Apparent diffusion coefficient metrics for treatment response assessment in soft tissue sarcoma: Reproducibility and prognostic value of whole-tumor and sub-regional analysis

Objective To evaluate the inter-observer reproducibility and predictive value of apparent diffusion coefficient (ADC) metrics derived from diffusion-weighted MRI for assessing histopathologic response to neoadjuvant radiotherapy in soft tissue sarcoma (STS). Materials and methods This retrospective, single-institution cohort study included 23 patients with histologically proven STS treated with neoadjuvant radiotherapy. Patients underwent 1.5 T or 3.0 T MRI, including diffusion-weighted sequences (b = 200, 800 s/mm 2 ). Three independent radiologists measured mean tumor ADC, high-ADC, and low-ADC regions at baseline and post-radiotherapy. Reproducibility was assessed using the Intraclass Correlation Coefficient (ICC). Response prediction and clinicopathological associations were evaluated using Mann–Whitney U tests, logistic regression, and Spearman rank correlations. Results Single-timepoint ADC measurements demonstrated high inter-observer reproducibility, particularly for mean tumor ADC (ICC = 0.91–0.95). Reproducibility for ADC changes over time was lower; agreement for mean tumor ADC (ICC = 0.79) outperformed subregional values (ICC = 0.06–0.63). Post-treatment mean tumor ADC was significantly elevated in responders (median 2.39 × 10⁻ 3 mm 2 /s) versus non-responders (1.50 × 10⁻ 3 mm 2 /s, p = 0.02). High follow-up ADC values (mean and high-ADC subregions) were the strongest predictors of pathological response (OR = 20.3–33.1, all p < 0.05). Follow-up ADC values inversely correlated with the percentage of viable tumor cells ( p < 0.05). Conclusion Post-treatment ADC measurements were highly reproducible and best predicted pathological response in our cohort. Longitudinal changes were less robust and require further research into the effect of standardized DWI protocols and more reliable techniques for sub-regional analysis.