A Lymph Node Metastasis Predictor (LN-MASTER) in Rectal Cancer

Overview

In this study, we aim to develop and validate an easy-to-use machine learning prediction model to preoperatively identify the lymph node metastasis status for rectal cancer patients by using these clinical data from three hospitals.

In this study, participants were recruited from the Cancer Hospital Chinese Academy of Medical Sciences and Peking Union Medical College (development set), Changhai Hospital, Naval Medical University (external validation set 1), and the Second Affiliated Hospital of Harbin Medical University (external validation set 2), between January 1, 2016, and December 31, 2020. According to the inclusion criteria, participants who (a) were in American Joint Committee on Cancer (AJCC) stages I -III rectal cancer and (b) underwent radical surgery were recruited. In contrast, the exclusion criteria were as follows: (a) other malignancies, (b) received treatment with endoscopic submucosal dissection (ESD), (c) metastatic lesions, (d) did not undergo lymph node dissection, (e) had unavailable assessed lymph node status, and (f) received neoadjuvant therapy. The lymph node metastasis (LNM) status was determined based on the pathological diagnosis of the surgical specimens. Clinicopathological features included sex, age, body mass index (BMI), comorbidity, distance from the lower edge of the tumor to the anus, carcinoembryonic antigen (CEA) levels, carbohydrate antigen 19-9 (CA19-9) levels, tumor size, degree of tumor differentiation, tumor histology, vascular or lymphatic vessel invasion, AJCC T stage, clinical diagnosis of LNM, and the pathological diagnosis of LNM. Among these, sex, age, BMI, and comorbidities of each participant, such as diabetes, hypertension, hyperlipidemia, and other chronic systemic diseases, were extracted from the electronic hospital information system. Preoperative CEA and CA19-9 levels were obtained from hematological examinations at the time of rectal cancer diagnosis. The distance from the lower edge of the tumor to the anus, differentiation degree, and tumor histology were recorded based on the results of endoscopy and endoscopic biopsies. The tumor diameter and clinical diagnosis of LNM were defined using preoperative pelvic MRI or CT. The diagnosis of vascular invasion, lymphatic vessel invasion, and LNM was based on postoperative pathological diagnosis.

Conditions

Interventions

Eligibility

Outcomes