key: cord-290195-8uaai9nv
authors: Stebbing, Justin; Krishnan, Venkatesh; de Bono, Stephanie; Ottaviani, Silvia; Casalini, Giacomo; Richardson, Peter J.; Monteil, Vanessa; Lauschke, Volker M.; Mirazimi, Ali; Youhanna, Sonia; Tan, Yee‐Joo; Baldanti, Fausto; Sarasini, Antonella; Terres, Jorge A. Ross; Nickoloff, Brian J.; Higgs, Richard E.; Rocha, Guilherme; Byers, Nicole L.; Schlichting, Douglas E.; Nirula, Ajay; Cardoso, Anabela; Corbellino, Mario
title: Mechanism of baricitinib supports artificial intelligence‐predicted testing in COVID‐19 patients
date: 2020-05-30
journal: EMBO Mol Med
DOI: 10.15252/emmm.202012697
sha: 
doc_id: 290195
cord_uid: 8uaai9nv

Baricitinib, is an oral Janus kinase (JAK)1/JAK2 inhibitor approved for the treatment of rheumatoid arthritis (RA) that was independently predicted, using artificial intelligence (AI)‐algorithms, to be useful for COVID‐19 infection via a proposed anti‐cytokine effects and as an inhibitor of host cell viral propagation. We evaluated the in vitro pharmacology of baricitinib across relevant leukocyte subpopulations coupled to its in vivo pharmacokinetics and showed it inhibited signaling of cytokines implicated in COVID‐19 infection. We validated the AI‐predicted biochemical inhibitory effects of baricitinib on human numb‐associated kinase (hNAK) members measuring nanomolar affinities for AAK1, BIKE, and GAK. Inhibition of NAKs led to reduced viral infectivity with baricitinib using human primary liver spheroids. These effects occurred at exposure levels seen clinically. In a case series of patients with bilateral COVID‐19 pneumonia, baricitinib treatment was associated with clinical and radiologic recovery, a rapid decline in SARS‐CoV‐2 viral load, inflammatory markers, and IL‐6 levels. Collectively, these data support further evaluation of the anti‐cytokine and anti‐viral activity of baricitinib and supports its assessment in randomized trials in hospitalized COVID‐19 patients.

atypical neutrophilic dermatosis with lipodystrophy and elevated temperature (CANDLE). Now, we provide biochemical and cellular evidence confirming predictions focused on anti-cytokine signaling and potential anti-viral effects for baricitinib, along with a case series, supporting its potential utility in hospitalized COVID-19 patients.

The anti-cytokine and anti-inflammatory activity of baricitinib was evaluated with a focus on cytokines relevant to COVID-19 infection. We evaluated the in vitro pharmacology of baricitinib across relevant leukocyte subpopulations coupled to its in vivo pharmacokinetics to determine its effect on distinct cytokine pathways. Concentration response curves combined with exposure data from baricitinib-treated healthy volunteers demonstrate that baricitinib affects cytokine-dependent phosphorylated STAT (pSTAT) inhibition to varying degrees (Fig. 1A) . Baricitinib inhibited signaling of cytokines implicated in COVID-19 infection, including IL-2, IL-6, IL-10, IFN-γ, and G-CSF, with lower IC 50 values translating to a greater overall inhibition of cytokine-induced JAK/STAT signaling during the dosing interval (Fig. 1A) . Furthermore, baricitinib treatment resulted in a significant reduction (p<0.05) from baseline in plasma IL-6 at week 12 in patients with active RA who had an inadequate response to methotrexate from a phase 2b (Tanaka, Emoto et al., 2016) , randomized, placebo-controlled, dose-ranging study (Fig. 1B) .

Next, we validated the proposed biochemical inhibitory activity of baricitinib on the numb-associated kinase (NAK) family members AAK1, BIKE, GAK, and STK16, some of which are hypothesized to facilitate viral propagation of coronavirus in epithelial cells (Bekerman et al., 2017 , Owczarek et al., 2018 , Richardson et al., 2020a . Baricitinib activity demonstrated affinity against AAK1 (8.2nM), BIKE (20nM), and GAK (120nM) ( Fig.   2A and 2B); these values are within the exposure range of the approved 2-mg (US, EU) and 4-mg (EU) oncedaily doses of baricitinib for the treatment of RA (Shi et al., 2014) . The pharmacokinetics of baricitinib show the unbound fraction of free bioavailable drug in RA patient sera as being 326 nM area under the curve (AUC) and 652 nM (AUC) for baricitinib 2-mg and 4-mg, respectively (data on file). We compared the relative binding affinities of different JAK inhibitors (JAKis) for these NAKs. Notably, amongst JAKis approved for the treatment of RA, baricitinib uniquely demonstrated high affinity for AAK1, BIKE, and GAK, whereas tofacitinib and

This article is protected by copyright. All rights reserved upadacitinib did not demonstrate high affinity for these kinases (Fig. 2B) . The binding affinity of baricitinib for AAK1 and GAK is similar to the binding affinity of baricitinib for JAK1 (5.9nM) and JAK2 (5.7nM) (Fridman et al., 2010) .

