Identification of HIF1A as a therapeutic target during SARS-CoV-2-associated lung injury

Abstract

Hypoxia-inducible factors (HIFs) promote lung protection and pathogen eradication during acute lung injury. We, therefore, tested the theory that pharmacologic stabilization of HIFs dampens lung injury during SARS-CoV-2 pneumonia. Initial studies in murine SARS-CoV-2 models showed improved outcomes after treatment with the FDA-approved HIF stabilizer vadadustat. Subsequent studies in genetic models implicated alveolus-expressed Hif1a in mediating lung protection. Therefore, we performed a randomized, double-blinded, multicenter phase II trial in patients admitted for SARS-CoV-2 infection and concomitant hypoxia (SpO2 ≤ 94%). Patients (n = 448) were randomized to oral vadadustat (900 mg/day) or placebo for up to 14 days. Safety events were similar between the 2 groups. Vadadustat treatment induced surrogate HIF target genes. The primary outcome of severe lung injury requiring high oxygen support on day 14 occurred in 43 patients in the vadadustat group and 53 patients in the placebo group (estimated probability, 13.3% vs. 16.9%). Among patients with baseline fraction of inspired oxygen of 80% or higher (n = 106), the estimated probability of the primary outcome was 12.1% (vadadustat) versus 79.1% (placebo), indicating an even greater benefit in patients with more severe baseline hypoxia. HIF1A is a likely therapeutic target during SARS-CoV-2-associated lung injury. Robust clinical trials of HIF stabilizers during pathogen-associated lung injury are warranted.

Keywords: Drug therapy; Hypoxia; Inflammation; Therapeutics.

Figure 1. The HIF stabilizer vadadustat provided lung protection during SARS-CoV-2 infection in mice.

(A) Schematic diagram of vadadustat treatment in HIF reporter ODD-luc mice. (B) Representative ex vivo bioluminescence imaging of luciferase activity using IVIS imager at 2 hours after vadadustat i.p. injection. (C) Quantification of bioluminescence intensity at 2 hours and 4 hours after vadadustat i.p. injection. Data are presented as mean ± SD. (D) Schematic diagram of vadadustat treatment in C57BL/6 mice. (E) Hif1A or Hif2A immunoblotting was performed on protein isolated from whole lung tissue after treatment with vadadustat. Each column represents 1 animal. (F and G) Quantification of Hif1a and Hif2a protein after treatment with vadadustat for 3 days. Data are presented as mean ± SD. (H) Schematic diagram of WA1 infection (280 PFU) in K18-hACE2 mice treated with vadadustat. (I) Kaplan-Meier plots of K18-hACE2 mice with vehicle or vadadustat treatment. P values were calculated with the Mantel-Cox test. (J) Blinded histological injury scores of the lungs were quantified as described in the Methods. Data are represented as mean ± SEM. (K) Representative H&E staining images of lung tissue from vehicle- and vadadustat-treated mice. Scale bars: 50 μm. (L) Schematic diagram of MA10 infection (200 PFU) in BALB/c mice treated with vadadustat. (M) Kaplan-Meier plots of BALB/c mice with vehicle or vadadustat treatment. P values were calculated with the Mantel-Cox test. *P < 0.05, **P < 0.01, ***P < 0.001 by 1-way ANOVA with Dunnett’s multiple-comparison test (C) or 2-tailed Student’s t test (F, G, and J).

Figure 2. Selective role of HIF1A-mediated lung protection during SARS-CoV-2 infection.

