeted inhibition involving the addition of alpelisib (BYL719), a PI3Kα-specific inhibitor, to the BRAF inhibitor encorafenib and the anti-EGFR antibody cetuximab has also been evaluated. A total of 102 patients was randomized in the study (triplet n = 52; doublet n = 50), a planned PFS analysis comparing the triplet to the doublet after 73 events showed a HR of 0.69 (95% CI 0.43–1.11; P = .064) with median PFS of 5.4 months (95% CI 4.1–7.2) for the triple inhibition and a confirmed ORR of 27% (95% CI 16%–41%). With 35 events, interim OS analysis (triplet v doublet) showed a HR of 1.21 (95% CI 0.61–2.39), with a median OS of 15.2 months for the triplet [73]. Based on this evidence, the phase III multicenter randomized BEACON trial is currently ongoing to evaluate the combination of encorafenib and cetuximab plus or minus binimetinib (a MEK inhibitor) versus investigator’s choice of either irinotecan/cetuximab or FOLFIRI/cetuximab in patients with BRAF V600E mCRC progressed after one or two prior regimens in the metastatic setting [80].

Combining standard cytotoxic chemotherapy with dual target inhibition is an additional strategy to increase the activity of BRAF inhibitor combinations. The first of such trials evaluated the combination of vemurafenib and cetuximab with irinotecan in BRAF-mutant mCRCs. In this phase Ib trial, six of 17 evaluable patients (35%) achieved a radiographic response and median PFS was 7.7 months [81]. In the Southwest Oncology Group (SWOG) 16 trial the addition of vemurafenib to cetuximab and irinotecan has been evaluated. Promising results were released at ASCO GI 2017: patients (N = 49) receiving the three-drug combination achieved a median PFS of 4.4. months compared to 2.2 months in those (N = 50) receiving the two-drug combination, with an HR of 0.42 (95% CI 0.26–0.66, P = .0 0 02) [82].

Several other promising targeted therapies designed to block resistance pathways to BRAF inhibitors or to overcome common acquired resistance mechanisms to BRAF inhibitor combinations are currently under investigation. Among those, ERK inhibitors represent a therapeutic class that is likely to have an important role in future targeted therapy combinations for BRAF-mutant mCRC [83,84]. As underlined above, BRAF mutation is often associated with MSI, the role of immunotherapy in the treatment of MSI-H mCRC, will be discussed in a separate paragraph.

4. HER2 a promising biomarker in mCRC

4.1. HER2 in preclinical models

The role of human epidermal growth factor receptor 2 (HER2/neu) as driver oncogene in CRC and as potential biomarker for targeted treatment in the metastatic setting is a relatively new concept. HER2 is a member of the EGRF family. Its activation stimulates the RAS-RAF-ERK and the PI3K-PTEN-AKT pathways regulating cell proliferation and apoptosis.

First data emerged in 2011 when possible determinants of resistance to anti-EGFRs and new therapeutic targets were investigated in a large mCRC patient-derived xenograft cohort. HER2 amplification was detected in a subset of cetuximab-resistant, KRAS/NRAS/BRAF/PIK3CA wild-type cases, thus possibly acting as negative predictive factor to anti-EGFR. Even more interestingly, a proof-of-concept study in the subgroup of HER2-amplified xenopatients showed tumor regression in cases treated with a combination of anti-HER2 and anti-EGFR [85].

4.2. HER2 in clinical trials

Above-reported results were subsequently challenged in mCRC patients in the Italian HERACLES study. First of all, in the HERACLES diagnostic the evaluation on more than 1,0 0 0 cases allowed to identify strict criteria for the definition of HER2 amplification/positivity [86] specifically for mCRC. Indeed, prior experiences reported a wide range of HER2 expression rates in CRC [87,88], due to selection bias and to the use of different diagnostic methods and scoring systems [89–92]. The anti-tumor activity of trastuzumab and lapatinib was then evaluated in patients with HER2-positive chemorefractory mCRC patients in the phase II trial. Among 914 KRAS wild-type screened patients, 48 (5%) were identified with HER2-positive tumours and 27 were eligible for the trial. A RR of 30% was reached, so the primary end point of the trial was met with one patient achieving a complete response and overall high rates of response duration and no unexpected adverse events [93]. Moving from such promising results the HERACLES trials is still ongoing enrolling patients to receive pertuzumab plus TDM1. In the HERACLES rescue trial, patients experiencing disease progression after trastuzumab and lapatinib are receiving TDM1 monotherapy. Confirmatory results of HER2 as possible target in mCRC were also provided by the phase II MyPathway trial. Response rate among 34 HER2-positive patients receiving trastuzumab plus pertuzumab was 38.2% with a median duration of response of 10.3 weeks.

