CXCR4 in colorectal cancer: from prognostic marker to immunotherapy guide?

Colorectal cancer (CRC) remains a major global health burden, and despite advances in systemic therapy, outcomes for metastatic disease remain poor. A central challenge in CRC immuno-oncology is the limited applicability of immune checkpoint blockade (ICB): only a small subset of patients—those with mismatch repair deficiency (dMMR)—derive substantial benefit, leaving the majority of patients with microsatellite stable (MSS) disease without effective immunotherapeutic options.1 2 This unmet need has driven intensive efforts to identify additional biomarkers that can better stratify patients and guide immunotherapy strategies.
Among emerging candidates, chemokine signalling pathways have gained increasing attention. The CXCL12–CXCR4 axis plays a critical role in tumour biology, including cell migration, metastasis and immune cell trafficking within the tumour microenvironment (TME).3 Elevated CXCR4 expression has been consistently associated with aggressive disease and poor prognosis across multiple cancer types, including CRC.4 However, its role in shaping the immune landscape—and particularly its relevance to immunotherapy response—has remained incompletely understood.
In this issue of BMJ Oncology, Gholami and colleagues provide one of the most comprehensive analyses to date of CXCR4 mRNA expression in CRC, leveraging a large real-world dataset of more than 15 000 tumors with integrated genomic, transcriptomic and immunologic profiling.5 Their findings both confirm and extend prior observations. Consistent with existing literature, high CXCR4 expression is associated with adverse clinical features and poorer overall survival, particularly in primary tumours.4 At the same time, the study reveals a more nuanced and potentially clinically relevant insight: tumours with high CXCR4 expression exhibit features of an ‘inflamed’ microenvironment, including higher tumour mutational burden (TMB), increased programmed death-ligand 1 (PD-L1) expression and enhanced immune cell infiltration.
Most strikingly, in a subset of patients treated with pembrolizumab, high CXCR4 expression was associated with improved survival outcomes. This apparent paradox—where a marker of aggressive biology also predicts better response to immunotherapy—highlights the complex dual role of CXCR4 within the TME. On one hand, CXCR4 signalling may facilitate tumour progression and immune evasion; on the other, it may also reflect or contribute to an immune-active environment that is more amenable to checkpoint inhibition. Similar observations linking CXCR4 expression to immunologically active tumours have been reported in other malignancies, including lung and pancreatic cancers.6 7 These findings suggest that CXCR4 could function not merely as a prognostic marker but as a context-dependent biomarker that helps identify tumours with latent immunogenicity.
The clinical implications of this work are significant. If validated prospectively, CXCR4 expression could be incorporated into multiparameter biomarker frameworks alongside TMB, PD-L1 and MMR status to refine patient selection for immunotherapy. This may be particularly relevant for the large population of MSS CRC patients, where current biomarkers are insufficient.1 2 Furthermore, the study reinforces interest in therapeutic strategies targeting the CXCR4–CXCL12 axis, either alone or in combination with ICB, as a means to modulate the TME and overcome resistance.8
At the same time, several limitations should be carefully considered. The study is retrospective and relies on real-world data derived from a commercial molecular profiling platform, which introduces potential selection bias and limits clinical annotation. The use of insurance claims as a surrogate for overall survival, while pragmatic, may lack the precision of prospectively collected clinical endpoints. In addition, the immunotherapy-treated cohort is relatively small, and important confounders—such as prior lines of therapy—are not fully characterised. From a biological perspective, the reliance on bulk RNA sequencing and computational deconvolution limits spatial resolution, precluding distinction between intratumoral and peritumoral immune infiltration.9
Importantly, the dichotomisation of CXCR4 expression using quartiles, while methodologically straightforward, may not capture the full biological continuum of expression levels. Future studies should explore alternative modelling approaches and seek to define clinically actionable thresholds.
Looking ahead, several key questions emerge. First, prospective validation in well-annotated clinical cohorts will be essential to establish CXCR4 as a predictive biomarker for ICB. Second, mechanistic studies are needed to disentangle the dual role of CXCR4 in tumour progression and immune activation, particularly in different metastatic niches such as the liver. Third, integration of CXCR4 into composite biomarker models—potentially augmented by spatial transcriptomics or single-cell profiling—may provide a more precise framework for patient stratification. Finally, clinical trials combining CXCR4 inhibition with checkpoint blockade may help translate these insights into therapeutic benefit.8
In summary, this study advances our understanding of the complex interplay between tumour biology and immune responsiveness in CRC. By positioning CXCR4 at the intersection of prognosis and immunotherapy response, it opens new avenues for biomarker development and therapeutic innovation. Whether CXCR4 will ultimately reshape clinical decision-making will depend on rigorous validation, but the current work provides a compelling rationale for continued investigation.