(A) Mice were inoculated with 3 × 10⁴ PFU of the murine-adapted SARS-CoV-2 strain (MA10) via oropharyngeal aspiration, and clinical outcomes were monitored over 7 days. (B and C) The survival rate in SARS-CoV-2–infected mice with whole-body deletion of Hif1a (Hif1a^fl/fl UBCCreER) or (C) Hif2a-deleted mice (Hif2a^fl/fl UBCCreER) compared to their respective Hif^fl/fl litter mates. P values were obtained using the Mantel-Cox test. (D) Mice with a specific deletion of Hif1a in alveolar epithelial cells (Hif1a^fl/fl SPCCreER) and their Cre-inducible counterpart (SPCCreER) were infected with 3 × 10³ PFU of the MA10 strain via oropharyngeal aspiration or mock infected, monitored for clinical outcomes and euthanized on day 4 to harvest BALF and lung tissue. (E) Albumin concentration in BALF was measured by ELISA. Data are represented as mean ± SEM. Two-tailed Student’s t test. (F and G) Viral load in BALF and lung tissue was detected by plaque assay. Gaussian distribution was assayed using the Shapiro-Wilk test. Unpaired 2-tailed Student’s t test or Mann-Whitney U test was applied to parametric or nonparametric data, respectively. (H) The lungs of infected SPCCreER and Hif1a^fl/fl SPCCreER mice 4 days after infection were collected, fixed, and paraffin embedded. H&E staining was performed, and images were taken at ×10 magnification (n = 5 or 8, respectively; representative images are shown). Scale bars: 200 μm. (I) The lung injury score was performed blindly. In the bar-and-whisker plots, the bounds of the boxes represent the 25%–75% interquartile range, the lines within the boxes represent the median, the whiskers represent data min/max, and there are no outlying values. Two-tailed Student’s t test. (J) Inflammatory molecules were measured using a multiplex array in the BALF from SPCCreER and Hif1a^fl/fl SPCCreER SARS-CoV-2– or mock-infected mice. Volcano plot resulting from an unpaired 2-tailed Student’s t test with Welch’s correction comparing both groups. Molecules that were highly differentially secreted are emphasized in red. The column graphs represent individual results for IL-6, G-CSF, and IP-10 (n = 10–12). Unpaired 2-tailed Student’s t tests with Welch’s correction or Mann-Whitney U test was applied to parametric or nonparametric data. Normality was established using the Shapiro-Wilk test. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 3. CONSORT diagram of study enrollment.

Figure 4. Vadadustat demonstrated a reassuring safety profile in hospitalized patients with SARS-CoV-2–associated lung injury.

Forest plot of the favorable safety profile of vadadustat (900 mg daily) compared with placebo. Each category presents the estimated relative risk (RR) and corresponding 95% Bayesian credible interval (CrI), with dots indicating point estimates and error bars representing the CrIs. The overall pooled safety estimate, shown by the diamond, demonstrated vadadustat’s comparable safety relative to placebo across all organ systems, adverse event grades, and relatedness categories. Importantly, no increased risk was observed for thromboembolic events or serious adverse events (grade 3 and higher). These findings provide strong reassurance regarding the safety of vadadustat in hospitalized patients.

Figure 5. Efficacy of vadadustat in hospitalized patients with SARS-CoV-2 infection and concomitant hypoxia.

(A) Erythropoietin level in the plasma measured by ELISA showing changes over 14 days. Mean slopes with 95% Bayesian credible intervals (CrIs) are shown, highlighting a 3-fold increase in erythropoietin levels in the vadadustat group (β = 3, 95% CrI = 1.94 to 4.03) compared with placebo (β = 0.85, 95% CrI = 0.61 to 1.08) with a >99% posterior probability (PP) of treatment and time interaction. (B) Proportion of patients with severe lung injury requiring high oxygen support (NIAID-OS score ≥ 6) on day 14 (primary outcome). The vadadustat group showed a reduced absolute probability (13.3%) compared with placebo (16.9%), with an absolute risk difference (ARD) of –3.6% (95% CrI = –8.4% to 0.9%). PPs of benefit for ARD < 0% and ARD ≤ –2.5% are 94% and 69%, respectively. (C) Proportion of patients with NIAID-OS score ≥ 6 on day 7 (key secondary outcome). Vadadustat treatment demonstrated a higher likelihood of clinical improvement compared with placebo, with an ARD of –4.2% (95% CrI = –9.0% to –0.1%) and a PP of benefit (ARD < 0%) of 97%. (D) Inflammatory mediators in the plasma measured by MILLIPLEX Multiplex Assays (IL-17E, IP-10, M-CSF, and TNF-α) between day 0 and day 7. Vadadustat treatment resulted in greater reductions in systemic inflammation compared with placebo, with significant differences noted for each marker. P values were obtained by Mann-Whitney U test. In the bar-and-whisker plots, the bounds of the boxes represent the 25%–75% interquartile range, the lines within the boxes represent the median, the whiskers represent data min/max, and there are no outlying values. (E) Subgroup analysis by FiO₂ levels upon hospital admission. PPs of clinical benefit for vadadustat (relative to placebo) were highest in patients with baseline FiO₂ ≥ 80% (PP > 99%), followed by lower probabilities for FiO₂ 60%–79% (PP = 92%), 40%–59% (PP = 95%), and <40% (PP = 66%).