ESMO 2017 Special Report:
Emerging Prognostic and Predictive Biomarkers in Colorectal Cancer

Colorectal cancer (CRC) is a heterogeneous disease characterized by diverse clinical outcomes and responses to cancer therapy. Biomarkers are critical for developing individualized treatment plans by providing a better understanding of the disease course (prognostic biomarkers) and anticipated therapeutic response (predictive biomarkers).

At the European Society for Medical Oncology (ESMO) 2017 Congress, held September 8-12, 2017, in Madrid, Spain, researchers presented new data on multiple biomarker—including consensus molecular subtypes (CMS), BRAF and KRAS mutation status, microsatellite-instability (MSI) status, epidermal growth factor receptor (EGFR) copy number, and other genetic, molecular, and clinical characteristics. Scroll down to read more about these emerging prognostic and predictive biomarkers in CRC.

Prognostic biomarkers

  • Developing a Low-Cost, Accessible Tool for Assigning CMS Category in CRC
  • Differential Prognostic Impact of BRAF and KRAS Mutations Across Patient Subgroups
  • New Tumor Biomarker Profile Identifies High Relapse Risk in Stage II-III CRC

Predictive biomarkers

  • EGFR Copy Number Predicts Anti-EGFR Benefit in RAS Wild-Type Advanced CRC
  • CMS Predicts Bevacizumab Benefit in First-Line Metastatic Colorectal Cancer

Developing a Low-Cost, Accessible Tool for Assigning CMS Category in CRC

In 2015, the international CRC Subtyping Consortium (CRCSC) introduced a new classification system for CRC.1 The consensus molecular subtypes (CMS) define 4 tumor subtypes with distinct genetic, prognostic, and clinical features (Table 1).1,2 Approximately 13% of CRC tumors display mixed features of more than one CMS category, suggesting intratumoral heterogeneity or a transition phenotype.1

 

Despite its promising clinical role in CRC tumor assessment, the CMS classification system has been limited to date by the lack of a cost-effective assay. Angura Sadanandam, PhD, and colleagues from The Institute of Cancer Research in London, UK, provided an update on the development of a low-cost multi-gene platform for identifying CMS subtypes.3

Refining an established gene panel

Dr. Sadanandam and colleagues previously described a 786-gene panel, the CRCAssigner-786, that reliably stratifies CRC tumors into CMS categories.4 Identifying a CMS subtype requires microarray and RNA-Seq platforms, which are costly, time-consuming, and impractical for routine clinical use.5

The first step in reducing the complexity of CMS classification involved reducing the standard 786-gene signature to a shorter panel of 38 genes (CRCAssigner-38). Next, the team developed a custom assay (NanoCRCAssigner) to assess tumor subtypes based on the 38-gene panel. The NanoCRCAssigner utilizes the nCounter platform, a low-cost and highly reproducible tool for molecular analysis. The research team them compared the performance of the standard (microarray/RNA-Seq) and low-cost (NanoCRCAssigner) platforms in assigning CMS subtypes to a validation cohort of primary untreated CRC samples (n = 17).

High concordance with CMS subtypes

Results showed a high concordance (>70%) between the standard and low-cost protocols for stratifying CRC tumors into CMS subtypes (Pearson’s correlation coefficient, 0.98). The CRCAssigner-38 panel was further validated in additional fresh-frozen CRC tumor samples (n = 51) and formalin-fixed, paraffin-embedded (FFPE) CRC tumor samples (n = 24).

In summary, these findings demonstrate the potential feasibility of stratifying CRC tumors into CMS categories using a 38-gene assay and fresh-frozen or FFPE tumor samples. The next step for this research group involves additional validation of the 38-gene panel in larger cohorts of CRC tumor samples, with the goal of bringing an inexpensive biomarker assay for identifying CMS category into clinical practice.

Return to top

Differential Prognostic Impact of BRAF and KRAS Mutations Across Patient Subgroups

The prognostic implications of BRAF and KRAS mutations in patients with CRC appear to vary by tumor microsatellite instability (MSI) status and CMS classification, according to new findings from a prospective study.6 Jørgen Smeby, MD, and colleagues from Oslo University Hospital in Oslo, Norway, examined the potential implications of these differences on the clinical management of CRC.

The study included the gene expression profiles of 1,197 primary tumors from a series of consecutive patients (i.e., the Oslo series) who were treated surgically for stage I-IV CRC. The analysis focused on hotspots for KRAS and BRAF mutations:

  • KRAS exon 2: codon 12 and 13
  • KRAS exon 3: codon 61
  • BRAF exon 15: codon 600

In total, 339 tumors (31%) had KRAS mutations, 192 (17%) had BRAF mutations, and 570 (52%) were wild-type for both mutations. All tumors were analyzed for MSI status, and a subgroup of 317 tumors from the Oslo series underwent additional genetic profiling using exon-level microarrays. The study population was further enriched with data from 514 patients with CRC in the publicly available GSE39582 dataset.

Interaction with MSI status and CMS category

Across all patients with CRC, the presence of KRAS and BRAF mutations predicted significantly worse overall survival relative to wild-type tumors (Table 2). When survival outcomes were examined by MSI status, however, KRAS and BRAF mutations retained their negative prognostic impact only in those patients with microsatellite stable (MSS) tumors. In contrast, these mutations had no prognostic significance in patients with MSI tumors.

In the next phase of the analysis, the research team identified another layer of interaction between KRAS and BRAF mutation status, MSS tumors, and CMS category. Results showed that KRAS mutations predicted significantly worse OS outcomes relative to KRAS wild-type tumors only in those patients with MSS tumors classified as CMS2 (HR, 1.60; 95% CI, 1.11-2.30; p = .011). In the BRAF analysis, BRAF mutations predicted worse OS relative to wild-type tumors only in those patients with MSS tumors classified as CMS1 (HR, 4.96; 95% CI, 1.74-14.12; p = .003).

Overall findings from this study reveal the presence of tumor subtype-specific prognostic associations, with KRAS mutations predicting worse outcomes in MSS tumors in CMS2, and BRAF mutations predicting worse outcomes in MSS tumors in CMS1. If validated in larger trials, these findings may have important implications for the evaluation and biomarker-directed management of patients with CRC.

Return to top

New Tumor Biomarker Profile Identifies High Relapse Risk in Stage II-III CRC

Researchers have validated a new biomarker profile—the combination of increased CD8+ T-cell infiltration and very high PD-L1 expression—that identifies 10% of patients with stage II and III CRC who have a high risk of relapse despite the frequent presence of MSI, which is typically associated with a favorable prognosis.7 Marwan Fakih, MD, and colleagues from the City of Hope National Medical Center in Duarte, California, presented findings from the study.

Biomarker profile discovery

To identify new prognostic markers, the research team first examined the expression levels of multiple tumor markers in samples from patients with stage III colon cancer who underwent primary tumor resection at the City of Hope Comprehensive Cancer Center between 1989 and 2014. The discovery cohort included 35 patients with disease recurrence within 5 years of treatment (cases), and 36 patients without recurrent disease after 5 years (controls).

Fakih and colleagues stratified patients into 3 groups based on the relative density of CD8+ T cells and PD-L1+ cells within the tumor microenvironment. In this risk-stratification model, ‘CD8-high’ was defined as intratumor infiltration of CD8+ T cells above the median, and ‘PD-L1-high’ was defined as PD-L1 expression exceeding the 90th percentile. Most of the PD-L1 expression in the tumor microenvironment occurred on tumor macrophages. Based on these expression profiles, relapse rates in the 3 groups were:

  • CD8-high/PD-L1-low: 7%
  • CD8-high/PD-L1-high: 100%
  • CD8-low/irrespective of PD-L1: 80%

Of note, 4 of 7 patients (57%) in the CD8-high/PD-L1-high group also had MSI-high (MSI-H) or mismatch repair-deficient (dMMR) tumors. These findings suggest that not all patients with MSI-H/dMMR tumors have favorable outcomes in stage III CRC.

Biomarker profile validation

Next, the researchers validated the biomarker signature in patients with stage II/III CRC from 2 major oncology databases: 385 cases from The Cancer Genome Atlas (TCGA) and 828 cases from the National Center for Biotechnology Information-Gene Expression Omnibus (NCBI-GEO). In both validation cohorts, patients with CD8-high/PD-L1-high tumors had worse clinical outcomes than those with CD8-high/PD-L1-low tumors. As with the discovery cohort, more than 50% of patients in the CD8-high/PD-L1-high groups also had MSI-H or dMMR tumors.

In the TCGA database, 12% of patients with stage II/III CRC had CD8-high/PD-L1-high tumors, and 6.9% of these were MSI-H tumors. Patients with CD8-high/PD-L1-high tumors or CD8-low tumors had significantly worse OS at 5 years than patients with CD8-high/PD-L1-low disease (HR, 3.56; p = .0095). The 5-year OS rates in each group were:

  • CD8-high/PD-L1-low: 86.1%
  • CD8-high/PD-L1-high: 59.6%
  • CD8-low/irrespective of PD-L1: 57.0%

In the NCBI-GEO database, the CD8-high/PD-L1-high tumor profile accounted for 9.7% of all cases of stage II/III CRC. Among CD8-high/PD-L1-high tumors, 57.5% were MSI-H. Recurrence-free survival was significantly worse at 5 years for patients with CD8-high/PD-L1-high disease (6.9%) or CD8-low disease (67.1%) relative to those with CD8-high/PD-L1-low disease (73.6%; HR, 1.65; p = .02).

Implications for practice

The combination of high levels of CD8+ T cell infiltration and very high PD-L1 expression defines a subpopulation of patients with stage II-III CRC who are at increased risk of relapse. Further, low CD8+ T cell infiltration is associated with worse clinical outcomes, regardless of PD-L1 expression.

Importantly, these data demonstrate that not all patients with MSI-H/dMMR tumors will do well, challenging the role of MSI-H/dMMR status as a biomarker of good prognosis in CRC. Examining the tumor CD8/PD-L1 expression profile may be a useful tool for stratifying patients with MSI-H/dMMR tumors into lower-risk and higher-risk categories, as well as identifying patients with stage II-III CRC who may be candidates for anti-PD-1 immunotherapy.

