Immunotherapy has transformed the treatment landscape for many cancers, but colorectal cancer remains a stubborn exception for most patients. Only a minority of people with metastatic colorectal cancer experience lasting benefit from checkpoint inhibitors, and researchers have long sought to understand why the majority fail to respond. A new brief report published in Cancer Immunology, Immunotherapy by a team of Italian researchers led from the University of Genoa now adds a provocative piece to this puzzle. The study describes two patients with metastatic colorectal cancer whose tumors carried microsatellite instability, a feature that normally predicts good responses to immunotherapy, yet who also displayed a set of unfavorable prognostic factors that would typically raise concern. Despite these obstacles, both patients achieved durable responses to the PD-1 blocker pembrolizumab, and the biological features of their tumors point to an unexpected player in the therapeutic effect: natural killer cells.
The two patients described in the report shared several characteristics that are generally associated with poorer outcomes in colorectal cancer. Their disease had recurred and spread to distant sites, and one of the tumors carried the BRAF V600E mutation, a genetic alteration linked to aggressive behavior and resistance to many therapies. Most strikingly, immunohistochemical analysis of tumor samples obtained at surgical resection revealed markedly reduced expression of HLA class I molecules on the surface of the tumor cells. HLA-I proteins are the molecular display stands that present fragments of tumor antigens to cytotoxic T cells, the primary assassins recruited by checkpoint blockade. When these molecules are lost or downregulated, T cells essentially cannot see the tumor, a mechanism considered one of the major routes by which cancers escape immunotherapy.
This is where the biology becomes genuinely intriguing. The very alteration that blinds T cells, the loss of HLA-I expression, is precisely the signal that activates natural killer cells. NK cells are innate lymphocytes that patrol the body looking for cells that have dropped their self-identifying HLA molecules, a tactic commonly used by virus-infected cells and malignant cells alike. In other words, the tumor in these patients had traded invisibility to T cells for visibility to NK cells. The Genoa-based team, including corresponding authors Emanuela Marcenaro, Mariella Della Chiesa and Simona Sivori, hypothesize that this trade-off may have created a therapeutic window that pembrolizumab was able to exploit, even though pembrolizumab is conventionally understood as a drug that reinvigorates exhausted T cells.
The mechanism by which a T-cell-directed drug might unleash NK cells centers on a receptor-ligand pair more often discussed in the context of T-cell biology. Examination of the resected tumor samples showed that the tumor cells expressed PD-L1, the ligand for the PD-1 checkpoint receptor, and that a substantial fraction of the tumor-infiltrating NK cells themselves carried PD-1 on their surface. PD-1 expression on NK cells has been reported in several tumor settings and is generally interpreted as a sign of functional exhaustion, a brake placed on the killer cells by the immunosuppressive tumor microenvironment. If NK cells in these tumors were being restrained through the PD-1 pathway, then pembrolizumab, by blocking PD-1, could in principle release that brake and restore NK-cell-mediated tumor killing.
This interpretation offers a coherent explanation for a clinical outcome that would otherwise be difficult to reconcile. Microsatellite instability generates abundant neoantigens and typically produces tumors that are densely infiltrated by immune cells, which is why MSI-high metastatic colorectal cancer is the one molecular subtype for which pembrolizumab is an approved first-line option. However, the markedly reduced HLA-I expression observed in these two patients should have prevented cytotoxic T cells from recognizing any of those neoantigens. The durable responses achieved under pembrolizumab treatment therefore suggest that an HLA-independent effector mechanism was at work, and the presence of PD-1-positive NK cells within the tumor, combined with PD-L1-positive tumor cells, provides a plausible cellular and molecular substrate for such a mechanism.
The findings carry practical implications that extend beyond the two cases themselves. The authors propose that assessing both PD-L1 and HLA-I expression on tumor cells at diagnosis could help identify which patients stand to benefit most from pembrolizumab, used alone or in combination with inhibitors of HLA-I-specific inhibitory NK receptors. This second component of the proposal draws on a well-characterized layer of NK-cell regulation. NK cells bear inhibitory receptors that recognize specific HLA class I allotypes and keep the killer cells in check as long as those molecules are present on healthy tissue. Tumors that retain partial HLA-I expression can still co-opt these inhibitory pathways to dampen NK attack, and drugs that block these HLA-I-specific inhibitory receptors are being explored as a way to fully unleash NK-cell activity. In a tumor with near-total HLA-I loss, such combination strategies could theoretically deepen the NK-cell response that checkpoint blockade alone initiates.