Data on HER2 as possible predictive factor of resistance to anti-EGFRs derived from retrospective series with similar results. Among 170 KRAS wild-type patients receiving cetuximab and irinotecan, HER2-positive cases demonstrated worse PFS, OS as compared to HER2-negative cases (HR for PFS 3.65 [95%CI 1.57–8.46], P = .0026; HR for OS 5.05 [95% CI 2.17–11.77], P = .0 0 02). Such results were further confirmed in a more recent experience [94]. Moreover, HER2 amplification detected on tissue or on ctDNA plasma samples was identified as possible mechanism of acquired resistance in RAS/BRAF wild-type, HER2-negative patients progressed during anti-EGFRs [95].

Based on a strong preclinical rationale and supported by clinical confirmations HER2 evaluation might become a routine test in patients with mCRC candidate to receive anti-EGFRs and/or anti-HER2 treatments. The identification of specific criteria for HER2 amplification detection in CRC will allow a greater consistency of data in the near future.

5. Microsatellite instability

5.1. Definition and testing

MSI is the result of inactivation of the DNA mismatch repair (MMR) system and is characterized by a high frequency of frameshift mutations in microsatellite DNA. A portion of MSI tumors is due to germline mutations in one of the MMR genes (MLH1, MSH2, MSH6, or PMS2), which results in hereditary Lynch syndrome. However, the majority (80%) of MSI cases are sporadic, often due to hypermethylation of the MLH1 gene promoter [96,97]. Sporadic cases are often associated with BRAF V600E mutations, CIMP [98]. The remaining cases of sporadic MSI are largely explained by multiple somatic mutations in the MMR genes. These recently characterized “double somatic” MSI cases have two or more MMR gene mutations resulting in MSI, but no identifiable germline MMR mutation [99]. Double somatic colorectal cancer is associated with a higher frequency of somatic mutations in PIK3CA [100].

Initially, germline MSI (Lynch syndrome) was largely recognized by clinical criteria. However, half of Lynch syndrome patients diagnosed by an identified germline mutation fail to meet Amsterdam criteria [97]. These clinical criteria have largely been replaced by two main tumor-based approaches that can detect both Lynch syndrome patients and sporadic cases of MSI. DNA MSI testing is a polymerase chain reaction (PCR)-based approach often using a validated 10-marker panel, where the presence of instability at 30% or more loci is considered diagnostic of MSI, while <30% is considered microsatellite stable (MSS). Alternatively or in conjunction, immunohistochemical staining of MLH1, MSH2, MSH6, and PMS2 can be performed where loss of expression represents MMR deficiency (dMMR), a surrogate for MSI [101]. If either MSI or dMMR is detected, further evaluation is recommended to rule out Lynch syndrome, rather than sporadic MSI.

5.2. MSI and prognosis

MSI is widely accepted as a positive prognostic factor in CRC and MSI CRC patients have a better prognosis than those with stage-matched MSS [102]. Similarly, the frequency of MSI is greater in earlier stage CRC as compared to advanced CRC, with MSI comprising only ~5% of mCRC cases compared to 12% in localized cases [103].