Return to top

EGFR Copy Number Predicts Anti-EGFR Benefit in RAS Wild-Type Advanced CRC

Tumor EGFR copy number is emerging as a potential clinical tool for predicting panitumumab benefit in patients with RAS wild-type advanced CRC. Jenny Seligmann, MBChB, PhD, and colleagues from Leeds Institute of Cancer and Pathology, University of Leeds, Leeds, UK, presented results of the preplanned EGFR copy number analysis from the phase III PICCOLO trial.8

The challenge: identifying patients who benefit from panitumumab

The PICCOLO trial compared irinotecan plus panitumumab with irinotecan alone in 460 patients with advanced CRC who progressed after fluoropyrimidine treatment, with or without oxaliplatin, and had no prior exposure to anti-EGFR therapy.9 Among all patients with KRAS wild-type tumors, the addition of panitumumab to irinotecan did not improve overall survival (OS) (HR, 1.01; p = .91).9 These preliminary trial findings, reported in 2013, highlighted the need for better tools to identify patients who might benefit from anti-EGFR therapy. Prospectively planned biomarker studies from PICCOLO now offer promising options.

The first biomarker analysis from PICCOLO identified a correlation between anti-EGFR treatment benefit and the EGFR ligands epiregulin (EREG) and amphiregulin (AREG).10 Among patients with RAS wild-type tumors, high EGFR ligand expression predicted a significant improvement in progression-free survival (PFS) with panitumumab compared with irinotecan alone (8.3 months vs. 4.4 months; HR, 0.38; p < .001). Patients with low EGFR ligand expression, by comparison, gained no PFS benefit from adding panitumumab (3.2 months vs. 4.0 months; HR, 0.93; p = .73).10 “The problem with EGFR ligand expression as a biomarker is that EGFR ligand expression is measured on a continuous spectrum,” Dr. Seligmann said. “It is difficult to find the cut-off point where tumors are clearly ‘positive’ or ‘negative’.”

Current study design: EGFR copy number

In the current study, Dr. Seligmann and colleagues examined EGFR copy number in 275 patients from the PICCOLO trial who had sufficient tumor tissue for DNA analysis. The study population included 219 patients with RAS wild-type tumors. All patients were stratified into 2 groups:

  • EGFR gain (>2 copies)
  • EGFR normal (2 copies)

In total, 196 patients (71.3%) had EGFR gain, and 79 patients (28.7%) had normal EGFR copy number. The analysis ruled out EGFR copy number as a general prognostic marker in advanced CRC. Among patients treated with irinotecan alone, there was no correlation between EGFR copy number status and clinical outcome, including PFS (HR, 1.0; p = .98) or OS (HR, 0.97; p = .87).

EGFR copy number gain correlated strongly with higher EGFR ligand expression, the established marker of panitumumab benefit. Compared with EGFR normal tumors, those with EGFR gain had significantly higher levels of expression of AREG (p = .0006) and EREG (p < .0001), and a trend toward higher expression of HER3 (p = .067).

Role as a predictive marker of panitumumab benefit

Results of the analysis support the role of EGFR copy number as a predictive marker of panitumumab benefit. Among those with RAS wild-type tumors, EGFR gain predicted a significant improvement in PFS with panitumumab compared with irinotecan alone (HR, 0.60; p = .002). In contrast, however, the addition of panitumumab had no effect on PFS in patients with RAS wild-type tumors and normal EGFR copy number (HR, 1.23; p = .45).

Additional results showed that the utility of EGFR copy number may extend beyond those with RAS wild-type tumors. Across the entire study population, including patients with RAS and BRAF mutations, EGFR copy number consistently differentiated between patients who were and were not likely to benefit from panitumumab (p = .03 for interaction).

In summary, EGFR copy number is consistent with EGFR ligand expression as a potential predictive marker of panitumumab benefit. In contrast with EGFR ligand expression, however, EGFR copy number offers a clearly defined binary option (gain vs. normal) that may be more useful in the clinical setting. Importantly, normal EGFR copy number may identify up to one-third of patients with RAS wild-type tumors who are not likely to benefit from treatment with panitumumab.

Return to top

CMS Predicts Bevacizumab Benefit in First-Line Metastatic Colorectal Cancer

For patients with stage IV mCRC, tumor CMS category appears to be a useful biomarker for predicting response to first-line bevacizumab.11 Jennifer Mooi, MD, of the Olivia Newton-John Cancer Research Institute in Heidelberg, Victoria, Australia, and colleagues presented results from a biomarker substudy of the phase III MAX clinical trial.11

In 2010, the Australasian Gastrointestinal Trials Group presented primary results from the MAX trial showing that bevacizumab significantly improved PFS by approximately 40% when added to capecitabine, with or without mitomycin, in the first-line treatment of mCRC.12 In the present study, Dr. Mooi and colleagues evaluated the role of CMS category in predicting which patients with mCRC were more likely to benefit from the addition of bevacizumab to standard first-line chemotherapy.

CMS category and overall survival

The MAX trial enrolled 471 patients with previously untreated, unresectable mCRC. The analysis of tumor CMS category showed the following distribution: 18% CMS1; 47% CMS2; 12% CMS3; and 23% CMS4. As expected with a well-defined prognostic biomarker, CMS correlated with survival irrespective of treatment arm (log-rank p = .006). The median OS (95% CI) for each patient subgroup was:

  • CMS1: 8.8 months (6.5-16.0 months)
  • CMS2: 24.2 months (19.1-27.4 months)
  • CMS3: 17.6 months (11.3-24.6 months)
  • CMS4: 21.4 months (15.8-23.1 months)

CMS category and bevacizumab benefit

Results showed a significant interaction between CMS group and treatment response (p = .03). Patients with CMS2 and CMS3 tumors were likely to benefit from the addition of bevacizumab to capecitabine, whereas patients with CMS1 and CMS4 tumors were less likely to benefit from bevacizumab (Table 3).

Results from the CMS substudy of the phase III MAX trial confirm the prognostic value of tumor CMS category in patients with mCRC, with CMS1 and CMS2 predicting reduced and prolonged survival, respectively. Moreover, results support the potential role of CMS as a predictive biomarker of bevacizumab benefit. Future research in this area will focus on validating the predictive value of CMS, as well as understanding the biological basis driving the preferential benefit of bevacizumab in CMS2 and CMS3 tumors in patients with mCRC.

Return to top

Summary

Biomarkers play a critical role in the evaluation and management of patients with CRC. Prognostic biomarkers enable clinicians to identify which patients will have a more aggressive disease course, while predictive biomarkers support clinical decision-making by clarifying which patients are most likely to benefit—and which patients are not expected to benefit—from specific therapies.

To test your skills in biomarker interpretation in CRC, click here to begin an interactive, case-based, CME-certified activity, Evolving Standards in Colorectal Cancer: Case-Based Insights from ESMO 2017.

References

  1. Guinney J, Dienstmann R, Wang X, et al. The consensus molecular subtypes of colorectal cancer. Nat Med. 2015;21:1350-1356. https://www.ncbi.nlm.nih.gov/pubmed/26457759
  2. Thanki K, Nicholls ME, Gajjar A, et al. Consensus molecular subtypes of colorectal cancer and their clinical implications. Int Biol Biomed J. 2017;3:105-111. https://www.ncbi.nlm.nih.gov/pubmed/28825047
  3. Sadanandam A, Eason K, Fontana E, et al. Development and validation of multiplex biomarker assay to stratify colorectal cancer (CRC) patient samples into subtypes. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 562P. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  4. Sadanandam A, Lyssiotis CA, Homicsko K, et al. A colorectal cancer classification system that associates cellular phenotype and responses to therapy. Nat Med. 2013;19:619-625. https://www.ncbi.nlm.nih.gov/pubmed/23584089
  5. Ragulan C, Eason K, Nyamundanda G, et al. A low-cost multiplex biomarker assay stratifies colorectal cancer patient samples into clinically-relevant subtypes. bioRxiv. 2017:Epub ahead of print. https://www.biorxiv.org/content/early/2017/08/16/174847
  6. Smeby J, Sveen A, Bergsland CH, et al. Prognostic impact of BRAF and KRAS mutations according to the consensus molecular subtypes of colorectal cancer. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 548P. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  7. Fakih M, Ouyang C, Wang C, et al. High PD-L1 expression and high CD8+ T-cell infiltration identifies a new subpopulation of colorectal cancer with high risk of relapse and poor outcome. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 558P. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  8. Seligmann J, Wood H, Richman S, et al. Epidermal growth factor receptor (EGFR) copy number as a biomarker of prognosis and panitumumab benefit in RAS-wt advanced colorectal cancer. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 545P. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  9. Seymour MT, Brown SR, Middleton G, et al. Panitumumab and irinotecan versus irinotecan alone for patients with KRAS wild-type, fluorouracil-resistant advanced colorectal cancer (PICCOLO): a prospectively stratified randomised trial. Lancet Oncol. 2013;14:749-759. https://www.ncbi.nlm.nih.gov/pubmed/23725851
  10. Seligmann JF, Elliott F, Richman SD, et al. Combined epiregulin and amphiregulin expression levels as a predictive biomarker for panitumumab therapy benefit or lack of benefit in patients with RAS wild-type advanced colorectal cancer. JAMA Oncol. 2016;2:633-642. https://www.ncbi.nlm.nih.gov/pubmed/26867820
  11. Mooi J, Wirapati P, Asher R, et al. Consensus molecular subtypes (CMS) as predictors of benefit from bevacizumab in first line treatment of metastatic colorectal cancer: Retrospective analysis of the MAX clinical trial. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 479O. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  12. Tebbutt NC, Wilson K, Gebski VJ, et al. Capecitabine, bevacizumab, and mitomycin in first-line treatment of metastatic colorectal cancer: results of the Australasian Gastrointestinal Trials Group Randomized Phase III MAX Study. J Clin Oncol. 2010;28:3191-3198. https://www.ncbi.nlm.nih.gov/pubmed/20516443

ESMO 2017 Special Report:
Emerging Prognostic and Predictive Biomarkers in Colorectal Cancer

Colorectal cancer (CRC) is a heterogeneous disease characterized by diverse clinical outcomes and responses to cancer therapy. Biomarkers are critical for developing individualized treatment plans by providing a better understanding of the disease course (prognostic biomarkers) and anticipated therapeutic response (predictive biomarkers).

