key: cord-0971846-x6edm0pv
authors: Cross, Gail; Ho, Jessica SY.; Zacharias, William; Jeyasekharan, Anand D.; Marazzi, Ivan
title: Emergency drug use in a pandemic: harsh lessons from COVID-19
date: 2021-09-21
journal: Cell
DOI: 10.1016/j.cell.2021.09.025
sha: 7062439240b2ff091fccec160160ab176d8adbca
doc_id: 971846
cord_uid: x6edm0pv

The scientific and clinical communities have both experienced several harsh lessons on clinical care management and drug development during the COVID-19 pandemic. Here, we discuss several key lessons learned and describe a framework within which our two communities can work together and invest in to improve future pandemic responses.

Over a year has passed since severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was first identified. Since then, Coronavirus Disease 2019 has claimed more than 4 million lives worldwide, and affected billions more through reduced economic opportunity, healthcare access, educational progress and community interactions. The pandemic spurred an unprecedented effort amongst scientific and medical communities to develop interventions to improve patient outcomes and alter the trajectory of the outbreak. However, vaccines and dexamethasone aside, other therapeutic interventions have had minimal impact on overall mortality from COVID-19. Further, in the chaotic early days of the pandemic, the principles of evidencebased medicine were often brushed aside -circumstantial evidence formed the basis of clinical management strategies and anecdotes and hunches were the rationale for scientifically weak hypotheses in research. There was widespread duplication of research effort due to poor coordination, leading to wasted funding and time.

Here we contextualize the shared experience of scientific and clinical communities during the COVID-19 pandemic, examine some mis-steps, and the lessons we have (re)learned. We frame these as a future-ready battleplan for the ongoing threat of emerging viral diseases; a plan that makes the best use of available resources and minimizes risk to patients.

Outbreaks in China and Italy defined the COVID-19 clinical syndrome early; patients presented with respiratory symptoms (cough, shortness of breath), systemic symptoms (fever, malaise and myalgias), with some setting off on a trajectory to acute respiratory distress syndrome (ARDS), severe pulmonary fibrotic changes, refractory hypoxemia, multiorgan failure, refractory shock and ultimately death. These features suggested a pathophysiology analogous to that of other severe respiratory infections, and especially similar to severe acute respiratory syndrome coronavirus (SARS-CoV), and Middle East respiratory syndrome coronavirus (MERS-CoV).

Despite this, much of the research activity in the early months of the pandemic involved attempts to differentiate COVID-19 from other respiratory viral illnesses through nuances in clinical and laboratory parameters. Whilst some of this knowledge ultimately showed some prognostic value and identified potential therapeutic targets (as discussed later), this effort overshadowed the fact that severe COVID-19 was an inflammatory ARDS. There was extensive argument about whether COVID-associated respiratory failure should be managed as ARDS (Gattinoni et al. 2020) . Eventually decades-old evidence that low-stretch tidal volume ventilation is the primary method of ARDS treatment (Network 2000; Thompson et al. 2017 ) prevailed, with improved ICU mortality, but only after many died from ARDS: a well-recognized disease process. Furthermore, it was clear early in the pandemic that patients died primarily due to immune dysfunction. Imbalanced immune activation was responsible for morbidity and mortality from SARS-CoV and MERS-CoV (Tay et al. 2020 ). Parallels could have been drawn about SARS-CoV-2 after early retrospective reports indicating the progression from mild to moderate to severe COVID19 is associated with features of hyperinflammatory response and sepsis (Zhou et al. 2020) . These data were later reinforced by multiple studies J o u r n a l P r e -p r o o f showing patients who died of COVID19 ARDS evinced a pattern of distinct immune activation related to viral response and inflammation in the alveolus (Grant et al. 2021) . A clear and coordinated clinical focus on immune dysfunction in COVID-19 could have helped design better trials, with greater emphases on immunomodulatory strategies. In line with this, the World Health Organization International COVID-19 trial is set to restart with a focus on three immunomodulatory drugs.

A key lesson from this experience is that pattern recognition based on similarity to known clinical syndromes is critical as a foundation upon which additional therapeutic modalities can be evaluated-even in times of uncertainty.

Principles of evidence based clinical care must not be abandoned in the midst of a pandemic Both antiviral and immunomodulatory drugs were considered and utilized early in the pandemic. However, much of this was administered off-label, outside of observational studies and certainly outside of clinical trials. The most prominent example of this was hydroxychloroquine, used widely and repeatedly, based on highly politicized and poorly scrutinized research findings, eventually doing more harm than good (Axfors et al. 2021 ). More recently we have seen this play out with the antiparasitic drug, Ivermectin-which was considered a bona fide therapeutic option in many settings based on in vitro data, despite high quality meta-analyses showing little or no effect (Popp et al. 2021) . The use of investigational drugs as immediate therapeutic or prophylactic options has, in fact, limited their ability to be tested through robust clinical trials (Reardon 2021) .

