Diagnostic value of centrally restricted diffusion in differentiating radiation necrosis from tumor progression in brain metastases: A single-center observational study

Abstract

Background

Distinguishing radiation necrosis (RN) from true progression (TP) in irradiated brain metastases is challenging. We evaluated the diagnostic performance of the centrally restricted diffusion sign on diffusion-weighted imaging (DWI).

Methods

From August 2014 to August 2024, we screened 321 patients with histologically confirmed brain metastases treated with radiation therapy and follow-up MRI for new or enlarging necrotic lesions ≥1 cm. Two board-certified neuroradiologists independently assessed the centrally restricted diffusion sign—central hyperintensity on b1000 images with corresponding ADC reduction—by rigidly co-registering DWI to postcontrast 3D T1-weighted sequences. Quantitative analysis included mean ADC measurement within manually drawn regions of interest in the necrotic core and the contrast-enhancing rim. Final diagnoses were established by histopathology (n = 17) or multidisciplinary consensus (n = 90).

Results

Of 107 patients (median age, 62 years; 57.9 % male), 62 had TP and 45 had RN. Median interval from radiotherapy completion to index MRI was 10.8 months. Overall survival was longer in patients with RN (median not reached) than in those with TP (17.5 months; P < 0.0001). Interobserver agreement for the centrally restricted diffusion sign was moderate (κ =0.55). The sign appeared in 34/45 RN cases and 13/62 TP cases (P < 0.0001). For RN diagnosis, sensitivity was 75.6 %, specificity 79.0 %, and accuracy 77.6 %. Quantitative ADC metrics did not enhance performance.

Conclusion

The centrally restricted diffusion sign on DWI may aid differentiation of RN from TP in irradiated brain metastases, despite moderate interrater reliability.