At the European Society for Medical Oncology (ESMO) 2017 Congress, held September 8-12, 2017, in Madrid, Spain, researchers presented new data on multiple biomarker—including consensus molecular subtypes (CMS), BRAF and KRAS mutation status, microsatellite-instability (MSI) status, epidermal growth factor receptor (EGFR) copy number, and other genetic, molecular, and clinical characteristics. Scroll down to read more about these emerging prognostic and predictive biomarkers in CRC.

Prognostic biomarkers

Predictive biomarkers

Developing a Low-Cost, Accessible Tool for Assigning CMS Category in CRC

In 2015, the international CRC Subtyping Consortium (CRCSC) introduced a new classification system for CRC.1 The consensus molecular subtypes (CMS) define 4 tumor subtypes with distinct genetic, prognostic, and clinical features (Table 1).1,2 Approximately 13% of CRC tumors display mixed features of more than one CMS category, suggesting intratumoral heterogeneity or a transition phenotype.1

 

Despite its promising clinical role in CRC tumor assessment, the CMS classification system has been limited to date by the lack of a cost-effective assay. Angura Sadanandam, PhD, and colleagues from The Institute of Cancer Research in London, UK, provided an update on the development of a low-cost multi-gene platform for identifying CMS subtypes.3

Refining an established gene panel

Dr. Sadanandam and colleagues previously described a 786-gene panel, the CRCAssigner-786, that reliably stratifies CRC tumors into CMS categories.4 Identifying a CMS subtype requires microarray and RNA-Seq platforms, which are costly, time-consuming, and impractical for routine clinical use.5

The first step in reducing the complexity of CMS classification involved reducing the standard 786-gene signature to a shorter panel of 38 genes (CRCAssigner-38). Next, the team developed a custom assay (NanoCRCAssigner) to assess tumor subtypes based on the 38-gene panel. The NanoCRCAssigner utilizes the nCounter platform, a low-cost and highly reproducible tool for molecular analysis. The research team them compared the performance of the standard (microarray/RNA-Seq) and low-cost (NanoCRCAssigner) platforms in assigning CMS subtypes to a validation cohort of primary untreated CRC samples (n = 17).

High concordance with CMS subtypes

Results showed a high concordance (>70%) between the standard and low-cost protocols for stratifying CRC tumors into CMS subtypes (Pearson’s correlation coefficient, 0.98). The CRCAssigner-38 panel was further validated in additional fresh-frozen CRC tumor samples (n = 51) and formalin-fixed, paraffin-embedded (FFPE) CRC tumor samples (n = 24).

In summary, these findings demonstrate the potential feasibility of stratifying CRC tumors into CMS categories using a 38-gene assay and fresh-frozen or FFPE tumor samples. The next step for this research group involves additional validation of the 38-gene panel in larger cohorts of CRC tumor samples, with the goal of bringing an inexpensive biomarker assay for identifying CMS category into clinical practice.

Return to top

Differential Prognostic Impact of BRAF and KRAS Mutations Across Patient Subgroups

The prognostic implications of BRAF and KRAS mutations in patients with CRC appear to vary by tumor microsatellite instability (MSI) status and CMS classification, according to new findings from a prospective study.6 Jørgen Smeby, MD, and colleagues from Oslo University Hospital in Oslo, Norway, examined the potential implications of these differences on the clinical management of CRC.

The study included the gene expression profiles of 1,197 primary tumors from a series of consecutive patients (i.e., the Oslo series) who were treated surgically for stage I-IV CRC. The analysis focused on hotspots for KRAS and BRAF mutations:

  • KRAS exon 2: codon 12 and 13
  • KRAS exon 3: codon 61
  • BRAF exon 15: codon 600

In total, 339 tumors (31%) had KRAS mutations, 192 (17%) had BRAF mutations, and 570 (52%) were wild-type for both mutations. All tumors were analyzed for MSI status, and a subgroup of 317 tumors from the Oslo series underwent additional genetic profiling using exon-level microarrays. The study population was further enriched with data from 514 patients with CRC in the publicly available GSE39582 dataset.

Interaction with MSI status and CMS category

Across all patients with CRC, the presence of KRAS and BRAF mutations predicted significantly worse overall survival relative to wild-type tumors (Table 2). When survival outcomes were examined by MSI status, however, KRAS and BRAF mutations retained their negative prognostic impact only in those patients with microsatellite stable (MSS) tumors. In contrast, these mutations had no prognostic significance in patients with MSI tumors.

In the next phase of the analysis, the research team identified another layer of interaction between KRAS and BRAF mutation status, MSS tumors, and CMS category. Results showed that KRAS mutations predicted significantly worse OS outcomes relative to KRAS wild-type tumors only in those patients with MSS tumors classified as CMS2 (HR, 1.60; 95% CI, 1.11-2.30; p = .011). In the BRAF analysis, BRAF mutations predicted worse OS relative to wild-type tumors only in those patients with MSS tumors classified as CMS1 (HR, 4.96; 95% CI, 1.74-14.12; p = .003).

Overall findings from this study reveal the presence of tumor subtype-specific prognostic associations, with KRAS mutations predicting worse outcomes in MSS tumors in CMS2, and BRAF mutations predicting worse outcomes in MSS tumors in CMS1. If validated in larger trials, these findings may have important implications for the evaluation and biomarker-directed management of patients with CRC.

Return to top

New Tumor Biomarker Profile Identifies High Relapse Risk in Stage II-III CRC

Researchers have validated a new biomarker profile—the combination of increased CD8+ T-cell infiltration and very high PD-L1 expression—that identifies 10% of patients with stage II and III CRC who have a high risk of relapse despite the frequent presence of MSI, which is typically associated with a favorable prognosis.7 Marwan Fakih, MD, and colleagues from the City of Hope National Medical Center in Duarte, California, presented findings from the study.

Biomarker profile discovery

To identify new prognostic markers, the research team first examined the expression levels of multiple tumor markers in samples from patients with stage III colon cancer who underwent primary tumor resection at the City of Hope Comprehensive Cancer Center between 1989 and 2014. The discovery cohort included 35 patients with disease recurrence within 5 years of treatment (cases), and 36 patients without recurrent disease after 5 years (controls).

Fakih and colleagues stratified patients into 3 groups based on the relative density of CD8+ T cells and PD-L1+ cells within the tumor microenvironment. In this risk-stratification model, ‘CD8-high’ was defined as intratumor infiltration of CD8+ T cells above the median, and ‘PD-L1-high’ was defined as PD-L1 expression exceeding the 90th percentile. Most of the PD-L1 expression in the tumor microenvironment occurred on tumor macrophages. Based on these expression profiles, relapse rates in the 3 groups were:

  • CD8-high/PD-L1-low: 7%
  • CD8-high/PD-L1-high: 100%
  • CD8-low/irrespective of PD-L1: 80%

Of note, 4 of 7 patients (57%) in the CD8-high/PD-L1-high group also had MSI-high (MSI-H) or mismatch repair-deficient (dMMR) tumors. These findings suggest that not all patients with MSI-H/dMMR tumors have favorable outcomes in stage III CRC.

Biomarker profile validation

Next, the researchers validated the biomarker signature in patients with stage II/III CRC from 2 major oncology databases: 385 cases from The Cancer Genome Atlas (TCGA) and 828 cases from the National Center for Biotechnology Information-Gene Expression Omnibus (NCBI-GEO). In both validation cohorts, patients with CD8-high/PD-L1-high tumors had worse clinical outcomes than those with CD8-high/PD-L1-low tumors. As with the discovery cohort, more than 50% of patients in the CD8-high/PD-L1-high groups also had MSI-H or dMMR tumors.

In the TCGA database, 12% of patients with stage II/III CRC had CD8-high/PD-L1-high tumors, and 6.9% of these were MSI-H tumors. Patients with CD8-high/PD-L1-high tumors or CD8-low tumors had significantly worse OS at 5 years than patients with CD8-high/PD-L1-low disease (HR, 3.56; p = .0095). The 5-year OS rates in each group were:

  • CD8-high/PD-L1-low: 86.1%
  • CD8-high/PD-L1-high: 59.6%
  • CD8-low/irrespective of PD-L1: 57.0%

In the NCBI-GEO database, the CD8-high/PD-L1-high tumor profile accounted for 9.7% of all cases of stage II/III CRC. Among CD8-high/PD-L1-high tumors, 57.5% were MSI-H. Recurrence-free survival was significantly worse at 5 years for patients with CD8-high/PD-L1-high disease (6.9%) or CD8-low disease (67.1%) relative to those with CD8-high/PD-L1-low disease (73.6%; HR, 1.65; p = .02).

Implications for practice

The combination of high levels of CD8+ T cell infiltration and very high PD-L1 expression defines a subpopulation of patients with stage II-III CRC who are at increased risk of relapse. Further, low CD8+ T cell infiltration is associated with worse clinical outcomes, regardless of PD-L1 expression.

Importantly, these data demonstrate that not all patients with MSI-H/dMMR tumors will do well, challenging the role of MSI-H/dMMR status as a biomarker of good prognosis in CRC. Examining the tumor CD8/PD-L1 expression profile may be a useful tool for stratifying patients with MSI-H/dMMR tumors into lower-risk and higher-risk categories, as well as identifying patients with stage II-III CRC who may be candidates for anti-PD-1 immunotherapy.

