Temporal changes in skeletal muscle index dataset for colorectal malignancy recurrence study
Description
This dataset comprises de-identified, retrospective demographic and imaging data for patients diagnosed with AJCC stage III colorectal cancer who underwent curative surgical resection at a single institution between 2000 and 2020. For each patient, skeletal muscle index (SMI) measurements were derived from routine abdominal CT scans at the third lumbar vertebral level. SMI was assessed at two timepoints: initially from the staging CT scan at diagnosis for a preoperative baseline, and again at the most recent follow-up scan prior to recurrence to capture changes preceding radiologic evidence of disease. Patients without recurrence were paired to those with recurrence through propensity score matching incorporating the duration between scans, neoadjuvant chemotherapy/radiation status, adjuvant chemotherapy/radiation status and steroid exposure. Key findings from the associated study indicate that a decline in skeletal muscle index postoperatively is independently associated with an increased risk of colorectal cancer recurrence. Receiver operating characteristics and decision curves suggest that SMI can stratify patients into clinically meaningful risk groups to guide surveillance. Note that version 2 rematched patients based on new inclusion/exclusion criteria and covariates. Some anonymous ID’s no longer correspond with those from version 1.
Files
Steps to reproduce
Records are organized by an anonymous patient ID with a prefix of either “R” for patients who experienced recurrence or “NR” for those without recurrence. For non-recurrence cases, IDs include a numeric suffix to indicate distinct follow-up scan timepoints, since these patients often undergo multiple surveillance scans. For example, NR100.1 and NR100.2 represent the same patient but correspond to follow-up CT scans obtained at different intervals after surgery. This structure enables more precise control during propensity score matching, as follow-up intervals can be closely aligned between recurrence and non-recurrence groups.
Institutions
- University of Vermont College of MedicineVT, Burlington