To extend these activities on NAKs, we evaluated the effect of baricitinib in reducing viral infectivity in 3D primary human liver spheroids (Bell, Hendriks et al., 2016) infected with purified SARS-CoV-2 and treated with baricitinib. Specifically, we used a 3D spheroid model of primary human liver cells in which hepatocytes retain their transcriptomic, proteomic, and metabolomic phenotype and functionality for multiple weeks (Bell et al., 2016 , Bell, Lauschke et al., 2017 , Vorrink, Ullah et al., 2017 , including expression of AAK1 and GAK (Fig. 2C ).

The overlay of the viral antigen 1A9 with the ACE-2 protein as detected by immunofluorescence, after 5 days of SARS-CoV-2 infection in these spheroids provides a visual confirmation of virus infectivity in this system ( Fig. 2C ). SARS-CoV-2 was able to infect human primary liver spheroids in this experimental paradigm, as demonstrated by a day's post-infection (dpi) dependent increase in viral RNA ( Fig. 2D and 2E ). Baricitinib did not result in liver cell injury up to concentrations around 8-fold higher than the clinically observed AUC values at 4 mg (AUC = 652nM; Fig. 2F ). Importantly, pretreatment of spheroids with physiologically relevant concentrations of baricitinib (400 nM and 800 nM) significantly (p<0.05) reduced viral load by 30-40%, corroborating the proposed inhibitory effects of baricitinib on AAK1 and GAK-mediated viral propagation ( Fig.   2G and 2H). These results suggest that host cells that express these NAKs may serve as a target for baricitinibmediated reduction in viral propagation.

Following the recent publications by Stebbing et al.(Richardson et al., 2020a, Stebbing et al., 2020) , COVID-19 patients were treated with baricitinib in a pilot study in Milan, Italy. Four patients with bilateral COVID-19 pneumonia, who presented with varying degrees of disease severity (Table 1) , were included in this pilot study; 3 individuals (Patients B, C, and D) were clinically unstable with moderate-tosevere disease. All four patients were admitted to the ward from the emergency department in March 2020.

As shown in Table 1 , Patient A was a female nurse, aged 29. Patient B, a 76-year-old male, had significant comorbidities (a former smoker, arterial hypertension (AH), chronic obstructive pulmonary disease [COPD], coronary artery disease (CAD), and had undergone an aortic aneurysm Endurant II graft repair on February 25, 2020). Patient C, a 57-year-old male, had co-morbidities including AH and COPD (non-smoker); he

suddenly deteriorated a few hours following hospitalization, and was placed on continuous positive airway pressure at the time baricitinib was initiated. Patient D, a 51-year-old male had a high body mass index (BMI) of 35. All four patients had detectable plasma IL-6 levels ( Fig. 3A) with markedly raised inflammatory markers (C-reactive protein [CRP] ), as would be expected in patients with COVID-19 pneumonia.

As shown in Figure 3A , all four patients showed improvement with baricitinib treatment in signs and symptoms such as cough, fever, and reduction in plasma IL-6 levels, along with a reduction in the SARS-CoV-2 RNA viral load, as detected by the real-time reverse-transcriptase-polymerase chain reaction (RT-PCR) signal from the nasopharyngeal carriage. Stringent criteria were used for RNA detection of SARS-CoV-2 in nasopharyngeal carriage and peripheral blood. Real-time RT-PCR was performed on three distinct viral gene targets (Corman, Landt et al., 2020) (Appendix Table S1 ), using the most sensitive target (N gene) and a cut-off using Ct values >40 for the analyses illustrated in Figure 3A , in contrast to others, for example in the hydroxychloroquine study (Ct values ≥ 35) (Gautret, Lagier et al., 2020) . Only two of the patients (Patients A and C) had detectable viral RNA in their peripheral blood ( Fig. 3A and Appendix Table S1 ).

There were nominal changes in lymphocyte counts throughout the course of treatment with baricitinib in all four patients (Fig. 3B ). In addition, changes in IgG (Fig. 3C) , and IgM suggest adequate levels consistent with the confirmed sero-conversion for 3/4 of the patients after initiation of baricitinib (Fig. 4) . Most importantly, all four patients achieved sero-conversion as evidenced by the presence of neutralizing antibodies against the S1 and S2 spike proteins of SARS-CoV-2 after baricitinib exposure (Fig. 4 ). Neutrophil and total white blood cell counts tracked with improvement in disease severity (Fig. 5A ). In addition, all four patients demonstrated improvement in their CRP, ferritin, and D-dimer levels (Fig. 5A) . Notably, Patient C, who had the lowest PaO 2 at baseline, showed radiographic improvement in lymphocytic infiltrates when comparing the computer tomography scan from day 1 to day 19 (Fig. 5C ). Overall, a total of 10 days of dosing for patients A, B, and D and 12 days for Patient C, with baricitinib 4-mg once-daily orally in Patients A, C and D, and 2-mg in Patient B (in accordance with the label guidelines) was sufficient to document in all patients improved lung function, resolution of their illness, and reductions in viral load, plasma IL-6, ferritin, and CRP levels. Baricitinib treatment was associated with a rapid and consistent improvement in clinical, radiologic, virologic, inflammatory, and cytokine measures in this diverse group of patients, including 3 (Patients B, C, and D) who were deemed to be moderately-to-severely unwell and at high risk of deteriorating rapidly (i.e. those apart