Return to top

EGFR Copy Number Predicts Anti-EGFR Benefit in RAS Wild-Type Advanced CRC

Tumor EGFR copy number is emerging as a potential clinical tool for predicting panitumumab benefit in patients with RAS wild-type advanced CRC. Jenny Seligmann, MBChB, PhD, and colleagues from Leeds Institute of Cancer and Pathology, University of Leeds, Leeds, UK, presented results of the preplanned EGFR copy number analysis from the phase III PICCOLO trial.8

The challenge: identifying patients who benefit from panitumumab

The PICCOLO trial compared irinotecan plus panitumumab with irinotecan alone in 460 patients with advanced CRC who progressed after fluoropyrimidine treatment, with or without oxaliplatin, and had no prior exposure to anti-EGFR therapy.9 Among all patients with KRAS wild-type tumors, the addition of panitumumab to irinotecan did not improve overall survival (OS) (HR, 1.01; p = .91).9 These preliminary trial findings, reported in 2013, highlighted the need for better tools to identify patients who might benefit from anti-EGFR therapy. Prospectively planned biomarker studies from PICCOLO now offer promising options.

The first biomarker analysis from PICCOLO identified a correlation between anti-EGFR treatment benefit and the EGFR ligands epiregulin (EREG) and amphiregulin (AREG).10 Among patients with RAS wild-type tumors, high EGFR ligand expression predicted a significant improvement in progression-free survival (PFS) with panitumumab compared with irinotecan alone (8.3 months vs. 4.4 months; HR, 0.38; p < .001). Patients with low EGFR ligand expression, by comparison, gained no PFS benefit from adding panitumumab (3.2 months vs. 4.0 months; HR, 0.93; p = .73).10 “The problem with EGFR ligand expression as a biomarker is that EGFR ligand expression is measured on a continuous spectrum,” Dr. Seligmann said. “It is difficult to find the cut-off point where tumors are clearly ‘positive’ or ‘negative’.”

Current study design: EGFR copy number

In the current study, Dr. Seligmann and colleagues examined EGFR copy number in 275 patients from the PICCOLO trial who had sufficient tumor tissue for DNA analysis. The study population included 219 patients with RAS wild-type tumors. All patients were stratified into 2 groups:

  • EGFR gain (>2 copies)
  • EGFR normal (2 copies)

In total, 196 patients (71.3%) had EGFR gain, and 79 patients (28.7%) had normal EGFR copy number. The analysis ruled out EGFR copy number as a general prognostic marker in advanced CRC. Among patients treated with irinotecan alone, there was no correlation between EGFR copy number status and clinical outcome, including PFS (HR, 1.0; p = .98) or OS (HR, 0.97; p = .87).

EGFR copy number gain correlated strongly with higher EGFR ligand expression, the established marker of panitumumab benefit. Compared with EGFR normal tumors, those with EGFR gain had significantly higher levels of expression of AREG (p = .0006) and EREG (p < .0001), and a trend toward higher expression of HER3 (p = .067).

Role as a predictive marker of panitumumab benefit

Results of the analysis support the role of EGFR copy number as a predictive marker of panitumumab benefit. Among those with RAS wild-type tumors, EGFR gain predicted a significant improvement in PFS with panitumumab compared with irinotecan alone (HR, 0.60; p = .002). In contrast, however, the addition of panitumumab had no effect on PFS in patients with RAS wild-type tumors and normal EGFR copy number (HR, 1.23; p = .45).

Additional results showed that the utility of EGFR copy number may extend beyond those with RAS wild-type tumors. Across the entire study population, including patients with RAS and BRAF mutations, EGFR copy number consistently differentiated between patients who were and were not likely to benefit from panitumumab (p = .03 for interaction).

In summary, EGFR copy number is consistent with EGFR ligand expression as a potential predictive marker of panitumumab benefit. In contrast with EGFR ligand expression, however, EGFR copy number offers a clearly defined binary option (gain vs. normal) that may be more useful in the clinical setting. Importantly, normal EGFR copy number may identify up to one-third of patients with RAS wild-type tumors who are not likely to benefit from treatment with panitumumab.

Return to top

CMS Predicts Bevacizumab Benefit in First-Line Metastatic Colorectal Cancer

For patients with stage IV mCRC, tumor CMS category appears to be a useful biomarker for predicting response to first-line bevacizumab.11 Jennifer Mooi, MD, of the Olivia Newton-John Cancer Research Institute in Heidelberg, Victoria, Australia, and colleagues presented results from a biomarker substudy of the phase III MAX clinical trial.11

In 2010, the Australasian Gastrointestinal Trials Group presented primary results from the MAX trial showing that bevacizumab significantly improved PFS by approximately 40% when added to capecitabine, with or without mitomycin, in the first-line treatment of mCRC.12 In the present study, Dr. Mooi and colleagues evaluated the role of CMS category in predicting which patients with mCRC were more likely to benefit from the addition of bevacizumab to standard first-line chemotherapy.

CMS category and overall survival

The MAX trial enrolled 471 patients with previously untreated, unresectable mCRC. The analysis of tumor CMS category showed the following distribution: 18% CMS1; 47% CMS2; 12% CMS3; and 23% CMS4. As expected with a well-defined prognostic biomarker, CMS correlated with survival irrespective of treatment arm (log-rank p = .006). The median OS (95% CI) for each patient subgroup was:

  • CMS1: 8.8 months (6.5-16.0 months)
  • CMS2: 24.2 months (19.1-27.4 months)
  • CMS3: 17.6 months (11.3-24.6 months)
  • CMS4: 21.4 months (15.8-23.1 months)

CMS category and bevacizumab benefit

Results showed a significant interaction between CMS group and treatment response (p = .03). Patients with CMS2 and CMS3 tumors were likely to benefit from the addition of bevacizumab to capecitabine, whereas patients with CMS1 and CMS4 tumors were less likely to benefit from bevacizumab (Table 3).

Results from the CMS substudy of the phase III MAX trial confirm the prognostic value of tumor CMS category in patients with mCRC, with CMS1 and CMS2 predicting reduced and prolonged survival, respectively. Moreover, results support the potential role of CMS as a predictive biomarker of bevacizumab benefit. Future research in this area will focus on validating the predictive value of CMS, as well as understanding the biological basis driving the preferential benefit of bevacizumab in CMS2 and CMS3 tumors in patients with mCRC.

Return to top

Summary

Biomarkers play a critical role in the evaluation and management of patients with CRC. Prognostic biomarkers enable clinicians to identify which patients will have a more aggressive disease course, while predictive biomarkers support clinical decision-making by clarifying which patients are most likely to benefit—and which patients are not expected to benefit—from specific therapies.

To test your skills in biomarker interpretation in CRC, click here to begin an interactive, case-based, CME-certified activity, Evolving Standards in Colorectal Cancer: Case-Based Insights from ESMO 2017.

References

  1. Guinney J, Dienstmann R, Wang X, et al. The consensus molecular subtypes of colorectal cancer. Nat Med. 2015;21:1350-1356. https://www.ncbi.nlm.nih.gov/pubmed/26457759
  2. Thanki K, Nicholls ME, Gajjar A, et al. Consensus molecular subtypes of colorectal cancer and their clinical implications. Int Biol Biomed J. 2017;3:105-111. https://www.ncbi.nlm.nih.gov/pubmed/28825047
  3. Sadanandam A, Eason K, Fontana E, et al. Development and validation of multiplex biomarker assay to stratify colorectal cancer (CRC) patient samples into subtypes. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 562P. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  4. Sadanandam A, Lyssiotis CA, Homicsko K, et al. A colorectal cancer classification system that associates cellular phenotype and responses to therapy. Nat Med. 2013;19:619-625. https://www.ncbi.nlm.nih.gov/pubmed/23584089
  5. Ragulan C, Eason K, Nyamundanda G, et al. A low-cost multiplex biomarker assay stratifies colorectal cancer patient samples into clinically-relevant subtypes. bioRxiv. 2017:Epub ahead of print. https://www.biorxiv.org/content/early/2017/08/16/174847
  6. Smeby J, Sveen A, Bergsland CH, et al. Prognostic impact of BRAF and KRAS mutations according to the consensus molecular subtypes of colorectal cancer. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 548P. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  7. Fakih M, Ouyang C, Wang C, et al. High PD-L1 expression and high CD8+ T-cell infiltration identifies a new subpopulation of colorectal cancer with high risk of relapse and poor outcome. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 558P. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  8. Seligmann J, Wood H, Richman S, et al. Epidermal growth factor receptor (EGFR) copy number as a biomarker of prognosis and panitumumab benefit in RAS-wt advanced colorectal cancer. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 545P. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  9. Seymour MT, Brown SR, Middleton G, et al. Panitumumab and irinotecan versus irinotecan alone for patients with KRAS wild-type, fluorouracil-resistant advanced colorectal cancer (PICCOLO): a prospectively stratified randomised trial. Lancet Oncol. 2013;14:749-759. https://www.ncbi.nlm.nih.gov/pubmed/23725851
  10. Seligmann JF, Elliott F, Richman SD, et al. Combined epiregulin and amphiregulin expression levels as a predictive biomarker for panitumumab therapy benefit or lack of benefit in patients with RAS wild-type advanced colorectal cancer. JAMA Oncol. 2016;2:633-642. https://www.ncbi.nlm.nih.gov/pubmed/26867820
  11. Mooi J, Wirapati P, Asher R, et al. Consensus molecular subtypes (CMS) as predictors of benefit from bevacizumab in first line treatment of metastatic colorectal cancer: Retrospective analysis of the MAX clinical trial. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 479O. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  12. Tebbutt NC, Wilson K, Gebski VJ, et al. Capecitabine, bevacizumab, and mitomycin in first-line treatment of metastatic colorectal cancer: results of the Australasian Gastrointestinal Trials Group Randomized Phase III MAX Study. J Clin Oncol. 2010;28:3191-3198. https://www.ncbi.nlm.nih.gov/pubmed/20516443

ESMO 2017 Special Report:
Emerging Prognostic and Predictive Biomarkers in Colorectal Cancer

Colorectal cancer (CRC) is a heterogeneous disease characterized by diverse clinical outcomes and responses to cancer therapy. Biomarkers are critical for developing individualized treatment plans by providing a better understanding of the disease course (prognostic biomarkers) and anticipated therapeutic response (predictive biomarkers).

