Abstract
Multiple monoclonal antibodies have been shown to be effective for both prophylaxis and therapy for SARS-CoV-2 infection. Here we aggregate data from randomized controlled trials assessing the use of monoclonal antibodies in preventing symptomatic SARS-CoV-2 infection. We use data on changes in the in vivo concentration of monoclonal antibodies, and the associated protection from COVID-19, over time to model the dose-response relationship of monoclonal antibodies for prophylaxis. We estimate that 50% protection from COVID-19 is achieved with a monoclonal antibody concentration of 939-fold of the in vitro IC50 (95% CI: 135 – 2073). This relationship provides a quantitative tool allowing prediction of the prophylactic efficacy and duration of protection for new monoclonal antibodies administered at different doses and against different SARS-CoV-2 variants. Finally, we compare the relationship between neutralization titer and protection from COVID-19 after either monoclonal antibody treatment or vaccination. We find no evidence for a difference between the 50% protective titer for monoclonal antibodies and vaccination, although vaccination is predicted to be capable of achieving a higher maximum level of protection.
Introduction
Vaccination has been shown to be highly effective at preventing both symptomatic and severe COVID-19 (reviewed in1). However, vaccination is less effective in many immune compromised and elderly individuals where immunogenicity and clinical data show considerably impaired responses to vaccination2,3. Multiple monoclonal antibody products have been shown to be effective as pre- and post-exposure prophylaxis against pre-Omicron variants4-6, as well as when administered therapeutically early in infection7-11. We recently analysed the available data on antibody treatment of symptomatic SARS-CoV-2 infection to determine the dose-response relationship between the antibody dose administered (after conversion to a neutralizing dose equivalence) and the protection from progression to hospitalisation12. However, the dose-response curve for monoclonal antibody administration as prophylaxis of COVID-19 has not yet been determined, since there have to-date been fewer randomized control trials assessing prophylactic effect. Here we adopt an alternative approach, comparing the loss of antibody in vivo with the loss of efficacy of monoclonal antibodies over time following administration. Using this temporal data, we estimate the relationship between antibody concentration and protection, which may provide a valuable clinical tool for predicting the efficacy of new monoclonal products and existing products against new variants12. Finally, we assess whether neutralizing antibodies mediate protection or merely correlate with protection in vaccinated individuals by comparing the relationship between neutralization titer and protection after vaccination13 and in naïve individuals receiving monoclonal antibodies. Together this work provides a quantitative framework for dissecting the mechanisms of protection in vaccination and informing the use of critical immunotherapies.
Results
Aggregating studies of monoclonal antibodies as prophylaxis
We searched MEDLINE, PubMed, Embase and the Cochrane COVID-19 Study Register for randomized placebo-control trials of SARS-CoV-2-specific monoclonal antibodies (mAbs) used as pre-exposure and peri-exposure prophylaxis for COVID-19. We included only studies where both protection from symptomatic infection and pharmacokinetic information of the monoclonal antibody were provided within the same study. We identified five eligible studies assessing monoclonal antibodies as pre-exposure and peri-exposure prophylaxis for COVID-194-6,14,15 (see Table S1). The antibodies used in these studies were casirivimab/imdevimab (3 studies), bamlanivimab, and tixagevimab/cilgavimab. One of these studies did not provide data on the pharmacokinetics of the antibody (bamlanivimab)15 and was excluded. Of the remaining four studies, two reported a break-down of cases in treatment and control arms by time since administration and two studies had data on the timing of cases that could be extracted from the publication4-6,14. These four studies assessed protection at a time before the omicron variants were the dominant circulating variants and against variants where the antibody had been shown to be able to neutralize the virus in vitro16. The reported overall efficacies in the included studies ranged from between 76.7 – 92.4%. We identified a trend for lower efficacies with increasing time since administration (Figure 1).
