The monoclonal antibody LY-CoV555 (bamlanivimab) increases viral clearance after a single infusion in high-risk outpatients. 1, 2 In previous studies, variant escape mutation after bamlanivimab monotherapy occurred in less than 12% of patients within 29 days. 1, 2 However, immunocompromised patients were not included in these clinical trials. Here, we report virological and clinical follow-up of the first two immunocompromised patients treated with bamlanivimab in our university hospital (Lyon University Hospital, Lyon, France). Patient 1 was a 62-year-old man with angioimmunoblastic T-cell lymphoma and patient 2 was an 87-year-old man with mantle cell lymphoma. A single 700 mg dose of bamlanivimab was administered 3 days after COVID-19 onset and resulted in transient clinical response in both patients (appendix p 1). Patient 1 had two recurrences of fever (from day 4 until day 6 and from day 15 until day 20 after bamlanivimab treatment) and required oxygen supplementation during each episode. Convalescent plasma was infused into patient 1 from day 17 to day 20, leading to clinical improvement and decreased inflammatory syndrome. Patient 2 developed fever (on day 5) and respiratory degradation (from day 8) and died on day 17. Viral isolates from patient 1 and patient 2 belonged to clades 20A. EU2 and 20I/501Y. V1, respectively. At the time of bamlanivimab infusion (day 0), only one non-synonymous intra-host single nucleotide variant (ihSNV) was detected in the Spike in patient 1 (Thr286Ala; frequency 3%), thus no variant was detected at position 484 in the Spike (Glu484) in either patient (appendix p 1). The Glu484Gln and Glu484Lys (patient 1 and patient 2) and Glu484Ala mutations (patient 1), which are known to be involved in neutralising antibody escape and resistance to bamlanivimab, 5 appeared for both patients within 7 days. Glu484Lys and Glu484Gln became predominant in patient 1 (90% by day 19) and patient 2 (55% by day 5), respectively. In patient 1, viral loads of SARS-CoV-2 were negatively associated with the proportion of wild-type Glu484. Although the viral load declined in patient 2 at day 5, plasma was positive (4· 3 log copies/mL) at day 14, suggesting failure of treatment. Other ihSNVs in the Spike were detected during follow-up (Gln493Arg in patient 1 and Ser494Pro in both patients; appendix p 2). These mutations were also selected for in vitro by bamlanivimab. 5 Immunocompromised patients are prone to developing high intrahost diversity. 3–5 Bamlanivimab rapidly selected Spike mutations at position 484 escape variants in two immunocompromised patients. Such variants were not observed within immunocompromised patients admitted to hospital but not treated with bamlanivimab at our hospital in 2021 (n= 69, including 23 patients with multiple samples). Although the US Food and Drug Administration revoked emergency use authorisation for bamlanivimab alone, combination treatment with monoclonal antibodies might also select escape variants in immunocompromised patients. This report emphasises the importance of longitudinal SARS-CoV-2 sequencing in this population.

We declare no competing interests. BL is a member of the French Scientific Committee for SARS-CoV-2. We thank laboratory technicians and Quentin Semanas for their active participation and help with this study. This study was done as part of the global surveillance of influenza viruses by WHO and by the Centre National de Référence des Virus des Infections Respiratoires, supported by Santé Publique France in France. Patients were included in a temporary authorisation for use cohort for bamlanivimab infusion with …