Large‐scale electronic health record research introduces biases compared to traditional manually curated retrospective research. We used data from a community‐acquired pneumonia study for which we had a gold standard to illustrate such biases. The challenges include data inaccuracy, incompleteness, and complexity, and they can produce in distorted results. We found that a naïve approach approximated the gold standard, but errors on a minority of cases shifted mortality substantially. Manual review revealed errors in both selecting and characterizing the cohort, and narrowing the cohort improved the result. Nevertheless, a significantly narrowed cohort might contain its own biases that would be difficult to estimate.

Importance COVID-19 is causing high mortality worldwide. Developing models to quantify the risk of poor outcomes in infected patients could help develop strategies to shield the most vulnerable during de-confinement. Objective To develop and externally validate COVID-19 Estimated Risk (COVER) scores that quantify a patient's risk of hospital admission (COVER-H), requiring intensive services (COVER-I), or fatality (COVER-F) in the 30-days following COVID-19 diagnosis. Design Multinational, distributed network cohorts. Setting We analyzed a federated network of electronic medical records and administrative claims data from 13 data sources and 6 countries, mapped to a common data model. Participants Model development used a patient population consisting of >2 million patients with a general practice (GP), emergency room (ER), or outpatient (OP) visit with diagnosed influenza or flu-like symptoms any time prior to 2020. The model was validated on patients with a GP, ER, or OP visit in 2020 with a confirmed or suspected COVID-19 diagnosis across four databases from South Korea, Spain and the United States. Outcomes Age, sex, historical conditions, and drug use prior to index date were considered as candidate predictors. Outcomes included i) hospitalization with pneumonia, ii) hospitalization with pneumonia requiring intensive services or death, and iii) death in the 30 days after index date. Results Overall, 43,061 COVID-19 patients were included for model validation, after initial model development and validation using 6,869,127 patients with influenza or flu-like symptoms. We identified 7 predictors (history of cancer, chronic obstructive pulmonary disease, diabetes, heart disease, hypertension, hyperlipidemia, and kidney disease) which combined with age and sex could discriminate which patients would experience any of our three outcomes. The models achieved high performance in influenza. When transported to COVID-19 cohorts, the AUC ranges were, COVER-H: 0.73-0.81, COVER-I: 0.73-0.91, and COVER-F: 0.82-0.90. Calibration was overall acceptable, with overestimated risk in the most elderly and highest risk strata. Conclusions and relevance A 9-predictor model performs well for COVID-19 patients for predicting hospitalization, intensive services and death. The models could aid in providing reassurance for low risk patients and shield high risk patients from COVID-19 during de-confinement to reduce the virus' impact on morbidity and mortality.

Psoriasis vulgaris (PV) results from activation of IL‐23/Th17 immune pathway and is further amplified by cytokines/chemokines from skin cells. Among skin‐derived pro‐inflammatory cytokines, IL‐36 family members are highly upregulated in PV patients and play a critical role in general pustular psoriasis. However, there is limited data showing crosstalk between the IL‐23 and IL‐36 pathways in PV. Herein, potential attenuation of skin inflammation in the IL‐23‐induced mouse model of psoriasiform dermatitis by functional inhibition of IL‐36 receptor (IL‐36R) was interrogated. Anti‐mouse IL‐36R monoclonal antibodies (mAbs) were generated and validated in vitro by inhibiting IL‐36α‐induced secretion of CXCL1 from NIH 3T3 cells. Antibody target engagement was demonstrated by inhibition of CXCL1 production in a novel acute model of IL‐36α systemic injection in mice. In addition, anti‐IL‐36R mAbs inhibited tissue inflammation and inflammatory gene expression in an IL‐36α ear injection model of psoriasiform dermatitis demonstrating engagement of the target in the ear skin. To elucidate the possible role of IL‐36 signalling in IL‐23/Th17 pathway, the ability of anti‐IL‐36R mAbs to inhibit skin inflammation in an IL‐23 ear injection model was assessed. Inhibiting the IL‐36 pathway resulted in significant attenuation of skin thickening and psoriasis‐relevant gene expression. Taken together, these data suggest a role for IL‐36 signalling in the IL‐23/Th17 signalling axis in PV.

IL-36 cytokines signal through the IL-36 receptor (IL-36R) and a shared subunit, IL-1RAcP (IL-1 receptor accessory protein). The activation mechanism for the IL-36 pathway is proposed to be similar to that of IL-1 in that an IL-36R agonist (IL-36α, IL-36β, or IL-36γ) forms a binary complex with IL-36R, which then recruits IL-1RAcP. Recent studies have shown that IL-36R interacts with IL-1RAcP even in the absence of an agonist. To elucidate the IL-36 activation mechanism, we considered all possible binding events for IL-36 ligands/receptors and examined these events in direct binding assays. Our results indicated that the agonists bind the IL-36R extracellular domain with micromolar affinity but do not detectably bind IL-1RAcP. Using surface plasmon resonance (SPR), we found that IL-1RAcP also does not bind IL-36R when no agonist is present. In the presence of IL-36α, however, IL-1RAcP bound IL-36R strongly. These results suggested that the main pathway to the IL-36R·IL-36α·IL-1RAcP ternary complex is through the IL-36R·IL-36α binary complex, which recruits IL-1RAcP. We could not measure the binding affinity of IL-36R to IL-1RAcP directly, so we engineered a fragment crystallizable-linked construct to induce IL-36R·IL-1RAcP heterodimerization and predicted the binding affinity during a complete thermodynamic cycle to be 74 μm The SPR analysis also indicated that the IL-36R antagonist IL-36Ra binds IL-36R with higher affinity and a much slower off rate than the IL-36R agonists, shedding light on IL-36 pathway inhibition. Our results reveal the landscape of IL-36 ligand and receptor interactions, improving our understanding of IL-36 pathway activation and inhibition.

