The predictive capacity of apparent diffusion coefficient (ADC) in response assessment of brain metastases following radiation.

To investigate the predictive capacity of the apparent diffusion coefficient (ADC) as a biomarker of radiation response in brain metastases. Seventy brain metastases from 42 patients treated with either stereotactic radiosurgery or whole brain radiotherapy were imaged at baseline, 1 week, and 1 month post-treatment using diffusion-weighted MRI. Mean and median relative ADC for metastases was calculated by normalizing ADC measurements to baseline ADC. At 1 year post-treatment, or last available follow-up MRI, volume criteria determined final tumour response status. Uni- and multivariate analysis was used to account for factors associated with tumour response at 1 week and 1 month. A generalized estimating equations model took into consideration multiple tumours per subject. Optimal thresholds that distinguished responders from non-responders, as well as sensitivity and specificity were determined by receiver operator characteristic analysis and Youden's index. Lower relative ADC values distinguished responders from non-responders at 1 week and 1 month (P < 0.05). Optimal cut-off values for response were 1.060 at 1 week with a sensitivity and specificity of 75.0 and 56.3 %, respectively. At 1 month, the cut-off was 0.971 with a sensitivity and specificity of 70.0 and 68.8 %, respectively. A multivariate general estimating equations analysis identified no prior radiation [odds ratio (OR) 0.211 and 0.137, P = 0.033 and 0.0177], and a lower median relative ADC at 1 week and 1 month (OR 0.619 and 0.694, P = 0.0036 and 0.005), as predictors of tumour response. Lower relative ADC values at 1 week and 1 month following radiation distinguished responders from non-responders and may be a promising biomarker of early radiation response.