Patients with mCRC will eventually face challenges in its medical management following failure of standard treatment. However, limited choices with good performance still exist in third or further line therapy. Previous studies have found that mCRC patients harboring MSS show little response to immunotherapy , likely due to a complex tumor microenvironment that counteracts antitumor immunity via a combination of low antigenic tumor cells and an immunosuppressive tumor microenvironment . However, the REGONIVO study demonstrated promising anticancer activity in mCRC patients with MSS who were administered with a PD-1 inhibitor (nivolumab) and regorafenib; these findings have a profound impact on cancer care for mCRC patients. As of January 2020, three additional antibody therapeutics developed by Chinese companies (tislelizumab, sintilimab, and camrelizumab) became available in China. These compounds are highly selective, fully humanized monoclonal antibodies that block the interaction between PD-1 and its ligands. Based on the REGONIVO combination strategy, our center attempted a similar regime using fruquintinib in combination with various PD-1 inhibitors as a third line therapy for mCRC patients with MSS. We retrospectively reviewed those patients and analyzed the efficacy of their treatments.
We obtained medical records from 35 patients at the time of analysis. Broadly, our therapy regime had efficacy. DCR was 60%, and ORR was 11.4% with four PRs observed. In the REGONIVO study, ORR was observed in 40% of patients; among 50 patients, including 25 CRC and 25 gastric cancers, eight patients had PRs, resulting in an ORR of 33% in MSS CRC patients. By contrast, our response rates were non-superior to the REGONIVO study. Notably, in the North American population, the efficacy of the combination treatment also differed from the Japanese population. More specifically, in the North American version of REGONIVO, five patients (7.1%) out of 70 had a PR, and 22 (31.4%) had SD. The LEAP-005 study (NCT03797326) evaluated the efficacy and safety of lenvatinib plus pembrolizumab in patients with previously treated CRC. They found an ORR of 22% (95% CI: 9–40) and a DCR of 47% in 32 CRC patients. Such discrepancies may be due to the REGONIVO study being a dose-finding and dose-expansion phase 1b trial with an aim of exploring safety and recommended doses. Accordingly, response rates should be verified in further investigations using a larger cohort. Second, the efficacy of the combination treatment varies depending on the specific anti-PD-1 or angiogenesis agent and the population. Third, patients in trials normally have PS scores of 0–1. Real-world studies do not limit PS scores, which means that our study included patients with worse PS scores, which may have affected the treatment responses.
We obtained a median PFS of 3.8 months. This result seems disappointing. The PFS was not superior to that from the REGONIVO study, and much similar to other studies associated with third or subsequent line treatment for mCRC. The PFS in the REGONIVO study was 6.3 months, 3.7 months in the FRESCO study, and 2.0 months in the TAS-102 study. We acknowledge the non-superior PFS of fruquintinib and anti-PD-1 treatment in such comparisons. Real world studies are complex, with various factors that reflect actual clinical practice, whereas clinical trials exclude poor condition. Furthermore, although regorafenib and fruquintinib are the same type of oral anti-angiogenesis agents, their underlying mechanisms at the functional site are different. Regorafenib is multi-targeted, while fruquintinib is highly selective for VEGFR1, VEGFR2, and VEGFR3. The molecular properties of PD-1-targeted antibodies are another factor. We acknowledge the limitations in comparing nivolumab and other types of PD-1 antibodies. However, nivolumab and pembrolizumab differ in the extent and spatial location of their binding sites with the flexible PD-1 loops . Based on the available characterization data on anti-PD-1 antibodies, the molecular behavior between nivolumab and other anti-PD-1 antibodies likely differ [22, 23]. We also consider, to some extent, the effect of fruquintinib dose on efficacy. Some patients in our study took fruquintinib at 3 mg due to their tolerance, while the recommended dose on a continuous regimen is 5 mg QD [24, 25]. Therefore, the dose may have affected clinical efficacy.
Notably, the median OS in our study was much higher than in previously reported studies. The OS benefit in our real-world study was 14.6 months, while that of the FRESCO study was 9.3 months, which was the longest OS reported prior to our study. We cannot directly compare the efficacy of our study with the FRESCO or REGONIVO studies; this outcome needs to be confirmed in a larger group.
The PFS in our study appear to be no better than the results from the FRESCO study. Accordingly, we further assessed whether clinical characteristics were correlated with clinical outcome. Gender, tumor location, metastatic organs, and KRAS status did not significantly differ, which was inconsistent with the REGONIVO study. All patients responding in the REGONIVO study  were male with lung metastases and had PS scores of 0. Therefore, the results from the REGONIVO study may have been biased by the small sample and need further confirmation in a larger population.
One shortcoming of our study is the limited population and the absence of a side effects profile and data on PD-L1 expression. We did not have sufficient data to calculate a safety profile; what we know is that we found no severe adverse-related deaths. PD-L1 is believed to indicate response to anti-PD-1 antibodies in several tumors, However, data are lacking in PD-L1 expression; thus, we were unable to assess PD-L1 as a potential biomarker for CRC patients.
In conclusion, we found that fruquintinib in combination with anti-PD-1 had clinical activity in mCRC refractory to standard chemotherapy. However, this benefit was not observed across all prespecified patient subgroups. Further research is needed to assess OS benefits using a larger group.