A significant dose-response relationship between protection and mAb concentration
To investigate whether declining efficacy with time was indicative of a dose-response relationship between mAb concentration and efficacy, we compared the antibody concentrations reported within different time intervals in each study with the reported efficacy at the corresponding time interval (Figure 2). To compare between antibodies of different potencies, we normalized antibody concentration using the in vitro IC50 for each antibody from a meta-analysis of in vitro studies (Table S2 and Figure S1, as previously reported12 using data from Tao et al.17). We found a significant relationship between efficacy and mAb concentration (as a fold of the in vitro IC50) (p=0.016, generalized linear mixed model (GLMM) and chi-squared test, see Methods). We note that this relationship is predominately driven by observations of one study, where the longest time course of protection was collected (i.e. the relationship is only significant when the Herman et al.14 data is included, Table S3). Fitting a logistic dose-response relationship to this data, we estimated a peak efficacy of 91% (95% CI: 86 – 99%), and concentration for 50% of efficacy of 939-fold in vitro IC50 (95% CI: 135 – 2073) (Table S4). This equates to a plasma antibody concentration of 3.7 mg/L concentration for tixagevimab/cilgavimab and 3.1 mg/L for casirivimab/imdevimab being required for 50% protection against COVID-19.
Predicting monoclonal antibody efficacy against new variants
A major challenge in the COVID-19 pandemic has been in decision-making around whether pharmaceuticals shown to be effective against ancestral SARS-CoV-2 should continue to be used when new variants emerge. This is especially true for monoclonal antibody therapies, where recommendations on the use of mAb therapeutics have changed numerous times with the emergence of each Omicron subvariant18,19. This has been particularly difficult when a mAb loses partial, but not complete, recognition of a new variant. If we assume that the relationship defined here between antibody concentration and efficacy for ancestral virus will continue to hold for variants of concern, as it has for vaccine-induced neutralizing antibodies1,20, we can use the dose-response relationship in Figure 2 to estimate the loss of efficacy and duration of protection of monoclonal antibodies to new variants. For example, tixagevimab/cilgavimab administered intramuscular at a dose of 300 mg is predicted to maintain >50% protection for 280 days (95% CI: 171 – 546 days) against the ancestral variant where the in vitro IC50 is 3.99 ng/mL and the half-life of antibody 95 days (Figure 3, Table S2 and Table S5). However, if the in vitro IC50 against a new variant were increased (but still detectable), as is the case for some mAbs against the Omicron subvariants (Table S2), the model allows us to predict the reduction in the time above 50% protection. In a meta-analysis of published studies, tixagevimab/cilgavimab was found to maintain detectable in vitro neutralization titers against the BA.1, BA.2 and BA.4/5 Omicron subvariants (see Table S2, meta-analysis as described in12, using data from Tao et al.17). However, the increase in IC50 for this antibody combination against all of these subvariants is predicted to cause antibody efficacy to drop well below 50% (for BA.1, BA.2 and BA.4/5 the efficacy at peak concentration is less than 1%). For example, the IC50 of tixagevimab/cilgavimab against BA.2 is 9.7-fold (95% CI: 2.9-32.8) higher than the IC50 to ancestral virus. Using the dose-response relationship here (Figure 2), this would predict that the duration of >50% protection of tixagevimab/cilgavimab against BA.2 will be reduced to 0 days, but with very wide confidence bands due to the uncertainty in the concentration associated with 50% protection (95% confidence intervals (95% CI: 0 – 234 days)). Similarly, casirivimab/imdevimab has larger increases in the IC50 against these subvariants (Table S2) and is not predicted to maintain >50% efficacy to these variants (the efficacy at peak concentration is below 1% for BA.1, BA.2 and BA.4/5).