5.3. MSI as a predictive biomarker

MSI is also useful in guiding treatment planning. Interpretation of retrospective data on MSI in terms of negative predictive factor to 5-fluorouracil (5-FU)–based adjuvant treatment is not unequivocal [104–110]. The most solid data derive from the ACCENT database, where the impact of MSI in stage II and III CRC among 17 adjuvant trials comparing surgery alone versus surgery followed by 5-FU–based therapy was assessed. Patients with stage II and III MSI tumors receiving surgery alone showed better outcomes than those with MSS (5-year OS for stage II: 90 v 78%; HR 0.37, 95% CI 0.17–0.81, P = .013; 5-year OS for stage III: 59% v 54%; HR 0.84, 95% CI 0.49–1.43, P = .51). In stage III CRC patients, a significant survival benefit from the addition of 5-FU monotherapy following surgery was seen both in patients with MSS (5-year time to recurrence [TTR] = 64% v 47%) and MSI tumors (5-year TTR = 72% v 60%). In conclusion, the authors suggest that chemotherapy is not recommended for patients with stage II MSI tumors due to their excellent prognosis, while stage III patients should receive adjuvant treatment irrespective of MSI status [111]. The etiology of the MSI may affect the predicted benefit from 5-FU. When Sinicrope et al retrospectively evaluated stage II and III colon cancer patients who received adjuvant 5-FU or placebo, individuals with MSI CRCs due to germline mutations (ie, Lynch syndrome) versus those with sporadic MSI tumors, individuals with Lynch syndrome-associated cancers did have improved disease-free survival with 5-FU [112]. This suggests that understanding the molecular profile of a CRC and the etiology of the observed molecular features are important for determining the utility of predictive biomarkers. It is currently unknown whether double somatic MSI cancers have differential benefit from modern chemotherapy regimens.

The role of MSI as a predictive marker within modern combination regimens, such as FOLFOX and FOLFIRI, has less evidence [113–115]. The lack of prediction capability for MSI in combination regimens as compared to single-agent 5-FU may be partly due to variability in multivariate models, as newer models often account for a greater number of relevant variables, such as KRAS and BRAF status [116]. Doublet chemotherapy treatment may also abrogate the prognostic differences in MSI and MSS CRCs. For example, the added benefit from oxaliplatin in individuals with MSS tumors may attenuate the improved prognosis associated with MSI status [117]. Although MSI was retrospectively shown to predict improved disease-free survival with adjuvant irinotecan and 5-FU (IFL regimen) in the CALGB (Alliance) 89803 trial, these results were inconsistently demonstrated in other exploratory analyses [118–120]. Thus, MSI is largely accepted as a prognostic marker, but its role as a predictive biomarker for combination chemotherapy is more controversial.

5.4. MSI as key driver of immunotherapy response

Recently, MSI assessment gained a prominent role in the metastatic setting due to the promising, although preliminary, results derived from immunotherapy trials. In the phase II KEYNOTE 016 trial, the activity of pembrolizumab in patients with refractory metastatic tumors was tested. Among 28 MSI mCRC patients RR, disease control rate (DCR), median PFS, and median OS were significantly improved compared to MSS patients (RR 50% v 0%, and DCR 89% vs 16%, respectively; HR for PFS 0.135, P <.001, HR for OS 0.247, P = .001) [121,122]. Median PFS as well as median OS was not reached at the time of data presentation for dMMR tumors. Based on these results pembrolizumab is currently under investigation in pretreated patients (KEYNOTE 164) [123] and in the first-line setting (KEYNOTE 177) [124].

The combination of the anti-CTLA4 ipilimumab and the anti-PD1 nivolumb is under investigation in the CheckMate-142 trial [125]. Preliminary results demonstrated a RR of 25.5% in patients receiving nivolumab (N = 47) and 33.3% in those receiving combination ipilimumab plus nivolumab (N = 27) [125]. Data regarding a larger cohort of patients treated with nivolumab only (N=72) showed encouraging results for RR, 12-month PFS rate, and 12-month OS rate (31%, 48.4%, and 73.8%, respectively). Responses were observed regardless of tumor or immune cell PD-L1 expression, BRAF, KRAS mutation status, or clinical history of Lynch syndrome. Centrally revised data identified two patients experiencing CR.