This article is protected by copyright. All rights reserved from the 29-year-old nurse). Transient increases in liver aminotransferases were observed in all four patients, with no changes in other liver enzymes or bilirubin (Fig. 5B) ; however, these liver enzyme elevations improved within 72 hours without interrupting baricitinib treatment, suggesting that these elevations may not be causally related to baricitinib treatment, but may be reflective of disease severity (Fig. 5B ). In general, these data demonstrate that treatment with baricitinib in these four patients with bilateral COVID-19 pneumonia was well tolerated. Collectively, this limited case series provides preliminary evidence that baricitinib treatment may lower inflammatory burden and may result in a reduction in disease severity in COVID-19 patients.

The pharmaceutical interventions for the treatment of COVID-19 patients proposed to date include testing anti-viral mechanisms, either in combination or alone, along with anti-malarials and immunomodulators (Siddiqi & Mehra, 2020) . One such potential pharmacological approach is the use of immunomodulators in the subset of patients who develop a cytokine storm associated with pulmonary involvement including ARDS that leads to a rapid deterioration of the co-morbid conditions in such patients (Appendix Figure S1 ) , Siddiqi & Mehra, 2020 . In this report, we demonstrate that baricitinib is an inhibitor of cytokines implicated in ICU-bound COVID-19 patients, which have been collated from several reports, including IL-2, IFN-γ, IL-6, IL-10, and G-CSF , Ruan et al., 2020 .

Elevated IL-6 and hyperferritinemia were predictors of death in COVID-19 patients in China (Ruan et al., 2020 . The relevance of IL-6 in COVID-19 associated ARDS was recently shown in an open-label study where blockade with tocilizumab, an IL-6R antibody, resulted in a rapid recovery of peripheral oxygen saturation and recovery from febrile signs and symptoms (Xu, Han et al., 2020) . Baricitinib has been previously shown to inhibit IL-6-induced MCP-1 production from human peripheral blood mononuclear cells (Fridman et al., 2010) . In addition to a marked and rapid reduction in levels of human serum CRP in adult RA patients previously described (Tanaka et al., 2016) , baricitinib treatment also reduced the mean change from baseline of plasma IL-6 in adult RA patients. The anti-inflammatory effects of baricitinib have also been demonstrated by the reduction of serum levels of IFN-γ, IP-10, GM-CSF, and MCP-1 in pediatric patients with steroid-dependent chronic inflammation, resulting in control of disease activity and the ability to wean or taper steroids (Sanchez, Reinhardt et al., 2018) .

Another pharmacologic strategy could include targeting members of the NAK family (AAK1, GAK, BIKE, and STK16) that activate the AP-2 scaffolding protein vital to viral entry and propagation (Bekerman et al., 2017 , Owczarek et al., 2018 . Previous reports demonstrated inhibiting such kinases was effective in reducing cellular infectivity for viruses that rely on the activity of AP-2 such as HCV, Dengue, HIV, SARS, and Ebola (Bekerman et al., 2017 , Chaudhuri, Lindwasser et al., 2007 , Neveu, Ziv-Av et al., 2015 , Owczarek et al., 2018 .

While baricitinib demonstrated potent inhibition of several NAKs that phosphorylate the AP-2 adaptor and reduced viral infectivity in primary human liver spheroids, it is still unknown whether this inhibition inhibits SARS-CoV-2 propagation in COVID-19 infected patients.

The aforementioned 4 COVID-19 patients offer support to the mechanism of action of baricitinib as an anticytokine and a potential anti-viral agent. Reductions in inflammatory markers of COVID-19 disease, such as IL-6, CRP, and ferritin are consistent with expected anti-inflammatory activity. While the consistent reduction in viral RNA observed in all four patients (measured from their nasopharyngeal carriage) is suggestive of reduction in viral load, this could be a consequence of the natural ability in these patients to clear the virus.

At the time of manuscript submission, a few weeks have elapsed since cessation of baricitinib. Active clinical, biochemical, and virologic surveillance of each patient to monitor any sequelae, including the possible appearance of an immune reconstitution inflammatory syndrome or molecular evidence of de novo SARS-CoV-2 replication in the respiratory tract or peripheral blood, clearly shows that all four patients were devoid of viral load and signs and symptoms of COVID-19.

