Changes in CT morphology can be an independent response marker for patients receiving regorafenib for colorectal liver metastases: retrospective pilot study

By searching a database of medical records at Toranomon Hospital, we identified 29 patients who had received regorafenib for metastatic colorectal cancer after standard chemotherapy including fluorouracil, oxaliplatin and irinotecan with or without molecularly targeted drugs between the period of January 2013 and April 2016. Among these, 10 patients who had liver metastases evaluable with pre- and post-chemotherapy contrast enhanced computed tomography (CECT) were retrospectively reviewed. All the analyses in the current study were performed in accordance with the ethical guidelines for clinical studies at Toranomon Hospital and were approved by the Institutional Review Board.

The imaging analysis was performed using CECT scans, and the images were reviewed by a radiologist (WG) who was blinded to the clinical data. The response to chemotherapy was determined according to RECIST ver.1.1 [12] and the morphologic response criteria. [4] The morphologic criteria were defined as follows: group 1, homogeneous low attenuation with a thin, sharply defined tumor-liver interface; group 3, heterogeneous attenuation with a thick, poorly defined tumor-liver interface; and group 2, intermediate morphology that could not be rated as group 1 or 3 (Additional file 1: Figure S1). An optimal morphologic response was defined as a change in morphology from group 3 or 2 to group 1, and a suboptimal morphologic response was defined as a change in morphology from group 3 to group 2 or the absence of remarkable changes in morphology, as previously described. [4, 6] In patients with multiple tumors, the morphologic response was assigned based on the response seen in the majority of tumors.

Continuous variables were compared using the Mann-Whitney U test, and categorical variables were compared using the chi-squared test or the Fisher exact test, where appropriate. Interobserver agreements in image reading were evaluated using the kappa value. The OS period and the PFS period were determined from the date of initial treatment with regorafenib until the date of death or the initial tumor progression, respectively. All the cases without specific prognostic events were censored at the date of the last follow-up examination. The survival curves were generated using the Kaplan-Meier method and were compared using a log-rank test. Data on serum tumor markers were collected for an exploratory analysis. The statistical analyses were performed using JMP (version 11.0; SAS Institute Inc., USA). All the statistical tests were two-sided, and significance was set at P < 0.05.

The baseline demographics and clinical characteristics of the 10 patients who had evaluable liver metastases are listed in Table 1. The median age was 63 years (range, 39–83 years), and 8 patients were male (80%) and 2 patients were female (20%). The primary tumor site was the colon in 2 patients (20%) and the rectum in 8 patients (80%). Extrahepatic disease was observed in all the patients. All the patients received fluorouracil-based chemotherapy before regorafenib, and the number of prior chemotherapy regimens was 1 in 1 patient (10%), 2 in 3 patients (30%), 3 in 3 patients (30%), and 4 in 3 patients (30%). These prior chemotherapy regimens included oxaliplatin, irinotecan, bevacizumab, cetuximab, panitumumab and TAS-102. Bevacizumab was used in all the patients, and oxaliplatin was used in nine patients (90%). Eight patients (80%) were treated with irinotecan regimens. Cetuximab was used in 1 patient (10%), and panitumumab was used in 4 patients (40%). All the patients whose RAS status was wild-type received either cetuximab or panitumumab as a prior chemotherapy regimen. Three patients received TAS-102 before regorafenib. The median follow-up duration was 4.9 months from the initiation of regorafenib treatment (range, 2 to 12.5 months). The median number of treatment cycles was 2 (range, 1–15). The reasons for discontinuation were progressive disease (PD) in 3 patients and adverse events in 4 patients; 1 patient was continuing to receive treatment at the time of the analysis.

Among the 10 patients, optimal morphologic responses were observed in 3 patients, and 7 patients did not show any changes in CT morphology during the treatment. According to the RECIST, the response rate was 0%: 4 patients had SD, and 6 patients had PD. In all 3 patients who achieved an optimal morphologic responses, the size of the tumor nodule had increased at the time of the initial diagnosis of an optimal morphologic response (+ 6.0% [SD], + 16.0% [SD], and + 31.5% [PD], respectively) (Figs. 1, 2). Although these patients had histories of bevacizumab as a first- or second-line treatment, no morphologic changes were seen at the time of the former treatment with bevacizumab.

Serum tumor markers, such as carcinoembryonic antigen (CEA), were analyzed in all the patients at the time of radiographic evaluation during regorafenib treatment. The median rate of CEA increase was 2.02 (range, 0.75–32.3) in 7 patients with a suboptimal morphologic response, whereas it was 0.95 (range, 0.72–1.59) in 3 patients with an optimal morphologic response. The median duration of the morphologic response was 3.4 months, and there were no significant increases in the CEA levels during the period of the morphologic responses in all 3 patients.

In the entire cohort, the median PFS was estimated as 1.6 months and the median OS was 4.9 months after the initiation of regorafenib. When stratified according to the RECIST, a difference in PFS between the SD and PD patients was not evident (median PFS, 2.7 months vs. 1.6 months; P = 0.30). However, when compared according to whether or not a morphologic response was observed, patients with optimal morphologic responses showed a significantly longer PFS than those with suboptimal responses (median PFS, 4.9 months vs. 0.7 month; P = 0.028) (Fig. 3). Similar analyses for OS showed no prognostic differences between the patients with an SD according to the RECIST and those with PD (median OS, 4.7 months vs. 4.9 months; P = 0.99), while those with optimal morphologic responses tended to survive longer than those with suboptimal responses (median OS, not reached vs. 4.6 months; P = 0.15). Univariate Cox-regression confirmed that PFS or OS was not correlated with the other potential prognostic factors including age, performance status, number of chemotherapy lines, history of treatment with FOLFOX or FOLFIRI, history of TAS102 administration, presence of toxicity, original tumor stage, or original tumor location.

Table 2 compares the background characteristics between the patients with an optimal response and those with a suboptimal response. Although a statistical comparison is difficult because of the limited number of cases, the group with an optimal morphologic response tended to be younger and had completed a larger number of treatment cycles without discontinuation because of toxicity before disease progression.