It is important to emphasize the limits of the evidence. This is a brief report describing two patients, and the authors themselves are careful to frame their conclusions as hypothesis-generating rather than definitive. The observation that PD-1-positive NK cells were present in the tumor microenvironment, and that tumor cells expressed PD-L1, is consistent with a model in which PD-1/PD-L1 blockade contributed to NK-cell-mediated antitumor activity, but it does not prove causation. Direct demonstration would require functional studies showing that NK cells isolated from such tumors kill autologous tumor targets in a manner enhanced by PD-1 blockade, as well as larger cohorts in which HLA-I status, PD-L1 expression and NK-cell infiltration are systematically correlated with clinical outcomes. The authors explicitly call for further investigation in larger and multicenter patient cohorts to test these ideas.
Even so, the report arrives at a moment of intense interest in the role of innate immunity in checkpoint inhibitor responses. HLA-I loss is increasingly recognized as a common mechanism of acquired resistance to PD-1 and PD-L1 inhibitors across tumor types, and it has traditionally been viewed as an almost uniformly bad prognostic sign. Studies in melanoma and lung cancer have shown that tumors losing HLA-I under immune pressure often escape T-cell control, and strategies to restore antigen presentation are an active area of drug development. The Genoa study flips this narrative in an interesting way for microsatellite-unstable colorectal cancer, suggesting that in the right immunological context, HLA-I downregulation may redirect the burden of tumor control from the adaptive to the innate immune system, and that checkpoint blockade may have a role in supporting that innate response.
The study also highlights the value of careful immunophenotyping of tumor specimens in patients who respond unexpectedly well or unexpectedly poorly to immunotherapy. Both patients in the report gave written informed consent under a protocol approved by the Ethics Committee of the Liguria Region in Genoa, and their tumor samples, obtained at surgical resection, allowed the team to characterize HLA-I expression, PD-L1 expression and the phenotype of tumor-infiltrating lymphocytes at a level of detail that routine clinical testing would not capture. As immunotherapy expands into new settings, this kind of granular analysis of exceptional responders has repeatedly proven to be one of the most productive routes to new biological insight, and the present report follows that tradition.
For patients with metastatic colorectal cancer, the immediate clinical message is one of cautious optimism rather than practice change. Pembrolizumab remains the standard of care for MSI-high metastatic disease, and nothing in this report alters that. What the study adds is a potential explanation for why some patients within this category do dramatically better than others, and a concrete, testable biomarker strategy: measuring HLA-I and PD-L1 on tumor cells at diagnosis, and perhaps assessing NK-cell infiltration and PD-1 expression, could eventually help oncologists predict who will benefit from pembrolizumab alone and who might need combination approaches that engage NK cells more directly. If larger studies confirm the hypothesis raised by these two cases, the long-standing division of labor in cancer immunology, in which T cells are seen as the primary effectors of checkpoint blockade and NK cells as supporting actors, may need to be redrawn for at least one important tumor type.
Subject of Research: NK-cell-mediated response to PD-1 blockade in HLA-I-deficient microsatellite-unstable metastatic colorectal cancer
Article Title: Response to pembrolizumab in metastatic colorectal cancer with microsatellite instability and markedly reduced HLA-I expression suggests NK cell involvement
Article References: Response to pembrolizumab in metastatic colorectal cancer with microsatellite instability and markedly reduced HLA-I expression suggests NK cell involvement. (n.d.). https://doi.org/10.1007/s00262-026-04594-7
Image Credits: AI Generated
DOI: 10.1007/s00262-026-04594-7
Keywords: NK cells, colorectal cancer, microsatellite instability, pembrolizumab, PD-1/PD-L1 blockade, HLA-I expression, immune-checkpoint inhibitors, tumor immunology, immunotherapy, BRAF V600E, cancer immunotherapy, tumor microenvironment
News Source: Nathaniel Bowman. (October 13, 2026). When Tumors Hide Their Flags, Natural Killer Cells May Take Over Against Colorectal Cancer. Scienmag.
