Our multivariate analysis identified total dose until the second cycle <3180 mg, age <65 years, PS 2, hepatic metastasis, and regorafenib initial dose ≤120 mg as prognostic factors of regorafenib. In groups divided by median dose, regorafenib total dose was associated with OS. It should be noted that a particular cut-off value for cumulative regorafenib dose was presented because it was not reported previously.
In this study, patients who dropped-out early due to adverse events or progressive disease, and we therefore considered the potential for confounding bias. We examined the study population except for early drop-out cases in which patients discontinued treatment until cycle 2 because of severe adverse events or progressive disease in the same multivariate analysis. In exception cases, total dose until the second cycle <3180 mg (HR 1.97, 95% CI, 1.00-3.86, P = 0.0496) was extracted as a statistically significant independent poor prognostic factor (Table S2). These results clearly demonstrate the clinical significance of the cumulative dose of regorafenib in the early cycles with regard to treatment efficacy in patients with mCRC.
A total of 122 of 176 patients (69.3%) in this study were treated with regorafenib at an initial dose of 160 mg because the study duration ranged from the time regorafenib went on the market to the close of observation. However, the number of patients treated with an initial dose ≤120 mg is currently increasing as a means of preventing discontinuation due to intolerable toxicity. In a recent meta-analysis, treatment with regorafenib at the standard dose of 160 mg was associated with a significant increase in adverse events related to permanent discontinuation, dose interruptions, and dose reductions . Optimizing treatment by means such as personalizing the regorafenib dose and schedule adjustments is common in clinical practice, and many physicians have adopted an empirical approach to manage toxicity as a result of phase III studies . A recent observational cohort study suggested that individualized dosing strategies in patients with mCRC might lead to improved clinical outcomes . In the CORRELATE prospective observational study, the regorafenib toxicity profile was similar to that reported in phase III trials. The starting dose for almost half of the patients in that study was less than the approved 160 mg dose, and the median OS and progression-free survival were in the ranges observed in phase III trials . In the ReDOS study, the dose-escalation group achieved cycle 3 of treatment, but the standard-dose group did not . The results of these studies indicate that optimizing the initial dose is associated with outcome and toxicity, although a relationship between cumulative dose and outcome was not reported. Furthermore, schedule adjustments or discontinuation/restarting, which often occur in real-world settings, were not considered except for the CORRELATE study. Our study shows that cumulative dose until the second cycle in a real-world setting is associated with OS. The association was not statistically significant with the initial dose distribution divided based on median total dose, whereas initial dose was extracted as a prognostic factor in the multivariate analysis. These results indicate that the initial dose should not be decreased arbitrarily and that an individualized starting dose should be considered, consistent with other studies.
The respective incidences of HFSR, liver dysfunction, and hypertension were 80%, 31%, and 60% in the Japanese population in the CORRECT study , in contrast to 93.1%, 25.5%, and 35.2%, respectively, in this study. The frequency of hypertension in this study was lower than previously reported, whereas that of HFSR was higher. The rates of adverse events of ≥ grade 3 were similar to other studies. In groups separated by median total dose, all grades of HFSR were statistically significant, although the frequency of HFSR was generally over 90% in both groups. These results indicate that HFSR is likely to occur in mCRC patients treated with regorafenib. The data also indicate that the incidences of skin rash and emergency hospitalization in patients with a total dose until the second cycle <3180 mg are clearly higher than in patients in the other group. The results show that skin rash and emergency hospitalization are direct causes of discontinuation or dose reduction. It is therefore important to identify patients who are likely to develop severe adverse effects.
Many researchers have examined ways to optimize the dose of regorafenib, but there are no significant real-world data available. We assessed adherence to regorafenib in order to examine real-world doses. It has not been previously reported that cumulative dose is associated with survival time in view of real-world adherence data. Our study indicates that total dose until the second cycle ≥3180 mg prolongs OS. This value may represent a cut-off point. A regorafenib initial dose of 80 mg continuing until second cycle at the standard schedule would lead to a cumulative dose of 3360 mg in the absence of discontinuation or dose reduction. That is the indicator for regorafenib treatment design in terms of dose escalation, dose reduction, or schedule adjustment.
Since regorafenib was approved, many studies have examined whether pharmacokinetic and pharmacodynamic parameters such as dose setting are associated with efficacy or adverse events. In general, regorafenib is metabolized by cytochrome P450 3A4 in the liver to its active metabolites, M-2 and M-5. Kubota et al examined the area under the unbound plasma concentration–time curve (AUCu) for these compounds . Higher AUCu values for M-2 and M-5 on day 1 were associated with significantly shorter progression-free survival than higher AUCu values for total plasma or unchanged drug. Moreover, the relative dose intensity during cycle 1 in patients with higher AUCu values for M-2 or M-5 was lower than that for patients with lower AUCu values. These results suggest that the standard dose was too high and that active metabolites played a significant role in patients’ decisions whether to continue treatment. In terms of genetic factors, Kubota et al reported a significant association between the ABCG2 421A/C genotype and AUCu values for the active metabolites, whereas another study reported that other genetic factors were not associated with regorafenib pharmacokinetics . Thus, whether genetic factors actually affect regorafenib efficacy and toxicity remains unclear and should be examined in future studies.
There were three major limitations to this study. The first limitation was the retrospective single-institution design, which caused us to overlook some clinical data or consider selection bias, as our focus was on real-world data regarding adherence to regorafenib. Our results were thus not completely clear. Therefore, prospective analyses should be conducted in the future. The second limitation involved the outcome measures used. It is possible that OS was affected by prior chemotherapy or other patient factors, even though we used a multivariate analysis and minimized confounders as much as possible. The third limitation involved the number of cases. Although the study included patients over a 5-year period, we were not able to calculate the appropriate the number of cases to include, which could have caused us to over- or underestimate our results.