The study by Arai et al1 provides significant insights into the role of CD47 gene expression in colorectal cancer (CRC) and its implications for immunotherapy. The authors present compelling evidence that CD47 expression is associated with the activation of several oncogenic pathways and an immune-engaged tumor microenvironment (TME). However, while this work advances our understanding of CD47’s role in CRC, several key limitations and areas for improvement warrant attention, particularly in terms of study design, statistical analysis, and clinical applicability.

First, one of the main concerns of this study lies in the retrospective nature of the analysis and the heterogeneity of the patient cohort. The study population, which includes a large sample size of 14,287 patients, consists of a diverse group of patients with CRC, including those with various stages of disease, treatment histories, and molecular subtypes. This inherent variability may introduce confounding factors that can obscure the true effect of CD47 expression on tumor biology and patient outcomes. For example, the lack of stratification by disease stage or treatment modality limits the ability to discern whether CD47 expression directly correlates with therapeutic response or if it is an indirect result of treatment-induced immune modulation. Additionally, the retrospective nature of the study introduces potential biases that are difficult to control for, including variations in data collection and patient management over time. Future prospective studies with more homogeneous cohorts and better-defined patient groups would be crucial to confirm these findings and improve the clinical relevance of the results.

Second, the statistical methods employed in this study, while robust in many areas, could be further enhanced to address certain aspects of CD47’s prognostic value. The authors categorize patients into CD47-high and CD47-low groups using a median cut-off, which, while a common approach, is relatively simplistic and may overlook important nuances in the data. Given the continuous nature of gene expression data, the use of a more granular approach—such as quartile or continuous analysis—could provide a better understanding of the dose-dependent effect of CD47 on clinical outcomes. The quartile analysis performed in the study revealed a weak association between CD47 expression and prognosis, suggesting that a dichotomous approach may not fully capture the complexity of CD47’s role in CRC progression. More advanced statistical models, such as Cox regression or machine learning algorithms, could help adjust for confounders and provide more precise prognostic estimates based on CD47 expression levels.

The survival analysis presented does not consistently show a strong prognostic impact of CD47 expression on overall survival (OS) across all patients, with a non-significant result for the primary OS analysis. This raises the question of whether CD47 expression is truly a reliable prognostic marker in CRC or if its role is more nuanced, perhaps depending on tumor microenvironmental conditions or the patient’s immune status.2 Given the heterogeneity of survival outcomes in CRC, future studies should incorporate additional clinical covariates, such as tumor stage, microsatellite instability status, and mutational burden, to better assess the interplay between CD47 expression and these critical prognostic factors.

Furthermore, while the study provides a detailed analysis of immune cell infiltration and the expression of adaptive immune checkpoint genes in CD47-high and CD47-low tumors, several limitations are worth noting. First, the authors use a computational approach (Microenvironment Cell Population-counter) to estimate immune cell populations in the TME, which, although a valuable tool, may not fully capture the complexity of the immune landscape. Single-cell RNA sequencing or spatial transcriptomics could provide more detailed insights into the heterogeneity of immune cell populations within different tumor regions, potentially uncovering localized immune evasion mechanisms that are missed in bulk tissue analysis.3 What is more, the immune profiling in this study does not account for potential temporal dynamics in the TME, as immune cell infiltration can vary throughout the course of disease progression and therapy.4 Longitudinal studies that track immune changes over time in response to treatment could provide a clearer picture of how CD47 expression modulates immune surveillance and response to therapy.

Additionally, the study presents a promising hypothesis for therapeutic strategies targeting CD47 in combination with traditional immune checkpoint inhibitors (ICIs) and angiogenesis inhibitors. However, the clinical applicability of this combination therapy remains speculative without further validation in clinical trials. Although the authors suggest that CD47 inhibition could enhance the effectiveness of ICIs, they do not provide sufficient evidence of a direct benefit in terms of patient survival or treatment efficacy. The study’s focus on survival outcomes from angiogenesis inhibitors and ICIs, while valuable, does not address the full spectrum of potential combination therapies that could leverage CD47 inhibition. Moreover, the analysis does not explore the potential for resistance mechanisms to emerge with dual targeting of the innate and adaptive immune systems, a well-known challenge in cancer immunotherapy.

At last, the authors acknowledge that their study lacks comprehensive clinical data, such as treatment regimens and disease stage, which limits the ability to draw definitive conclusions about CD47 as a predictive biomarker for immunotherapy. A more detailed examination of how CD47 expression influences therapeutic outcomes—perhaps in conjunction with other biomarkers such as programmed death-ligand 1 or tumor mutational burden—would help to refine the patient populations most likely to benefit from CD47-targeted therapies.

To address these limitations, future studies should focus on:

Prospective cohort designs to minimize biases and better define patient subgroups.

Granular statistical modeling, incorporating continuous data analysis and advanced survival modeling techniques to refine prognostic predictions.

Integration of single-cell technologies to further dissect immune responses and TME heterogeneity.

Clinical trials evaluating the combination of CD47 inhibitors with ICIs and angiogenesis inhibitors in a well-defined cohort of patients, particularly those with microsatellite stable CRC, which is less responsive to current immune therapies.

In conclusion, while Arai et al’s study offers valuable insights into the molecular underpinnings of CD47 expression in CRC and its potential as a therapeutic target, several limitations need to be addressed in future research. Enhancing study design, refining statistical methods, and validating clinical outcomes are crucial steps in translating these findings into actionable therapeutic strategies for patients with CRC.

Not applicable.

Not applicable.