Another formulation of this question is to ask “What is the maximum increase of IC50 (drop in neutralization titer) that can be tolerated while still maintaining a minimum duration of protection?”. For example, if we wish to provide a period of at least 30 days with >50% protection, then tixagevimab/cilgavimab and casirivimab/imdevimab can tolerate at most 6.2-fold (95% CI: 2.8 – 43.2) and 18.5-fold (95% CI: 8.4 – 129.1) increases in in vitro IC50, respectively. Figure 3 shows the predicted duration of >50% protection for casirivimab/imdevimab and tixagevimab/cilgavimab for any given change in in vitro IC50. This analysis provides a tool to determine how regularly monoclonal antibodies may need to be re-administered to provide high confidence that subjects will maintain more than 50% efficacy against the circulating variant.
Comparing monoclonal antibody prophylaxis with vaccine-induced protection
Multiple lines of evidence have established that neutralizing antibody titers correlate with protection from COVID-19 in vaccinated individuals13,21-23. An important question in whether neutralizing antibodies are mechanistic in mediating this protection, or merely correlate with protection24. Similarly, if antibodies are able to directly mediate protection, identifying the magnitude of their contribution to overall protection (compared to other mechanisms) is an important question. One way to address this is to compare the level of protection achieved after administration of antibodies alone with that achieved after vaccination. Antibody administration alone should reflect the antibody-related contribution to protection, while vaccination should incorporate both antibody- and other-mechanism-based protection. To address this, we first analysed whether prophylaxis against COVID-19 after passive antibody administration is achieved at similar levels of neutralisation to protection observed after vaccination.
Figure 4a compares the efficacy of high-potency mRNA vaccines25,26 and monoclonal antibody prophylaxis in prevention of symptomatic SARS-CoV-2 infection with the ancestral variant. We find that the observed mean protection achieved with mAbs of 91.9% (95% CI: 84.2-96.4) was not significantly different to the efficacy achieved with vaccination of 94.5% (95% CI: 91.6-96.7) (p=0.31, GLMM, see Methods). However, it may be that this was only possible by administering monoclonal antibodies at much higher neutralizing antibody titers compared with the neutralising antibody titers achieved in vaccinated individuals. Thus, we next compared the level of protection achieved for a given neutralizing antibody titer after either vaccination (from Khoury et al.13) or after treatment with monoclonal antibodies (Figure 4). To compare titers between vaccination and monoclonal antibody administration, we normalized the titers of each to a scale relative to the geometric mean titer of neutralizing antibodies seen in convalescent individuals against ancestral virus, after the first wave of COVID-19 (a ‘fold-convalescent’ scale, as previously described for the studies of vaccination13, and as described in the Methods for mAbs). In order to test whether the neutralizing antibody titer associated with a given level of protection for vaccines and mAbs were consistent or different, we fitted these data together using the same parameters for the monoclonal antibody and vaccine dose-response relationships. We then tested whether the model fitted better with separate parameters for mAb treatment and vaccination (likelihood ratio test, see Methods and Figure S3). This analysis showed that the best-fit model was one where the same dose-response relationship existed for both vaccination and mAbs but with the estimated maximum protection being higher for vaccination, which was estimated at 98.9% (95% CI: 94.5-100%) and 91.4% (95% CI: 85.1-96.1%) for vaccines and monoclonals respectively (p=0.011, Figure 4, Figure S3, Table S6). Once allowing for different maximal efficacy, we found a trend towards mAbs requiring a higher estimated in vitro neutralisation titer than vaccination to produce 50% protection (Figure S4), but this was not significant (0.23 (95% CI: 0.17-0.49) vs 0.49 (95% CI: 0.10-3.08) fold convalescent for vaccines and mAbs, respectively, p=0.33, Figure S3). Given the limited data on monoclonal antibody efficacy at low antibody concentrations, there is limited power to detect whether the level of antibodies required for 50% protection is different for mAbs compared with vaccination. Together, we see that the protection from COVID-19 achieved after vaccination and monoclonal antibody administration are observed at similar neutralisation titers, with a trend to monoclonal antibodies predicted to require higher (∼2-fold) in vivo neutralization titers to achieve the same protection, and mAbs may reach a lower maximal protection against COVID-19, compared with vaccination.
