BACKGROUND:Soluble species of multimeric amyloid-beta including globular amyloid-beta oligomers (AβOs) and linear amyloid-beta protofibrils are toxic to neurons. Sabirnetug (ACU193) is a humanized monoclonal antibody, raised against globular species of soluble AβO, that has over 650-fold greater binding affinity for AβOs over monomers and appears to have relatively little binding to amyloid plaque. OBJECTIVES:To assess safety, pharmacokinetics, and exploratory measures including target engagement, biomarker effects, and clinical efficacy of sabirnetug in participants with early symptomatic Alzheimer's disease (AD; defined as mild cognitive impairment and mild dementia due to AD). DESIGN:Randomized, double-blind, placebo-controlled, ascending dose first-in-human phase 1 study. SETTING:Fifteen study centers in the United States. PARTICIPANTS:Sixty-five participants with early symptomatic AD. INTERVENTION:Participants received one infusion of sabirnetug 2 mg/kg, 10 mg/kg, 25 mg/kg, 60 mg/kg, or placebo (Part A) or three infusions of sabirnetug 10 mg/kg, 25 mg/kg, 60 mg/kg, or placebo (Part B). MEASUREMENTS:Safety, tolerability, serum pharmacokinetics, and central target engagement of single and multiple doses of sabirnetug, cerebrospinal fluid (CSF) concentrations of sabirnetug, and amyloid plaque load, as determined by positron emission tomography. RESULTS:Sabirnetug was generally well tolerated. A larger percentage of participants receiving sabirnetug (56.3%) versus placebo (42.9%) had at least one treatment emergent adverse event, with approximately 29% in each group considered related to study drug. Most events were mild-to-moderate in severity. Of 48 participants given sabirnetug, five developed amyloid related imaging abnormalities - edema/effusion, including one instance that was mildly symptomatic in a participant who had received one dose sabirnetug 60 mg/kg. Notably, none of the six apolipoprotein E Ɛ4 homozygotes who received sabirnetug developed amyloid related imaging abnormalities - edema/effusion or - hemorrhage/hemosiderin deposition. Infusion reactions, such as rash, pain, or erythema, were not frequent (6.3% for sabirnetug versus 0.0% for placebo). Sabirnetug exposure was dose proportional in both serum and CSF. Target engagement, defined as drug bound to AβOs in CSF, was shown to be dose and exposure dependent. Over three months, approximately 25% and 20% reduction in amyloid plaques, respectively, were observed in participants receiving three infusions of sabirnetug 60 mg/kg every four weeks and 25 mg/kg every two weeks. CONCLUSIONS:The Phase 1 INTERCEPT-AD study provided safety, tolerability, dosing, and target engagement data that supported the design of the ongoing ALTITUDE-AD study (NCT06335173).

Assessing the pharmacokinetics of monoclonal antibodies (mAbs) in relevant animal models is essential for designing improved formulations and developing mAb delivery platforms. We have established the pig, a large natural host animal for influenza with many similarities to humans, as a robust model for testing the therapeutic efficacy of anti-influenza mAbs and evaluating mAb delivery platforms. Here, we compared the pharmacokinetic characteristics of two anti-influenza hemagglutinin mAbs, human 2-12C and porcine pb27, in Göttingen minipigs and Landrace × Large White outbred pigs. Minipigs offer the advantage of a more stable weight, whereas outbred pigs are more readily available but exhibit rapid growth. Outbred pigs and minipigs showed similar pharmacokinetics and a similar porcine pb27 half-life (half-life of 15.7 days for outbred pigs and 16.6 days for minipigs). In contrast, the half-life of human 2-12C was more rapid in two of the minipigs but not in the outbred pigs, correlating with the development of antidrug antibodies in the two minipigs. Our results demonstrate that both outbred pigs and minipigs are appropriate models for pharmacokinetic studies and the evaluation of mAb delivery platforms, potentially bridging the gap between small animals and human trials.

