This report of prolonged SARS-CoV-2 infection with COVID-19 in two patients with a defective humoral response to SARS-CoV-2, evidenced by hypogammaglobulinaemia and the absence of detectable nucleocapsid binding IgG, reinforces the importance of antibodies in clearance of SARS-CoV-2 infection. Several reports are now available describing a prolonged course of COVID-19 in immunocompromised patients, largely in the context of primary or acquired antibody deficiency (12–28). The clinical course in our patients was consistent with other reports, with episodes of hospitalisation with typical COVID-19 illness - featuring hypoxia, radiological evidence of pneumonitis and signs of inflammation - on a background of persistent debilitating symptoms of fever, fatigue, dyspnoea and malaise. Interestingly, these latter symptoms resolved promptly upon viral clearance, implying that they were predominantly driven by ongoing viral infection. This latter aspect highlights the clinical rationale for excluding ongoing viral infection in immunocompromised patients with prolonged ‘post-COVID-19’ symptoms.
We describe evidence of therapeutic efficacy of casirivimab and imdevimab (REGN-COV2) leading to clearance of persistent infection in these immunocompromised patients. Of particular note in Case 1, this was achieved despite a prior failure of CP therapy. The response to REGN-COV2 was evident in the rapid and sustained loss of viral RNA (vRNA) from the upper respiratory tract, accompanied by resolution of clinical symptoms and systemic inflammation, leading to full clinical recovery - all in close temporal association to the receipt of REGN-COV2. Given that unremitting clinical illness and detectable SARS-CoV-2 vRNA prior to exposure to REGN-COV2 had persisted for over 190 days in Case 1 and 32 days in Case 2, we postulate a causal association. The seminal randomised controlled trial of REGN-COV2 reported efficacy in the management of patients with COVID-19 in the outpatient setting, although did not report specifically on immuncompromised patient groups (patients on immunoglobulin supplementation were excluded) (32). Other mAb therapies - e.g. bamlanivimab together with etesevimab, and sotrovimab - are also approved under emergency use authorisation in the US (reviewed in (30)). Preliminary unpublished results of the RECOVERY study demonstrate significant efficacy of REGN-COV2 in seronegative, but not seropositive hospitalised patients with COVID-19. In this study, seronegative patients had twice the baseline mortality rate of seropositive patients (30% vs 15%). To our knowledge, only a limited number of case reports have described therapeutic efficacy of mAb therapies in immunocompromised patient groups, however those that do are generally consistent with our observations. Luitel and colleagues described a patient with hypogammaglobulinaemia, treated with immunoglobulin replacement, who had a clinical response to REGN-COV2 monotherapy (27). Interestingly, this patient had evidence of vRNA in bronchoalveolar lavage which was absent from the upper respiratory tract. Nguyen and colleagues reported effectiveness of REGN-COV2 in combination with remdesivir in a patient with X-linked agammaglobulinaemia (XLA) and persistent COVID-19 (23). Conversely, Choi and colleagues reported use of REGN-COV2 in a rituximab-treated patient who was critically ill with COVID-19 and aspergillus coinfection and died 11 days after administration (34). Finally, Kavanagh Williamson and colleagues reported a patient with hypogammaglobulinaemia secondary to chronic lymphocytic leukaemia in whom chronic infection for 290 days was successfully cleared by REGN-COV2 monotherapy (28). Interestingly, the kinetic of resolution appeared relatively slow - taking 45 days, as opposed to 3 days in our case - although in the former the possibility of coinfection with B.1.1.7 lineage occurring after REGN-COV2 could not be excluded. Collectively these findings are promising and support the argument for further studies of the clinical effectiveness of mAb therapy in antibody-deficient patient cohorts.
