Background and importance Uncertainty exists regarding the comparative effectiveness of cetuximab versus bevacizumab in metastatic colorectal cancer (mCRC) due to conflicting evidence of efficacy of previous randomised clinical trials and the absence of quality of life (HRQoL) studies in this setting.

Aim and objectives In order to assess clinical effectiveness and patient reported tolerability of the different targeted treatment options simultaneously, we conducted a mainly retrospective head-to-head multi-cohort study. The main retrospective study was designed to compare real world clinical outcomes from both antibodies. Concurrently, we nested in it a smaller prospective cohort study for the purpose of measuring patient reported outcomes (PROs).

Material and methods Retrospective cohorts were defined by treatment line, and subgroups by (K)RAS status and tumour sidedness. Among other effectiveness outcomes, we compared response rates, progression free survival (PFS) and overall survival (OS). PROs were measured prospectively through EORTC disease specific instruments. Methods and reporting followed STROBE guidelines and SISAQOL/SPIRIT-PRO recommendations.

Results Between 2010 and 2018, 311 mCRC patients were included in the overall analysis, of whom 44 were further allocated to PROs nested cohorts. Except for (K)RAS mutation status, baseline characteristics were balanced across groups. In the full analyses, PFS (firstline: HR=0.85; p=0.26; secondline: HR=1.16; p=0.51) and OS (firstline: HR=0.83; p=0.26; secondline: HR=0.88; p=0.58) were similar between treatment arms. In subgroup analyses (firstline), we found a survival difference favouring bevacizumab in right-sided tumours (PFS: HR=0.52; p=0.025; OS: HR=0.60; p=0.11), but not in left-sided or (K)RAS wild-type tumours. Response rates were higher for bevacizumab in patients with right-sided primaries and similar across other comparisons. During the first 12 weeks of treatment, a higher proportion of patients in the cetuximab arm experienced clinically meaningful (≥10%) deterioration of HRQoL comparing with the bevacizumab cohort: 53.8% versus 18.2% at 6 weeks and 66.7% versus 12.5% at 12 weeks. We also observed progressively increased scoring on the symptom scales in the cetuximab cohort during follow-up.

Conclusion and relevance This study provides evidence suggesting that bevacizumab and cetuximab containing regimens result in similar clinical effectiveness outcomes in mCRC, except for right-sided tumours, where bevacizumab performed substantially better. Cetuximab led to a progressive negative impact on HRQoL compared with baseline and bevacizumab. These findings should be further explored in randomised studies.

Conflict of interest No conflict of interest