WHO's Therapeutics and COVID-19 Living Guideline on mAbs needs to be reassessed

An essential line of defence in a global living-with-COVID-19 policy is formed by effective therapeutic strategies for vulnerable patients,¹ many of whom are excluded from treatment with nirmatrelvir–ritonavir (sold as Paxlovid, Pfizer) by virtue of their comorbidities or interacting medications. Preliminary data suggest that monoclonal antibodies (mAbs) is highly effective for these groups,² and WHO, in its Therapeutics and COVID-19: Living Guideline,³ has previously conditionally recommended the use of sotrovimab (sold as Xevudy, Vir Biotechnology and GlaxoSmithKline) or casirivimab–imdevimab (sold as Ronapreve, Regeneron) for people at high risk of hospitalisation. However, in a Sept 16, 2022, update,⁴ WHO issued a “strong recommendation against” use of these mAbs, stating that they “[do] not neutralize the currently circulating variants of SARS-CoV-2 and their subvariants”.⁴

This guidance requires an urgent reassessment. Based on analysis of both the existing literature and data presented here, mAbs neutralise circulating variants and remain the best treatment option for many vulnerable patients, offering a high benefit-to-risk ratio.

To measure neutralising capabilities of mAbs, we used live-virus microneutralisation assays on Good Clinical Practice-compliant high-throughput platform,5, 6, 7 calibrated to WHO International Standards8, 9 for anti-SARS-CoV-2 immunoglobulin (WHO/BS/2020.2403 and WHO/BS/2022.2427, which we contributed to the calibration of) against sequence-validated batches of nine SARS-CoV-2 variants. Further methodological details on mAbs, virus variants and culture, high-throughput live virus microneutralisation assay, and data analysis, statistics, and availability are available in the appendix.

This work was a benchmarking exercise between the Crick COVID Surveillance Unit and the Legacy study, an ongoing collaboration between University College London Hospitals and the Francis Crick Institute in London, UK. Legacy was approved by London Camden and Kings Cross Health Research Authority Research and Ethics committee (IRAS number 286469) and is sponsored by University College London.

We assayed a panel of mAbs: sotrovimab; casivirimab and imdevimab (sold as Ronapreve, Regeneron); and tixagevimab and cilgavimab (sold as Evusheld, AstraZeneca). EC₅₀ values were calculated by fitting a four-parameter dose–response curve to 288 independent data points, generated from three independent repeats of 12 independent titrations, each consisting of two technical replicates of a four-point dilution series against live SARS-CoV-2 variants (appendix p 2). This approach enabled refined determination of EC₅₀ values and corresponding 95% CIs (appendix pp 3–4). We found that sotrovimab, imdevimab, and cilgavimab neutralised omicron BA.2, BA.2.12.1, BA.4, and BA.5.

Many reports focus their narrative and data presentation on the reduction of neutralisation of a given mAb relative to ancestral SARS-CoV-2; however, this metric is less helpful in assessing efficacy since starting absolute titres versus ancestral virus vary widely, and mAbs formulation, dosing, and administration also vary. Similarly, in the absence of real-world efficacy data, consideration of the EC₅₀ value relative to pharmacokinetic data (eg, measured serum concentrations of mAbs post-administration) requires assumptions regarding tissue penetration and mechanism of action.

In this context, technical aspects of neutralisation assays matter: many laboratories continue to use ACE2-overexpressing cells, despite such cells showing an aberrant lack of neutralisation of SARS-CoV-2 by certain classes of mAbs (including sotrovimab¹⁰) and at a fundamental level, comparison across laboratories is hampered by the use of different cell lines that may be infected by SARS-CoV-2 variants to different extents. Here, calibration with the WHO International Standard for anti-SARS-CoV-2 immunoglobulin and reporting of neutralisation titres in International Units (appendix p 5) would be useful to facilitate such comparisons. Additionally, reporting of CIs (rather than point estimates) is essential to evaluate the significance of any possible changes in neutralisation—especially when considering EC₉₀ values, which lie close to the plateau of the dose–response curve and are inherently noisy, both in cell-based assays and in fitting of a dose–response curve.

With the emergence of the omicron BA.2 subvariant in early 2022, the US Food and Drug Administration (FDA) found a decrease in microneutralisation titre EC₉₀ of 25–48-fold relative to ancestral SARS-CoV-2.¹¹ On April 5, 2022, the FDA withdrew approval of sotrovimab¹² on the basis of their assay results and pharmacokinetic modelling suggesting that the authorised dose was unlikely to be effective against BA.2. In wider reporting, including medical literature,¹³ the prevailing (but, in our view, erroneous) view became that sotrovimab is ineffective against BA.2 and BA.5, with the dose aspect of the FDA's statement set aside.

We found that sotrovimab neutralised BA.2 with an EC₅₀ of 1849 ng/mL (95% CI 1429–2391), representing a 22-fold reduction versus ancestral SARS-CoV-2, with the EC₅₀ remaining 64-fold below the mean peak serum concentration of sotrovimab (appendix p 6) and 13-fold below the mean serum concentration 28 days post-administration. These and other data led UK health authorities, in contrast to US health authorities, to retain sotrovimab for the treatment of extremely clinically vulnerable patients who are at risk of progression to severe COVID-19, given the absence of an alternative approved mAb and clinical data showing a reduction in real-world efficacy.