At the European Society for Medical Oncology (ESMO) 2017 Congress, held September 8-12, 2017, in Madrid, Spain, researchers presented new data on multiple biomarker—including consensus molecular subtypes (CMS), BRAF and KRAS mutation status, microsatellite-instability (MSI) status, epidermal growth factor receptor (EGFR) copy number, and other genetic, molecular, and clinical characteristics. Scroll down to read more about these emerging prognostic and predictive biomarkers in CRC.

Prognostic biomarkers

Predictive biomarkers

Developing a Low-Cost, Accessible Tool for Assigning CMS Category in CRC

In 2015, the international CRC Subtyping Consortium (CRCSC) introduced a new classification system for CRC.1 The consensus molecular subtypes (CMS) define 4 tumor subtypes with distinct genetic, prognostic, and clinical features (Table 1).1,2 Approximately 13% of CRC tumors display mixed features of more than one CMS category, suggesting intratumoral heterogeneity or a transition phenotype.1

 

Despite its promising clinical role in CRC tumor assessment, the CMS classification system has been limited to date by the lack of a cost-effective assay. Angura Sadanandam, PhD, and colleagues from The Institute of Cancer Research in London, UK, provided an update on the development of a low-cost multi-gene platform for identifying CMS subtypes.3

Refining an established gene panel

Dr. Sadanandam and colleagues previously described a 786-gene panel, the CRCAssigner-786, that reliably stratifies CRC tumors into CMS categories.4 Identifying a CMS subtype requires microarray and RNA-Seq platforms, which are costly, time-consuming, and impractical for routine clinical use.5

The first step in reducing the complexity of CMS classification involved reducing the standard 786-gene signature to a shorter panel of 38 genes (CRCAssigner-38). Next, the team developed a custom assay (NanoCRCAssigner) to assess tumor subtypes based on the 38-gene panel. The NanoCRCAssigner utilizes the nCounter platform, a low-cost and highly reproducible tool for molecular analysis. The research team them compared the performance of the standard (microarray/RNA-Seq) and low-cost (NanoCRCAssigner) platforms in assigning CMS subtypes to a validation cohort of primary untreated CRC samples (n = 17).

High concordance with CMS subtypes

Results showed a high concordance (>70%) between the standard and low-cost protocols for stratifying CRC tumors into CMS subtypes (Pearson’s correlation coefficient, 0.98). The CRCAssigner-38 panel was further validated in additional fresh-frozen CRC tumor samples (n = 51) and formalin-fixed, paraffin-embedded (FFPE) CRC tumor samples (n = 24).

In summary, these findings demonstrate the potential feasibility of stratifying CRC tumors into CMS categories using a 38-gene assay and fresh-frozen or FFPE tumor samples. The next step for this research group involves additional validation of the 38-gene panel in larger cohorts of CRC tumor samples, with the goal of bringing an inexpensive biomarker assay for identifying CMS category into clinical practice.

Return to top

Differential Prognostic Impact of BRAF and KRAS Mutations Across Patient Subgroups

The prognostic implications of BRAF and KRAS mutations in patients with CRC appear to vary by tumor microsatellite instability (MSI) status and CMS classification, according to new findings from a prospective study.6 Jørgen Smeby, MD, and colleagues from Oslo University Hospital in Oslo, Norway, examined the potential implications of these differences on the clinical management of CRC.

The study included the gene expression profiles of 1,197 primary tumors from a series of consecutive patients (i.e., the Oslo series) who were treated surgically for stage I-IV CRC. The analysis focused on hotspots for KRAS and BRAF mutations:

  • KRAS exon 2: codon 12 and 13
  • KRAS exon 3: codon 61
  • BRAF exon 15: codon 600

In total, 339 tumors (31%) had KRAS mutations, 192 (17%) had BRAF mutations, and 570 (52%) were wild-type for both mutations. All tumors were analyzed for MSI status, and a subgroup of 317 tumors from the Oslo series underwent additional genetic profiling using exon-level microarrays. The study population was further enriched with data from 514 patients with CRC in the publicly available GSE39582 dataset.

Interaction with MSI status and CMS category

Across all patients with CRC, the presence of KRAS and BRAF mutations predicted significantly worse overall survival relative to wild-type tumors (Table 2). When survival outcomes were examined by MSI status, however, KRAS and BRAF mutations retained their negative prognostic impact only in those patients with microsatellite stable (MSS) tumors. In contrast, these mutations had no prognostic significance in patients with MSI tumors.

In the next phase of the analysis, the research team identified another layer of interaction between KRAS and BRAF mutation status, MSS tumors, and CMS category. Results showed that KRAS mutations predicted significantly worse OS outcomes relative to KRAS wild-type tumors only in those patients with MSS tumors classified as CMS2 (HR, 1.60; 95% CI, 1.11-2.30; p = .011). In the BRAF analysis, BRAF mutations predicted worse OS relative to wild-type tumors only in those patients with MSS tumors classified as CMS1 (HR, 4.96; 95% CI, 1.74-14.12; p = .003).

Overall findings from this study reveal the presence of tumor subtype-specific prognostic associations, with KRAS mutations predicting worse outcomes in MSS tumors in CMS2, and BRAF mutations predicting worse outcomes in MSS tumors in CMS1. If validated in larger trials, these findings may have important implications for the evaluation and biomarker-directed management of patients with CRC.

Return to top

New Tumor Biomarker Profile Identifies High Relapse Risk in Stage II-III CRC

Researchers have validated a new biomarker profile—the combination of increased CD8+ T-cell infiltration and very high PD-L1 expression—that identifies 10% of patients with stage II and III CRC who have a high risk of relapse despite the frequent presence of MSI, which is typically associated with a favorable prognosis.7 Marwan Fakih, MD, and colleagues from the City of Hope National Medical Center in Duarte, California, presented findings from the study.

Biomarker profile discovery

To identify new prognostic markers, the research team first examined the expression levels of multiple tumor markers in samples from patients with stage III colon cancer who underwent primary tumor resection at the City of Hope Comprehensive Cancer Center between 1989 and 2014. The discovery cohort included 35 patients with disease recurrence within 5 years of treatment (cases), and 36 patients without recurrent disease after 5 years (controls).

Fakih and colleagues stratified patients into 3 groups based on the relative density of CD8+ T cells and PD-L1+ cells within the tumor microenvironment. In this risk-stratification model, ‘CD8-high’ was defined as intratumor infiltration of CD8+ T cells above the median, and ‘PD-L1-high’ was defined as PD-L1 expression exceeding the 90th percentile. Most of the PD-L1 expression in the tumor microenvironment occurred on tumor macrophages. Based on these expression profiles, relapse rates in the 3 groups were:

  • CD8-high/PD-L1-low: 7%
  • CD8-high/PD-L1-high: 100%
  • CD8-low/irrespective of PD-L1: 80%

Of note, 4 of 7 patients (57%) in the CD8-high/PD-L1-high group also had MSI-high (MSI-H) or mismatch repair-deficient (dMMR) tumors. These findings suggest that not all patients with MSI-H/dMMR tumors have favorable outcomes in stage III CRC.

Biomarker profile validation

Next, the researchers validated the biomarker signature in patients with stage II/III CRC from 2 major oncology databases: 385 cases from The Cancer Genome Atlas (TCGA) and 828 cases from the National Center for Biotechnology Information-Gene Expression Omnibus (NCBI-GEO). In both validation cohorts, patients with CD8-high/PD-L1-high tumors had worse clinical outcomes than those with CD8-high/PD-L1-low tumors. As with the discovery cohort, more than 50% of patients in the CD8-high/PD-L1-high groups also had MSI-H or dMMR tumors.

In the TCGA database, 12% of patients with stage II/III CRC had CD8-high/PD-L1-high tumors, and 6.9% of these were MSI-H tumors. Patients with CD8-high/PD-L1-high tumors or CD8-low tumors had significantly worse OS at 5 years than patients with CD8-high/PD-L1-low disease (HR, 3.56; p = .0095). The 5-year OS rates in each group were:

  • CD8-high/PD-L1-low: 86.1%
  • CD8-high/PD-L1-high: 59.6%
  • CD8-low/irrespective of PD-L1: 57.0%

In the NCBI-GEO database, the CD8-high/PD-L1-high tumor profile accounted for 9.7% of all cases of stage II/III CRC. Among CD8-high/PD-L1-high tumors, 57.5% were MSI-H. Recurrence-free survival was significantly worse at 5 years for patients with CD8-high/PD-L1-high disease (6.9%) or CD8-low disease (67.1%) relative to those with CD8-high/PD-L1-low disease (73.6%; HR, 1.65; p = .02).

Implications for practice

The combination of high levels of CD8+ T cell infiltration and very high PD-L1 expression defines a subpopulation of patients with stage II-III CRC who are at increased risk of relapse. Further, low CD8+ T cell infiltration is associated with worse clinical outcomes, regardless of PD-L1 expression.

Importantly, these data demonstrate that not all patients with MSI-H/dMMR tumors will do well, challenging the role of MSI-H/dMMR status as a biomarker of good prognosis in CRC. Examining the tumor CD8/PD-L1 expression profile may be a useful tool for stratifying patients with MSI-H/dMMR tumors into lower-risk and higher-risk categories, as well as identifying patients with stage II-III CRC who may be candidates for anti-PD-1 immunotherapy.