Axicabtagene ciloleucel (axi-cel, Yescarta) is an autologous, anti-CD19, chimeric antigen receptor (CAR) T-cell therapy approved for patients with relapsed and refractory non-Hodgkin’s lymphoma. Substantial inter-individual variability in cellular kinetics has been observed with CAR-T therapies and factors impacting CAR-T cellular kinetics remain poorly understood. This work reports a population cellular kinetic model of axi-cel in relapsed and patients with refractory non-Hodgkin’s lymphoma and investigated the impact of covariates on early and late kinetic phases of CAR-T exposure. A population cellular kinetic model (NONMEM® version 7.4) for axi-cel was developed using data from 410 patients (2050 transgene observations) after a single intravenous infusion of 2 × 106 anti-CD19 CAR+ T cells/kg in patients with non-Hodgkin’s lymphoma (ZUMA-1, ZUMA-5, and ZUMA-7 clinical studies). A large panel of covariates was assessed to decipher the variability of CAR-T cell kinetics including patient characteristics, product characteristics, and disease types. Axi-cel cellular kinetics were well described by a piecewise model of cellular growth kinetics characterized by an exponential growth phase followed by a triphasic decline phase including a long-term persistence phase. The final cellular kinetic model retained in vitro doubling time during CAR-T cell manufacturing and total number of T cells infused as covariates impacting the duration of the growth phase, which, however, did not substantially influence maximum concentration, area under the concentration–time curve over the first 28 days, or long-term persistence. A statistically significant relationship was observed between maximum concentration and the probability to receive tocilizumab and/or corticosteroids. No covariates considered in this study were found to significantly and substantially predict the exposure profile of axi-cel. Tocilizumab and steroid use were related to maximum concentration, but they were used reactively to treat toxicities that are associated with a higher maximum concentration. Further CAR-T kinetic analyses should consider additional factors to explain the observed variability in cellular kinetics or help establish a dose–exposure relationship. NCT02348216 (ZUMA-1), NCT03105336 (ZUMA-5), and NCT03391466 (ZUMA-7).

Licensed rabies virus vaccines based on whole inactivated virus are effective in humans. However, there is a lack of detailed investigations of the elicited immune response, and whether responses can be improved using novel vaccine platforms. Here we show that two doses of a lipid nanoparticle-formulated unmodified mRNA vaccine encoding the rabies virus glycoprotein (RABV-G) induces higher levels of RABV-G specific plasmablasts and T cells in blood, and plasma cells in the bone marrow compared to two doses of Rabipur in non-human primates. The mRNA vaccine also generates higher RABV-G binding and neutralizing antibody titers than Rabipur, while the degree of somatic hypermutation and clonal diversity of the response are similar for the two vaccines. The higher overall antibody titers induced by the mRNA vaccine translates into improved cross-neutralization of related lyssavirus strains, suggesting that this platform has potential for the development of a broadly protective vaccine against these viruses.

Diversity in specificity of polyclonal antibody (pAb) responses is extensively investigated in vaccine efficacy or immunological evaluations, but the heterogeneity in antibody avidity is rarely probed as convenient tools are lacking. Here we have developed a polyclonal antibodies avidity resolution tool (PAART) for use with label-free techniques, such as surface plasmon resonance and biolayer interferometry, that can monitor pAb-antigen interactions in real time to measure dissociation rate constant (kd) for defining avidity. PAART utilizes a sum of exponentials model to fit the dissociation time-courses of pAb-antigens interactions and resolve multiple kd contributing to the overall dissociation. Each kd value of pAb dissociation resolved by PAART corresponds to a group of antibodies with similar avidity. PAART is designed to identify the minimum number of exponentials required to explain the dissociation course and guards against overfitting of data by parsimony selection of best model using Akaike information criterion. Validation of PAART was performed using binary mixtures of monoclonal antibodies of same specificity but differing in kd of the interaction with their epitope. We applied PAART to examine the heterogeneity in avidities of pAb from malaria and typhoid vaccinees, and individuals living with HIV-1 that naturally control the viral load. In many cases, two to three kd were dissected indicating the heterogeneity of pAb avidities. We showcase examples of affinity maturation of vaccine induced pAb responses at component level and enhanced resolution of heterogeneity in avidity when antigen-binding fragments (Fab) are used instead of polyclonal IgG antibodies. The utility of PAART can be manifold in examining circulating pAb characteristics and could inform vaccine strategies aimed to guide the host humoral immune response.

