Interim FDG-PET Imaging as an Early Response Biomarker for Pathologic Response to Chemoradiation in Esophageal Cancer

Abstract

To evaluate predictors of complete pathologic response to chemoradiation (CRT) in esophageal cancer (EC) treatment using fluorodeoxyglucose (FDG) positron emission tomography (PET) imaging changes before, during, and after CRT. Previous research shows that change in maximum standardized uptake value (SUVmax) does not predict for response to CRT. We evaluated other PET parameters in this prospective clinical trial. Forty-two EC patients were enrolled in this ongoing prospective imaging study. For this report, we report on the 20 patients who had surgery after neoadjuvant CRT. Outcome was assessed pathologically based on tumor regression grade (TRG, 1 = pathologic complete response, 2 = near complete response, and ≥3 = more than minimal residual). FDG PET imaging was acquired before treatment (baseline, BL), approximately 2 weeks after starting chemoradiation (interim, IM), and at first follow-up (FU) 4-6 weeks after completing CRT. The 3D tumor was contoured on the PET images using gradient edge technique using MIM software (MIM Software Inc. Cleveland, OH). Summary statistics for SUVmax, SUVmean, metabolic tumor volume (MTV), and total lesion glycolysis (TLG) were determined from the contoured volume at each time point. Relative changes in each tumor PET parameter from BL to IM and FU scan were calculated and compared by outcome status defined as TRG1 vs. TRG2/TRG≥3 using Wilcoxon rank-sum test. Multiple logistic regression was used to evaluate receiver operating curve (ROC) characteristics for predicting pathologic response status based on analysis of DSUVmax, DSUVmean, DMTV, and DTLG values. TRG 1, 2, and 3 after CRT and surgery were seen in 5 (29%), 9 (43%), and 6 pts (29%), respectively. Tumors that had a complete pathologic response exhibited significantly greater change in TLG (P = 0.0117) from the baseline to interim scan, with a threshold of 59% decrease. Change in SUVmax (P = 0.232, P = 0.832), SUVmean (P = 0.086, P = 0.550), and MTV from BL to IM or BL to FU scans were not predictive of pathologic CR, respectively. Predicting outcome status based on change in TLG at the IM scan yielded AUROC of 0.888 (95% CI: 0.742-1.000). Youden’s optimal sensitivity (specificity) were 1.000 (0.813) for DTLG at the IM scan. Our results indicate that DTLG on PET imaging after 2 weeks of CRT may predict pathologic complete response in esophageal cancer. Early response assessment with FDG-PET is potentially useful for stratification of EC pts to treatment intensification or non-surgical management.