Of note, both pembrolizumab monotherapy and ipilimumab/nivolumab combination treatment showed a manageable toxicity profile. Moreover, similar to previous experience with immunotherapy in melanoma, survival curves clearly showed a trend towards a plateau in the tail of the curves, suggesting the unprecedented possibility of long-term responders in a setting of pretreated chemorefractory mCRC patients.

The high activity of immune checkpoint inhibitors agents in MSI patients might be explained by the identification in MSI tumor of a high burden of somatic mutations that can be recognized by the patient’s immune system. As a supplementary proof, MSI-high tumors were found to be characterized by a dense immune infiltration and a cytokine-rich environment [126]. If the available preliminary results of phase II trials are validated, immunotherapy treatment of MSI-high mCRC is expected to become one of the main therapeutic tools for these patients.

6. CpG island methylator phenotype

6.1. Definition and evaluation

CpG island methylator phenotype (CIMP) is found in approximately 20% of CRCs and is defined by an exceptionally high frequency of aberrantly methylated CpG islands across the genome [127]. In normal cells, CG-rich regions of the DNA are normally unmethylated. These so-called CpG islands are found in approximately half of gene promoters. In contrast, in CRC cells, these CpG islands are often aberrantly methylated. CIMP CRCs are distinct from other CRCs because they have an exceptionally higher frequency of CpG island methylation, which is thought to promote carcinogenesis by regulating transcription and silencing tumor-suppressor genes [128].

The potential of CIMP as a biomarker has been limited by a lack of consensus regarding which assay to use [127]. The most common panel is the Weisenberger panel, which evaluates methylation at the CACNA1G, IGF2, NEUROG1, RUNX3, and SOCS1 loci [59]. At each locus, a percent of methylated reference (PMR) is calculated, with PMR ≥4% typically accepted as defining methylation [129]. However, assays with additional loci, such as MLH1, or different thresholds, such as PMR ≥10%, are also frequently seen in the literature [130,131]. Furthermore, while most commonly a binary classification scheme is used with “CIMP-positive” defined as methylation at ≥3/5 tested loci, a three-group scheme may also be selected, dividing the CIMP-negative group into CIMP-low (methylation at 1–2/5 loci) and CIMP negative (no loci are methylated). Overall, the lack on uniformity of the CIMP assay has hindered the uptake of CIMP as a useful biomarker in CRC.

6.2. CIMP as prognostic marker

Despite the lack of a “gold-standard” CIMP assay, there does appear to be an association between CIMP and poor prognosis [127,132,133]. However, even when using a consistent assay, this has not been universally demonstrated [134], which may be explained by population differences as the frequency of CIMP in CRC has been noted in 8%–33% of patients, while an international collaborative analysis reported a frequency of 18% [135,136]. Additionally, the inclusion/exclusion of rectal primaries, staging system, and addition of molecular markers (eg, BRAF) to statistical models could all influence the interpretation of CIMP as a significant, independent prognostic biomarker.

6.3. CIMP and chemotherapy

The initial studies evaluating CIMP as a predictive biomarker focused on the interaction with 5-FU. Despite many analyses, the data remain conflicted whether CIMP-positive tumors receive benefit from 5-FU [132,137–141]. However, the majority of these results suggest no benefit or an adverse response to 5-FU–based adjuvant chemotherapy in patients with CIMP-positive tumors.

On the other hand, CIMP may be a useful biomarker for combination chemotherapy. In a recent exploratory retrospective analysis of a randomized trial of 5-FU/leucovorin (5-FU/LV) alone versus with irinotecan (IFL regimen) in stage III colon cancer patients (CALGB/Alliance 89803), Shiovitz et al found that patients with CIMP-positive colon cancers that were also MMR-intact did better with IFL than with 5-FU/LV alone (5-year OS of 66% v 46%, respectively) [133]. Furthermore, in this updated data analysis from this trial, CIMP was a stronger prognostic feature than MSI. Other studies assessing the use of CIMP to predict response to FOLFOX therapy did not observe CIMP to be a significant predictor [135]. At this time, CIMP appears to have strong potential to be a prognostic marker for CRC, but its use as a predictive marker for conventional chemotherapy will require further investigation and validation of the observed benefit for irinotecan is needed.