Background Odronextamab is a hinge-stabilized, human CD20×CD3 IgG4-based bispecific antibody that binds CD20-expressing cells and CD3 on T cells, targeting CD20+ cells via T-cell-mediated cytotoxicity, independent of T-cell receptor-mediated recognition. Clinical studies of odronextamab in adult patients with relapsed/refractory CD20+ B-cell malignancies are ongoing. Encouraging efficacy has been reported in patients participating in the Phase I study ELM-1 (NCT02290951; Bannerji R, et al. Lancet Haematol. 2022;9:e327-39). In the US, approximately 800 new cases of pediatric non-Hodgkin lymphoma (NHL) are diagnosed each year. Improved treatment options are needed for pediatric patients, particularly those with relapsed/refractory disease following chemoimmunotherapy, who have poor outcomes. A study of odronextamab in relapsed/refractory pediatric patients with B-cell NHL (B-NHL) is being planned. The objective of this analysis was to determine intravenous (IV) regimens for testing in pediatric patients that can achieve similar exposures to those achieved in adults. Methods The analysis was conducted in 6 steps: 1) A population pharmacokinetic (PK) model was developed with data from adults receiving IV odronextamab and assessed effects of body weight (WT) on drug clearance (CL) and volume of distribution (V); 2) A virtual pediatric population was created with demographics derived from the 2017-2018 National Health and Nutrition Examination Survey database; 3) Pediatric model parameters were extrapolated from those of adults by applying effects of WT and age on CL and V as appropriate; 4) Odronextamab PK profiles were simulated in virtual pediatric patients aged 6 months to 18 years in 4 weight bands (≥40 kg, 20-39 kg, 10-19 kg, 6-9 kg) using the extrapolated pediatric population model parameters; 5) Odronextamab drug amounts were adjusted in each weight band for each weekly (QW) dosing period, and every 2 weeks (Q2W) dosing period with the aim of matching adult patient exposures; and 6) Pediatric regimens were selected for testing based on 2 criteria. Firstly, for safety, median maximal concentration (Cmax) during Cycle 1 step-up dosing within 80-125% of the adult value. Secondly, for efficacy, median minimal concentration (Cmin) at steady state during Cycle 2 onwards not less than the adult value. Results Step-up dosing is used in Weeks 1-3 of treatment of adult patients with odronextamab to mitigate the risk of cytokine release syndrome, a common adverse event with T-cell engaging therapies. Applying a simulation approach, dosing regimens tiered by WT bands were proposed for pediatric patients (Table). Children with WT ≥40 kg could receive the adult's regimen; those weighing <40 kg could receive reduced doses according to their WT bands. The regimens and model assumptions will be tested in a pediatric study with patients weighing ≥20 kg enrolled first. The proposed regimens were predicted to achieve odronextamab exposures similar to those with the adult regimen. No significant difference in Cmax was predicted with the step-up doses; the ratios of median Cmax for the pediatric WT groups relative to adults fell within 0.831-1.24. The median Cmin values at steady state were not below those observed in adults. Additional exploratory simulations were performed to assess uniform weight-based dosing (mg/kg) for the overall pediatric population. This approach was inferior to the regimens tiered by WT bands due to above-target concentrations among individuals weighing ≥20 kg and below-target concentrations in those weighing 6-9 kg. Conclusion The modeling and simulation approach allowed the identification of dosing regimens that might be suitable for investigational use in children with B-NHL. Based on predicted PK profiles, pediatric dosing regimens tiered by WT bands may preserve the efficacy of odronextamab observed in adults, without increasing the risk of acute toxicity. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

Aim: To assess the potential of interleukin-6 (IL-6) signaling blockade in the lung to treat SARS-CoV-2 infection via model-based simulation by exploring soluble IL-6 receptor (sIL-6R) sequestration by tocilizumab (TCZ) and IL-6 sequestration by siltuximab (SIL). Methods: Literature values of IL-6, the IL-6 antagonist SIL, sIL-6R, the IL-6R antagonist TCZ, and their respective binding constants were used to develop a model to predict the impact of treatment on IL-6 signaling. Models were used to generate simulated bronchoalveolar lavage (BAL) concentrations for normal subjects, subjects at risk of developing acute respiratory distress syndrome (ARDS), and subjects with ARDS were simulated under four conditions: without treatment, treatment with TCZ, treatment with SIL, and treatment with TCZ + SIL. Results: With TCZ intervention, IL-6 levels are unaffected and sIL-6R is reduced somewhat below the Normal case. IL-6:sIL-6R complex only slightly decreased relative to the no-intervention case. With SIL intervention, sIL-6R levels are unaffected and IL-6 is greatly reduced below the Normal case. IL-6:sIL-6R complex is greatly decreased relative to the no-intervention case. With TCZ + SIL intervention, IL-6 and sIL-6R levels are reduced below the Normal case and achieve suppression equivalent to monotherapy results for their respective targets. IL-6:sIL-6R complex reduction is predicted to be greater than monotherapy. This reflects sequestration of both components of the complex and the nonlinear binding equilibrium. Conclusion: Co-administration of both IL-6 and IL-6R sequestering products such as SIL and TCZ may be necessary to effectively treat COVID-19 patients who have or are at risk of developing ARDS.