It is understandable that in the face of substantial mortality and overwhelmed health systems with little access to clinical trials, some clinicians resorted to therapies without strong clinical evidence. Yet herein lies a key lesson from the COVID-19 pandemic -the need to default to core principles of evidence-based medicine. Especially amidst uncertainty, evidence-based practice allow the definition and testing of therapeutic interventions most likely to work, whilst we prove or disprove alternative hypotheses. In the face of an emerging deadly viral pathogen, sticking to these principles is more critical than ever. Pandemics give rise to a huge number of people at risk, providing a unique opportunity to conduct robust clinical drug trials. Conversely, the ad hoc use of experimental therapies results not only in an opportunity cost, but puts vast number at risk of receiving unproven therapies that are either ineffective, harmful or both. However, key to maintaining public "trust" in the evidence based approach is to design therapeutic trials that have the highest likelihood of success, through the best use of solid scientific knowledge available at the time.

This pandemic witnessed an unprecedented number of drugs clinically tested in less than 2 years, but few have borne out to be effective. Here we discuss possible reasons for this inefficiencies in clinical drug development, and highlight areas for possible improvement.

Antiviral drugs target key processes during the viral life cycle, such as entry, replication, translation, or egress, are typically selective for viral proteins/processes with few off-target effects, theoretically making them ideal viral pandemic therapeutics. Despite a large number of antiviral drugs being tested clinically, only one thus far, Remdesivir, has shown significant clinical benefit (Beigel et al. 2020) , and its utility in reducing death remains uncertain (Rochwerg et al. 2020). One of the reasons why many antiviral candidates failed is that in the panicked push for effective "emergency response" drugs, there was reduced scrutiny of pre-clinical experimental designs. Unlike Remdesivir, which had a known mechanism-of-action (MOA) and activity against related coronaviruses, many other candidate drugs were identified through unbiased screening assays. Whilst such assays have their use, they have several caveats; for one, drug potencies can vary greatly based on the assay used. A large proportion of early testing involved the in vitro use of a single cell lines (e.g. Vero-E6). Such cell lines often have altered immune responses to infection and do not fully recapitulate viral growth in vivo. Secondly, the MOA and in vivo bioavailability for drugs identified in unbiased screens were often unknown, making it difficult to ascertain off target effects, and ultimately leading to many being false positives. Both limitations could be overcome through the focusing on repurposed drugs with known MOA, and pharmacokinetics/pharmacodynamics (PK/PD), focusing on molecules that can be administered orally, and tested in multiple cell lines and where available, animal models.

Another contributing factor was the failure to consider viral dynamics during COVID-19 disease. Several experimental drugs were used by pre-treating or treating experimental animals very early during infection (e.g., less than a day post infection). This is unrealistic in clinical practice: most people are unaware of when they become infected, SARS-CoV-2 has a long incubation period during which a person is asymptomatic, and viral burden peaks very early in the symptomatic phase. Drugs that show efficacy in pre-treated animals would only be useful in humans if given in the pre-symptomatic phase of illness or prophylactically, which is improbable even when health systems are functioning optimally. Prophylactic antiviral therapy for COVID-19 seems implausible as this requires administration to be regular or immediately prior to a known exposure, whereas exposure to COVID-19 is typically unpredictable. Regular drug administration requires a huge volume of drug to be manufactured and even if such a prophylactic antiviral was readily available, the challenge of dispensing sufficient drug in a short time while balancing the risk of adverse events from a mass administration is daunting.

For future pandemics, it would be reasonable to clinically evaluate only anti-viral agents that have shown efficacy against similar infectious pathogens in vivo, as was done for Remdesivir. Testing molecules discovered through in vitro assays alone without considering the natural history of the disease, PK/PD and viral replication kinetics is risky and unlikely to succeed. Attention must be paid to these variables, even if doing so may require time and patience. This is likely to be a thorough, more efficacious and ultimately safer approach for patients. These principles however are different for immunomodulatory drugs.

