Failure of bamlanivimab with selection of E484K mutation in an allogeneic stem cell transplant recipient with nosocomial SARS-CoV-2 infection

Abstract

We report the case of an allogeneic stem cell transplant recipient with nosocomial acquisition of SARS-CoV-2 infection who received antispike neutralizing monoclonal antibody bamlanivimab 2 days after diagnosis of SARS-CoV-2 infection but progressed to severe COVID-19 pneumonia and died with the selection of E484K/Q resistance mutations to bamlanivimab.

Background

Bamlanivimab (also known as LY3819253 or LY-CoV555) is the first antispike neutralizing monoclonal antibody approved in France for the treatment of mild to moderate COVID-19. ¹ Bamlanivimab monotherapy at the dose of 700 mg reduced hospitalization rates when administered within 3 days after diagnosis of SARS-CoV-2 infection. ² However, there is a potential risk of emergence of mutations in the spike protein receptor binding domain (RBD) especially among immunocompromised patients.

Case report

A 62-year-old man with a history of relapsing acute myeloid leukemia was admitted in February 2021 for transient headache and confusion 2 months after allogeneic stem cell transplantation. Immunosuppressive treatment included low dose steroids, mycophenolate mofetil, and cyclosporine. Neurological symptoms were finally attributed to cyclosporine-related toxicity and cyclosporine was stopped. Prednisolone at 1 mg per kg was introduced for graft versus host disease prophylaxis and the patient was discharged from hospital. At admission (D0), the patient was briefly in contact with another patient (contact > 2 m, during less than 2 h wearing a mask) later diagnosed with SARS-CoV-2 infection. Nasopharyngeal swabs performed for detection of SARS-CoV-2 by RT–PCR at D0 and day 5 (D5) were negative. Eleven days after exposure (D11), while the patient was asymptomatic, the nasopharyngeal test for SARS-CoV-2 was positive (31 cycle threshold – ct). Bamlanivimab (700 mg intravenous infusion) was administered 48 h later at D13 (Figure 1A). At this time point, RT–qPCR was positive in nasopharyngeal swab (13 ct) and plasma (35 ct), but the patient was still asymptomatic and was discharged home. He was re-admitted 7 days later (D20) for dyspnea without fever while he was still under steroid therapy. Peripheral oxygen saturation was 89% and chest CT-scan showed widespread bilateral infiltrates. Treatment included a combination of high dose steroids, high flow oxygen, and a 5-day course of remdesivir, followed by tocilizumab given clinical and biological deterioration. At D24, nasopharyngeal swab and plasma were still positive for SARS-CoV-2 (24 and 34 ct, respectively) and PCR screening for variants revealed E484K mutation selection (VirSNiP assay, TIB MOLBIO). No N-antibodies were detected. As the patient continued to deteriorate, hyperimmune convalescent plasma was administered (4 units) on D28 and D29. Unfortunately, the patient developed acute respiratory failure requiring admission to the intensive care unit where relapse of acute leukemia was diagnosed and he subsequently died of septic shock at D44.

Figure 1.

(A) Clinical course, nasopharyngeal and plasma SARS-CoV-2 PCR Ct values. Blue box represent days of whole genome sequencing. (B) Whole genome maximum-likelihood phylogenetic tree with patient sequences at three time points (D13, D24, and D35) along with representative sequences from France collected from February to March 2021. Nucleotide sequence alignment was performed with MAFFT. (Best-fit nucleotide substitution GTR+G+I was used for the datasets using model selection in IQ-Tree2 followed by maximum likelihood phylogenetic tree construction using with 1000-bootstrap replicate. ggtree R package was used for tree visualization. (C) Map of synonymous and non-synonymous minority and majority mutations in the sequences at D13, D24, and D35 compared to D13 consensus sequence. Synonymous mutations are represented with “+” symbol, non-synonymous with “*” symbol. Color gradient represents variant frequency among viral population (red =100%, yellow =10%). For the spike region, frequency of non-synonymous mutations at each time point was represented.

We performed whole genome sequencing (QIAseq SARS-CoV-2 primer panel, Qiagen) on Illumina MiSeq platform in three respiratory samples with Ct<30 cycles (D13, D24, and D35). Phylogenetic tree including our three sequences and a set of 1222 whole genome SARS-CoV-2 sequences from GISAID (https://www.gisaid.org/) collected in France between February and March 2021, revealed a cluster of our three strains (bootstrap 100%) included in the GH Gisaid clade (20A.EU.2 nextclade) (Figure 1B). Phylogenetic analysis was consistent with intra-host viral evolution with no evidence for co-infection or super-infection. At bamlanivimab initiation (D13), viral population was homogeneous without mutations in the spike region known to confer immune escape. We compared sequences obtained at D24 and D35 with the D13 consensus sequence and we evidenced minority and majority mutations (threshold 10%) (Figure 1C).

We described non synonymous mutations predominantly in the orf1ab and the spike genes, with 1 and 4 mutations at D24, 16 and 7 mutations at D35, respectively.

Previously reported resistance mutations to remdesivir in the RNA-dependant RNA polymerase gene (RdRp) were not detected.^3,4 In our case, 1 mutation (W215S, 25%) in the RdRp gene was selected at D35 under remdesivir pressure. This mutation occurred in the NiRAN domain (Nidovirus RdRp associated nucleotidyl transferase domain) where no resistance mutation to remdesivir has yet been reported. ⁵

In the spike region, the E484K mutation was selected at D24 and persisted at D35 in 44% and 26% of the viral population, respectively. At D35, we also described a mutation E484Q at a minority level (10%). Three other mutations in the receptor binding domain (RBD) were reported at D24 under bamlanivimab pressure: L362V, Q493T, and S494P at 99%, 28%, and 34% of viral population, respectively. We reported an expansion of Q493T and S494P mutations at D35 (57 and 58%, respectively). Several studies reported immune escape to neutralizing antibodies in case of selection of E484K and E484Q mutations. ⁶ Selection of E484K and S494P was also reported previously in two patients who received bamlanivimab.^7,8 Moreover, E484K, E484Q, and S494P mutations are known to confer resistance to bamlanivimab in vitro.^8–10

Other non-synonymous mutations in the spike region (P9L, N30D, G1131E, and L1203F) were selected at D35 under convalescent plasma pressure. To our knowledge, none of them have been reported to confer immune escape.

In addition, we calculated the ratio of non-synonymous (amino acid altering substitutions) to synonymous substitutions (substitutions that do not alter amino acids) (dN/dS) which have been extensively used as an indicator of selection pressure. From D24, spike region appeared to be under positive selection (dN/dS>1) (x). Recent studies performed in more than 10,000 individuals to evaluate within-host SARS-CoV-2 diversity found that most areas of the genome appeared to be under purifying selection (dN/dS <1), including S. ¹¹ In our patient, monoclonal antibody and hyperimmune convalescent plasma may explain positive selection which conducted rapidly to escape mutations in the spike region.

Conclusion

Bamlanivimab monotherapy in immunocompromised patients diagnosed with SARS-CoV-2 infection may rapidly select resistant SARS-CoV-2 variants with potential risk of dissemination.