Immunotherapy for Colorectal Cancer

A Review of Current and Novel Therapeutic Approaches

Aaron J. Franke; William Paul Skelton IV; Jason S. Starr; Hiral Parekh; James J. Lee; Michael J. Overman; Carmen Allegra; Thomas J. George


J Natl Cancer Inst. 2019;111(11):1131-1141. 

In This Article

Biomarkers of Immune Response

Although promising immunotherapy advancements in CRC continue to evolve and generate enthusiasm, to optimize treatment efficacy, overcome resistance to immune checkpoint blockade, and appropriately select for patients who will likely benefit from immunotherapy, the development of rational therapeutic combinations remains critical. There are several ongoing studies investigating potential targetable pathways involved in the host immune response to CRC. For this field to substantively evolve, correlative studies from clinical trials will be essential to identify biomarkers that can serve as immune response surrogates in mCRC. The following section briefly reviews a limited number of candidate biomarkers and molecular classification systems that have demonstrated the potential to exploit inherent tumor biology and immunophenotype to further our knowledge in this area. A more detailed review on the topic of immunotherapy biomarker selection, validation, and development is beyond the scope of this article and has been recently published elsewhere.[38–41]


Patients harboring MSI-H:dMMR tumors represent a unique population of mCRC patients that currently appear to benefit the most from immune-based therapies.[42] High DNA microsatellite instability, defined as instability at two or more loci (or ≥30% of loci if a larger panel of markers is used), results in a high number of DNA replication errors and represents hallmark consequences both of hereditary and sporadic alterations in MMR genes.[43]

The utility of MSI-H:dMMR as a predictive biomarker to anti-PD-1 therapy (pembrolizumab) in mCRC was highlighted in the KEYNOTE-016 trial, which showed a notably higher overall response rate (ORR) in MSI-H:dMMR compared with microsatellite stable (MSS):proficient MMR (pMMR) tumors: 40% (95% confidence interval [CI] = 12% to 74%) vs 0% (95% CI = 0% to 19%), respectively.[8] One molecular rationale for the discrepancies between these subgroups is differences in tumor mutational burden. Using whole-exome sequencing, investigators identified an average of 1782 somatic mutations in MSI-H:dMMR tumors vs 73 in MSS:pMMR (P = .01). Interestingly, a high tumor mutational burden was associated with prolonged progression-free survival (PFS) in MSI-H patients (hazard ratio [HR] = 0.63, 95% CI = 0.42 to 0.93, P = .02).

These data, in addition to similar data for nivolumab with or without ipilimumab, led to the recent FDA approval of these agents in MSI-H:dMMR–refractory mCRC patients (Table 1). However, despite MSI-H:dMMR accounting for 15%–20% of all sporadic CRC, this subset of patients represents only a minority (~5%) of mCRC, and ongoing efforts to expand the application of immunotherapy in MSS:pMMR mCRC are eagerly awaited.


The utility of PD-L1 expression as a surrogate of tumor immunogenicity and predictor of response to checkpoint inhibitors remains controversial, but despite conflicting evidence of its clinical significance, it is recognized as one of the most extensively studied candidate biomarkers to date. Studies have shown that upregulation of PD-L1 in the TME is associated with increased effector T-cell infiltration, and that PD-L1–"positive" tumors have a higher likelihood of clinical benefit with checkpoint inhibitors. In a large meta-analysis of 21 trials (non-CRC primary tumors), the ORR in PD-L1–positive tumors was 34.1% (95% CI = 27.6% to 41.3%) vs 19.9% (95% CI = 15.4% to 25.3%) in PD-L1– negative tumors.[44] However, in mCRC the available data have demonstrated no predictive role of PD-L1 expression with regard to clinical outcomes. This is in contrast to other tumor sites (ie, melanoma and non-small cell lung cancer), for which PD-L1 expression has been shown to predict response to immune checkpoint inhibition.

A small subgroup analysis (n = 23) from a multicohort phase-Ib trial of pembrolizumab in advanced solid tumors (KEYNOTE-028) reported a poor ORR of 4% with only one PD-L1–positive mCRC patient, who was also noted to be MSI-H:dMMR, achieving a partial response.[45] In the phase II trial of pembrolizumab (KEYNOTE-016) for patients with refractory mCRC (included both MSI-H and MSS), no statistically significant association was found between PD-L1 expression and PFS or OS.[8] A recent update from the CheckMate-142 trial demonstrated a promising ORR for MSI-H:dMMR patients both in the monotherapy (nivolumab alone) and combination immunotherapy arm (nivolumab + ipilimumab); however, ORR appeared irrespective of PD-L1 expression level.[46]

The observed discordance between PD-L1 expression and response to immune checkpoint blockade may be related to the dynamic nature of this cell surface biomarker, with variable expression at any one time point depending on interactions within the local TME. It is also important to note that different pharmDx kits were used in these studies (IHC 22C3 for pembrolizumab, IHC 28–8 for nivolumab). In addition, the lack of standardized metrics and consensus on "positivity" thresholds may have also obfuscated the true clinical potential of this predictive biomarker.


The density and distribution of the immune response to tumor cells, represented by TILs, is another potential predictive biomarker in CRC. Some studies have shown peritumoral immune infiltration to be a more useful and predictive surrogate marker of disease progression than the standardized American Joint Committee on Cancer TNM staging system.[15,23] Researchers developed the Immunoscore as a risk stratification tool performed by quantification of two lymphocyte population densities in the core of the tumor and at the invasive margin. Although the role of this scoring system to predict response to immunotherapy agents has not yet been widely accepted (owing in part to the central pathologic review and score calculation), multivariable analysis revealed that Immunoscore was a superior prognostic biomarker compared with MSI-H status in predicting disease-specific recurrence and survival in mCRC.[47] Additional evidence supporting the consensus Immunoscore as a reliable biomarker to estimate the risk of recurrence in stage I–III CRC was provided by Pagès et al., who recently published the results from a large (n = 2648 tumor samples) multicenter study conducted by an international consortium of expert pathologists and immunologists.[48] The prognostic value of the Immunoscore was validated to predict statistically significant (P < .0001) differences in time to recurrence (HR = 0.33, 95% CI = 0.23 to 0.47), DFS (HR = 0.50, 95% CI = 0.39 to 0.64), and OS (HR = 0.56, 95% CI = 0.43 to 0.73). Importantly, the Immunoscore had a larger relative prognostic value than pathologic TNM staging, lymphovascular invasion, tumor differentiation, and MSI status (and based on predictive models, appears independent of those factors as well). The relevance of the Immunoscore in mCRC has been less studied, though limited studies have suggested its continued prognostic significance in patients with metastases.