Responses to CP have been more widely reported in the literature. Generally these reports demonstrated effectiveness in immunocompromised patients (17, 18, 20–22, 35). Less frequently, CP failed to clear infection (15, 19), as in Case 1. In these reports, there were also temporal associations between administration of CP and the accumulation of viral mutations, suggesting that suboptimal antibody pressure may have driven viral evolution (15, 19). This is clearly a cause for concern both from the perspective of the individual patient and for wider public health, arguing for caution in the use of CP in such patients. However in Case 1, viral sequencing provided inconclusive evidence of an association between CP administration and the acquisition of additional variants, since additional spike variants were only detected at three months but not at one month following CP receipt. The del141-143LGV and G769A variants map to the N-terminal domain (NTD) and downward helix of spike respectively. We note similarity of the former to the recurrent 141-144LGVY deletion, a ‘recurrent deletion’ region of NTD (36). This is reported in immunocompromised patients (15) and is associated with a reduction in nAb efficacy (36). The significance of G769A is not yet known. Although several factors may have accounted for the lack of efficacy of CP in our case and in other published cases, it is also consistent with the lack of efficacy of CP in hospitalised patients in large scale randomised clinical trials such as the RECOVERY study (31). One factor that may be influence the effectiveness of CP is the variable quantity of SARS-CoV-2 spike binding and/or neutralising antibody present within the plasma product. An advantage of REGN-COV2, or other highly neutralising monoclonal antibody (mAb) products, is that the antibody content is consistent and well-defined, however the effective dose remains to be defined. Interestingly, in our cases there was limited evidence of a clinical or virological response to remdesivir in vivo. Similar findings have been observed elsewhere (23, 28), in situations where in vitro sensitivity of the SARS-CoV-2 isolate has been demonstrated (28), although in other reports a response to remdesivir was seen, albeit followed by viral recrudescence (14, 19, 34, 37, 38). It remains an open question whether remdesivir is needed as an adjunct to REGN-COV2 or other mAb therapies in the management of chronic COVID-19 in immunocompromised patients.
In both cases, resolution of inflammation occurred following viral clearance in the absence of steroid therapy. This suggests that sustained viral replication makes a relevant contribution to the inflammatory response underpinning COVID-19. Thus while immunomodulation with agents such as dexamethasone and/or tocilizumab is a logical strategy in immunocompetent patients, supported by clear evidence from randomised controlled trials (39, 40), there may also be a clinical need for antiviral therapies and/or interventions designed to enhance antiviral responses in immunocompromised patients. This is relevant since patients with pre-existing immunocompromise are at increased risk of infectious complications of agents such as corticosteroids or anticytokine therapies (23).
We acknowledge there are some important limitations to this report. In both cases, due to the co-administration of remdesivir, we were unable to definitively prove that resolution occurred solely as a consequence of REGN-COV2, although the absence of a virological response to previous courses of remdesivir along with the reported effectiveness of REGN-COV2 monotherapy (27, 28), suggests it is likely that REGN-COV2 made the dominant contribution. T-cell responses may provide a degree of protection against progression to severe disease in antibody-deficient patients (14). We did not formally assess SARS-CoV-2-specific T-cell responses in these patients, although we documented normal numbers of T-cell subsets in Case 1 and the efficacy of CD20/CD3 bispecific therapy in lymphoma was evidence of a functional antitumour T cell response. Finally, whilst we did not undertake sub-genomic RNA PCR analysis or viral culture to prove definitively that persistent detection of viral RNA reflected ongoing replication, this is nevertheless a reasonable assumption, particularly given the rapid clearance following REGN-COV2.
In summary, we report two cases of persistent SARS-CoV-2 infection in antibody-deficient patients that were apparently cleared rapidly following administration of REGN-COV2. Future studies should seek to determine the functional relevance of spike protein variants such as those identified in Case 1. These cases highlight the potential benefit of REGN-COV2 in therapy for persistent SARS-CoV-2 infection in antibody-deficient individuals, including following failure of CP treatment. Studies are warranted to assess the clinical effectiveness of mAb therapy in patients who are unable to generate functional antibodies against SARS-CoV-2, to clarify the dosage needed, to assess the value of antiviral coadministration, and to establish the potential risk of variant development if therapy is ineffective. Another important question for future studies is whether regular administration of REGN-COV2 or other mAb therapies may have value as primary prophylaxis against SARS-CoV-2 infection in vulnerable patient populations.