Scientific Papers

Preoperative contrast-enhanced CT imaging and clinicopathological characteristics analysis of mismatch repair-deficient colorectal cancer | Cancer Imaging

In recent years, anti-PD-1 immunotherapy has shown promising results in improving survival for both metastatic and non-metastatic MSI-H/dMMR CRC patients [6]. The potential of MSI/MMR status to guide personalized therapy, predict prognosis, and assess the efficacy of targeted immunotherapy is becoming increasingly recognized.

Our study aimed to analyse differences between dMMR and pMMR CRC in terms of clinicopathological and CT characteristics. Our findings suggest that increased dMMR risk is most highly associated with the right hemi-colon, HR (1/3–2/3 group and > 3/2 group) and the number of LNs with LD ≥ 8 mm. We also found that the dMMR protective factors correlated strongly with CEA positivity and the number of LN metastasis.

Interestingly, the number of LN metastasis served as a protective factor, while the number of LNs with LD ≥ 8 mm on CT played a vital role as a risk factor. For dMMR CRC, we propose that enlarged lymph nodes observed on CT may be attributed to the robust immune response of the primary tumour rather than to lymph node metastasis. This finding is not unprecedented, as previous studies have reported a correlation between lymph nodes and immune response in primary tumours [16,17,18,19]. Lal [17] et al. demonstrated that high LN yields in stages II and III colon cancer resection were significantly regulated by broad B- and T-cell adaptive immune responses. Furthermore, MSI-H/dMMR CRC has been shown to have marked “Crohn’s-like” lymphocyte infiltration [2, 20] and tends to have less extensive nodal metastases [21, 22]. These studies suggest that the size and number of LNs may increase in dMMR CRC. However, reactive proliferative LNs and LN metastases are challenging to differentiate on CECT due to similar enhancement patterns and morphological features. Typically, both types exhibit isolated and homogeneous enhancement, often with a round shape. Imaging methods that rely on lymph node size, enhancement pattern and morphological features to estimate the probability of LN metastasis are unreliable and may result in false-positive results, particularly for dMMR CRC. As a result, clinical N stage may be overestimated for dMMR CRC, and lymph nodes should be carefully considered as target lesions to assess the efficacy of chemotherapy or immunotherapy. These findings are crucial for assessing the efficacy of imaging methods for anti-PD-1 therapy and clinical staging.

Although the precise role of PD-1-positive T cells in LNs is still unclear, recent evidence suggests that these cells may play a crucial role in PD-1 blockade-mediated antitumor immunity by enriching tumour-specific progenitor T cells in LNs [23, 24]. In animal models, antitumor immunity is unable to halt tumour progression when lymphocyte migration from LNs is blocked or tumour-draining LNs are dissected [25]. Despite some guidelines recommending dissection of at least 12 lymph nodes, increased LN yield does not increase the number of LN metastases [17]. Our study also suggests that LN yield is not associated with lymph node metastasis (p > 0.05). In fact, excessive LN yield might lead to poor prognosis of patients with dMMR CRC [18]. Compared to pMMR CRC, dMMR CRC has a lower tendency for lymph node and distant metastasis [26, 27]. Our MLR results also confirm that LN metastasis is a protective factor for dMMR. Therefore, non-metastatic LN dissection should be carefully considered in dMMR CRC [18, 25]. We believe that combining multi-dimensional information, such as imaging and clinicopathological data, with intraoperative visualization, such as fluorescence molecular imaging [28], will help avoid excessive lymph node dissection in some patients, especially dMMR CRC patients. This approach may be an essential research direction for future studies.

In our research, we found that a tumour in the right hemi-colon was a risk factor in patients with dMMR but that a tumour in the left hemi-colon was a protective factor. The right hemi-colon is more likely to have MCs and shows low differentiation in dMMR CRC. This is consistent with previous studies [2, 29, 30]. It has been proposed that right hemi-colon cancer exhibits higher levels of infiltration of CD4 + T cells and CD8 + T cells than left hemi-colon cancer [31], which is similar to the aforementioned discussion of the relationship between LN and the immune response.

HR was the most significant risk factor for dMMR in our study. Both the mucinous and necrotic components of the tumour show hypoattenuation on CECT. Univariate analysis revealed that MCs and MA were risk factors for dMMR, which was consistent with previous studies [32, 33]. However, MCs and MA were excluded from the MLR analysis, as we believe that HR may be a better predictor for dMMR. Indeed, HR not only reflects the tumour composition but also provides an accurate representation of the degree of mucous or necrosis present.

CRC patients are routinely tested for CEA as a tumour marker for diagnosis and surveillance. In our study, pre-surgical CEA positivity was a protective factor. However, the correlation between CEA and MMR status remains controversial [13, 34, 35]. Based on our study and a review of the existing literature, we found that qualitative analysis of CEA may provide some predictive value for MMR status. However, the reasons why qualitative analysis is superior to quantitative analysis remain unclear. It is possible that quantitative levels of CEA are not associated with the presence of mismatch repair defects but that qualitative judgment of CEA correlates with mismatch repair defects, suggesting that qualitative judgment of CEA has greater value in predicting MMR status. The inconsistencies might be partially attributed to differences in population and sample size. Further research is needed to explore the underlying mechanisms for this observation. Therefore, the association of MMR status with CEA must be interpreted with caution and requires validation using a larger sample size.

Similarly, Zeng [36] et al. conducted a study on preoperative gastric cancer microsatellite instability prediction using imaging and radiomic features, as well as clinical data derived from contrast-enhanced CT. They developed a nomogram based on age, CT-reported N stage, and radiomic score. Although our study also considered clinical indicators, we incorporated additional pathological features such as pathological T stage, N stage, and tumour differentiation, with pathological results as a reference. Furthermore, our study had a larger sample size. Additionally, our research encompassed both qualitative and quantitative investigations. Specifically, we explored the degree of tumour enhancement, the low-density ratio within the tumour, and lymph node involvement. While this study may lack certain quantitative tumour features, we provide detailed analysis of lymph node characteristics. In contrast to Zeng et al., who relied on radiomics to construct a nomogram, our study focused on CT imaging features combined with clinical and pathological data, making it highly applicable and easily replicable for other researchers.

To our knowledge, this is the study to compare clinical, pathological, and CECT features of primary CRC and LNs to predict MSI/MMR status. Our results suggest that in vivo, imaging features of a tumour may be better than clinicopathological features in revealing the characteristics of the tumour itself with regard to some aspects. However, several limitations should be noted. First, this was a retrospective, single-centre study, and the findings need to be confirmed in a large-scale, prospective study. During the study period, we were only able to collect a small number of samples from the dMMR group. Therefore, we aimed to minimize the number of features in our prediction model to ensure reliable results while reducing the risk of overfitting. We are pleased to report that the DCA of our prediction model demonstrated some clinical utility, despite the small sample size of the dMMR group. Additionally, we acknowledge the potential biases that may exist in our study. Patients who received neoadjuvant therapy were excluded from the analysis due to the potential influence of treatment on DNA mismatch repair (MMR) status, as these individuals often present with advanced-stage disease. Furthermore, patients who did not undergo surgery were excluded to avoid inclusion bias resulting from the presence of distant metastasis identified through preoperative imaging or advanced-stage disease that rendered them unsuitable for surgical intervention. Concurrent malignancies were also excluded to minimize potential confounding effects on tumour-related blood markers, such as CEA. Lastly, patients with multiple primary colorectal cancers were excluded to mitigate the impact on N staging caused by the presence of multiple lesions.

Source link