CoChat

Computed Tomography-Derived Sarcopenia and Two- Versus Three-Dimensional Body Composition for Prognostication in Colorectal Cancer Patients Receiving Chemoradiotherapy: An Automated Deep Learning Segmentation Study with External Reproducibility Assessment

Open access

The DOI matches its Crossref record; 55 of the 55 references it cites were found in a registry or an index; none of the 89 sources its claims cite is checked for support yet.

Abstract

Background: Skeletal muscle depletion predicts poor outcomes in gastrointestinal cancers, but whether volumetric three-dimensional (3D) body composition adds anything over the standard two-dimensional (2D) single-slice approach in colorectal cancer (CRC) patients receiving chemoradiotherapy is unclear. We quantified CT-derived sarcopenia and directly compared 2D and 3D body composition metrics from a fully automated deep learning pipeline. Methods: We retrospectively analyzed 368 patients with CRC. Body composition was quantified automatically from CT using SMAT-BC, a pipeline combining TotalSegmentator-based vertebral localization with an nnU-Net Residual Encoder XL network incorporating a Transformer bottleneck for four-class tissue segmentation. Single-slice L3 (2D) and L1–L5 volumetric (3D) indices were derived. The primary endpoint was overall survival (OS); recurrence-free survival (RFS) was secondary. Cox proportional hazards models with bootstrap optimism correction were used. Measurement reproducibility was assessed in 25 external TCGA-COAD cases. Results: Sarcopenia was strongly associated with both overall and recurrence-free survival (unadjusted OS HR 2.16, 95% CI 1.55–3.00; unadjusted RFS HR 1.83, 95% CI 1.36–2.47; both raw p < 0.001; adjusted OS HR 1.89, 95% CI 1.57–2.28; adjusted RFS HR 1.44, 95% CI 1.22–1.70; FDR-adjusted p < 0.0001 for both). L3 single-slice indices were strongly correlated with volumetric indices (r = 0.91 for muscle index) and added discriminant value over the clinical model for overall survival (optimism-corrected C-index: clinical 0.602 (95% CI 0.578–0.626), clinical + 2D 0.631 (95% CI 0.609–0.654), clinical + 3D 0.599 (95% CI 0.574–0.624); DeLong p = 0.002 for clinical + 2D vs. clinical, FDR-adjusted p = 0.006). The 2D- and 3D-augmented models yielded overlapping bootstrap confidence intervals and were not clinically meaningfully different in this cohort (OS ΔC = +0.032, 95% CI +0.013 to +0.051; FDR-adjusted p = 0.006; RFS ΔC = +0.008, 95% CI −0.011 to +0.027; FDR-adjusted p = 0.612). Conclusions: Automated CT-derived sarcopenia is an independent predictor of survival in CRC patients receiving chemoradiotherapy. In our cohort, single-slice L3 measurement matched or exceeded volumetric discrimination, but the 2D- and 3D-augmented models yielded overlapping bootstrap confidence intervals for both endpoints: for overall survival, the DeLong FDR-adjusted p value for the 2D-versus-3D contrast was 0.006, and 0.612 for recurrence-free survival. Because no non-inferiority margin was pre-specified, the 2D–3D comparison is presented as exploratory, and we make no formal claim of non-inferiority or equivalence for either endpoint. The findings support single-slice L3 measurement as an efficient biomarker for risk stratification but warrant external validation in larger prospectively designed cohorts.

Overview

CoChat Summary

TL;DR

In 368 colorectal cancer patients receiving chemoradiotherapy, fully automated CT-derived sarcopenia independently predicted worse overall and recurrence-free survival. Single-slice L3 body composition matched or exceeded volumetric L1–L5 discrimination in this cohort, but the 2D-versus-3D comparison was exploratory and external assessment was limited to measurement reproducibility in 25 TCGA-COAD cases.

