Definition and verification of novel metastasis and recurrence related signatures of ccRCC: A multicohort study. Issue 2 (30th August 2022)

• Record Type: Journal Article
• Authors: Jiang, Aimin; Pang, Qingyang; Gan, Xinxin; Wang, Anbang; Wu, Zhenjie; Liu, Bing; Luo, Peng; Qu, Le; Wang, Linhui
• Is Part Of: Cancer innovation. Volume 1:Issue 2(2022)
• Page Start: 146
• Page End: 167
• Publication Date: 2022-08-30
• Subjects: clear cell renal cell carcinoma -- metastasis -- recurrence -- machine learning -- multiple omics -- single‐cell sequencing; Cancer -- Research -- Periodicals; Oncology -- Periodicals; Cancer -- Treatment -- Technological innovations; Cancer -- Diagnosis -- Technological innovations; Cancer -- Research; Cancer -- Treatment -- Technological innovations; Oncology; Periodicals; 616.994
• DOI: 10.1002/cai2.25
• Languages: English
• ISSNs: 2770-9183

Abstract

Abstract: Background: Cancer metastasis and recurrence remain major challenges in renal carcinoma patient management. There are limited biomarkers to predict the metastatic probability of renal cancer, especially in the early‐stage subgroup. Here, our study applied robust machine‐learning algorithms to identify metastatic and recurrence‐related signatures across multiple renal cancer cohorts, which reached high accuracy in both training and testing cohorts. Methods: Clear cell renal cell carcinoma (ccRCC) patients with primary or metastatic site sequencing information from eight cohorts, including one out‐house cohort, were enrolled in this study. Three robust machine‐learning algorithms were applied to identify metastatic signatures. Then, two distinct metastatic‐related subtypes were identified and verified; matrix remodeling associated 5 (MXRA5), as a promising diagnostic and therapeutic target, was investigated in vivo and in vitro. Results: We identified five stable metastasis‐related signatures (renin, integrin subunit beta‐like 1, MXRA5, mesenchyme homeobox 2, and anoctamin 3) from multicenter cohorts. Additionally, we verified the specificity and sensibility of these signatures in external and out‐house cohorts, which displayed a satisfactory consistency. According to these metastatic signatures, patients were grouped into two distinct and heterogeneous ccRCC subtypes named metastatic cancer subtype 1 (MTCS1) and type 2 (MTCS2). MTCS2 exhibited poorer clinical outcomes and metastatic tendencies than MTCS1. In addition, MTCS2 showed higher immune cell infiltration and immune signature expression but a lower response rate to immune blockade therapy than MTCS1. The MTCS2 subgroup was more sensitive to saracatinib, sunitinib, and several molecular targeted drugs. In addition, MTCS2 displayed a higher genome mutation burden and instability. Furthermore, we constructed a prognosis model based on subtype biomarkers, which performed well in training and validation cohorts. Finally, MXRA5, as a promising biomarker, significantly suppressed malignant ability, including the cell migration and proliferation of ccRCC cell lines in vitro and in vivo. Conclusions: This study identified five robust metastatic signatures and proposed two metastatic probability clusters with stratified prognoses, multiomics landscapes, and treatment options. The current work not only provided new insight into the heterogeneity of renal cancer but also shed light on optimizing decision‐making in immunotherapy and chemotherapy. Abstract : Multicohorts of clear cell renal cell carcinoma (ccRCC) datasets and machine learning algorithms were used to identify and verify metastatic biomarkers. Two distinctive ccRCC subgroups were uncovered and compared at multiomics level and targeting matrix remodeling associated 5 (MXRA5) could be treated as an effective approach to suppress ccRCC progression.