Target audience Physicians intended to get more information from conventional and gadolinium enhanced MR imaging in monitoring the therapy to patients in PCNSL.

Purpose To investigate the objective response rate(ORR) and identify MR findings as predictors to evaluate the therapeutic response in PCNSL.

Methods We retrospectively reviewed the MR images and clinical records of 40 patients who were diagnosed with PCNSL from January 2007 to February 2015 at Guangdong General Hospital. Response to treatment was assessed by contrast enhanced brain MRI. The imaging characteristics were tabulated and tested for objective response in entire series. Objective response rates, clinical characteristics and MR findings were compared using the χ2 test or the Fisher’s exact test for categorical variables, according to the sample size. The Binary Logistic Regression model was used in multivariate analysis. All Data were analyzed using SPSS 17.0 for Windows. P<0.05 was used as the criteria for the statistical difference among groups.

Results Of the forty assessable patients, twenty-nine (72%) achieved an object response. 6 patients (15%) had progressive disease (PD) during treatment, lymphoma remained unchanged in 1 cases (3%), 3 patients (8%) died before or during the initial treatment course. According to the planned therapeutic modality, 14 patients (64%) achieved an OR after chemotherapy (CHT) alone, 13 patients (87%) achieved an OR after chemotherapy→radiotherapy (CHT→RT), and 2 patients (100%) achieved an OR after radiotherapy→chemotherapy （RT→CHT） (P<0.001). Higher objective response rates were observed from patients with MR characteristics of: 1) lesion size≤3cm; 2) single lesion; 3) superficial location; 4) homogenous enhancement (P=0.020, 0.030, 0.033, 0.029, respectively).

Discussion The present study focused on MR appearance-related predictors, which influence treatment response in PCNSL. Efforts to identify predictors of response in PCNSL have produced observations in some series, but most of them focused on the clinical characteristics of patients. Our study investigated MR appearance, the findings show that tumor size, number, location and homogenous enhancement were independent predictors of objective response. Lesions with smaller size, single number, superficial location and homogenous enhancement predicted a higher OR. In addition, it is notable that the planned therapeutic strategy played a prognostic role in monitoring treatment response in patients of PCNSL.

Conclusion The current study identified MR features as predictors to evaluate the therapeutic response in PCNSL. It might apply more serviceable information for the patients with PCNSL. Chemotherapy→Radiotherapy might be a recommended therapeutic strategy in clinical trials.