Discussion
Here we demonstrate a relationship between the monoclonal antibody concentration and efficacy to prevent COVID-19. Further, we estimate the concentration of antibody required to have a high confidence of maintaining at least 50% protection. Although we acknowledge that this relationship contains considerable uncertainty due to the sparseness of data, our model fitting enabled us to quantify this uncertainty and estimate that if a treated population can maintain a mean in vivo monoclonal antibody concentration of > 939-fold (95% CI: 135 – 2073) of the in vitro IC50 of the antibody to the circulating variant, they should maintain > 50% efficacy against COVID-19. Analysis of the dose-response curve for monoclonal antibodies allows prediction of the level and duration of protection against different SARS-CoV-2 variants (Figure 3). Our results suggest that both casirivimab/imdevimab and tixagevimab/cilgavimab combinations would provide >50% protection against the ancestral SARS-CoV-2 strain for 151 and 280 days respectively, and that at the current doses this protective interval is dramatically reduced to 56 and 0 days by even a 10-fold loss of neutralization to a new variant. Counterintuitively, although antibodies with a longer half-life are expected in general to provide protection for longer, these are also expected to lose more days of protection for a given fold-increase in IC50 (to a new variant) than compared to mAbs with shorter half-lives (Figure 3). The higher susceptibility of therapeutics with longer half-lives to fold-shifts in the IC50 has been discussed previously for antimalarial products27, and can be explained by considering that when antibodies lose 2-fold neutralization against a new variant it is equivalent to the mAb losing one half-life of time above a threshold. Therefore, a mAb with a 100-day half-life will lose 100 days above a specified threshold, whereas a mAb with only a 30-day half-life will lose 30 days above the same threshold.
The estimated in vivo concentration of antibody required for 50% protection from COVID-19 is much higher than the level of antibody required to neutralize virus in vitro (approx. 900-fold), suggesting that in vivo neutralization may be much less efficient than the observed neutralization in vitro. This difference between in vitro IC50 and the in vivo 50% protective titer is not unexpected, given the major differences between infection in these environments. For example, in vitro neutralization assays usually involve pre-incubation of antibody and virus for an hour before exposure to cells. Similarly, the in vitro IC50 in plaque reduction assays estimates the antibody concentration required to neutralize 50% of virions. However, the dose required to completely neutralize large inocula may be considerably higher28. In addition, in vivo antibody titers are assessed in the serum. However, antibody concentration at the mucosa is lower than the plasma level29, and thus higher (serum) titers may be required to achieve neutralization on mucosal surfaces.
We and others have previously shown that neutralizing antibodies are a correlate of protection from COVID-1913,21,30-32. A major question in understanding vaccine-mediated immunity is whether neutralizing antibodies are simply a surrogate marker of protection or are mechanistic in protecting individuals from symptomatic infection13. To-date it has only been possible to consider this question indirectly. For example, we have noted that the drop in neutralizing antibodies against new variants and over time both provide good predictions of the change in efficacy of vaccines over time and against new variants1,33. However, this study of monoclonal prophylaxis studies in unvaccinated individuals provides a direct means of assessing the dose-response function of neutralizing antibodies in otherwise naïve individuals. We have been able to show that, for a given neutralization titer, the protection induced by vaccination or monoclonal antibodies is comparable (Figure 4), although higher maximal efficacy was estimated with vaccination. Strictly, this higher maximum efficacy for vaccines was only estimated to occur at neutralization titers above those seen in the data observed after vaccination. This suggests that neutralizing antibodies are sufficient to explain the majority of protection from symptomatic infection induced by vaccination. A caveat to this analysis is the limited data on efficacy of monoclonal antibodies at low concentrations (Figure 2).