Given many unknowns concerning this novel SARS-CoV-2 virus, it is necessary to consider safety concerns alongside potential efficacy. As an effective immunomodulatory agent used for the treatment of RA, baricitinib was associated with an increased risk of infections such as upper respiratory tract infections and herpes zoster (Eli Lilly and Company, 2019). Baricitinib treatment in RA patients has been associated with increased incidence of deep vein thrombosis (DVT), and it is recommended that appropriate prophylactic measures be considered with use of baricitinib in hospitalized patients with higher risk of DVT (Eli Lilly and Company, 2019). This is especially relevant given recent findings of thrombotic complications in COVID-19 patients (Klok, Kruip et al., 2020) . As baricitinib is an inhibitor of IFN-responsive genes (Sanchez et al., 2018) , its potential impact on the subsequent development of protective humoral and cell-mediated anti-viral immunity needs to be assessed in the context of COVID-19 infection. Of relevance, RA patients on long-term Accepted Article baricitinib treatment achieved satisfactory humoral responses to pneumococcal conjugate vaccination suggesting that humoral responses remain partially intact (Winthrop, Bingham et al., 2019) .

Changes seen in lymphocyte subsets in this limited clinical case series suggest that baricitinib treatment does not reduce T cell subsets (CD3 + and CD4 + ) or NK cells. It is too early to ascribe observed reductions of SARS-CoV-2 RNA loads in the nasopharyngeal carriage to a putative antiviral activity of baricitinib, since the underlying T cell subset levels were maintained in these treated patients. This is consistent with previous reports wherein T cell and various B cell subsets remained within normal reference ranges after initiation of baricitinib treatment in RA patients (Tanaka, McInnes et al., 2018) . It is, however, reassuring to underscore that after treatment with an immunomodulatory agent such as baricitinib, we could measure IgG anti-S1 and anti-S2 neutralizing antibodies. Most importantly, three out of four patients, while under baricitinib exposure, achieved seroconversion, as evidenced by the appearance of IgG antibodies with neutralizing activity against SARS-CoV-2. Moreover, in all four patients, a progressive rise in the titer of these neutralizing antibodies was observed during follow-up. However, the frequency and time course of this occurrence would require prospective validation in randomized controlled trials (Ottoviani & Stebbing, 2020) . Translational research from these studies will also help delineate the relative contributions of the humoral, cellular, and innate immune responses to SARS-CoV-2.

Baricitinib has a well-established safety profile in patients with RA over long-term treatment; in 10,127

patients-year of exposure and up to 6.9 years duration, the incidence rate of serious infection was 2.8/100patient years and stable over time (Genovese, Smolen et al., 2019) . Furthermore, in 89 individuals treated with baricitinib in 4 separate single center studies across Northern Italy, no over-arching adverse safety signals have been observed with transient use (<14 days) of baricitinib in COVID-19 patients (Francesco Menichetti, Fabio Lena, Fabrizio Cantini, and Pinuccia Omodeo, personal communications) . Vigilance for detecting serious adverse events and appropriate management of potential complications is essential given the inhibition on IFN-responsive genes by baricitinib (Sanchez et al., 2018) . Therefore, the impact of baricitinib on the subsequent development of protective humoral and cell-mediated anti-viral immunity in COVID-19 patients must be evaluated in randomized clinical trials (Ottoviani & Stebbing, 2020) . Importantly, baricitinib is administered orally once a day, and has a low cytochrome P 450 inhibitory activity with a low drugdrug interaction risk; its short half-life (approximately 12 hours in RA patients) and renal elimination as the main clearance mechanism make it fast to wash-out if necessary (Eli Lilly and Company, 2019).

This article is protected by copyright. All rights reserved Collectively, these data provide preliminary evidence that baricitinib could be tested as an effective intervention strategy to stem the cytokine storm and viral propagation seen in hospitalized COVID-19

patients. The finding that baricitinib is a potent AAK1/BIKE/GAK inhibitor that may reduce host cell infectivity, along with reaffirmation of its anti-cytokine profile, provide reasons to study this intervention in randomized clinical trials. We cannot over-interpret the findings from the case series described in this manuscript.

Therefore, results from such randomized trials, notably the adaptive National Institute of Allergy and Infectious disease (ADAPTIVE COVID-19 ACCT2) will be central to define efficacy and safety of baricitinib and to manage effective clinical care as this outbreak continues to expand across the globe.

Leukocyte preparation and experimental design were largely performed as previously described (McInnes et al., 2019) . Whole blood samples from healthy donors (N=6) were apheresed, and leukocyte-enriched fractions were transferred to the company Primity Bio (Fremont, CA, USA). Immediately following apheresis, approximately 600,000 cells were plated in 100 µL into 96-well plates and incubated with baricitinib using an Stanford University (reference numbers eProtocol # 38735 and IRB Registration # 6208). The IRB provided a Waiver of Consent because leukocyte-enriched fractions are by-products of the blood donation.

Patient samples were obtained from the double-blind, randomized, placebo-controlled, phase 2b study NCT01469013. Patients had moderate-to-severe active adult onset RA, despite stable methotrexate treatment. Patients (N=145) were randomized (2:1:1:1:1) to placebo or once-daily oral 1-mg, 2-mg, 4-mg, or 8-mg baricitinib for 12 weeks (Tanaka et al., 2016) . Plasma IL-6 levels were analyzed using an enzyme immunoassay. Treatment effects were estimated using a mixed effects repeated measures model (SAS Proc Mixed) using log10 transformed IL-6 levels with an unstructured covariance matrix. NCT01469013 was conducted in accordance with ethical principles of the Declaration of Helsinki and Good Clinical Practice guidelines. All investigation sites received approval from the appropriate authorized institutional review board or ethics committee. All patients provided written informed consent before the study-related procedures were undertaken.

