Metastatic colorectal cancer remains one of the most common causes of cancer death worldwide. According to estimates cited by the study’s authors from the International Agency for Research on Cancer, colorectal cancer accounts for a substantial share of new cancer diagnoses globally, and Japan, where the study was conducted, maintains detailed national cancer statistics reflecting the disease’s heavy burden. Once patients progress through initial chemotherapy regimens and targeted combinations, they enter a refractory stage where only a handful of options remain. Regorafenib, an oral multikinase inhibitor that blocks tumor angiogenesis and oncogenic signaling pathways, and trifluridine/tipiracil, an oral nucleoside antimetabolite combination that thwarts tumor DNA repair, are the two pillars of this late-line setting. Both extend survival compared with placebo in randomized trials, but no definitive evidence had established which sequence delivers better outcomes when both drugs are ultimately available to a patient.

The OSERO study, led by Takeshi Kawakami of the Shizuoka Cancer Center with Eiji Oki of Kyushu University serving as principal investigator, was designed to close that evidence gap prospectively rather than retrospectively. Sponsored by the Kyushu Study group of Clinical Cancer and funded by Bayer Yakuhin, the study enrolled patients across 42 Japanese institutions who were scheduled to begin later-line treatment with regorafenib or trifluridine/tipiracil, with or without bevacizumab. Patients were assigned to three cohorts: cohort A received regorafenib monotherapy, cohort B received trifluridine/tipiracil alone, and cohort C received trifluridine/tipiracil plus bevacizumab, an anti-VEGF antibody that starves tumors of their blood supply. The primary endpoint was overall survival, the most clinically meaningful measure of whether a treatment sequence prolongs life.

In total, 468 patients were analyzed, with 154 in the regorafenib cohort, 82 in the trifluridine/tipiracil cohort, and 232 in the trifluridine/tipiracil plus bevacizumab cohort. The raw numbers told an initially striking story. Patients starting with regorafenib had a median overall survival of 12.2 months, compared with 7.5 months for those starting with trifluridine/tipiracil alone and 10.6 months for those receiving the combination with bevacizumab. Rates of proceeding to a scheduled second-line treatment also differed markedly: 65.6 percent of regorafenib starters went on to further therapy, compared with 37.8 percent of the trifluridine/tipiracil-alone group and 62.9 percent of the bevacizumab combination group.

But raw comparisons in observational research are notoriously misleading. Clinicians do not assign these drugs at random; they weigh each patient’s performance status, prior treatment history, organ function, and the aggressiveness of the disease. Patients selected for one drug may be systematically healthier or sicker than those selected for another, a phenomenon known as confounding by indication. To address this, the investigators applied stabilized inverse probability of treatment weighting, a statistical technique derived from propensity score modeling. Each patient was assigned a weight reflecting the probability of receiving the treatment they actually received given their baseline characteristics, effectively creating a pseudo-population in which treatment assignment was independent of measured confounders. This method, widely used in pharmacoepidemiology, cannot eliminate bias from unmeasured variables, but it represents the strongest adjustment feasible outside a randomized trial.

After this adjustment, the survival difference between the two principal strategies evaporated. Median overall survival was 12.2 months for the regorafenib-first sequence and 10.6 months for the trifluridine/tipiracil-plus-bevacizumab sequence, with a hazard ratio of 1.03 and a 95 percent confidence interval of 0.81 to 1.30, and a P value of 0.837. A hazard ratio so close to 1.0, with a confidence interval comfortably spanning unity, indicates essentially no detectable survival advantage for either sequence. In practical terms, the total length of survival a patient could expect from the two-drug sequence appeared similar regardless of which agent was deployed first.