Given the biologic mechanism of hypermethylation, therapy targeting hypermethylation has been gaining traction for CIMP-positive CRCs. There is preclinical evidence for the use of hypomethylating agents in CIMP CRC [142]. Azacitadine is an inhibitor of histone deacetylase (HDAC) and potent hypomethylating agent. However, even when used in conjunction with CAPOX chemotherapy, there was no correlation between response rate or other efficacy end points and CIMP positivity [143]. This novel area may benefit from additional investigation into the synergy between hypomethylating agents and cytotoxic chemotherapy, including dosing, timing, and patient selection. Second-generation hypomethylating agents, ie, SG110, are also under investigation. Early-phase studies suggest an improved safety profile compared to first-generation hypomethylating agents. Many trials investigating combinations of hypomethylating agents and chemotherapy or immunomodulatory agents are ongoing and results are strongly awaited.

7. Conclusions

For years, most efforts in translational research aimed at the identification of biomarkers able to drive treatment selection. In the last few years, increasing research has demonstrated that CRC is not one unique entity, but rather molecularly heterogeneous. This heterogeneity, both among CRCs and within a CRC, likely affects the observed variability in treatment responses, especially in the metastatic setting. Thus, the early and accurate identification of biological features such as RAS, BRAF mutations, HER2 amplification, and MSI is crucial in order to drive treatment choices from the time of diagnosis of metastatic disease. Table 1 summarizes the above-reported main findings.

Table 1.

Main findings on presented biomarkers.

Biomarker	Required in clinical practice	Prognostic value	Predictive value	Therapeutic options under investigation
RAS mutations	Yes	Mild	Strong (resistance to anti-EGFRs)	Anti-EGFR + anti-PIK3CA Anti-panHER + anti-MEK Anti-ERK + anti-MEK Cobimetinib + atezolizumab
BRAF V600E mutation	Yes	Strong	Mild (resistance to anti-EGFRs)	Anti-BRAF + anti-MEK Anti-BRAF + anti-EGFRs Anti-BRAF + anti-EGFRs + antiMEK Anti-BRAF + anti-EGFRs + anti-PIK3CA Anti-BRAF + anti-EGFRs + chemotherapy
HER2	No	–	Mild (resistance to anti-EGFRs)	Anti-Her2 combinations
MSI	Yes	Strong	Low (chemotherapy in early stage)	Anti-PDL1 Anti-PDL1 + anti CTLA4
CIMP	No	Low	Low	Hypomethylating agents

The correct initial assessment of the above reported pathological features is of clinical significance. Only certified and expert pathologists and/or molecular biologists should provide results. The report must include the adopted technique, a description of the analyzed tissue origin (primary or metastasis) and of percentage of available tumor cells. Moreover, the final results must be clearly and univocally stated. In cases of doubt of misleading information clinical oncologist should cooperate with the pathologist and/or molecular biologist to interpret the provided results. Biomarker evaluations need to be standardized since variability of results might limit the applicability and consistency of these biomarkers to the clinical setting, with the HER2 experience representing a good example of cooperative validation for the definition of a biomarker.

Once a specific molecular feature is identified, it is important to tailor the best treatment option for the patient and to evaluate the opportunity of inclusion in a clinical trial. Indeed, the most exciting results presented in this review derive from ongoing clinical trials. Promising new areas include HER2-targeted therapy, immunotherapy for MSI patients, and epigenetic therapy for CIMP. Unfortunately, no conclusive data are available for targeted treatment in BRAF mutant tumors. From a general perspective, future clinical trials should include biomarker information in the trial design and interpretation to better tailor therapy to an individual patient.

Finally, it is important to note that during the course of the disease, treatments might exert selective pressure leading to the acquisition of new molecular features that are known to be possible causes of acquired resistance to therapies. Liquid biopsies have been proposed as possible tool able to catch such dynamisms, but are not currently adopted in the clinical practice. However, the collection of plasma samples from ongoing clinical trials is a more and more adopted practice and prospective/confirmatory data on liquid biopsies are eagerly awaited.