AAK1, BIKE, GAK, and STK16 binding assays were performed with DNA-tagged recombinant human proteins derived from HEK-293 cells or E. coli. AAK1 Kinase-tagged T7 phage strains were prepared in an E. coli host derived from the BL21 strain. E. coli were grown to log-phase and infected with T7 phage and incubated with shaking at 32°C until lysis. The lysates were centrifuged and filtered to remove cell debris. The remaining kinases (BIKE, GAK, and STK16) were produced in HEK-293 cells and subsequently tagged with DNA for qPCR detection. Streptavidin-coated magnetic beads were treated with biotinylated small molecule ligands for 30 minutes at room temperature to generate affinity resins for kinase assays. The liganded beads were blocked with excess biotin and washed with blocking buffer (SeaBlock; Thermo Scientific Pierce, 1% BSA, 0.05% Tween 20, 1 mM DTT) to remove unbound ligand and to reduce non-specific binding. Binding reactions were assembled by combining kinases, liganded affinity beads, and test compounds in 1x binding buffer (20% SeaBlock, 0.17x PBS, 0.05% Tween 20, 6 mM DTT). Test compounds were prepared as 111x stocks in 100% DMSO. Equilibrium affinity constants (Kds) were determined using an 11-point 3-fold compound dilution series with three DMSO control points.

This article is protected by copyright. All rights reserved All compounds for Kd measurements were distributed by acoustic transfer (non-contact dispensing) in 100% DMSO. The compounds were then diluted directly into the assays such that the final concentration of DMSO was 0.9%. All reactions performed in polypropylene 384-well plate. Each was a final volume of 0.02 mL. The assay plates were incubated at room temperature with shaking for one hour and the affinity beads were washed with wash buffer (1x PBS, 0.05% Tween 20). The beads were then re-suspended in elution buffer (1x PBS, 0.05% Tween 20, 0.5 µM non-biotinylated affinity ligand) and incubated at room temperature with shaking for 30 minutes. The kinase concentration in the eluates was measured by qPCR.

Protein levels of AAK1 and GAK were pulled out from tandem mass tag (tmt)-based proteomics datasets SARS-CoV-2 (Genbank accession number MT093571) was isolated from a nasopharyngeal sample of a patient in Sweden on Vero E6 cells. Cryopreserved PHH (BioIVT, USA) were thawed and seeded into 96-well ultra-low attachment plates (Corning) with 1,500 cells per well as previously described (Bell et al., 2016) . Spheroids were pre-exposed to baricitinib for 24h starting five days after cell seeding. At day six of culture, cells were exposed to SARS-CoV-2 at a multiplicity of infection of 0.1 in triplicate for 48h. After 48h, spheroids were washed with PBS, pooled (32 wells/condition) and lyzed using TrizolTM (Thermofisher). RNA was extracted using Direct'zol mini-kit (Zymo-research) and relative level of viral RNA was determined by qRT-PCR as previously described (Monteil, Kwon et al., 2020) .

Primary human hepatocyte spheroid cryo-sections were washed twice with PBS for 10 min at room temperature. Afterwards, samples were blocked with PBTA buffer (5% BSA, 0.25% Triton X-100, 0.01% NaN3 in PBS) for two hours at room temperature prior to an overnight incubation at 4°C with the monoclonal

This article is protected by copyright. All rights reserved primary antibody anti-1A9 (diluted in PBTA to a final concentration 5µg/ml). Samples were washed 3x15 min with PBS at room temperature before incubation with the secondary antibody (donkey anti-mouse diluted in PBTA 1:500) for 2 hours at room temperature. Subsequently, samples were washed three times with PBS (15 min each) at room temperature and mounted with DAPI Gold Anti fade.

On the basis of the two reports of AI-derived discovery of baricitinib in COVID-19 (Richardson et al., 2020a , the Sacco Baricitinib Study Group and Imperial College developed a protocol to assess baricitinib in a small case series of patients, acknowledging potential risks of pulmonary infections (Favalli, Biggioggero et al., 2020 , Richardson, Corbellino et al., 2020b based on its mechanism of action while leveraging the hypothesis for a potential therapeutic benefit in COVID-19. Here, we describe the cases from the independent research conducted. Ethics and local Institutional Review Board approval was authorized on

March 16, 2020 (number 14581/2020), which provided permission to treat three patients with bilateral COVID-19 pneumonia with baricitinib. Three patients (Patients A, B, and C) were immediately recruited that day, having been admitted to the wards from the emergency department. A supplementary approval was sought on March 25, 2020 and obtained to treat a fourth patient (Patient D) for validation purposes.

Inclusion criteria for compassionate use of an investigational oral medicine (albeit one that was approved in another indication) for patients with signs of severe illness at diagnosis or secondary clinical aggravation (respiratory symptoms or general signs) was based on World Health Organization (WHO) criteria for severe pneumonia caused by SARS-CoV-2 (WHO, 2020a).