CD137 (TNFRSF9, 4-1BB) agonist antibodies (mAb) have demonstrated potent antitumor activity with memory response while causing hepatotoxicity in mouse models. In clinical trials, the degrees of liver toxicity of anti-CD137 vary from grade 4 transaminitis (urelumab) to nonexistent (utomilumab). To exploit the antitumor potential of CD137 signaling, we identified a new class of CD137 agonist mAbs with strong antitumor potency without significant transaminitis in vivo compared with CD137 agonists previously reported. These mAbs are crossreactive to mouse and cynomolgus monkey and showed crosslinking-dependent T-cell costimulation activity in vitro. Antitu-mor efficacy was maintained in Fc gamma receptor (FcgR) IIIdeficient mice but diminished in FcgRIIB-deficient mice, suggesting the critical role for FcgRIIB to provide cross-linking in vivo. Interestingly, a single dose of an affinity-reduced variant was sufficient to control tumor growth, but a higher affinity variant did not improve efficacy. These observations suggest that binding epitope and FcgR interaction, but not necessarily high affinity, are important for antitumor efficacy and reduced liver toxicity of CD137 mAb. Our study suggests the possibility of CD137 agonist therapy with improved safety profile in humans.

No abstract

Human NEMO (NF-κB essential modulator) is a 419 residue scaffolding protein that, together with catalytic subunits IKKα and IKKβ, forms the IκB kinase (IKK) complex, a key regulator of NF-κB pathway signaling. NEMO is an elongated homodimer comprising mostly α-helix. It has been shown that a NEMO fragment spanning residues 44–111, which contains the IKKα/β binding site, is structurally disordered in the absence of bound IKKβ. Herein we show that enforcing dimerization of NEMO1–120 or NEMO44–111 constructs through introduction of one or two interchain disulfide bonds, through oxidation of the native Cys54 residue and/or at position 107 through a Leu107Cys mutation, induces a stable α-helical coiled-coil structure that is preorganized to bind IKKβ with high affinity. Chemical and thermal denaturation studies showed that, in the context of a covalent dimer, the ordered structure was stabilized relative to the denatured state by up to 3 kcal/mol. A full-length NEMO-L107C protein formed covalent dimers upon treatment of mammalian cells with H2O2. Furthermore, NEMO-L107C bound endogenous IKKβ in A293T cells, reconstituted TNF-induced NF-κB signaling in NEMO-deficient cells, and interacted with TRAF6. Our results indicate that the IKKβ binding domain of NEMO possesses an ordered structure in the unbound state, provided that it is constrained within a dimer as is the case in the constitutively dimeric full-length NEMO protein. The stability of the NEMO coiled coil is maintained by strong interhelix interactions in the region centered on residue 54. The disulfide-linked constructs we describe herein may be useful for crystallization of NEMO’s IKKβ binding domain in the absence of bound IKKβ, thereby facilitating the structural characterization of small-molecule inhibitors.

TRAIL can activate cell surface death receptors, resulting in potent tumor cell death via induction of the extrinsic apoptosis pathway. Eftozanermin alfa (ABBV-621) is a second generation TRAIL receptor agonist engineered as an IgG1-Fc mutant backbone linked to two sets of trimeric native single-chain TRAIL receptor binding domain monomers. This hexavalent agonistic fusion protein binds to the death-inducing DR4 and DR5 receptors with nanomolar affinity to drive on-target biological activity with enhanced caspase-8 aggregation and death-inducing signaling complex formation independent of FcγR-mediated cross-linking, and without clinical signs or pathologic evidence of toxicity in nonrodent species. ABBV-621 induced cell death in approximately 36% (45/126) of solid cancer cell lines in vitro at subnanomolar concentrations. An in vivo patient-derived xenograft (PDX) screen of ABBV-621 activity across 15 different tumor indications resulted in an overall response (OR) of 29% (47/162). Although DR4 (TNFSFR10A) and/or DR5 (TNFSFR10B) expression levels did not predict the level of response to ABBV-621 activity in vivo, KRAS mutations were associated with elevated TNFSFR10A and TNFSFR10B and were enriched in ABBV-621–responsive colorectal carcinoma PDX models. To build upon the OR of ABBV-621 monotherapy in colorectal cancer (45%; 10/22) and pancreatic cancer (35%; 7/20), we subsequently demonstrated that inherent resistance to ABBV-621 treatment could be overcome in combination with chemotherapeutics or with selective inhibitors of BCL-XL. In summary, these data provide a preclinical rationale for the ongoing phase 1 clinical trial (NCT03082209) evaluating the activity of ABBV-621 in patients with cancer. Significance: This study describes the activity of a hexavalent TRAIL-receptor agonistic fusion protein in preclinical models of solid tumors that mechanistically distinguishes this molecular entity from other TRAIL-based therapeutics.