Return to top

EGFR Copy Number Predicts Anti-EGFR Benefit in RAS Wild-Type Advanced CRC

Tumor EGFR copy number is emerging as a potential clinical tool for predicting panitumumab benefit in patients with RAS wild-type advanced CRC. Jenny Seligmann, MBChB, PhD, and colleagues from Leeds Institute of Cancer and Pathology, University of Leeds, Leeds, UK, presented results of the preplanned EGFR copy number analysis from the phase III PICCOLO trial.8

The challenge: identifying patients who benefit from panitumumab

The PICCOLO trial compared irinotecan plus panitumumab with irinotecan alone in 460 patients with advanced CRC who progressed after fluoropyrimidine treatment, with or without oxaliplatin, and had no prior exposure to anti-EGFR therapy.9 Among all patients with KRAS wild-type tumors, the addition of panitumumab to irinotecan did not improve overall survival (OS) (HR, 1.01; p = .91).9 These preliminary trial findings, reported in 2013, highlighted the need for better tools to identify patients who might benefit from anti-EGFR therapy. Prospectively planned biomarker studies from PICCOLO now offer promising options.

The first biomarker analysis from PICCOLO identified a correlation between anti-EGFR treatment benefit and the EGFR ligands epiregulin (EREG) and amphiregulin (AREG).10 Among patients with RAS wild-type tumors, high EGFR ligand expression predicted a significant improvement in progression-free survival (PFS) with panitumumab compared with irinotecan alone (8.3 months vs. 4.4 months; HR, 0.38; p < .001). Patients with low EGFR ligand expression, by comparison, gained no PFS benefit from adding panitumumab (3.2 months vs. 4.0 months; HR, 0.93; p = .73).10 “The problem with EGFR ligand expression as a biomarker is that EGFR ligand expression is measured on a continuous spectrum,” Dr. Seligmann said. “It is difficult to find the cut-off point where tumors are clearly ‘positive’ or ‘negative’.”

Current study design: EGFR copy number

In the current study, Dr. Seligmann and colleagues examined EGFR copy number in 275 patients from the PICCOLO trial who had sufficient tumor tissue for DNA analysis. The study population included 219 patients with RAS wild-type tumors. All patients were stratified into 2 groups:

  • EGFR gain (>2 copies)
  • EGFR normal (2 copies)

In total, 196 patients (71.3%) had EGFR gain, and 79 patients (28.7%) had normal EGFR copy number. The analysis ruled out EGFR copy number as a general prognostic marker in advanced CRC. Among patients treated with irinotecan alone, there was no correlation between EGFR copy number status and clinical outcome, including PFS (HR, 1.0; p = .98) or OS (HR, 0.97; p = .87).

EGFR copy number gain correlated strongly with higher EGFR ligand expression, the established marker of panitumumab benefit. Compared with EGFR normal tumors, those with EGFR gain had significantly higher levels of expression of AREG (p = .0006) and EREG (p < .0001), and a trend toward higher expression of HER3 (p = .067).

Role as a predictive marker of panitumumab benefit

Results of the analysis support the role of EGFR copy number as a predictive marker of panitumumab benefit. Among those with RAS wild-type tumors, EGFR gain predicted a significant improvement in PFS with panitumumab compared with irinotecan alone (HR, 0.60; p = .002). In contrast, however, the addition of panitumumab had no effect on PFS in patients with RAS wild-type tumors and normal EGFR copy number (HR, 1.23; p = .45).

Additional results showed that the utility of EGFR copy number may extend beyond those with RAS wild-type tumors. Across the entire study population, including patients with RAS and BRAF mutations, EGFR copy number consistently differentiated between patients who were and were not likely to benefit from panitumumab (p = .03 for interaction).

In summary, EGFR copy number is consistent with EGFR ligand expression as a potential predictive marker of panitumumab benefit. In contrast with EGFR ligand expression, however, EGFR copy number offers a clearly defined binary option (gain vs. normal) that may be more useful in the clinical setting. Importantly, normal EGFR copy number may identify up to one-third of patients with RAS wild-type tumors who are not likely to benefit from treatment with panitumumab.

Return to top

CMS Predicts Bevacizumab Benefit in First-Line Metastatic Colorectal Cancer

For patients with stage IV mCRC, tumor CMS category appears to be a useful biomarker for predicting response to first-line bevacizumab.11 Jennifer Mooi, MD, of the Olivia Newton-John Cancer Research Institute in Heidelberg, Victoria, Australia, and colleagues presented results from a biomarker substudy of the phase III MAX clinical trial.11

In 2010, the Australasian Gastrointestinal Trials Group presented primary results from the MAX trial showing that bevacizumab significantly improved PFS by approximately 40% when added to capecitabine, with or without mitomycin, in the first-line treatment of mCRC.12 In the present study, Dr. Mooi and colleagues evaluated the role of CMS category in predicting which patients with mCRC were more likely to benefit from the addition of bevacizumab to standard first-line chemotherapy.

CMS category and overall survival

The MAX trial enrolled 471 patients with previously untreated, unresectable mCRC. The analysis of tumor CMS category showed the following distribution: 18% CMS1; 47% CMS2; 12% CMS3; and 23% CMS4. As expected with a well-defined prognostic biomarker, CMS correlated with survival irrespective of treatment arm (log-rank p = .006). The median OS (95% CI) for each patient subgroup was:

  • CMS1: 8.8 months (6.5-16.0 months)
  • CMS2: 24.2 months (19.1-27.4 months)
  • CMS3: 17.6 months (11.3-24.6 months)
  • CMS4: 21.4 months (15.8-23.1 months)

CMS category and bevacizumab benefit

Results showed a significant interaction between CMS group and treatment response (p = .03). Patients with CMS2 and CMS3 tumors were likely to benefit from the addition of bevacizumab to capecitabine, whereas patients with CMS1 and CMS4 tumors were less likely to benefit from bevacizumab (Table 3).

Results from the CMS substudy of the phase III MAX trial confirm the prognostic value of tumor CMS category in patients with mCRC, with CMS1 and CMS2 predicting reduced and prolonged survival, respectively. Moreover, results support the potential role of CMS as a predictive biomarker of bevacizumab benefit. Future research in this area will focus on validating the predictive value of CMS, as well as understanding the biological basis driving the preferential benefit of bevacizumab in CMS2 and CMS3 tumors in patients with mCRC.

Return to top

Summary

Biomarkers play a critical role in the evaluation and management of patients with CRC. Prognostic biomarkers enable clinicians to identify which patients will have a more aggressive disease course, while predictive biomarkers support clinical decision-making by clarifying which patients are most likely to benefit—and which patients are not expected to benefit—from specific therapies.

To test your skills in biomarker interpretation in CRC, click here to begin an interactive, case-based, CME-certified activity, Evolving Standards in Colorectal Cancer: Case-Based Insights from ESMO 2017.

References

  1. Guinney J, Dienstmann R, Wang X, et al. The consensus molecular subtypes of colorectal cancer. Nat Med. 2015;21:1350-1356. https://www.ncbi.nlm.nih.gov/pubmed/26457759
  2. Thanki K, Nicholls ME, Gajjar A, et al. Consensus molecular subtypes of colorectal cancer and their clinical implications. Int Biol Biomed J. 2017;3:105-111. https://www.ncbi.nlm.nih.gov/pubmed/28825047
  3. Sadanandam A, Eason K, Fontana E, et al. Development and validation of multiplex biomarker assay to stratify colorectal cancer (CRC) patient samples into subtypes. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 562P. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  4. Sadanandam A, Lyssiotis CA, Homicsko K, et al. A colorectal cancer classification system that associates cellular phenotype and responses to therapy. Nat Med. 2013;19:619-625. https://www.ncbi.nlm.nih.gov/pubmed/23584089
  5. Ragulan C, Eason K, Nyamundanda G, et al. A low-cost multiplex biomarker assay stratifies colorectal cancer patient samples into clinically-relevant subtypes. bioRxiv. 2017:Epub ahead of print. https://www.biorxiv.org/content/early/2017/08/16/174847
  6. Smeby J, Sveen A, Bergsland CH, et al. Prognostic impact of BRAF and KRAS mutations according to the consensus molecular subtypes of colorectal cancer. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 548P. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  7. Fakih M, Ouyang C, Wang C, et al. High PD-L1 expression and high CD8+ T-cell infiltration identifies a new subpopulation of colorectal cancer with high risk of relapse and poor outcome. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 558P. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  8. Seligmann J, Wood H, Richman S, et al. Epidermal growth factor receptor (EGFR) copy number as a biomarker of prognosis and panitumumab benefit in RAS-wt advanced colorectal cancer. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 545P. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  9. Seymour MT, Brown SR, Middleton G, et al. Panitumumab and irinotecan versus irinotecan alone for patients with KRAS wild-type, fluorouracil-resistant advanced colorectal cancer (PICCOLO): a prospectively stratified randomised trial. Lancet Oncol. 2013;14:749-759. https://www.ncbi.nlm.nih.gov/pubmed/23725851
  10. Seligmann JF, Elliott F, Richman SD, et al. Combined epiregulin and amphiregulin expression levels as a predictive biomarker for panitumumab therapy benefit or lack of benefit in patients with RAS wild-type advanced colorectal cancer. JAMA Oncol. 2016;2:633-642. https://www.ncbi.nlm.nih.gov/pubmed/26867820
  11. Mooi J, Wirapati P, Asher R, et al. Consensus molecular subtypes (CMS) as predictors of benefit from bevacizumab in first line treatment of metastatic colorectal cancer: Retrospective analysis of the MAX clinical trial. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 479O. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  12. Tebbutt NC, Wilson K, Gebski VJ, et al. Capecitabine, bevacizumab, and mitomycin in first-line treatment of metastatic colorectal cancer: results of the Australasian Gastrointestinal Trials Group Randomized Phase III MAX Study. J Clin Oncol. 2010;28:3191-3198. https://www.ncbi.nlm.nih.gov/pubmed/20516443

ESMO 2017 Special Report:
Emerging Prognostic and Predictive Biomarkers in Colorectal Cancer

Colorectal cancer (CRC) is a heterogeneous disease characterized by diverse clinical outcomes and responses to cancer therapy. Biomarkers are critical for developing individualized treatment plans by providing a better understanding of the disease course (prognostic biomarkers) and anticipated therapeutic response (predictive biomarkers).