Written informed consent was obtained from each patient, and 4-mg oral baricitinib was administered oncedaily according to the label for treatment of RA with a 2-mg lower dose as per the label, for 10 -12 days.

Criteria for patient discharge with recovery were from the European Centre for Disease Prevention and

Infections of the International Severe Acute Respiratory and Emerging Infection Consortium (supported by WHO), which has been updated and used in response to COVID-19 (Dunning, Merson et al., 2014) .

This article is protected by copyright. All rights reserved

Clinical samples for SARS-CoV-2 diagnostic testing were obtained according to WHO guidelines (WHO, 2020b).

For each patient, a sampling strategy was implemented in which nasopharyngeal and blood samples were obtained regularly from hospital admission, and subsequently once every two or three days until patient discharge. At the time of manuscript submission, these four patients continue to be monitored for COVID-19 disease. Upper respiratory samples were nasopharyngeal swabs and blood samples were EDTA tubes adapted for RT-PCR. All samples were analyzed in the same center as the patients where procedures for RNA extraction, real-time RT-PCR (rRT-PCR), were undertaken.

Throat-swabs and plasma for each patient were processed using the automated ELITe InGenius® system and the GeneFinderTM COVID-19 Plus RealAmp Kit assay (ELITechGroup, France). The reaction mix was manually prepared (according to manufacturer's instruction) and loaded onto the system with other reagents, while RNA was extracted from 200 µL of sample and eluted in 100 µL; the final reaction volume consisted of 5 µL of RNA plus 15 µL of reagents mix. The RT-PCR set up according to manufacturer's instructions were: 50° C for 20 minutes, 95° C for five minutes plus 45 cycles at 95° C for 15 seconds and 58° C for 60 seconds. Three target genes, RNA-dependent RNA polymerase (RdRP), nucleocapsid protein (N), and Envelope membrane protein (E) were simultaneously amplified and tested. A cycle threshold value (Ct-value) less than 40 was defined as a positive test result, and a Ct-value greater than 40 was defined as a negative outcome according to our criteria. The quality of nasopharyngeal swabs was checked using the CELL Control r-gene kit (bioMérieux).

Oligonucleotide Sequence 

This article is protected by copyright. All rights reserved P) traceable to the Reference Measurement System, for CRP the high sensitive immunoturbidimetric assay, and for ferritin levels the chemiluminescent microparticle immunoassay, were respectively used.

On the wards, standard laboratory and clinical management according to two treating physicians' discretions (MC and GC) were used. The chest CT shown was performed using a single inspiratory phase in One commercial multidetector CT scanner (General Electric Healthcare Revolution64) with a breath-holding protocol (tube voltage 120 kVp, thickness 1.4 mm, increment of 1.4 mm, mean CTDIvol 19 mGy).

Detection of IgG antibodies against the S1/S2 antigens of SARS-CoV-2 using the LIAISON SARS-CoV-2 S1/S2 IgG Kit (DiaSorin S.p.A., Saluggia, Italy). The IgG antibody concentrations are expressed as arbitrary units (AU/ml) with values <12 being interpreted as negative; between 12 to 15 as equivocal; and greater than 15 as positive.

Dashed horizontal lines indicate the cut-off values.

The concentration response curve fitting for NAK binding assay as previously described was normalized by dividing by the mean of three replicates of the 0 nM treatment condition. Four-parameter logistic curves were fit with the normalized assay signal as the response and log 10 compound concentration as the independent variable using the R drc package (R version 3.6.0 and drc version 3.0). The curve top parameter was fixed to a value of one while all other parameters were estimated in the curve fitting. Equilibrium affinity constant (Kd) values were determined using the absolute half maximum efficacy concentration (EC 50 ) values estimated from the logistic model.

IC 50 values were determined by analyzing the mean fluorescence intensity (MFI) of cytokine-stimulated samples in the presence of the designated concentration of compound. For a given case (stimulation, cell type, and pSTAT combination), the MFI for unstimulated and stimulated cells was determined for each donor.

To ensure that a biologically relevant signal was induced, concentration-response curves (CRCs) were only

analyzed when a consistent response to stimulus was observed as described below. Data for baricitinib were analyzed with a mixed effect model having compound as a fixed effect and donor as a random effect.

Two sets of criteria for reporting an IC 50 value and computing a steady state activity (SSACTIVITY) curve were used: one at the case level and another at the individual curve level. At the case level, the median across the 6 donors of the minimum stimulated to unstimulated ratio over two replicates is computed, and it is required that the median minimum ratio be above 1.5. Out of a total of 85 cases, 43 were selected according to these criteria. Once a case met this criterion, four-parameter logistic curves were fit to the curve response concentration data. If an individual curve had top outside of the 80%-120% activity range or bottom outside of the (-20%, 20%) range, the 4PL was refitted with these constraints. After a CRC was fit, the following quality criteria was evaluated: (1) R2 above 0.8; (2) SE of ln(I IC 50 ) below 10; (3) estimated IC 50 is within a five-fold difference of the minimum and maximum experimented concentrations. A curve failing any of these criteria was discarded.