Early retrospective analyses showed that COVID-19 patients who had higher expression of inflammatory cytokines had poorer survival prognoses. This led to hypotheses that inhibiting specific cytokine mediated pathways would be beneficial, and consequently immunomodulatory therapies such as Anakinra (anti-IL1b) and Tocilizumab (anti-IL6) were repurposed for COVID-19. Whilst monotherapy with these agents has thus far proved unsuccessful, broad-spectrum immunomodulators such as Dexamethasone, or Baricitinib (anti-Janus kinase), or combinations of immunomodulatory drugs have shown efficacy in those with severe COVID-19 disease (Group 2021; Kalil et al. 2021) .

In hindsight, it may have been possible to predict which drugs were likely to be efficacious and in which target patient population. Since the inflammatory response to SARS-CoV-2 involved multiple inflammatory pathways and regulators driving large, complex, often non-overlapping transcriptional networks (Ho et al. 2021a) , it was more likely that broad-based rather than specific immunomodulatory targeting would be needed. Similarly, an immunomodulatory strategy would only be necessary or helpful for those with immune dysregulation (severe COVID-19) and could be harmful when the immune system is working effectively to control viral replication and disease (mild to moderate COVID-19). Some may argue that until specific drug were clinically evaluated in specific patient strata, we could not have known this to be true. However we propose that knowledge of disease pathogenesis could have filtered earlier into prioritization of key drugs repurposing in specific patient populations.

Finally, perhaps the hardest lesson we had to learn is that our arsenal of (usable) antivirals and immunomodulatory drugs is relatively empty and poorly equipped to deal with emerging pathogens. It is clearly time to re-think how antivirals and anti-inflammatory drugs are prioritized, designed and identified. A novel method of making vaccines (RNA-based vaccine) revolutionized our world during this pandemic. Novel drug design strategies, possibly embracing new technologies, are needed more than ever (Ho et al. 2021b) .

So how should we choose appropriate repurposed anti-inflammatory/anti-viral agents for clinical development during a pandemic, especially one with immunological features, such as COVID-19? Immunological effects are difficult to evaluate without in vivo studies. Given the unavoidable and lengthy delay in establishing the best humanized animal model, starting directly with human studies using repurposed drugs may be acceptable if done with close interaction between clinicians and scientists.

As randomized controlled trials (RCT) cost time and money, choosing the wrong drugs or wrong approach for an RCT has a large opportunity cost. Starting RCTs with agents that are well established in clinical practice (e.g., Dexamethasone) is acceptable and has succeeded. However, in this pandemic the initial choice for clinical evaluation of less well understood anti-inflammatory drugs was largely driven by expert clinical opinion (not necessarily with pre-clinical rationale) and through observational reports (which suffer from selection bias). The case of Hydroxychloroquine, with hundreds of wasted randomized trials attempting to prove/disprove its effects, attests to the need for a stringent framework for the clinical development of repurposed agents in a global pandemic. We propose the following key considerations to be addressed through pre-clinical studies and early single arm clinical studies, to "short-list" agents that are most likely to work prior to running large, randomized trials:

 Biological and pre-clinical rationale and/or evidence in other analogous clinical situations  Established safety profile, expected toxicities  Pharmacokinetics (PK) for modified dosing  Evidence of pharmacodynamics (PD) activity on-target (analogous to phase 0 studies in oncology)

The traditional clinical development model is an observational study followed by a single-arm trial assessing safety, and then randomized studies in the select patient population (phase 2, followed by phase 3). In oncology, a phase 1b followed by a phase 2 expansion in a specific patient cohort is a commonly used approach, even in the setting of drug repurposing. An important component of these single arm studies in the setting of cancer however is the inclusion of translational readouts to define the disease and response to treatment. Similarly, even within the exigencies of a viral pandemic, early single-arm studies testing efficacy and safety of repurposed drugs with known MOA and safety profiles should include as many biological assays of activity as possible, developed through close collaboration between clinicians and scientists. Clinical investigators must obtain input from virologists, immunologists and pharmacologists, to choose agents which will work during key viral kinetic time-points or which blunt specific (or broad) immune phenomena, and select drug doses and pharmacodynamic endpoints to best assess on-target activity. These initial studies may need to focus on PK/PD, or surrogate biochemical or virological end-points, prior to looking at hard clinical endpoints such as incidence of intubation, time in intensive care or 30-day mortality, which typically need a much larger sample size and well defined control arms.

Simultaneous animal studies can be used to refine these early trial protocols prior to agents being evaluated in phase 3 trials. When agents are ready for assessment in phase 3 trials, adaptive trial designs which permit the testing of multiple agents simultaneously or in rapid sequence is an efficient, cost-effective approach for otherwise time and resource intensive RCTs. Such a strategy will allow a shift from "logic-based medicine" to a sound evidence-based application, maximizing input from ongoing fundamental research and reducing the number of futile phase 3 trials testing drugs that were never likely to work.