The difference in protection at a given neutralization titer between vaccination and monoclonal antibody therapy may be due to the additional benefit in vaccinees of a polyclonal antibody response, other non-neutralizing functions of antibodies, recall of immune memory and/or other cellular immune responses. These functions may contribute to the estimated higher maximal protection for vaccines, as well as contribute to the estimated 2-fold higher neutralization titers required to achieve 50% protection with mAbs compared with vaccination (although this difference in IC50’s is not significant). Additionally, the analysis is heavily dependent on the circulating antibody levels and protection observed in the 1-6 months after administration of tixagevimab/cilgavimab and casirivimab/imdevimab antibodies. It is possible that mAb function had been degraded over this period of circulation in vivo (and thus if circulating antibody function were directly measured at this time it may have a lower neutralisation per microgram than would be observed earlier in infection). While our analysis has shown that neutralizing antibodies are sufficient for protection from COVID-19, it is not possible to conclude from this analysis that neutralizing antibodies are necessary for protection. We note that evidence in animal models supports the findings that neutralising antibodies mediate protective immunity34, with some showing an additional benefit of Fc-receptor function35.
The analysis presented here has a number of limitations. Firstly, our dose-response analysis requires comparison of the in vivo measured antibody concentrations and the estimated in vitro IC50s. We also assume the population average half-life of antibodies in the treated population, rather than assessing individual titers over time. Further, we are reliant upon summary data on monoclonal antibody protection broken down by time as reported in, or extracted from, the published studies, and thus could not account fully for subjects lost to follow-up (although fortunately these numbers are relatively small). To gain more robust estimates of the dose-response curve for the use of monoclonal antibodies for prophylaxis, studies with longer follow-up times would be greatly beneficial where this is ethical.
Another factor affecting our analysis is the different mode of delivery of the monoclonal antibodies. Casirivimab/imdevimab was administered subcutaneously in the Isa5, O’Brien6 and Herman14 studies, whereas tixagevimab/cilgavimab was administered intramuscularly in the Levin study4. Plasma antibody concentrations had much slower increases following intramuscular administration of tixagevimab/cilgavimab compared with subcutaneous administration of casirivimab/imdevimab, and thus it is possible that there is a delay until protective antibody concentrations are achieved (Figure 1). To avoid this difference and also to account for the risk of infection at the time of antibody administration, we omitted the earliest time interval from our analysis (the first 7 to 10 days of each study).
Studies of in vitro neutralization of different SARS-CoV-2 variants provide a range of different IC50’s for each antibody. We therefore used a meta-analysis of in vitro IC50’s to inform the mean IC50 for each antibody and to normalize antibody levels to the average convalescent titer for comparison with vaccine titers17. Only a subset of these studies reported the geometric mean neutralization titer of a cohort of convalescent plasma from early in the pandemic (n=13). This meta-analysis aggregated IC50’s estimated from a wide range of neutralization assays, and the definition of convalescent sera was specified differently in each study, introducing some potential confounders to these aggregated estimates. However, our multiple regression model included random effects for study, which can account for systematic differences in neutralization in the studies – for example due to assay differences.
Conclusion
Vaccination has provided a high level of population immunity to COVID-19. However, there remain a number of subgroups in which vaccination is either not possible or is ineffective (largely due to immunodeficiency). The use of monoclonal antibodies for prophylaxis in these cohorts has the potential to provide long-term protection from both symptomatic and severe COVID-19 for these vulnerable groups4-11,14,15. However, the frequent observation of novel SARS-CoV-2 variants that escape antibody recognition has raised significant challenges in predicting monoclonal antibody protection against new variants. Further work is required to obtain more data on protection at low antibody levels, as well as to validate predictions of prophylactic efficacy against SARS-CoV-2 variants. Within this context, the work presented here provides a quantitative and evidence-based framework for predicting monoclonal antibody efficacy that can be used in the assessment of novel therapeutics or in designing optimal regimes for new SARS-CoV-2 variants.