The individual 4PL CRCs were combined with population pharmacokinetic (PK) curves to calculate the average steady state daily percent inhibition. The PK profiles of baricitinib were estimated from a two-compartment model with zero-order absorption that was developed using data from healthy volunteers from three Phase 1 clinical trials with once-daily 2-20 mg or twice-daily 5-mg. Protein binding effects were accounted for by replacing the in vitro IC 50 with an adjusted IC 50 value computed by dividing the IC 50 value for each donor by the proportion of compound unbound. Protein-bound adjusted CRCs were constructed by replacing the in vitro IC 50 value with the adjusted value. The average daily percent inhibition for a subject was obtained by entering the steady-state PK concentrations into the adjusted CRCs, computing the area under this curve, and dividing it by 24 hours. Using the individual donor values for average daily percent inhibition, a mixed effect model having compound as a fixed effect and donor as a random effect was fit. The reported estimates for population average SSACTIVITY are taken as the least squares means from this model. No transformations were undertaken to keep the estimates of SSACTIVITY within the 0-100% range.

This article is protected by copyright. All rights reserved

Eli Lilly and Company supported the non-clinical studies (anti-cytokine activity of baricitinib and NAK binding assays) and IL-6 assay from NCT01469013. The Sacco Baricitinib Study Group supported the clinical case series. The authors would like to thank the patients who participated in the study. We also thank the contributions of the members of the Sacco Baricitinib Study Group, who are listed in the Appendix. JS wishes to dedicate the paper to his uncle Mark who died from COVID-19 pneumonia on April 1, 2020. 

The non-clinical studies (anti-cytokine activity of baricitinib and NAK binding assays) and IL-6 assay from NCT01469013 were designed and analyzed by representatives of Eli Lilly and Company. The clinical case series was not funded or approved by Eli Lilly and Company and discussions between Eli Lilly and Company and the Sacco Baricitinib Study Group did not occur until patients had been dosed and preliminary results obtained.

Similarly, Eli Lilly and Company performed the non-clinical in vitro analyses (anti-cytokine activity of baricitinib and NAK binding assays) and IL-6 assay from NCT01469013 independently. The Sacco Baricitinib Study Group funded the clinical case series, and representatives (including those from Imperial College, London) had a role in study design, data collection, and data analysis. All authors participated in data analysis and interpretation,

draft and final manuscript review, and provided critical comment, including the decision to submit the manuscript for publication with medical writing support from Eli Lilly and Company; all authors reviewed and approved the final submitted version. Venkatesh Krishnan had full access to all data and had final responsibility for the decision to submit for publication.

PROBLEM: There are few drugs that are useful to treat COVID-19, the largest medical crisis of this century. To try to solve this problem, using AI, we found that an oral, once-daily medicine, baricitinib, normally used to treat adult rheumatoid arthritis (RA), may be useful in both reducing viral propagation in cells, and to mitigate the cytokine signaling seen in the hyper inflammatory stage of the disease. We wished to understand this further in various models and in a case series of patients.

We found that baricitinib inhibited the signaling of cytokines we typically see as being present in hospitalized patients with COVID-19. Using samples from a previous randomized trial in RA, we showed statistically significant declines in IL-6 levels. In liver spheroids designed to investigate SARS-CoV-2 infectivity and separately in kinase assays, we showed that baricitinib could reduce cellular infection by blockade of numb-associated kinase members used in viral propagation. In a small series of patients in Northern Italy with bilateral COVID-19 pneumonia, baricitinib therapy was associated with improvement in clinical, radiologic and viral parameters along with a rapid decline in CRP and plasma IL-6 levels.

IMPACT: Finding new drugs in our armamentarium to treat COVID-19 would be enormously valuable. Here, we stitched together the anti-cytokine and anti-viral activity of baricitinib and studied it in a small number of hospitalized patients. This study represents rapid repurposing from AI to the laboratory to a potential bedside therapeutic and supports the testing of baricitinib in randomized controlled trials in COVID-19 patients.

Eli Lilly and Company data are available upon request at vivli.org. Access to data is provided after a proposal has been approved by an independent review committee identified for this purpose and after receipt of a signed data sharing agreement. Access to data and documents will be provided in a secure data sharing environment. For details on submitting a request, see the instructions provided at www.vivli.org. In your request, please include the following accession/reference numbers (Genbank accession number MT093571) so that we can identify the related publication and specific dataset that you're trying to access. (A) IC 50 s for baricitinib (orange) in cytokine-stimulated human CD4 + T cells, CD8 + T cells, NK cells, and monocytes are shown for IL-2/pSTAT5, IL-4/pSTAT6, IL-3/pSTAT5, IL-6/pSTAT3, IL-10/pSTAT3, IL-15/pSTAT5, IL-21/pSTAT3, G-CSF/pSTAT3, GM-CSF/pSTAT5, and IFNγ/pSTAT1. Average daily percent STAT inhibition was calculated using in vitro concentration response curves and published human exposure data. Cytokine treatments that did not result in sufficient pSTAT stimulation are denoted as "NS" in the radar plots. 

This article is protected by copyright. All rights reserved (B) Levels of aspartate aminotransferase, alanine aminotransferase, and bilirubin are shown for all patients.

