Radiomics-clinical nomogram for response to chemotherapy in synchronous liver metastasis of colorectal cancer: Good, but not good enough

Abstract

There remains a persistent unmet need to detect the disease nonresponse (nonDR) subgroup before adjuvant therapy in synchronous liver metastasis patients with colorectal cancer. Ma’s radiomics-clinical nomogram shows potential for the early detection of nonDR subgroups, but it is not good enough owing to at least three limitaions, which we address in this letter to the editor. First, the study did not explore RAS/BRAF mutations, HER2 amplifications, etc. to complement the current nomogram. Second, the nomogram was not validated in left- and right-sided tumors separately. Third, the most critical factor for determining the success of adjuvant therapy should be resectability rather than tumor size shrinkage, which was used in the study.

Core Tip: There remains a persistent unmet need to detect the disease nonresponse subgroup before adjuvant therapy in synchronous liver metastasis patients with colorectal cancer. Ma’s radiomics-clinical nomogram is currently not good enough, as the study did not explore the statuses of certain tumor genes, did not validate the nomogram in left- and right-sided tumors separately, and used tumor size shrinkage rather than resectability to judge the success of adjuvant therapy.

TO THE EDITOR

Ma et al[1] recently published a novel study investigating the effect of magnetic resonance imaging-radiomics in predicting chemotherapeutic response in synchronous liver metastasis (SLM) patients with colorectal cancer (CRC). They proposed a radiomics-clinical nomogram (including the radiomics score, CA19-9, and lymphatic staging) with an area under the curve of 0.809, suggesting high predictive accuracy.

We congratulate the authors for their creative work, as in decision-making for adjuvant therapy in CRC patients with unresectable SLM, there remains a persistent unmet need to detect the disease nonresponse (nonDR) subgroup. Early detection of nonDR patients, aided by the radiomics-clinical nomogram of Ma’s study, could result in substantial changes in subsequent therapeutic plans. For instance, in nonDR cases, more aggressive regimens could be applied instead of the frequently used FOLFOX or CAPOX, such as administration of bevacizumab to inhibit vascular endothelial growth factor or pembrolizumab for immunotherapy. Local regional therapies, including radiofrequency ablation and transcatheter arterial chemoembolization, could also be considered to treat SLM.

However, despite the aforementioned merit, there are at least three limitations to be discussed concerning this nomogram. First, although the authors explored tumor biomarkers, including CEA and CA19-9, to complement radiomics, the statuses of some critical tumor genes (e.g., RAS/BRAF mutations, HER2 amplification, and MSI/MMR status) were not examined, despite the relevant recommendation in the latest National Comprehensive Cancer Network guideline[2]. Second, it is noteworthy that the biological behaviors of CRC differed depending on the anatomical location[3]. For instance, right-sided CRC patients with SLM were unlikely to respond to cetuximab and panitumumab as first-line therapy. Therefore, the performance of Ma’s nomogram should be validated in right- and left-sided CRC separately. Last but not least, the most critical limitation was that the success of adjuvant therapy in CRC patients with SLM should be resectability, rather than tumor size shrinkage used in this study.

In conclusion, in CRC patients with SLM, Ma’s radiomics-clinical nomogram shows potential for clinical utilization. However, it is currently not good enough.