Methods
Search strategy for studies of COVID-19 prophylaxis with monoclonal antibodies
We searched MEDLINE, PubMed, Embase and the Cochrane COVID-19 Study Register for randomized control trials of monoclonal antibodies for the prevention of COVID-19. We identified five studies. We included only the four studies where monoclonal antibody concentration was also reported for the cohort (Table S1). These studies were in a mixture of true pre-exposure prophylaxis and peri-exposure prophylaxis settings (Table S1). Two studies, O’Brien et al.6 and Herman et al.14, reported results from the same clinical trial over different follow-up intervals (4 weeks and 8 months respectively). Thus, to avoid duplication of the same trial results, we integrated the results from these studies. In particular, the O’Brien trial reported outcomes on a weekly basis for 4 weeks whereas the Herman trial reported outcomes on a monthly basis for 8 months. Therefore, for these trials the weekly outcomes reported in O’Brien were used for the weeks 2 to 4 after administration (the initial week was omitted due to rising antibody levels in this period) and the results from Herman et al., were used for the months 2 to 8 only. In addition, antibody concentration data for the cohort was extracted from Figure S4 of O’Brien et al. from 0-168 days, whereas the Herman reported only pharmacokinetic model predictions of the concentration over the interval of 30-240 days. Therefore, the raw O’Brien et al. antibody concentration data was used from 0-168 days and the predicted concentrations from Herman et al. were used for the remaining interval, 168-240 days after treatment. This is indicated in Figure 1B by different a different line type (solid for in vivo concentration data from O’Brien et al. and dashed for modelled concentration data from Herman et al.).
Meta-analysis of monoclonal antibody IC50 and neutralization titer for early pandemic convalescent sera
A previous systematic review and meta-analysis of in vitro studies of monoclonal antibodies IC50 against ancestral SARS-CoV-2 and BA.1, BA.2 and BA.1.1 Omicron subvariants has been reported17, and provided the raw data from this review. In addition, we have previously extended this systematic review to include studies of in vitro IC50 against the BA.4/5 subvariants12. Thus, using this previously developed dataset of in vitro IC50’s we modified a previously performed meta-analysis of the in vitro IC50 of antibodies (bamlanivimab, casirivimab, cilgavimab, imdevimab, sotrovimab, tixagevimab, etesevimab, regdanvimab and bebtelovimab, as these were the antibodies of interest in our previous analysis12) to include tixagevimab/cilgavimab and casirivimab/imdevimab combinations (Table S2). This meta-analysis involved performing a mixed-effects linear regression (with censoring if IC50 above 10,000) using the lmec package36,37, as described previously12. From this meta-regression we derive a central estimate of the IC50 for each antibody-variant combination (casirivimab/imdevimab, tixagevimab/cilgavimab, see Figure S1) considered here, which we use to calculate the in vivo antibody concentration as a fold of the in vitro IC50 as described below. In addition, we searched all of the studies included in the systematic review described above for additional data on whether the study reported neutralizing antibody titers for a panel of convalescent subjects from early in the pandemic. We identified 10 studies in which a panel of convalescent sera (obtained from individuals with infections that occurred early in the pandemic, when ancestral viral linages dominated transmission) were assessed for their neutralizing titers38-48. Additionally, we identified a further 2 studies where a panel of convalescent sera from individuals infected pre-delta variant of concern were assessed49,50. We included the (inverse-transformed) geometric mean neutralization titers reported for these panels of convalescent sera in our meta-regression in order to estimate a mean neutralization titer of early pandemic convalescent sera across the same set of studies as was used to estimate a mean IC50 (Figure S1). This mean convalescent titer across the studies was used to estimate the neutralization titer (on the fold-convalescent scale) for each monoclonal antibody based on the concentration data reported in each prophylaxis study as is described below.