This article is protected by copyright. All rights reserved (C) Chest CT scan for Patient C on day one and 19 from symptom onset showing clinical improvement over time. Day one CT scan shows ground glass opacity (arrows) sub-pleurally in the lower lobes bilaterally (early stage one according to Pan et al (Pan, Ye et al., 2020) . Day 19 CT scan shows that consolidation was gradually absorbed with evident residual fibrosis and emphysema bubbles (arrows) in the site of the early lesions (absorption stage four according to Pan et al .

Data information: ALT, alanine aminotransferase; AST, aspartate aminotransferase; CRP, C-reactive protein;

CT, computed tomography; WBC, white blood cells. 

Anticancer kinase inhibitors impair intracellular viral trafficking and exert broad-spectrum antiviral effects

Characterization of primary human hepatocyte spheroids as a model system for drug-induced liver injury, liver function and disease

Comparisons of Stem Cell-Derived Hepatocytes, HepaRG Cells, and Three-Dimensional Human Hepatocyte Spheroids as Predictive In Vitro Systems for Drug-Induced Liver Injury

Interferon-mediated immunopathological events are associated with atypical innate and adaptive immune responses in patients with severe acute respiratory syndrome

Downregulation of CD4 by human immunodeficiency virus type 1 Nef is dependent on clathrin and involves direct interaction of Nef with the AP2 clathrin adaptor

Detection of 2019 novel coronavirus (2019-nCoV) by real-time RT-PCR

Expression of lymphocytes and lymphocyte subsets in patients with severe acute respiratory syndrome

Diagnostic detection of 2019-nCoV by real-time RT-PCR

Open source clinical science for emerging infections

Infection prevention and control for the care of patients with 2019-nCoV in healthcare settings. European Centre for Disease Prevention and Control

Baricitinib for COVID-19: a suitable treatment

Selective inhibition of JAK1 and JAK2 is efficacious in rodent models of arthritis: preclinical characterization of INCB028050

Hydroxychloroquine and azithromycin as a treatment of COVID-19: results of an open-label non-randomized clinical trial

Safety profile of baricitinib for the treatment of rheumatoid arthritis up to 7 Years: an updated Integrated safety analysis

HKUMed Detection of 2019 novel coronavirus (2019-nCoV) in suspected human cases by RT-PCR

Clinical features of patients infected with 2019 novel coronavirus in Wuhan

An interferon-gamma-related cytokine storm in SARS patients

Clathrin-dependent entry of severe acute respiratory syndrome coronavirus into target cells expressing ACE2 with the cytoplasmic tail deleted

Accepted Article This article is protected by copyright. All rights reserved

Comparison of baricitinib, upadacitinib, and tofacitinib mediated regulation of cytokine signaling in human leukocyte subpopulations

Penninger JM (2020) Inhibition of SARS-CoV-2 Infections in Engineered Human Tissues Using Clinical-Grade Soluble Human ACE2. Cell

AP-2-associated protein kinase 1 and cyclin G-associated kinase regulate hepatitis C virus entry and are potential drug targets

Lung pathology of fatal severe acute respiratory syndrome

What is the best drug to treat COVID-19? The need for randomized controlled trials. MED epub ahead of print

Early events during human coronavirus OC43 entry to the cell

Time Course of Lung Changes On Chest CT During Recovery From 2019 Novel Coronavirus

Baricitinib as potential treatment for 2019-nCoV acute respiratory disease

Comment to Baricitinib for COVID-19: a suitable treatment

Clinical predictors of mortality due to COVID-19 based on an analysis of data of 150 patients from Wuhan, China. Intensive Care Med Accepted Article This article is protected by copyright

JAK1/2 inhibition with baricitinib in the treatment of autoinflammatory interferonopathies

Rethinking drug design in the artificial intelligence era

The pharmacokinetics, pharmacodynamics, and safety of baricitinib, an oral JAK 1/2 inhibitor, in healthy volunteers

COVID-19 infection: the perspectives on immune responses

COVID-19 Illness in Native and Immunosuppressed States: A Clinical-Therapeutic Staging Proposal

COVID-19: combining antiviral and anti-inflammatory treatments

Japanese Patients with Active Rheumatoid Arthritis Receiving Background Methotrexate Therapy: A 12-week, Double-blind, Randomized Placebo-controlled Study

Endogenous and xenobiotic metabolic stability of primary human hepatocytes in long-term 3D spheroid cultures revealed by a combination of targeted and untargeted metabolomics

Clinical management of severe acute respiratory infection (SARI) when COVID-19 disease is suspected: Interim guidance V 1.2. World Health Organization. WHO/2019-nCoV/Clinical/2020.4. In WHO (2020b) Laboratory testing for coronavirus disease (COVID-19) in suspected human cases: interim guidance. World Health Organization. WHO/COVID-19/laboratory/2020.5

Evaluation of pneumococcal and tetanus vaccine responses in patients with rheumatoid arthritis receiving baricitinib: results from a long-term extension trial substudy

Effective treatment of severe COVID-19 patients with tocilizumab

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