Key findings

  • Sarcopenia was associated with worse survival: unadjusted OS HR 2.16 (95% CI 1.55–3.00) and RFS HR 1.83 (95% CI 1.36–2.47); adjusted OS HR 1.89 (95% CI 1.57–2.28) and RFS HR 1.44 (95% CI 1.22–1.70), with FDR-adjusted p < 0.0001 for both. Five-year OS was 51.9% vs. 74.8% and five-year RFS 35.9% vs. 57.5% in sarcopenic vs. non-sarcopenic patients (log-rank p < 0.001 for both).
  • L3 single-slice indices correlated strongly with volumetric indices: L3 muscle area vs. L1–L5 muscle volume r = 0.93, and L3 SMI vs. muscle volume index r = 0.91. For OS, optimism-corrected C-index was 0.602 (95% CI 0.578–0.626) for clinical, 0.631 (95% CI 0.609–0.654) for clinical + 2D, and 0.599 (95% CI 0.574–0.624) for clinical + 3D; clinical + 2D vs. clinical DeLong p = 0.002, FDR-adjusted p = 0.006.
  • Direct 2D-versus-3D contrast: OS ΔC = +0.032 (95% CI +0.013 to +0.051; FDR-adjusted p = 0.006); RFS ΔC = +0.008 (95% CI −0.011 to +0.027; FDR-adjusted p = 0.612). The 2D- and 3D-augmented models yielded overlapping bootstrap confidence intervals, and the paper makes no formal non-inferiority or equivalence claim.
  • External reproducibility in 25 TCGA-COAD cases: median SMI 45.0 cm²/m², sarcopenia prevalence 52.0%; external SMI distribution did not differ significantly from the internal cohort (Mann–Whitney p = 0.66), and L3 muscle area vs. L1–L5 muscle volume correlation was r = 0.94 (95% CI 0.86–0.97).

Methods

  • Retrospective single-center study of 368 patients with histologically confirmed colorectal cancer who received chemoradiotherapy at the Second Affiliated Hospital of Zhejiang University School of Medicine between January 2015 and December 2022; primary endpoint was overall survival, secondary endpoint recurrence-free survival.
  • Body composition was quantified automatically from contrast-enhanced abdominal CT using SMAT-BC, combining TotalSegmentator-based L1–L5 vertebral localization with an nnU-Net Residual Encoder XL network incorporating a Transformer bottleneck for four-class tissue segmentation; mid-L3 single-slice (2D) and L1–L5 volumetric (3D) indices were derived.
  • Cox proportional hazards models with bootstrap optimism correction (500 replications), Harrell C-index, DeLong tests, and Benjamini–Hochberg FDR correction were used; measurement reproducibility was assessed in 25 external TCGA-COAD cases processed in inference-only mode.

Limitations

  • Stated: retrospective single-center design with possible residual confounding; the external TCGA-COAD cohort (n = 25) was used only for measurement reproducibility, not survival prediction, and its wide correlation CI (0.86–0.97) reflects the small sample size; sarcopenia thresholds are not universally standardized and muscle strength/physical function were not assessed; clinical variable models were not penalized for number of predictors and are exploratory; no non-inferiority margin was pre-specified, so the 2D–3D comparison is exploratory.
  • Stated/evident: performance status was operationalized as a binary proxy rather than formal ECOG; scanner vendor metadata was not retained for the present cohort; external-cohort per-vertebra quality-control logs were not recorded; exact training/validation/test split sizes were not recorded; regimen-specific chemotherapy data were unavailable.
  • Evident: the cross-sectional design cannot directly test proposed longitudinal advantages of 3D assessment, and external prognostic validation in a larger multi-institutional cohort with linked outcomes is still needed.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.3390/healthcare14172846

Cites

55 of 55 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 SMAT-BC automated body composition pipeline. DICOM CT is converted to NIfTI, vertebrae are localized using TotalSegmentator, four-class tissue segmentation is performed using an nnU-Net Residual Encoder XL network with a Transformer bottleneck and HU-guided constraints, and 2D L3 plus 3D L1-L5 body composition indices are produced for downstream Cox survival modeling.
Figure 1 SMAT-BC automated body composition pipeline. DICOM CT is converted to NIfTI, vertebrae are localized using TotalSegmentator, four-class tissue segmentation is performed using an nnU-Net Residual Encoder XL network with a Transformer bottleneck and HU-guided constraints, and 2D L3 plus 3D L1-L5 body composition indices are produced for downstream Cox survival modeling.CC BY · From the PDF
Figure 2 Kaplan-Meier estimates of ( A ) overall survival and ( B ) recurrence-free survival stratified by CT-derived sarcopenia. Shaded bands denote 95% pointwise confidence intervals. Numbers at risk (sarcopenic/non-sarcopenic) at 0, 12, 24, 36, 48, and 60 months were 170/198, 152/188, 121/174, 89/156, 58/121, and 32/78 for overall survival and 170/198, 138/178, 96/152, 68/121, 41/92 and 22/58 for recurrence-free survival.
Figure 2 Kaplan-Meier estimates of ( A ) overall survival and ( B ) recurrence-free survival stratified by CT-derived sarcopenia. Shaded bands denote 95% pointwise confidence intervals. Numbers at risk (sarcopenic/non-sarcopenic) at 0, 12, 24, 36, 48, and 60 months were 170/198, 152/188, 121/174, 89/156, 58/121, and 32/78 for overall survival and 170/198, 138/178, 96/152, 68/121, 41/92 and 22/58 for recurrence-free survival.CC BY · From the PDF
Image 3
CC BY · From the PDF
Figure 3 Box-and-whisker plots of two- and three-dimensional muscle indices stratified by sarcopenia status ( A , B ) and vital status ( C , D ). Boxes show the interquartile range; central lines, the median; and whiskers, 1.5 times the interquartile range. Individual patients are overlaid. Per-panel n and medians are annotated; significance is based on the Mann-Whitney U test (**** p < 0.0001; ns, not significant).
Figure 3 Box-and-whisker plots of two- and three-dimensional muscle indices stratified by sarcopenia status ( A , B ) and vital status ( C , D ). Boxes show the interquartile range; central lines, the median; and whiskers, 1.5 times the interquartile range. Individual patients are overlaid. Per-panel n and medians are annotated; significance is based on the Mann-Whitney U test (**** p < 0.0001; ns, not significant).CC BY · From the PDF
Image 5
CC BY · From the PDF
Image 6
CC BY · From the PDF
Figure 4 Forest plot of multivariable Cox model for overall survival incorporating two-dimensional body composition. Squares denote hazard ratios; horizontal lines denote 95% confidence intervals; dashed line marks hazard ratio of 1.
Figure 4 Forest plot of multivariable Cox model for overall survival incorporating two-dimensional body composition. Squares denote hazard ratios; horizontal lines denote 95% confidence intervals; dashed line marks hazard ratio of 1.CC BY · From the PDF
Figure 5 Concordance index of clinical, clinical plus two-dimensional, and clinical plus threedimensional body composition models for overall survival (optimism-corrected) and recurrence-free survival (apparent estimates).
Figure 5 Concordance index of clinical, clinical plus two-dimensional, and clinical plus threedimensional body composition models for overall survival (optimism-corrected) and recurrence-free survival (apparent estimates).CC BY · From the PDF