At the European Society for Medical Oncology (ESMO) 2017 Congress, held September 8-12, 2017, in Madrid, Spain, researchers presented new data on multiple biomarker—including consensus molecular subtypes (CMS), BRAF and KRAS mutation status, microsatellite-instability (MSI) status, epidermal growth factor receptor (EGFR) copy number, and other genetic, molecular, and clinical characteristics. Scroll down to read more about these emerging prognostic and predictive biomarkers in CRC.

Prognostic biomarkers

  • Developing a Low-Cost, Accessible Tool for Assigning CMS Category in CRC
  • Differential Prognostic Impact of BRAF and KRAS Mutations Across Patient Subgroups
  • New Tumor Biomarker Profile Identifies High Relapse Risk in Stage II-III CRC

Predictive biomarkers

  • EGFR Copy Number Predicts Anti-EGFR Benefit in RAS Wild-Type Advanced CRC
  • CMS Predicts Bevacizumab Benefit in First-Line Metastatic Colorectal Cancer

Developing a Low-Cost, Accessible Tool for Assigning CMS Category in CRC

In 2015, the international CRC Subtyping Consortium (CRCSC) introduced a new classification system for CRC.1 The consensus molecular subtypes (CMS) define 4 tumor subtypes with distinct genetic, prognostic, and clinical features (Table 1).1,2 Approximately 13% of CRC tumors display mixed features of more than one CMS category, suggesting intratumoral heterogeneity or a transition phenotype.1

 

Despite its promising clinical role in CRC tumor assessment, the CMS classification system has been limited to date by the lack of a cost-effective assay. Angura Sadanandam, PhD, and colleagues from The Institute of Cancer Research in London, UK, provided an update on the development of a low-cost multi-gene platform for identifying CMS subtypes.3

Refining an established gene panel

Dr. Sadanandam and colleagues previously described a 786-gene panel, the CRCAssigner-786, that reliably stratifies CRC tumors into CMS categories.4 Identifying a CMS subtype requires microarray and RNA-Seq platforms, which are costly, time-consuming, and impractical for routine clinical use.5

The first step in reducing the complexity of CMS classification involved reducing the standard 786-gene signature to a shorter panel of 38 genes (CRCAssigner-38). Next, the team developed a custom assay (NanoCRCAssigner) to assess tumor subtypes based on the 38-gene panel. The NanoCRCAssigner utilizes the nCounter platform, a low-cost and highly reproducible tool for molecular analysis. The research team them compared the performance of the standard (microarray/RNA-Seq) and low-cost (NanoCRCAssigner) platforms in assigning CMS subtypes to a validation cohort of primary untreated CRC samples (n = 17).

High concordance with CMS subtypes

Results showed a high concordance (>70%) between the standard and low-cost protocols for stratifying CRC tumors into CMS subtypes (Pearson’s correlation coefficient, 0.98). The CRCAssigner-38 panel was further validated in additional fresh-frozen CRC tumor samples (n = 51) and formalin-fixed, paraffin-embedded (FFPE) CRC tumor samples (n = 24).

In summary, these findings demonstrate the potential feasibility of stratifying CRC tumors into CMS categories using a 38-gene assay and fresh-frozen or FFPE tumor samples. The next step for this research group involves additional validation of the 38-gene panel in larger cohorts of CRC tumor samples, with the goal of bringing an inexpensive biomarker assay for identifying CMS category into clinical practice.

Return to top

Differential Prognostic Impact of BRAF and KRAS Mutations Across Patient Subgroups

The prognostic implications of BRAF and KRAS mutations in patients with CRC appear to vary by tumor microsatellite instability (MSI) status and CMS classification, according to new findings from a prospective study.6 Jørgen Smeby, MD, and colleagues from Oslo University Hospital in Oslo, Norway, examined the potential implications of these differences on the clinical management of CRC.

The study included the gene expression profiles of 1,197 primary tumors from a series of consecutive patients (i.e., the Oslo series) who were treated surgically for stage I-IV CRC. The analysis focused on hotspots for KRAS and BRAF mutations:

  • KRAS exon 2: codon 12 and 13
  • KRAS exon 3: codon 61
  • BRAF exon 15: codon 600

In total, 339 tumors (31%) had KRAS mutations, 192 (17%) had BRAF mutations, and 570 (52%) were wild-type for both mutations. All tumors were analyzed for MSI status, and a subgroup of 317 tumors from the Oslo series underwent additional genetic profiling using exon-level microarrays. The study population was further enriched with data from 514 patients with CRC in the publicly available GSE39582 dataset.

Interaction with MSI status and CMS category

Across all patients with CRC, the presence of KRAS and BRAF mutations predicted significantly worse overall survival relative to wild-type tumors (Table 2). When survival outcomes were examined by MSI status, however, KRAS and BRAF mutations retained their negative prognostic impact only in those patients with microsatellite stable (MSS) tumors. In contrast, these mutations had no prognostic significance in patients with MSI tumors.

In the next phase of the analysis, the research team identified another layer of interaction between KRAS and BRAF mutation status, MSS tumors, and CMS category. Results showed that KRAS mutations predicted significantly worse OS outcomes relative to KRAS wild-type tumors only in those patients with MSS tumors classified as CMS2 (HR, 1.60; 95% CI, 1.11-2.30; p = .011). In the BRAF analysis, BRAF mutations predicted worse OS relative to wild-type tumors only in those patients with MSS tumors classified as CMS1 (HR, 4.96; 95% CI, 1.74-14.12; p = .003).

Overall findings from this study reveal the presence of tumor subtype-specific prognostic associations, with KRAS mutations predicting worse outcomes in MSS tumors in CMS2, and BRAF mutations predicting worse outcomes in MSS tumors in CMS1. If validated in larger trials, these findings may have important implications for the evaluation and biomarker-directed management of patients with CRC.

Return to top

New Tumor Biomarker Profile Identifies High Relapse Risk in Stage II-III CRC

Researchers have validated a new biomarker profile—the combination of increased CD8+ T-cell infiltration and very high PD-L1 expression—that identifies 10% of patients with stage II and III CRC who have a high risk of relapse despite the frequent presence of MSI, which is typically associated with a favorable prognosis.7 Marwan Fakih, MD, and colleagues from the City of Hope National Medical Center in Duarte, California, presented findings from the study.

Biomarker profile discovery

To identify new prognostic markers, the research team first examined the expression levels of multiple tumor markers in samples from patients with stage III colon cancer who underwent primary tumor resection at the City of Hope Comprehensive Cancer Center between 1989 and 2014. The discovery cohort included 35 patients with disease recurrence within 5 years of treatment (cases), and 36 patients without recurrent disease after 5 years (controls).

Fakih and colleagues stratified patients into 3 groups based on the relative density of CD8+ T cells and PD-L1+ cells within the tumor microenvironment. In this risk-stratification model, ‘CD8-high’ was defined as intratumor infiltration of CD8+ T cells above the median, and ‘PD-L1-high’ was defined as PD-L1 expression exceeding the 90th percentile. Most of the PD-L1 expression in the tumor microenvironment occurred on tumor macrophages. Based on these expression profiles, relapse rates in the 3 groups were:

  • CD8-high/PD-L1-low: 7%
  • CD8-high/PD-L1-high: 100%
  • CD8-low/irrespective of PD-L1: 80%

Of note, 4 of 7 patients (57%) in the CD8-high/PD-L1-high group also had MSI-high (MSI-H) or mismatch repair-deficient (dMMR) tumors. These findings suggest that not all patients with MSI-H/dMMR tumors have favorable outcomes in stage III CRC.

Biomarker profile validation

Next, the researchers validated the biomarker signature in patients with stage II/III CRC from 2 major oncology databases: 385 cases from The Cancer Genome Atlas (TCGA) and 828 cases from the National Center for Biotechnology Information-Gene Expression Omnibus (NCBI-GEO). In both validation cohorts, patients with CD8-high/PD-L1-high tumors had worse clinical outcomes than those with CD8-high/PD-L1-low tumors. As with the discovery cohort, more than 50% of patients in the CD8-high/PD-L1-high groups also had MSI-H or dMMR tumors.

In the TCGA database, 12% of patients with stage II/III CRC had CD8-high/PD-L1-high tumors, and 6.9% of these were MSI-H tumors. Patients with CD8-high/PD-L1-high tumors or CD8-low tumors had significantly worse OS at 5 years than patients with CD8-high/PD-L1-low disease (HR, 3.56; p = .0095). The 5-year OS rates in each group were:

  • CD8-high/PD-L1-low: 86.1%
  • CD8-high/PD-L1-high: 59.6%
  • CD8-low/irrespective of PD-L1: 57.0%

In the NCBI-GEO database, the CD8-high/PD-L1-high tumor profile accounted for 9.7% of all cases of stage II/III CRC. Among CD8-high/PD-L1-high tumors, 57.5% were MSI-H. Recurrence-free survival was significantly worse at 5 years for patients with CD8-high/PD-L1-high disease (6.9%) or CD8-low disease (67.1%) relative to those with CD8-high/PD-L1-low disease (73.6%; HR, 1.65; p = .02).

Implications for practice

The combination of high levels of CD8+ T cell infiltration and very high PD-L1 expression defines a subpopulation of patients with stage II-III CRC who are at increased risk of relapse. Further, low CD8+ T cell infiltration is associated with worse clinical outcomes, regardless of PD-L1 expression.

Importantly, these data demonstrate that not all patients with MSI-H/dMMR tumors will do well, challenging the role of MSI-H/dMMR status as a biomarker of good prognosis in CRC. Examining the tumor CD8/PD-L1 expression profile may be a useful tool for stratifying patients with MSI-H/dMMR tumors into lower-risk and higher-risk categories, as well as identifying patients with stage II-III CRC who may be candidates for anti-PD-1 immunotherapy.