Estimation of antibody concentration on fold-IC50 scale and estimate of neutralization titer on fold-convalescent scale
Antibody concentrations in each study were extracted and normalized as a ratio to the IC50 from the meta-regression described above (Table S2, reported as antibody concentration, fold-IC50). Note that since these studies used antibody combination therapies, the total antibody concentration (sum of both antibody components) was used for the antibody concentration. When comparing the efficacy of these monoclonal antibody trials against that seen after vaccination, we used the estimated neutralization titer in individuals treated with mAbs, converted to a fold-of-convalescent scale since all vaccine trials has been aligned based on this scale previously13. This was estimated by taking the antibody concentration as a fold of the IC50 described here, and dividing by the mean convalescent neutralization titer (as described previously12).
Test of concentration effect on the efficacy of prophylactic mAb treatment
Since the concentration data is very sparse below a concentration of 1,000-fold in vitro IC50, we tested if there is a significant effect of the antibody concentration on the efficacy in the data. For this analysis, we used a log-binomial regression model, a generalised linear mixed model (GLMM) with a binomial error family and logarithmic link function (using the glmer function of the lme4 package51 in R version 4.2.152). The model includes random intercepts for different trials, the covariate “treatment” and the interaction of treatment with concentration. The significance of the interaction of treatment and concentration was tested with a chi-squared test (with the function drop1).
Dose-response fitting with maximum likelihood approach
To estimate the dose-response curve, we fitted a logistic function to the efficacy by concentration data (Figure 2, Figure 4). As we aimed to estimate the concentration that gives 50% protection, we scaled the logistic function such that this concentration is directly estimated as a parameter of the function. The efficacy for a treatment with mAb concentration c is then given by where m denotes the maximal efficacy, k is a slope parameter and c50 is the concentration that gives 50% efficacy. Note that this transformed logistic function will be ill-defined when m is below 50%.
We fitted this dose-response curve to the data with a maximum likelihood approach (as previously described12). Briefly, the likelihood function used in this optimisation was, where p denotes the parameters of the likelihood function, i.e. the three parameters of the efficacy function (m, k, c50) and the baseline risk bτfor each trial/time interval combination (τ), Binom is the probability mass function of the binomial distribution, and for each trial/time interval combination, and are the numbers of events (symptomatic infections) and total number of individuals in the control group, and are the numbers of events and total number of individuals in the treatment group in the corresponding trial time interval, and bτ is the baseline risk which is reduced by the efficacy of treatment for the treatment group in the same interval. The initial guess of the trial/time interval combinations baseline risk used in the optimization were . The parameter c(τ) is the (log10) concentration of monoclonal antibodies (in the fold IC50 scale) in the trial time interval τ.
The negative log-transform of this likelihood function was minimised using the nlm optimiser in the R statistical software package to estimate the log-transform of the model parameter k, c50 and the logit-transform of the parameter m. The optimiser was run 25 times using randomly generated initial parameters for the transformed parameters m, k, c50 drawn uniformly from the ranges logit(m) = [logit(0.6), logit(0.99)], log(k) = [log(0.1), log(100)] and log(c50) = [log(10), log(5000)].
Model fitting was used for parameter estimation and hypothesis testing, the latter using a likelihood ratio test of nested models. In all fitting, we excluded the earliest time interval of each study to account for the rapid change of the antibody concentration over this time interval and ensured exclusion of treatments which occurred after an unidentified infection (see Table S1 for the time intervals included in the analysis).
The confidence intervals of the fitted model and the fitted model parameters were estimated by bootstrapping. In particular, trial time intervals (i.e. each point in Figure 2) were randomly sampled with replacement 10,000 times. For each random sample of the data the model was refitted to the sampled data as described above (with 25 random initial parameter guesses used for each iteration, plus the best fitting model parameters estimated above) and the parameters that fit best to each bootstrapped dataset were stored. The 2.5% and 97.5% percentiles of the estimated parameters provided an estimate of the 95% confidence intervals. Similarly, evaluating the confidence intervals of the fitted model in Figure 2 was estimated by evaluating the model for each at each of the 10,000 bootstrapped parameter estimates and taking the 2.5% and 97.5% percentiles of the evaluated models at each antibody concentration.