Verification

Claims the paper makes

0 Valid · 0 Not valid · 43 Not checked · 150 Own results and interpretations

81 more claims

Create a free account to see all 193 claims and to verify the 89 references still open.

See all claims — free account
  • Probably not a substantial claim. 49
  • It may state two separate claims as one. 73
  • The quote only partly states this claim. 4

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

5 of 6 · the rest aren’t loaded

1 more finding

Create a free account to see every finding of this check.

See all findings — free account

Citations

Cites

Show all 55 references

Cited by

Still being checked.

Metadata

Authors
Wang, D., Sui, J., Chen, J., Chen, L., Yang, Q., Wang, Y., & Lin, S.
Affiliations
  • Second Affiliated Hospital of Zhejiang University (CN)
  • University of California San Diego (US)
Corresponding author
  • Wang, Da
  • Lin, Shuangxiang
Venue
Healthcare
Publisher
MDPI AG
Published
2026-09-04
Volume
14
Issue
17
Pages
2846
ISSN
2227-9032 (ISSN-L)
OpenAlex
W7208719856
Cited by
0
References
53
FWCI
0.00
Type
journal article
Language
en
Open access
Locations
  • Healthcare (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
Funders
  • National Natural Science Foundation of China (82303951)
  • Noncommunicable Chronic Diseases-National Science and Technology Major Project (2023ZD0502900, 2025ZD0544500)
  • Zhejiang Province Leading Geese Plan (2025C02197)
  • Key Research and Development Program of Zhejiang Province (2024C03143)
Topics
  • Health Sciences › Medicine › Physiology (medical) › Nutrition and Health in Aging
  • Health Sciences › Medicine › Radiology, Nuclear Medicine and Imaging › Radiomics and Machine Learning in Medical Imaging
  • Health Sciences › Medicine › Physiology (medical) › Body Composition Measurement Techniques
Keywords
colorectal cancer, chemoradiotherapy, nnU-Net, TotalSegmentator, outcome prediction, overall survival, recurrence-free survival, measurement reproducibility, TCGA-COAD, prognostic factors

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-11-21, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-21 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Computed Tomography-Derived Sarcopenia and Two- Versus Three-Dimensional Body Composition for Prognostication in Colorectal Cancer Patients Receiving Chemoradiotherapy: An Automated Deep Learning Segmentation Study with External Reproducibility Assessment. https://app.cochat.ai/papers/10.3390/healthcare14172846
The paper
Wang, D., Sui, J., Chen, J., Chen, L., Yang, Q., Wang, Y., & Lin, S. (2026). Computed Tomography-Derived Sarcopenia and Two- Versus Three-Dimensional Body Composition for Prognostication in Colorectal Cancer Patients Receiving Chemoradiotherapy: An Automated Deep Learning Segmentation Study with External Reproducibility Assessment. Healthcare. https://doi.org/10.3390/healthcare14172846

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.