The present registry-based retrospective analysis suggests that the combination of bevacizumab with XELOX had similar efficacy as an infusional regimen combining bevacizumab with FOLFOX. This result is in agreement with published results of randomised clinical trials that demonstrated comparable efficacies of XELOX and FOLFOX) alone or in combination with bevacizumab [2, 6, 7]. In addition, three registry-based studies examining bevacizumab efficacy in mCRC have been published, including the prospective Bevacizumab Expanded Access Trial (BEAT) study, the prospective Bevacizumab Regimens’ Investigation of Treatment Effects (BRiTE) observational study, and the retrospective Medicare-based analysis [7–9]. Nevertheless, to the best of our knowledge, the present study reports the largest cohort treated with XELOX/bevacizumab combination so far.
The age structure of patients in our study was similar to that reported in the NO16966, BRiTE, and BEAT studies. The present analysis used similar approach and was of comparable size as the Medicare analysis that, however, only included patients aged 65 year or older and only those with synchronous metastases [9]. This patient profile in the Medicare cohort may explain the difference in OS of almost 12 months compared to our study. The proportion of patients previously receiving adjuvant treatment was lower in the present study compared to the BEAT and BRiTE cohorts. Although the median PFS of patients in our cohort was similar to that reported in the above studies, the median OS was substantially longer, reaching 30 months for patients in the XELOX/bevacizumab subgroup. The favourable survival may be due to a more recent patient cohort in the present study and, possibly, patient selection. Gradual incremental improvements in OS have been observed in mCRC over the past decade because of the introduction of novel drugs and therapeutic strategies [10]. Importantly, the administration of the most expensive cancer drugs including bevacizumab has been centralised in the Czech Republic. It is possible that the centralisation of patients into cancer centres is partly responsible for the excellent survival results. In contrast, the median number of patients enrolled during a 16-month period per centre was only eight for the 248 sites participating in the BRiTE observational study [8].
For obvious reasons, the present analysis does not answer the question of the benefit of adding bevacizumab to an oxaliplatin-based regimen. The addition of bevacizumab has been shown to prolong OS in patients with irinotecan-based regimens, but the data on patients treated with the combination chemotherapy containing oxaliplatin are more ambiguous. In the NO 16966 trial that randomized patients using a 2 × 2 factorial design between XELOX and FOLFOX4 with or without bevacizumab, the addition of bevacizumab significantly prolonged PFS. However, statistically significant superiority could be demonstrated only in the subgroup of patients treated with XELOX but not FOLFOX. Adding bevacizumab to oxaliplatin-based chemotherapy resulted in a trend to prolongation of OS that did not reach statistical significance [2, 11]. The lack of survival improvement after adding bevacizumab to oxaliplatin-based chemotherapy that contrasted with a significant effect in patients treated with irinotecan-based regimens has also been reported in the retrospective Medicare analysis [9].
The prognostic factors identified in the present cohort are well established for the mCRC population. The risk of death was almost two times higher in patients with three or more metastatic sites at the start of bevacizumab therapy compared to patients with only one metastatic site. The presence of two metastatic sites was associated with almost 50% increase in the risk of death. The risk of death was 34% higher in patients with metastatic CRC at diagnosis compared to patients with recurrent disease.
Of note, while the overall response rates in the FOLFOX/bevacizumab and XELOX/bevacizumab cohorts were similar, in the FOLFOX/bevacizumab group, there were significantly less patients who had disease stabilisation (31.4% versus 40.5%, respectively) and conversely more patients who had progressive disease than in the XELOX/bevacizumab group (15.0% versus 6.5%, respectively).
The possible differential association between OS and KRAS status for the two backbone regimens that reached borderline statistical significance in our analysis is surprising. We found that there was a trend to improved survival in patients with KRAS wild-type tumours versus those with KRAS mutated tumours in the XELOX/bevacizumab subgroup but in the FOLFOX/bevacizumab subgroup. This finding may merit further research.
Patients who had received prior adjuvant chemotherapy treated within the NO 16966 trial had better PFS with FOLFOX alone than with FOLFOX/bevacizumab [5]. We have carried out a similar analysis on our dataset but detected no statistically significant survival differences between the two studied combinations for any clinically defined patient subgroup (Table 4).
The present analysis has obviously several weaknesses that are partly due to its retrospective nature. Selection bias cannot be excluded as fitter patients could have been preferentially allocated to XELOX chemotherapy. Data on initial performance status are missing in one-half of the patients and there is some imbalance in the proportion of primarily metastatic patients between the cohorts. The registry does not provide data on the removal of primary tumours in patients with primarily metastatic colorectal cancer and on variant FOLFOX regimens that may be used in some centres, although these variables would be unlikely to skew the results of survival analysis.
No centralised review of radiological response was performed and the data on PFS may be less reliable given the number of centres involved and different patterns of care, including radiological imaging, in each centre. We were not able to extract valid data on metastasectomies from the registry for the entire period of study. On the other hand, the survival data from the registry were checked against the national registry of deaths. In general, OS data are more reliable for this type of registry-based retrospective studies, and some studies, including the study recently published by Meyerhardt et al., analysed only OS [9].
The incidence of adverse events was lower in the present analysis than that reported in prospective trials. The registry was focused on bevacizumab and the attending physicians apparently tended to report only events associated with bevacizumab and not toxicities linked to the chemotherapy backbone. Also, asymptomatic thrombotic events detected only on imaging, such as visceral thrombosis were unlikely to be reported. Because the incidence of severe or even life-threatening toxicities such as thromboembolism or gastrointestinal perforation that usually lead to treatment interruption or modification is less likely to be affected by underreporting, only grade 3–5 adverse events are reported here. Another reason for the relatively low incidence of adverse events could be the selection bias that is inherent to registry-based studies.
On the other hand, the strong point of the present analysis is that it shows similar activity of the combination of bevacizumab and FOLFOX/XELOX chemotherapies in real-world medicine.