The safety data, however, revealed a sharp and clinically consequential divide. Grade 3 or higher adverse events occurring in more than 10 percent of patients differed by cohort. In the regorafenib group, the dominant toxicities were hand-foot syndrome, proteinuria, and hypertension, classic signatures of the drug’s anti-angiogenic and multikinase activity. Hand-foot syndrome, a painful inflammation of the palms and soles, can severely limit daily function and often forces dose reductions. In the trifluridine/tipiracil groups, both with and without bevacizumab, the principal toxicities were hematologic: neutropenia, a depletion of infection-fighting white blood cells, and anemia. These contrasting profiles mean that the two strategies impose different burdens on different patients, and the authors concluded that treatment selection may reasonably depend on the safety profile of each drug in the context of individual patient characteristics.

The findings arrive at a moment when the late-line colorectal cancer landscape is shifting. The addition of bevacizumab to trifluridine/tipiracil gained momentum after a series of Japanese investigator-initiated studies, including the C-TASK FORCE, TAS-CC3, TAS-CC4, and BiTS trials, demonstrated the feasibility and activity of the combination. That momentum culminated in the international phase 3 SUNLIGHT trial published in the New England Journal of Medicine in 2023, which showed that trifluridine/tipiracil plus bevacizumab improved survival over trifluridine/tipiracil alone in refractory disease, cementing the combination in international guidelines from the National Comprehensive Cancer Network and the European Society for Medical Oncology. Meanwhile, the FRESCO-2 trial introduced fruquintinib, another oral angiogenesis inhibitor, as a further option, and earlier work such as the Japanese REGOTAS propensity score analysis had attempted retrospective comparisons of regorafenib and trifluridine/tipiracil. Prior work by several of the same OSERO investigators had also suggested that survival benefits accumulate with the number of active drugs a patient receives, underscoring why sequencing questions matter so much.

What distinguishes OSERO is its prospective design. Most evidence comparing these agents has come from retrospective chart reviews or registry analyses, which are vulnerable to selection bias and incomplete data. By registering patients before treatment began, at the moment the clinician committed to a strategy, the investigators captured the true decision point of interest. The study was conducted in accordance with the Declaration of Helsinki, approved by the Kyushu University institutional review board, and all participants provided written informed consent. The breadth of participation, spanning 42 institutions from Shizuoka to Hokkaido to Kyushu, lends the results a real-world texture that single-center trials often lack, reflecting the heterogeneous patient populations and clinical judgments that define everyday oncology practice.

For clinicians, the message is one of flexibility rather than a new mandate. Because the two sequences yielded comparable adjusted survival, oncologists can tailor the choice to the patient in front of them: a patient with poorly controlled hypertension or painful hand-foot syndrome risk factors might be steered toward trifluridine/tipiracil with bevacizumab, while one with marginal bone marrow reserve might better tolerate regorafenib. For patients, the study offers reassurance that neither choice forecloses a meaningful survival opportunity, since both drugs can typically be delivered in sequence. The authors caution that, as an observational study, OSERO cannot rule out residual confounding, and randomized confirmation remains the gold standard. Yet with a randomized head-to-head sequencing trial unlikely to be mounted, this prospectively collected, statistically adjusted evidence from nearly 500 patients stands as the clearest guidance available, and it suggests that in the final chapters of metastatic colorectal cancer treatment, the order of the last two drugs matters far less than ensuring patients can safely receive both.

Subject of Research: Treatment sequencing of regorafenib and trifluridine/tipiracil in later-line metastatic colorectal cancer

Article Title: A prospective observational study of treatment sequencing with regorafenib and FTD/TPI in metastatic colorectal cancer: The OSERO study

Article References: A prospective observational study of treatment sequencing with regorafenib and FTD/TPI in metastatic colorectal cancer: The OSERO study. (n.d.). https://doi.org/10.1038/s41416-026-03620-w

Image Credits: AI Generated

DOI: 10.1038/s41416-026-03620-w

Keywords: metastatic colorectal cancer, regorafenib, trifluridine/tipiracil, bevacizumab, treatment sequencing, overall survival, observational study, propensity score, inverse probability weighting, chemotherapy, drug safety, clinical oncology