Purpose: Batiraxcept is a novel Fc fusion protein that inhibits the growth arrest-specific (GAS)/AXL signaling pathway by binding the sole activating GAS6 ligand, which reduces AXL signaled invasion and migration of highly metastatic cells in vitro and inhibits metastatic disease in nonclinical models of aggressive human cancers. We used dose-exposure (D-E) and exposure-response (E-R) modeling and simulation to identify a recommended Phase 2 dose (RP2D). The D-E modeling is expected to apply to any cancer and batiraxcept combination, while the E-R modeling may be disease specific. Patients and Methods: A population pharmacokinetic (PK)/pharmacodynamic (PD) model was applied to exposure data from healthy volunteers (n=24, Phase (P) 1 study, single ascending dose cohort [NCT03401528]) and platinum-resistant ovarian cancer (PROC) patients (n=42, P1b study [NCT03639246]) treated with either batiraxcept+paclitaxel or batiraxcept+pegylated liposomal doxorubicin to capture the D-E relationship of batiraxcept. A Cox Proportional Hazards Model was applied to progression-free survival (PFS) observed from the PROC patients, to capture the E-R relationship of batiraxcept. The models were used to study PFS in prospective studies of 5, 10, 15, and 20 mg/kg administered iv once every 2 weeks at the 10th, 50th (median), and 90th percentile of expected exposure metrics. In a separate P1b study in 2L+clear cell renal cell carcinoma (ccRCC), 23 patients were dosed with 15mg/kg or 20mg/kg batiraxcept in combination with standard of care cabozantinib [NCT04300140]. PK, PD, and clinical response data from both doses tested in ccRCC were compared to the modeling predictions generated from the PROC study. Results: The PROC E-R analysis predicted that a dose of 15mg/kg would provide exposure levels associated with longer PFS, and a 20mg/kg dose did not add additional benefit. Fifteen mg/kg was identified as the RP2D from the PROC P1b study and is the batiraxcept dose being evaluated in combination with standard of care weekly paclitaxel in the ongoing PROC P3 study [NCT04729608]. PK and PD data from the ccRCC P1b study compare favorably with simulations from the PROC D-E model. Consistent with simulations from the PROC E-R analysis, the 20mg/kg dose did not provide additional benefit above that seen with the 15mg/kg dose in the ccRCC P1b study, indicating these doses are at the plateau of drug effect. Conclusion: Consistent with emerging U.S. Food and Drug Administration guidance around a more thoughtful approach to dose selection in cancer (“Project Optimus”), E-R analysis supported the identification of batiraxcept 15mg/kg as the RP2D in the PROC P1b study. Furthermore, PK/PD and preliminary activity of batiraxcept 15mg/kg+cabozantinib are consistent with these data and support the batiraxcept RP2D of 15mg/kg across different tumor types and combinations with batiraxcept. Citation Format: Gail Mcintyre, Reshma Rangwala, Michael G. Dodds, Olivier Barriere. Exposure-response analysis of batiraxcept and application to recommended phase 2 dose in platinum-resistant ovarian and clear cell renal cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr CT506.

