Three-Millimeter Peritumoral MRI Margin Best Predicts Soft Tissue Sarcoma Progression

A 3-millimeter peritumoral margin provides optimal prognostic information for predicting progression-free survival in patients with soft tissue sarcoma, according to a retrospective multi-sequence magnetic resonance imaging study published in Cancer Imaging. Researchers Yu Zhang, Tingting Cui, Yifeng Zhu, Lijun Wei, Huayong Cai, and colleagues analyzed imaging and clinical data from 305 patients with soft tissue sarcomas located in the trunk or extremities to establish an integrated prognostic framework.

The primary objective of the investigation was to construct an integrated prognostic model combining intratumoral and peritumoral radiomic features with semantic magnetic resonance imaging characteristics. By evaluating these imaging features simultaneously, the investigators sought to improve risk stratification and the prediction of progression-free survival compared to models relying solely on conventional observations or isolated tumor regions.

Radiomic features were extracted from both intratumoral volumes and surrounding peritumoral tissues across three distinct expansion distances: 3 millimeters, 5 millimeters, and 10 millimeters. Imaging analysis incorporated dual sequences consisting of contrast-enhanced T1-weighted images and fat-suppressed T2-weighted images, allowing the research team to capture complementary tissue characteristics across both the primary lesion and adjacent margins.

To identify the most reliable prognostic variables, candidate radiomic features were filtered and selected through intraclass correlation coefficients, univariate Cox proportional hazards regression, and least absolute shrinkage and selection operator Cox regression. The investigators then constructed prognostic models combining selected radiomics scores with significant magnetic resonance imaging features, evaluating performance using Harrell’s concordance index, calibration curves, and decision curve analysis.

Among the three evaluated expansion distances, the 3-millimeter peritumoral margin demonstrated the highest predictive performance. Integrated models combining intratumoral and 3-millimeter peritumoral radiomic features from both contrast-enhanced T1-weighted and fat-suppressed T2-weighted sequences consistently outperformed single-region models. Subgroup analyses stratified by imaging phenotype confirmed that the 3-millimeter margin retained superior predictive accuracy across both cellular and non-cellular sarcoma phenotypes.

The final multimodal model, which combined dual-sequence radiomics scores with peritumoral enhancement, achieved a concordance index of 0.888 in the validation cohort. Subsequent correlation analyses revealed that peritumoral radiomics scores were most strongly associated with peritumoral T2 hyperintensity, while distinct correlation patterns across the two magnetic resonance sequences highlighted their complementary clinical value. The authors concluded that this non-invasive tool facilitates personalized risk stratification and may help guide adjuvant treatment decisions in soft tissue sarcoma management.

The methodological framework focused on extracting and evaluating radiomic features across distinct tumor regions in soft tissue sarcomas. Radiomic features were derived from both intratumoral volumes and peritumoral tissue boundaries using contrast-enhanced T1-weighted (CE-T1WI) and fat-suppressed T2-weighted (FS-T2WI) magnetic resonance sequences. The experimental design tested three explicit peritumoral expansion distances: 3 millimeters, 5 millimeters, and 10 millimeters. To filter the extracted radiomic variables, the investigators implemented a multi-stage feature selection process using intraclass correlation coefficients, univariate Cox regression, and least absolute shrinkage and selection operator (LASSO) Cox regression. The selected features were then used to construct dual-sequence Rad-scores integrating both imaging sequences across the evaluated anatomical margins.

Model construction integrated these Rad-scores with significant magnetic resonance imaging characteristics, including peritumoral enhancement. The research team evaluated prognostic performance using Harrell’s concordance index, calibration curves, and decision curve analysis to establish predictive accuracy for progression-free survival. Within the validation cohort, the final multimodal model combining dual-sequence Rad-scores with peritumoral enhancement demonstrated a concordance index of 0.888. Among all evaluated expansion boundaries, the 3-millimeter peritumoral margin yielded the highest predictive performance when compared against the 5-millimeter and 10-millimeter distances. Furthermore, integrated models that incorporated both intratumoral and 3-millimeter peritumoral features from contrast-enhanced T1-weighted and fat-suppressed T2-weighted sequences outperformed models constructed from a single region alone.

Subgroup analyses stratified by imaging phenotype demonstrated that the predictive superiority of the 3-millimeter peritumoral margin remained consistent across major soft tissue sarcoma presentations. When evaluated in cellular and non-cellular sarcoma phenotypes, the 3-millimeter margin consistently demonstrated superior performance in both groups. The authors noted that the optimality of the 3-millimeter peritumoral margin was thereby validated across the major soft tissue sarcoma imaging phenotypes identified in the study.

Correlation analyses between the peritumoral Rad-scores and semantic magnetic resonance imaging features provided additional insight into the imaging characteristics captured by the model. Peritumoral Rad-scores were most strongly associated with peritumoral T2 hyperintensity. Furthermore, differential correlation patterns emerged between the contrast-enhanced T1-weighted and fat-suppressed T2-weighted sequences, supporting their complementary roles within the multimodal framework. According to the authors, the resulting non-invasive model integrates intra- and peritumoral features with key magnetic resonance characteristics to facilitate personalized risk stratification, potentially informing adjuvant treatment decisions and supporting clinical translation in the management of soft tissue sarcomas.

Subject of Research: MRI radiomics prognostic modeling in soft tissue sarcoma

Article Title: The 3‑mm peritumoral margin provides optimal prognostic information in soft tissue sarcoma radiomics: a retrospective multi‑sequence MRI study

Article References: Zhang, Y., Cui, T., Zhu, Y., Wei, L., & Cai, H. (2026). The 3‑mm peritumoral margin provides optimal prognostic information in soft tissue sarcoma radiomics: a retrospective multi‑sequence MRI study. Cancer Imaging. https://doi.org/10.1186/s40644-026-01123-8

Image Credits: AI Generated

DOI: 10.1186/s40644-026-01123-8

Keywords: soft tissue sarcoma, radiomics, magnetic resonance imaging, peritumoral margin, progression-free survival, prognostic modeling, CE-T1WI, FS-T2WI, peritumoral, margin, provides, optimal