Apparent Diffusion Coefficient (ADC) as an Early Biomarker for Tumor Response to Radiation Therapy, a Meta-Analysis

Abstract

<h3>Purpose/Objective(s)</h3> MRI equipped linear accelerators are increasing the routine ability to acquire daily diffusion weight imaging (DWI) during a course of radiation therapy (RT). Multiple small series have examined DWI and apparent diffusion coefficient (ADC) as an early biomarker of tumor response to RT. We sought to comprehensively evaluate ADC change (ΔADC) by conducting a meta-analysis of published data. We hypothesized that quantifiable changes in ADC occur early after initiation of treatment and that a certain magnitude in ΔADC changes would be associated with response to RT. <h3>Materials/Methods</h3> A search was conducted for studies with ADC values acquired during and following RT. Studies were included that published mean ADC values of tumors prior to treatment (baseline), 1 and/or 2 weeks into treatment, and included therapeutic response data or progression reported in follow up. The mean percentage changes between studies were combined using the inverse variance weighted method, where variances were estimated from the study reported standard deviations, using a random effects model, to accommodate potential between study heterogeneity. A single estimate of percentage change in ADC, along with the standard deviation and sample size corresponding to estimate was used as the summary statistic of interest from each study, for the response and non-response groups separately. These were combined using the inverse variance weighting method to obtain an overall estimate of percentage ADC change for the two groups of interest, where the square of the reported standard deviation served as the variance estimate. The overall estimate was computed using the random-effects meta-analysis technique. <h3>Results</h3> Data from eight published studies were analyzed totaling 207 patients undergoing RT for brain metastases (n=66), esophageal (n=50), head-and-neck (n=34), cervical (n=33), and rectal (n=24) cancer. One week after the induction of RT, there was no significant change in ADC from baseline in non-responding lesions, +3.32%, 95% CI [-2.25, +8.89], or responding lesions, +6.61%, 95% CI [-0.27, +13.49]. Two weeks after induction, however, the ADC in responding lesions increased by an average of 16.60%, 95% CI [4.86, 28.34], while non-responding lesions remained unchanged from baseline with a mean ΔADC of +4.36%, 95% CI [-8.38, +17.10]. ADC was significantly increased from baseline at two weeks, but not at one week, and lesions with less than 4.86% increase in ADC from baseline at two weeks were unlikely to be responding to the current therapeutic regimen. <h3>Conclusion</h3> From this relatively small meta-analysis it was demonstrated that significant increases in mean ADC for responding lesions were observed as early as two weeks after the initiation of RT. No significant change from baseline ADC was observed in tumors that did not respond to therapy. These data suggest a strong potential role for ΔADC as an early universal biomarker should be investigated in prospective clinical trials as a method to personalize RT dose.