Return to top

EGFR Copy Number Predicts Anti-EGFR Benefit in RAS Wild-Type Advanced CRC

Tumor EGFR copy number is emerging as a potential clinical tool for predicting panitumumab benefit in patients with RAS wild-type advanced CRC. Jenny Seligmann, MBChB, PhD, and colleagues from Leeds Institute of Cancer and Pathology, University of Leeds, Leeds, UK, presented results of the preplanned EGFR copy number analysis from the phase III PICCOLO trial.8

The challenge: identifying patients who benefit from panitumumab

The PICCOLO trial compared irinotecan plus panitumumab with irinotecan alone in 460 patients with advanced CRC who progressed after fluoropyrimidine treatment, with or without oxaliplatin, and had no prior exposure to anti-EGFR therapy.9 Among all patients with KRAS wild-type tumors, the addition of panitumumab to irinotecan did not improve overall survival (OS) (HR, 1.01; p = .91).9 These preliminary trial findings, reported in 2013, highlighted the need for better tools to identify patients who might benefit from anti-EGFR therapy. Prospectively planned biomarker studies from PICCOLO now offer promising options.

The first biomarker analysis from PICCOLO identified a correlation between anti-EGFR treatment benefit and the EGFR ligands epiregulin (EREG) and amphiregulin (AREG).10 Among patients with RAS wild-type tumors, high EGFR ligand expression predicted a significant improvement in progression-free survival (PFS) with panitumumab compared with irinotecan alone (8.3 months vs. 4.4 months; HR, 0.38; p < .001). Patients with low EGFR ligand expression, by comparison, gained no PFS benefit from adding panitumumab (3.2 months vs. 4.0 months; HR, 0.93; p = .73).10 “The problem with EGFR ligand expression as a biomarker is that EGFR ligand expression is measured on a continuous spectrum,” Dr. Seligmann said. “It is difficult to find the cut-off point where tumors are clearly ‘positive’ or ‘negative’.”

Current study design: EGFR copy number

In the current study, Dr. Seligmann and colleagues examined EGFR copy number in 275 patients from the PICCOLO trial who had sufficient tumor tissue for DNA analysis. The study population included 219 patients with RAS wild-type tumors. All patients were stratified into 2 groups:

  • EGFR gain (>2 copies)
  • EGFR normal (2 copies)

In total, 196 patients (71.3%) had EGFR gain, and 79 patients (28.7%) had normal EGFR copy number. The analysis ruled out EGFR copy number as a general prognostic marker in advanced CRC. Among patients treated with irinotecan alone, there was no correlation between EGFR copy number status and clinical outcome, including PFS (HR, 1.0; p = .98) or OS (HR, 0.97; p = .87).

EGFR copy number gain correlated strongly with higher EGFR ligand expression, the established marker of panitumumab benefit. Compared with EGFR normal tumors, those with EGFR gain had significantly higher levels of expression of AREG (p = .0006) and EREG (p < .0001), and a trend toward higher expression of HER3 (p = .067).

Role as a predictive marker of panitumumab benefit

Results of the analysis support the role of EGFR copy number as a predictive marker of panitumumab benefit. Among those with RAS wild-type tumors, EGFR gain predicted a significant improvement in PFS with panitumumab compared with irinotecan alone (HR, 0.60; p = .002). In contrast, however, the addition of panitumumab had no effect on PFS in patients with RAS wild-type tumors and normal EGFR copy number (HR, 1.23; p = .45).

Additional results showed that the utility of EGFR copy number may extend beyond those with RAS wild-type tumors. Across the entire study population, including patients with RAS and BRAF mutations, EGFR copy number consistently differentiated between patients who were and were not likely to benefit from panitumumab (p = .03 for interaction).

In summary, EGFR copy number is consistent with EGFR ligand expression as a potential predictive marker of panitumumab benefit. In contrast with EGFR ligand expression, however, EGFR copy number offers a clearly defined binary option (gain vs. normal) that may be more useful in the clinical setting. Importantly, normal EGFR copy number may identify up to one-third of patients with RAS wild-type tumors who are not likely to benefit from treatment with panitumumab.

Return to top

CMS Predicts Bevacizumab Benefit in First-Line Metastatic Colorectal Cancer

For patients with stage IV mCRC, tumor CMS category appears to be a useful biomarker for predicting response to first-line bevacizumab.11 Jennifer Mooi, MD, of the Olivia Newton-John Cancer Research Institute in Heidelberg, Victoria, Australia, and colleagues presented results from a biomarker substudy of the phase III MAX clinical trial.11

In 2010, the Australasian Gastrointestinal Trials Group presented primary results from the MAX trial showing that bevacizumab significantly improved PFS by approximately 40% when added to capecitabine, with or without mitomycin, in the first-line treatment of mCRC.12 In the present study, Dr. Mooi and colleagues evaluated the role of CMS category in predicting which patients with mCRC were more likely to benefit from the addition of bevacizumab to standard first-line chemotherapy.

CMS category and overall survival

The MAX trial enrolled 471 patients with previously untreated, unresectable mCRC. The analysis of tumor CMS category showed the following distribution: 18% CMS1; 47% CMS2; 12% CMS3; and 23% CMS4. As expected with a well-defined prognostic biomarker, CMS correlated with survival irrespective of treatment arm (log-rank p = .006). The median OS (95% CI) for each patient subgroup was:

  • CMS1: 8.8 months (6.5-16.0 months)
  • CMS2: 24.2 months (19.1-27.4 months)
  • CMS3: 17.6 months (11.3-24.6 months)
  • CMS4: 21.4 months (15.8-23.1 months)

CMS category and bevacizumab benefit

Results showed a significant interaction between CMS group and treatment response (p = .03). Patients with CMS2 and CMS3 tumors were likely to benefit from the addition of bevacizumab to capecitabine, whereas patients with CMS1 and CMS4 tumors were less likely to benefit from bevacizumab (Table 3).

Results from the CMS substudy of the phase III MAX trial confirm the prognostic value of tumor CMS category in patients with mCRC, with CMS1 and CMS2 predicting reduced and prolonged survival, respectively. Moreover, results support the potential role of CMS as a predictive biomarker of bevacizumab benefit. Future research in this area will focus on validating the predictive value of CMS, as well as understanding the biological basis driving the preferential benefit of bevacizumab in CMS2 and CMS3 tumors in patients with mCRC.

Return to top

Summary

Biomarkers play a critical role in the evaluation and management of patients with CRC. Prognostic biomarkers enable clinicians to identify which patients will have a more aggressive disease course, while predictive biomarkers support clinical decision-making by clarifying which patients are most likely to benefit—and which patients are not expected to benefit—from specific therapies.

To test your skills in biomarker interpretation in CRC, click here to begin an interactive, case-based, CME-certified activity, Evolving Standards in Colorectal Cancer: Case-Based Insights from ESMO 2017.

References

  1. Guinney J, Dienstmann R, Wang X, et al. The consensus molecular subtypes of colorectal cancer. Nat Med. 2015;21:1350-1356. https://www.ncbi.nlm.nih.gov/pubmed/26457759
  2. Thanki K, Nicholls ME, Gajjar A, et al. Consensus molecular subtypes of colorectal cancer and their clinical implications. Int Biol Biomed J. 2017;3:105-111. https://www.ncbi.nlm.nih.gov/pubmed/28825047
  3. Sadanandam A, Eason K, Fontana E, et al. Development and validation of multiplex biomarker assay to stratify colorectal cancer (CRC) patient samples into subtypes. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 562P. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  4. Sadanandam A, Lyssiotis CA, Homicsko K, et al. A colorectal cancer classification system that associates cellular phenotype and responses to therapy. Nat Med. 2013;19:619-625. https://www.ncbi.nlm.nih.gov/pubmed/23584089
  5. Ragulan C, Eason K, Nyamundanda G, et al. A low-cost multiplex biomarker assay stratifies colorectal cancer patient samples into clinically-relevant subtypes. bioRxiv. 2017:Epub ahead of print. https://www.biorxiv.org/content/early/2017/08/16/174847
  6. Smeby J, Sveen A, Bergsland CH, et al. Prognostic impact of BRAF and KRAS mutations according to the consensus molecular subtypes of colorectal cancer. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 548P. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  7. Fakih M, Ouyang C, Wang C, et al. High PD-L1 expression and high CD8+ T-cell infiltration identifies a new subpopulation of colorectal cancer with high risk of relapse and poor outcome. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 558P. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  8. Seligmann J, Wood H, Richman S, et al. Epidermal growth factor receptor (EGFR) copy number as a biomarker of prognosis and panitumumab benefit in RAS-wt advanced colorectal cancer. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 545P. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  9. Seymour MT, Brown SR, Middleton G, et al. Panitumumab and irinotecan versus irinotecan alone for patients with KRAS wild-type, fluorouracil-resistant advanced colorectal cancer (PICCOLO): a prospectively stratified randomised trial. Lancet Oncol. 2013;14:749-759. https://www.ncbi.nlm.nih.gov/pubmed/23725851
  10. Seligmann JF, Elliott F, Richman SD, et al. Combined epiregulin and amphiregulin expression levels as a predictive biomarker for panitumumab therapy benefit or lack of benefit in patients with RAS wild-type advanced colorectal cancer. JAMA Oncol. 2016;2:633-642. https://www.ncbi.nlm.nih.gov/pubmed/26867820
  11. Mooi J, Wirapati P, Asher R, et al. Consensus molecular subtypes (CMS) as predictors of benefit from bevacizumab in first line treatment of metastatic colorectal cancer: Retrospective analysis of the MAX clinical trial. European Society of Medical Oncology (ESMO) 2017 Congress. September 8-12, 2017; Madrid, Spain. Abstract 479O. http://www.esmo.org/content/download/117241/2057634/file/ESMO-2017-Abstract-Book.pdf
  12. Tebbutt NC, Wilson K, Gebski VJ, et al. Capecitabine, bevacizumab, and mitomycin in first-line treatment of metastatic colorectal cancer: results of the Australasian Gastrointestinal Trials Group Randomized Phase III MAX Study. J Clin Oncol. 2010;28:3191-3198. https://www.ncbi.nlm.nih.gov/pubmed/20516443