It remains challenging to integrate clinical trial results into these assessments. Most trials use a primary endpoint of a positive COVID-19 test since they recruit too few participants to power the statistical analysis of progression to severe disease, which would be the more relevant metric for clinical practice for extremely clinically vulnerable patients and for public health policy to reduce the burden on health-care infrastructure. However, large-scale analyses of digital primary care health records from the UK National Health Service, linking clinical outcomes with patient metadata, have recently shown that sotrovimab was superior to treatment with molnupiravir in preventing COVID-19-related hospitalisation and death in extremely clinically vulnerable populations during the period of omicron BA.2 wave.²

This demonstration of sotrovimab's efficacy against BA.2 can be used in conjunction with our in-vitro neutralisation data to conservatively infer real-world efficacy against emerging variants of concern: those that are neutralised to the same extent as, or even better than, BA.2 (numerically, a lower EC₅₀) would be expected to remain effective. We found that sotrovimab neutralised BA.4, BA.5, and BA.2 to similar extents (EC₅₀=1490 ng/mL; 95% CI 881–2517), suggesting that sotrovimab would remain effective against BA.5. Similarly, a second-generation BA.2 variant, BA.2.12.1, was neutralised to a greater extent than parental BA.2 (EC₅₀=1211 ng/mL; 95% CI 844–1738), in line with preliminary pseudotyped lentivirus neutralisation data on a wider set of second-generation omicron sublineages, including BA.2.75.2.¹⁴ In light of this evidence, it would be reasonable to retain the use of sotrovimab, especially in extremely clinically vulnerable patients who test positive for COVID-19 and have few other options.

The lack of directionality in the degree of neutralisation by a mAb and successive variants is worth highlighting, perhaps most strikingly in the case of the imdevimab component of ronapreve. As omicron spread in the UK in December, 2021, we and others found that BA.1 fully escaped from casirivimab and imdevimab in vitro,7, 15, 16, 17 with no neutralisation whatsoever at concentrations up to 18 750 ng/mL. UK policy was then changed to restrict the usage of casirivimab–imdevimab to infection with a non-omicron variant on Dec 24, 2021, during the UK's transition from delta to omicron BA.1.¹⁸ However, we later found that the imdevimab component of Ronapreve was able to neutralise subsequent omicron BA.2, BA.2.12.2, BA.4, and BA.5 variants (appendix p 4). For BA.4 and BA.5, this value is 536-fold below the mean peak serum concentration and 92-fold below the mean serum concentration 28 days post-administration (appendix p 6), reflecting a greater degree of neutralisation by imdevimab than sotrovimab in vitro, despite only the latter remaining in clinical use.

Our results on imdevimab are consistent with other reports published in June–August, 2022.19, 20, 21 All together, the evidence does not support WHO's decision in September, 2022, to withdraw conditional recommendation of casirivimab–imdevimab for seronaive patients. Regulators might need to re-evaluate withdrawn mAbs in light of evidence of regained activity against the spike protein of a future variant.

In the case of sotrovimab, the combined evidence from our in-vitro neutralisation and real-world clinical efficacy data supports its continued use against circulating omicron variants, including BA.4 and BA.5. The ongoing evolution of SARS-CoV-2 variants and continued global transmission has resulted in a situation where new variants can replace one another within weeks. Although WHO's Living Guideline posits⁴ the “need for clinical trial evidence in order to confirm any clinical effectiveness of new monoclonal antibodies that reliably neutralize the circulating strains”, generating this evidence is challenging since the results would almost instantly become obsolete. This situation is compounded by the overall low rate of hospital admissions in the face of high levels of population immunity: trialists struggle to recruit sufficient participants to power analysis of efficacy of new therapies against progression to severe illness.

Overall, our results highlight how the regulatory environment for mAbs has not kept pace—a fact also illustrated by the cilgavimab component of Evusheld, which showed strong neutralisation against all omicron variants tested here (appendix p 4). Evusheld is now approved for treatment in Europe and Japan but remains approved only for prophylactic use in the USA and the UK, in part due to a dogmatic focus by some regulators on not licensing the same therapeutics for prophylaxis and treatment of acute COVID-19. Bebtelovimab (not studied here and not assessed in WHO's Living Guideline⁴) remains unavailable outside the USA,22, 23 where the FDA adopted a more relaxed approach to Emergency Use Authorisation based on data showing improved symptoms and somewhat reduced viral loads from phase 1 and 2 trials together with data in-vitro neutralisation.²⁴ That regulatory flexibility is limited: although long experience with antiviral therapies suggests mAb combination therapy (eg, across mAb classes or in combination with small-molecule antivirals) is preferrable to monotherapy, the regulatory and commercial backdrop makes these kinds of cross-company trials challenging. The net result is that effective monoclonals are available but not offered to extremely clinically vulnerable patients with COVID-19 who are at risk of progressing to severe disease.

We recommend, first, that WHO's Living Guideline⁴ be further updated to reflect the available data, and second, that more responsive regulatory approaches are developed to integrate high-quality, standardised live-virus neutralisation data with efficacy data from real-world clinical use. This approach might contribute positively towards determining an index of protection for a given mAb (or class of mAbs with shared Fc modifications considered as a single platform)—analogous to a correlate of protection for vaccine-induced antibodies.

At present there is an unrealistically high threshold to enter a therapeutic agent into clinical practice. The threshold to withhold or withdraw the same agent is much lower when based on in-vitro evidence for loss of potency alone. Such a situation disproportionately affects vulnerable patients whose other essential medications or comorbidities exclude COVID-19 therapeutics other than a neutralising mAb. This situation also strongly disincentivises development of novel antivirals that are needed to continue to offer protection to highly vulnerable populations.

For data and full R code see https://github.com/davidlvb/Crick-UCLH-Legacy-Monoclonals-2022-10