Lisocabtagene maraleucel (liso-cel) is a CD19-directed, defined composition, 4-1BB chimeric antigen receptor (CAR) T-cell product administered at equal target doses of CD8+ and CD4+ CAR+ T cells. Large between-subject variability has been noted with CAR T-cell therapies; patient characteristics might contribute to CAR T-cell expansion variability. We developed a population cellular kinetic model to characterize the kinetics of the liso-cel transgene, via quantitative polymerase chain reaction assessment after intravenous infusion of liso-cel, and to understand covariates that might influence liso-cel kinetics in individual patients. We employed nonlinear mixed-effects modeling to develop a population cellular kinetic model for liso-cel. The population cellular kinetic analysis was performed using 2524 post-infusion transgene observations from 261 patients with relapsed/refractory large B-cell lymphoma who were treated with a single dose of liso-cel in TRANSCEND NHL 001. Covariates for the analysis included baseline intrinsic factors such as age, baseline disease characteristics, and liso-cel and coadministration factors. Liso-cel cellular kinetics were well described by a piecewise model of cellular growth kinetics that featured lag, exponential growth, and biexponential decay phases. Population means (95% confidence interval) of lag phase duration, doubling time, time to maximum levels, initial decline half-life, and terminal half-life were 3.27 (2.71–3.97), 0.755 (0.667–0.821), 9.29 (8.81–9.70), 5.00 (4.15–5.90), and 352 (241–647) days, respectively. The magnitude of effect on liso-cel expansion metrics demonstrated that the covariate associations were smaller than the residual between-subject variability in the population. The covariates tested were not considered to have a meaningful impact on liso-cel kinetics. NCT02631044.

We applied a mathematical framework originally used to model the effects of multiple inhibitors on enzyme activity to guide the development a therapeutic antibody cocktail, LMN-201, to prevent and treat C. difficile infection (CDI). CDI causes hundreds of thousands of cases of severe, often recurrent diarrhea and colitis in the United States annually and is associated with significant morbidity and mortality worldwide. Current therapies for preventing recurrent CDI are only partially successful, and there are no options available to prevent initial bouts of CDI in at-risk populations. Almost all antibody therapies have been developed and administered as monotherapies. Antibody cocktails are relatively rare even though they have the potential to greatly increase efficacy. One reason for this is our limited understanding of how antibody interactions can enhance potency, which makes it difficult to identify and develop antibodies that can be assembled into optimally effective cocktails. In contrast to the view that antibody synergies depend on unusual instances of cooperativity or allostery, we show that synergistic efficacy requires nothing more than that the antibodies bind independently to distinct epitopes on a common target. Therefore, synergy may be achieved much more readily than is generally appreciated. Due to synergy the LMN-201 antibody cocktail, which targets the C. difficile exotoxin B (TcdB), is 300- to 3000-fold more potent at neutralizing the most clinically prevalent TcdB toxin types than bezlotoxumab, the only monoclonal antibody currently approved for treatment or prevention of CDI. The efficacy of LMN-201 is further enhanced by inclusion of a phage-derived endolysin that destroys the C. difficile bacterium, and which therefore has a complementary mechanism of action to the antibody cocktail. These observations may serve as a paradigm for the development of high potency biologic cocktails against targets that have proven challenging for single-agent therapies.

During a pandemic caused by a novel pathogen (NP), drug repurposing offers the potential of a rapid treatment response via a repurposed drug (RD) while more targeted treatments are developed. Five steps of model-informed drug repurposing (MIDR) are discussed: (i) utilize RD product label and in vitro NP data to determine initial proof of potential, (ii) optimize potential posology using clinical pharmacokinetics (PK) considering both efficacy and safety, (iii) link events in the viral life cycle to RD PK, (iv) link RD PK to clinical and virologic outcomes, and optimize clinical trial design, and (v) assess RD treatment effects from trials using model-based meta-analysis. Activities which fall under these five steps are categorized into three stages: what can be accomplished prior to an NP emergence (preparatory stage), during the NP pandemic (responsive stage) and once the crisis has subsided (retrospective stage). MIDR allows for extraction of a greater amount of information from emerging data and integration of disparate data into actionable insight.

Aim We hypothesized that viral kinetic modelling could be helpful to prioritize rational drug combinations for COVID-19. The aim of this research was to use a viral cell cycle model of SARS-CoV-2 to explore the potential impact drugs, or combinations of drugs, that act at different stages in the viral life cycle might have on various metrics of infection outcome relevant in the early stages of COVID-19 disease. Methods Using a target-cell limited model structure that has been used to characterize viral load dynamics from COVID-19 patients, we performed simulations to inform on the combinations of therapeutics targeting specific rate constants. The endpoints and metrics included viral load area under the curve (AUC), duration of viral shedding and epithelial cells infected. Based on the known kinetics of the SARS-CoV-2 life cycle, we rank ordered potential targeted approaches involving repurposed, low-potency agents. Results Our simulations suggest that targeting multiple points central to viral replication within infected host cells or release from those cells is a viable strategy for reducing both viral load and host cell infection. In addition, we observed that the time-window opportunity for a therapeutic intervention to effect duration of viral shedding exceeds the effect on sparing epithelial cells from infection or impact on viral load AUC. Furthermore, the impact on reduction on duration of shedding may extend further in patients who exhibit a prolonged shedder phenotype. Conclusions Our work highlights the use of model-informed drug repurposing approaches to better rationalize effective treatments for COVID-19.

