The notion that the immune system can be manipulated against cancerous cells is not something novel. Although immunotherapy has been used therapeutically in cancer for a couple of decades already, it is the development of immune checkpoint inhibitors (ICIs) that completely revolutionized the treatment of numerous solid tumors. Most importantly, ICIs not only opened a new dimension in the ways in which we can disrupt the intersection between the immune system and the tumor, but also expanded our overall understanding of tumor immunology. Contrary to tumors such as melanoma and lung cancer, ICIs in gastrointestinal (GI) malignancies, with the exception of cancers with microsatellite instability (MSI), remain modestly effective as monotherapy.
To begin with, it should be emphasized that GI malignancies are a group of different cancers with different biology and a distinct immune tumor microenvironment (TME). This is evident even within the GI MSI tumors for which currently ICIs are a U.S. Food and Drug Administration (FDA) histology agnostic approved therapy. The most obvious difference is seen between colorectal cancer (CRC) and non-CRC GI MSI tumors; CRC MSI tumors have consistently higher responses compared to non-CRC GI MSI cancers, ranging from 31% to 52% with pembrolizumab or nivolumab alone[3–5] and 55% with nivolumab combined with ipilimumab. On the other hand, in non-CRC MSI tumors responses to pembrolizumab differ between trials and range from 40% to 45% in gastric, cholangiocarcinoma, and small intestine carcinomas in the more recent KEYNOTE-158 and from 25% to 80% in an earlier phase 2 trial, in which, however, the number of patients was significantly smaller. Pancreatic MSI tumors had a response rate of 18.2% (N = 22) and 61% (N = 8) in those two trials, respectively.[3,7] Although comparisons between trials are certainly not advisable, this variability in responses may reflect to some degree at least the tumor variability and perhaps the lack of an in-depth understanding of these tumors despite their shared MSI. It is a plausible hypothesis that the MSI status does not necessarily confer the same immunogenicity across all tumors. For some tumors, histology may ultimately have a higher impact than the MSI status, hinting at the importance of other tissue-specific factors like the surrounding stroma that can influence the immune cell infiltration and thus the responses to ICIs. In addition, one aspect that warrants more research is the role of the host and whether there would be any variation in the tumor phenotype across histologies or within the same histology, between germline MSH mutations versus somatic MSI tumors.
Beyond MSI GI tumors, pembrolizumab is FDA approved for patients with gastric cancer and PD-L1 combined positive score (CPS) greater than or equal to 1, who have progressed on two other lines of chemotherapy. There is likely a role for pembrolizumab in earlier lines in a subset of patients with this disease. The results of KEYNOTE-062, a randomized phase 3 trial that compared pembrolizumab versus pembrolizumab plus chemotherapy versus chemotherapy alone in the first-line setting, in patients with HER2-negative cancer, locally advanced unresectable, or metastatic gastric/gastroesophageal junction (GEJ) adenocarcinoma, were recently presented. The study showed that pembrolizumab alone was not inferior in terms of overall survival (OS) when compared to chemotherapy alone, with a median OS of 10.6 versus 11.1 months, respectively. However, the effect was not consistent throughout the entire OS curve. In addition, the progression-free survival (PFS) was substantially shorter in the pembrolizumab-alone arm (2.0 vs. 6.4 months for chemotherapy and 6.9 months for the combination arm). This suggests that for a subset of patients with gastric/GEJ cancer, single-agent pembrolizumab in the first line can indeed be detrimental. Patients with high disease burden, peritoneal metastases, or rapidly progressive tumors may not be fit for second-line therapy if treated initially with pembrolizumab alone. Additional efforts to understand patient selection for single-agent pembrolizumab are definitely needed before this strategy can be implemented.
Immune checkpoint inhibitors recently showed promise in hepatocellular carcinoma (HCC) in combination with bevacizumab, an anti-VEGF (vascular endothelial growth factor) antibody. The remarkable results of the recent phase 3 IMbrave150 trial were not only on outcomes but also on quality-of-life maintenance when compared to the current standard-of-care sorafenib. These findings represent true progress after more than a decade in this chemorefractory disease, often challenging to clinically manage, and will likely alter the first-line therapy landscape. A recent work showed that patients with advanced HCC with WNT/CTNNB1 mutations were less likely to derive benefit from ICIs. We can hypothesize that upcoming studies in HCC will use sequencing to enrich trials with patients more likely to benefit from ICI therapy.
Another highlight regarding the great heterogeneity of GI malignancies can be seen in anal squamous cell cancer (SCCA). Anal squamous cell cancer is linked with HPV infection in the majority of cases, suggesting a potential role of enhanced ICI efficacy. However, contrary to other virus-associated solid tumors like head and neck squamous cell cancers, responses to ICIs in SCCA were modest, ranging from 24% in a phase 2 trial with nivolumab and 17% in the KEYNOTE-028 phase 1b pembrolizumab trial in advanced SCCA. Both agents are currently approved as subsequent systemic therapies following concurrent chemotherapy with radiation, in light of limited treatment options in this setting.
Though we have a better understanding of recognizing immunologically inflamed versus deserted tumors, we have yet to discover reliable biomarkers of response to ICIs across all tumors, beyond MSI. As each different cancer has its own TME, one would expect that TME characteristics conferring sensitivity to ICIs would be similar across histologies. However, this has not proven true with the most characteristic example of one of the first biomarkers developed, the level of PD-L1 expression. This again highlights our incomplete understanding of cancer immunology to this date. However, progress is constantly being made and our knowledge expands daily. One simple and emerging potential biomarker of response to ICI therapy in GI malignancies is the Immunoscore, which is based on quantification of CD3 and CD8 T lymphocytes in the tumor center and margins and has been shown to be prognostic for early-stage CRC. The Immunoscore, along with other components of TME, offers a rationale for predicting ICI response at least in CRC; however, it has yet to be validated in this context.
Although the immune checkpoint blockade in GI malignancies has not been as widely successful as in other solid tumors, multiple trials are currently under way combining ICI with other agents targeting angiogenesis or antigen presentation to increase ICI efficacy. Whether combination or sequential approaches will be more effective in overcoming or reversing the cancer immune evasion and improve immunotherapy efficacy remains unclear. What is now clear is that other facets of the immune cascade, as well the tumor itself with the surrounding microenvironment and stroma, will need to be accounted for in a successful application of ICI therapy in GI malignancies. Immune checkpoint inhibitors will continue to drastically affect the treatment landscape of multiple malignancies, but precision immunotherapy for each tumor subtype will likely be the way forward for most cancers.
Source of Support: Ignacio Garrido-Laguna disclosed consulting or advisory role for Array BioPharma and research support (institutional) from Novartis, Ignyta, Halozyme, Bayer, Bristol-Myers Squibb, Pfizer, Newlink Genetics, MedImmune, Lilly, Incyte, GlaxoSmithKline, OncoMed, ARMO BioSciences, and Glennmark. Vaia Florou: none. Conflict of Interest: None.