Unfortunately, logistic and economic decisions have severely hampered drug discovery in the last few decades. Until very recently, antiviral drug development has been considered slow compared to other diseases (~5% drugs approved between 1972 to 2017, many against one virus (HIV), compared to 17% cancer). Drugs are also available to only ~9-10 viruses out of the >200 known to cause human disease. Orphan viral diseases often have no coverage, setting us up for the next pandemic. To counter this, there needs to be a global effort to dedicate resources to (a) cataloging/sequencing new and emerging viruses/variants, (b) developing experimental models in vivo and in vitro for infectious diseases, (c) drug discovery for orphan viruses, and (d) multiply medicinal chemistry efforts on known molecules with the potential of pan-viral activities (chain terminator, protease inhibitors), (e) that are also easily administered (e.g. orally bioavailable).

Managing ethical issues and regulatory hurdles even before Pandemics hit An important lesson from COVID-19 is the need to have at the ready rapidly implementable clinical trial protocols testing repurposed drug "X" whose safety profiles are known and acceptable, whose MOA has biological plausibility, or at a minimum, has early in vivo/ ex vivo data in two independent systems supporting its use. Such trial protocols should have interim ethics and regulatory approval prior to a pandemic, which would then require final rapid review from regulators when the next pandemic hits. Forward planning would allow the testing of multiple candidate drugs simultaneously during the exponential growth in numbers seen during the initial stages of a pandemic. Most importantly, having such trials immediately implementable would limit the use of experimental therapies in poorly designed observational studies, or worse, ad-hoc use that puts patients at risk and is destined at best for uninformative retrospective analyses.

This requires the establishment of scientific/clinical cooperatives across state, national and global borders to coordinate clinical trials, and needs to be an international effort. It divides the work, divides the cost, builds capacity, and makes the impact instantaneously global. In addition, any candidate drug that gets tested simultaneously world-wide would put to rest concerns about its efficacy in specific social or racial groupsaspects which invariably come up in discussions, whether founded in science or not. The establishment of such global cooperatives would also expedite the process for drug approval globally. Currently, drug approval is meted out by national government agencies even though safety and efficacy are evaluated similarly throughout the world. Nation-specific drug approvals have resulted in delays in drug distribution to communities that need them the most. If trials are conducted globally, and subject to review by an international regulator, this could help expedite drug review processes within local jurisdictions. Such an international regulator is not unimaginable; as has been done for other global threats, such as nuclear arms regulation and climate change. Once suitable broad-acting drugs are identified, manufacturers could be regulated against holding market exclusivity or have this limited through government "buy-outs". The rapid production of generic drugs would reduce cost and ensure equitable global distribution. Ensuring equitable access to a drug of utility in a viral pandemic setting is one of the foremost lessons learned from the COVID-19 pandemic.

By instituting a global cooperative scientific effort to find drugs that work, the scientific community could influence the distribution of any treatment more equitably. Inequitable distribution is perilous both to those with and without access to the treatment. As long as a pandemic virus rages and mutates "elsewhere", the entire world remains at risk. Thus, a cooperative effort from the medical and scientific community, based on the principles outlined above is essential to both identify suitable therapeutic interventions, and to facilitate their distribution in the setting of a pandemic. Solidarity in the scientific community will also facilitate public education and legal reforms on misinformation, which are essential for the ultimate translation of research activity into lives saved.

Mortality outcomes with hydroxychloroquine and chloroquine in COVID-19 from an international collaborative meta-analysis of randomized trials

Remdesivir for the Treatment of Covid-19 -Final Report

COVID-19 pneumonia: different respiratory treatments for different phenotypes?

Circuits between infected macrophages and T cells in SARS-CoV-2 pneumonia

Tocilizumab in patients admitted to hospital with COVID-19 (RECOVERY): a randomised, controlled, open-label, platform trial

TOP1 inhibition therapy protects against SARS-CoV-2-induced lethal inflammation

Unconventional viral gene expression mechanisms as therapeutic targets

Baricitinib plus Remdesivir for Hospitalized Adults with Covid-19

Ventilation with Lower Tidal Volumes as Compared with Traditional Tidal Volumes for Acute Lung Injury and the Acute Respiratory Distress Syndrome

Ivermectin for preventing and treating COVID-19

Flawed ivermectin preprint highlights challenges of COVID drug studies

The trinity of COVID-19: immunity, inflammation and intervention

Acute Respiratory Distress Syndrome

Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study