As new treatment modalities are being explored for SARS-CoV-2, efforts to repurpose existing marketed drugs remain an attractive option, as these agents are readily available and have a known safety profile. It is important to recognize that these drugs have not been specifically developed or optimized for the treatment of SARS-CoV-2 infected patients. Success in repurposing efforts will depend on being mindful of first principles around clinical pharmacology and dosing strategies, noting that the dose regimens of existing drugs were developed for different indications. 'Getting the dose right' for antivirals being targeted against acute respiratory viruses requires knowledge of potency, pharmacokinetics and viral kinetics to guide rational use. This commentary aims to review those principles of clinical pharmacology that are critical to the successful design and implementation of an optimal dosing regimen for drugs repurposed against SARS-CoV-2, using lopinavir/ritonavir (LPV/r) as an example. We selected LPV/r as it is being considered for the treatment of SARS-CoV-2; however, the standard dosing regimen may not be optimal for this new indication. Antiviral treatment must be initiated in an infected patient as soon as possible. This limited window to initiate antiviral therapy against respiratory viruses is well known. For example, if treatment of influenza is not initiated within 48–72 h of symptom onset, drug efficacy is substantially reduced or eliminated entirely.1 This treatment window can be defined by the viral kinetic profile of the pathogen, and corresponds to the time of peak viral load for an individual patient,2 and generally occurs prior to the onset of the so-called proinflammatory cytokine storm3 which is associated with severe disease and poor clinical outcome. If treatment of a patient is initiated after peak viral load or after the onset of the cytokine storm, any antiviral is unlikely to be effective. The viral kinetic profile varies by respiratory virus, with respiratory syncytial virus (RSV) having a longer time to peak and more prolonged duration of viral shedding compared to influenza.4 The viral kinetic profile of SARS-CoV-2 is not yet well understood, but initial data suggest that the time to peak appears to occur approximately 8–10 days after infection, or 5 days post onset of symptoms (See Supporting Information). This would indicate a 3- to 5-day treatment opportunity from the time of symptom onset. Similarly, the length of treatment should cover the duration of viral shedding to maximize drug effect, reduce the risk of viral rebound and to minimize the spread to uninfected individuals. SARS-CoV-2 appears to maintain a longer duration of viral shedding compared to influenza, and therefore a longer treatment duration may be warranted (e.g., 14 to 28 days). Even longer treatment durations may be required in immunocompromised patients. Given the limited treatment window to successfully intervene, it is paramount that antiviral dose regimens be constructed to achieve high therapeutic concentrations at the effect site as rapidly as possible. This has prompted numerous antivirals for respiratory pathogens to consider loading doses. A loading dose may achieve therapeutic concentrations much more quickly, increasing the probability of successful treatment. For existing antivirals like LPV/r, pharmacokinetic models can be leveraged via simulation to design an optimal loading dose regimen. Repurposing a drug already in clinical use provides reassurance that the safety profile is already established. For drugs like LPV/r, there is extensive clinical experience from decades of use in the treatment of patients with HIV, including use in special populations such as children, pregnant women and patients with hepatic impairment. However, it is important to remember that its clinical use was studied in a different patient population under a different treatment paradigm. LPV/r is generally considered to be 'well tolerated' in comparison to other long-term HIV treatment options, but gastrointestinal disorders (e.g., nausea, vomiting and diarrhoea) are common side effects, and there is little experience with the use of LPV/r and other HIV treatments in the elderly population with co-morbidities who may be most at risk from COVID-19. Any repurposing activities therefore need to include a benefit:risk evaluation focused on the new proposed clinical application. Little publically available potency information has been reported for any repurposed drug against SARS-CoV-2, to our knowledge. Therefore, antiviral activity must generally be inferred from studies of related human coronaviruses, which is a current limitation in the repurposing efforts. In addition, usually an IC50 or EC50 is reported in the literature, whereas a more aggressive target such as EC90 may be more clinically appropriate. For lopinavir, in the absence of human serum, the EC50 against HIV-1 ranges from 0.006 to 0.017 μg/ml, whereas in the presence of 50% human serum, the EC50 is approximately 10-fold higher (0.04–0.18 