The centrally restricted diffusion sign on MRI for assessment of radiation necrosis in metastases treated with stereotactic radiosurgery

Abstract

Purpose Differentiation of radiation necrosis from tumor progression in brain metastases treated with stereotactic radiosurgery (SRS) is challenging. For this, we assessed the performance of the centrally restricted diffusion sign. Methods Patients with brain metastases treated with SRS who underwent a subsequent intervention (biopsy/resection) for a ring-enhancing lesion on preoperative MRI between 2000 and 2020 were included. Excluded were lesions containing increased susceptibility limiting assessment of DWI. Two neuroradiologists classified the location of the diffusion restriction with respect to the post-contrast T1 images as centrally within the ring-enhancement (the centrally restricted diffusion sign), peripherally correlating to the rim of contrast enhancement, both locations, or none. Measures of diagnostic accuracy and 95% CI were calculated for the centrally restricted diffusion sign. Cohen's kappa was calculated to identify the interobserver agreement. Results Fifty-nine patients (36 female; mean age 59, range 40 to 80) were included, 36 with tumor progression and 23 with radiation necrosis based on histopathology. Primary tumors included 34 lung, 12 breast, 5 melanoma, 3 colorectal, 2 esophagus, 1 head and neck, 1 endometrium, and 1 thyroid. The centrally restricted diffusion sign was seen in 19/23 radiation necrosis cases (sensitivity 83% (95% CI 63 to 93%), specificity 64% (95% CI 48 to 78%), PPV 59% (95% CI 42 to 74%), NPV 85% (95% CI 68 to 94%)) and 13/36 tumor progression cases (difference p < 0.001). Interobserver agreement was substantial, at 0.61 (95% CI 0.45 to 70.8). Conclusion We found a low probability of radiation necrosis in the absence of the centrally restricted diffusion sign.

Sample strategies for the assessment of the apparent diffusion coefficient in single large intracranial space-occupying lesions of dogs and cats

“…Excluding lesions considered homogeneous on the ADC map, the highest median ADC value was found in M2p and the lowest in M2c. This could be expected as low ADC values have been used as a biomarker reflecting hypercellularity, which is often found in the central part of neoplasia ( 13 ), which was sampled in M2c and M5. The statistically significant difference between M5 against M2p and M6 could be explained by the position of the ROI placement.…”

Tumor Habitat Analysis Using Longitudinal Physiological MRI to Predict Tumor Recurrence After Stereotactic Radiosurgery for Brain Metastasis

“…Using physiological MRI, a layered appearance (without matching) consisting of outer, middle, and inner layers of high CBV, low ADC, and high ADC values, respectively, was suggestive of RN [ 34 ], with sensitivity and specificity rates of approximately 100% for 16 patients. Nonetheless, the visual analysis of enhancement patterns comprising T1/T2 signal intensity mismatch [ 35 ] or a layered appearance of high CBV, low ADC, and high ADC values with brain metastasis [ 36 ] is limited because it is a subjective assessment. In contrast, a quantitative tumor habitat analysis using k-means clustering applied to co-registered isometric voxels during our study provided objective and robust measurements that could be used to assess RN and viable tumors.…”

Diagnostic accuracy of MRI techniques for treatment response evaluation in patients with brain metastasis: A systematic review and meta-analysis

“…One study excluded patients with lesions which could induce susceptibility artefacts such as metastasis from melanoma [29]. In the index test domain, 16 out of 37 studies showed a high risk of bias, because for these studies it was unknown if the results of the index test were interpreted with or without knowledge of the reference test [30,35,36,[40][41][42][43][44][45][46][47][48][49][50][51][52]. Also, in many studies (n = 17), a predefined cut-off value of the index test was missing [30,31,[33][34][35][36][37][38]41,42,[44][45][46][53][54][55][56].…”

“…Besides anatomical images, all conventional MRI studies used additional analytical techniques (such as radiomics features). Seven studies used DWI [33,34,37,42,49,54,60], twelve DSC perfusion MRI [29,38,39,41,47,54,[58][59][60][62][63][64], four dynamic contrast enhanced (DCE) perfusion MRI [31,37,43,54], four magnetic resonance spectroscopy (MRS) [30,51,59,60], and one study used arterial spin labelling (ASL) perfusion MRI [35]. Lastly, six data sets (five derived from three studies and one separate study) used combined MRI techniques to calculate sensitivity and specificity [33,34,50,62].…”