PURPOSE This study was performed to evaluate whether fluorine-18 fluorodeoxyglucose positron-emission tomography/computed tomography (FDG PET/CT) could predict treatment outcome of regorafenib in metastatic colorectal cancer (mCRC).

METHODS Previously treated refractory mCRC patients were enrolled into a prospective biomarker study of regorafenib. For this sub-study, the results of FDG PET/CT scans at baseline and after two cycles of treatment were analyzed. Various metabolic parameters obtained from PET images were analyzed in relation to treatment outcome.

RESULTS A total of 40 patients were evaluable for PET image analysis. Among various PET parameters, total lesion glycolysis (TLG) measured in the same target lesions for RECIST 1.1 analysis were the most significant in predicting prognosis, with the lowest p-value observed in TLG calculated using the margin threshold of 40 % (TLG_{40 %}). Further analysis using TLG_{40 %} showed significantly longer overall survival (OS) in patients with lower baseline TLG_{40 %} (<151.8) (p = 0.003, median 14.2 vs. 9.1 months in <151.8 and ≥151.8, respectively). Patients showing higher decrease in TLG_{40 %} after treatment showed significantly longer progression-free survival (PFS) (p = 0.001, median 8.0 vs. 2.4 months in %ΔTLG_{40 %} < -9.6 % and ≥ -9.6 %, respectively) and OS (p = 0.002, median 16.4 vs. 9.1 months in %ΔTLG_{40 %} < -9.6 % and ≥ -9.6 %, respectively). The same cutoff could discriminate patients with longer survival among the patients who were under the stable disease category according to RECIST 1.1 (median PFS 8.4 vs. 6.8 months, p = 0.020; median OS 18.3 vs. 11.5 months, p = 0.049).

CONCLUSION Measurement of TLG can predict treatment outcome of regorafenib in mCRC.