μg/ml), representing the significant impact of plasma protein binding on potency. While published reports for SARS-CoV-2 are not available, in vitro potency for related coronaviruses are assumed to be similar. The published IC50s or EC50s in the absence of human serum (i.e., unbound drug) for LPV for SARS or and MERS viruses have been reported to range from 4 to 15 μg/ml (see Supporting Information). The in vitro potency of LPV against coronaviruses would suggest that LPV may be more than 100 times less active against SARS-CoV-2 compared to HIV-1. Therefore, higher drug exposures and possibly a new dose regimen may be required to successfully treat coronaviruses with LPV/r. A basic principle of clinical pharmacology is that only free (unbound) drug is available to bind to the target or to distribute to the site of action, as drug bound to plasma proteins is sequestered and otherwise inaccessible. Like most HIV protease inhibitors, LPV demonstrates very high protein binding, being approximately 98.5% bound (1.5% free). Therefore, in considering whether a dose is appropriate for SARS-Cov-2, it is critical to consider free drug concentrations. For LPV, the steady-state total minimum plasma concentration (Cmin) at the standard HIV-1 treatment dose of 400/100 mg LPV/r is approximately 5 μg/ml, or 0.075 μg/ml of unbound (free) drug. This is a sufficient concentration to cover HIV-1 but is substantially below the 4–15 μg/ml EC50 reported for coronaviruses. It is important to consider drug concentrations at the site of infection, and currently, the lack of robust lung penetration data is an important gap that exists for many agents being considered for repurposing. In the absence of data, physiologically based pharmacokinetic (PBPK) modelling can be used to predict lung exposures, as it has been done for hydroxychloroquine.5 This work by Yao and colleagues for hydroxychloroquine demonstrates a potential framework for drug repurposing, via the generation IC50s against related coronaviruses, and use of PBPK to design a loading and maintenance dose regimen specifically targeted towards SARS-CoV-2 in the lung. In the case of LPV, lung penetration is complex and not well understood; however, typically it is the plasma free fraction that is available to penetrate into tissues. Therefore, given its potency, lung penetration of LPV would have to be high to provide concentrations in the therapeutic range. There are two case reports of LPV/r lung concentrations in HIV patients, and it appears that the ELF:plasma exposure of lopinavir may be up to 2-fold (~1.6 μg/ml) at standard doses. Given the uncertainty around LPV/r lung penetration, additional data are required to understand for certain if very high doses might be sufficient to reach therapeutic exposures in the lung. An important further goal of dose optimization is to ensure that most patients in the population, treated with a common dose regimen, will achieve therapeutic drug concentrations. Lopinavir demonstrates high pharmacokinetic variability, which means that doses will need to be relatively high to ensure a common dose regimen will achieve therapeutic concentrations in the majority of patients. Population PK modelling can be utilized to simulate exposures across a range of patients, to ensure that the majority of the target population will achieve target concentrations. The above ideas are synthesized using LPV/r as a case study for SARS-CoV-2. We utilized published population pharmacokinetic models that characterize the plasma exposures of LPV/r, incorporating the clinical pharmacology concepts noted above (see Supporting Information). We have thus used these models to simulate LPV plasma (total and unbound) and lung (ELF) exposures in an adult population and compared them to in vitro potency against SARS-CoV and MERS-CoV, to evaluate the potential of LPV as an effective antiviral for SARS-CoV-2 (Figure 1). These results suggest that the unbound plasma concentrations of LPV remain very low relative to the available EC50s. While limited data are available to inform the lung exposures, it may be possible to reach therapeutic exposures in some patients with higher doses. It is also important to note that the time to achieve high concentrations of LPV takes several days, which would suggest that a loading dose would be necessary to maximize any potential treatment effect. Based on the results of our assessment, standard LPV/r doses are at high risk of treatment failure. The time to reach full concentrations takes 36–48 h in the absence of a loading dose, hence losing valuable time in the treatment window. Doses of LPV/r are also limited by safety and tolerability, with significant GI-related adverse events, potential for QTc prolongation and substantial drug–drug interactions. To date, a number of animal and clinical trials have failed to show a benefit of LPV/r in coronaviruses, including most recently a large trial in severe hospitalized patients,6 and a small trial in mild-to-moderate hospitalized patients.7 The large study by Cao et al. evaluated a