Timing of protection
Using the dose-response curve, we can predict the protection over time and how long the protection remains above 50% protection (Figure 3). With the meta-analysis of IC50’s against different variants, the timing of protection can be predicted not only for the ancestral strain but also for variants (Figure 3).
We assumed that the concentration of mAbs declines exponentially from the time of the peak concentration for casirivimab/imdevimab and tixagevimab/cilgavimab (extracted from the data) with a half-life fitted to the data (linear fit to log-transformed concentration, see supplementary methods). With the concentration over time and the efficacy by concentration (Figure 2), we computed the protection over time. For variants, we scaled the concentration data by dividing the fold-IC50 concentration from the meta-analysis by the fold-increase in the IC50 and computed the protective efficacy in the same way as against ancestral. The time from treatment to 50% protection is then the time until the concentration falls to the concentration that gives 50% protection which we computed from the concentration over time (using a linear fit to the log-transformed concentration data, Figure S2). The uncertainty in the time to 50% protection is due to the uncertainty in the concentration that gives 50% protection. The upper and lower bounds for the time to 50% protection are the time to reach the upper or lower bound of the 95% CI of the concentration that gives 50% protection, respectively.
Comparison of the relationship between neutralizing antibodies and protection for vaccination and monoclonal antibody prophylaxis
To compare the mAb data with the data from vaccine studies (Figure 4a), we aimed to match the mAb data as closely as possible with the vaccine studies. Thus, we restricted the mAb data to 2-3 months after treatment, patients who are PCR-negative at baseline, and cases later than 1-2 weeks after treatment (the start of follow-up in vaccine studies) (Figure 4). We then used a generalised linear mixed model (GLMM) with a binomial error family and logarithmic link function (see above). The model included random intercepts for different trials and a treatment variable with the factors “control”, “mAb” and “vaccine”. The treatment effect of mAbs and vaccination was compared by testing if there is a significant difference between the coefficients for mAb treatment and vaccination (using the glht function from the multcomp package54).
To further compare the efficacy of vaccination and prophylactic mAb treatment, we normalized the concentration to a common ‘fold-convalescent’-scale (as previously described12,13. Using a maximum likelihood approach (see above), we fitted logistic dose-response curves to the monoclonal antibody and vaccine data (Figure 4). We tested whether there is a significant difference between the monoclonal antibody prophylaxis and vaccination by fitting all data with the same parameters for the two types of treatment (Figure 4A, black curve) and compared this fit to fits which have different parameters for antibody treatment and vaccination (e.g., Figure 4A red and blue curves). Different models were compared with a likelihood ratio test. The p-values for different model comparisons are reported in Figure S3 and parameter values for different fits in Table S6. Confidence intervals were calculated using bootstrapping, as described above.
Data Availability
All data was extracted from publicly available sources and all extracted data and code will be made available on GitHub upon publication.
Ethics statement
This work was approved under the UNSW Sydney Human Research Ethics Committee (approval HC200242).
Funding statement
This work is supported by an Australian government Medical Research Future Fund awards GNT2002073 (to MPD, SJK), MRF2005544 (to SJK and MPD), MRF2005760 (to MPD), an NHMRC program grant GNT1149990 (SJK and MPD), and the Victorian Government (SJK). DSK and SJK are supported by NHMRC fellowships. DC, ZKM and EMW and MPD are supported by NHMRC Investigator grants. KLC is supported by PhD scholarship from the Leukaemia Foundation, the Haematology Society of Australia and New Zealand (HSANZ) and Monash University. The Melbourne WHO Collaborating Centre for Reference and Research on Influenza is supported by the Australian Government Department of Health.
Competing Interests statement
The authors declare no competing interests.
Data Availability Statement
All data and code will be made available on GitHub upon publication.