very challenging patient population, and standard dose LPV/r was initiated late in the disease course, likely too late for there to be benefit. In all likelihood, no antiviral was likely to be effective in this difficult-to-treat population, particularly using the LPV/r dosing regimen that was optimized for HIV infection. The results of this study confirm that late treatment with standard doses of LPV/r are unlikely to have benefit for COVID-19. Li et al reported no LPV/r treatment effect in a small cohort of approximately 21 mild-to-moderate hospitalized patients, with a mean (range) start time of 4.3 days post symptom onset using standard dose LPV/r.7 Based on these studies, there is growing evidence that current approaches using standard LPV/r treatment for hospitalized patients are not effective, which is consistent with the clinical pharmacology data. This commentary highlights the critical need for the collection and sharing of the key data, which will be required to prioritize and evaluate potential repurposed drug candidates, and to design COVID-19 targeted dosing paradigms which specifically address the unique challenges associated with treating this disease. While there are uncertainties regarding the likelihood of achieving effective concentrations in the lung with LPV/r, it is clear that the delay in time before plasma concentrations achieve the effective EC50 levels decreases the opportunity for the standard dosing regimen to effectively block virus replication until many days after instituting therapy. If rapid viral inhibition is the purpose of implementing this therapy, then early aggressive intervention with the use of loading doses will be necessary, which could impact tolerability. The use of clinical pharmacology and model-based dosing design will be important to maximize the likelihood of successful deployment of existing antiviral drugs for COVID-19 and future pandemics. The authors acknowledge the helpful input, discussions and review of Steve Kern, Dan Hartman, Mark Milad and Terrence Blashke. Funding was provided in part by the Bill and Melinda Gates Foundation. PFS, DB and CR wrote and edited the manuscript, and MD and KY conducted the modelling and data analysis. The authors report no conflict of interest. Data S1. Supporting information Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

AIMS:We propose the use of in silico mathematical models to provide insights that optimize therapeutic interventions designed to effectively treat respiratory infection during a pandemic. A modelling and simulation framework is provided using SARS-CoV-2 as an example, considering applications for both treatment and prophylaxis. METHODS:A target cell-limited model was used to quantify the viral infection dynamics of SARS-CoV-2 in a pooled population of 105 infected patients. Parameter estimates from the resulting model were used to simulate and compare the impact of various interventions against meaningful viral load endpoints. RESULTS:Robust parameter estimates were obtained for the basic reproduction number, viral release rate and infected-cell mortality from the infection model. These estimates were informed by the largest dataset currently available for SARS-CoV-2 viral time course. The utility of this model was demonstrated using simulations, which hypothetically introduced inhibitory or stimulatory drug mechanisms at various target sites within the viral life-cycle. We show that early intervention is crucial to achieving therapeutic benefit when monotherapy is administered. In contrast, combination regimens of two or three drugs may provide improved outcomes if treatment is initiated late. The latter is relevant to SARS-CoV-2, where the period between infection and symptom onset is relatively long. CONCLUSIONS:The use of in silico models can provide viral load predictions that can rationalize therapeutic strategies against an emerging viral pathogen.

Drugs which independently inhibit a shared target or pathway can have synthetic activities that result in multiplicative instead of merely additive potencies. This characteristic of drug combinations can be quantified by expressing the potency of the combination as if it were a single agent. We show that by optimizing this quantity we can prospectively design drug cocktails with apparent potencies that far exceed any of its individual components. We illustrate the power of this approach, which is based on statistical design of experiments to select optimal drug combinations, and response surface methodology to determine optimal drug ratios, by building a drug cocktail comprised of three antibodies for treating C. difficile infection that is almost 1000-fold more potent than the current, clinically approved antibody monotherapy. High synthetic activities do not require unusual drug interactions, and therefore may be achievable much more readily than generally appreciated. One-Sentence Summary A development pathway is described for designing antibody cocktails with potencies that far exceed what is achievable with single antibodies