We report evidence of an age-dependent allelic association between interferon-induced transmembrane protein 3 (rs12252 allele) and severity of coronavirus disease 2019, highlighting the need for further studies and the potential for a personalized, genotype-based approach to identifying high-risk individuals.
Coronavirus disease 2019 (COVID-19) is a newly recognized illness caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which firstly emerged in December 2019 in the city of Wuhan in Hubei province, China. It has spread rapidly throughout Wuhan to other provinces in China and around the world. As of 29 February 2020, a cumulative total of 79 394 COVID-19 cases were reported in China, which led to 2838 deaths, according to the report of the World Health Organization–China Joint Mission on Coronavirus Disease 2019 (COVID-19) . There is an urgent unmet need to understand why severe disease develops in some people. Heritable differences are known to modulate individual susceptibility to and severity of infectious disease [2, 3].
We hypothesized that a genetic variant of the interferon-induced transmembrane protein 3 (IFITM3) gene (IFITM3) was associated with the severity of COVID-19, specifically the single-nucleotide polymorphism rs12252. This genetic variant is common in Asian populations, and homozygosity for the C allele has been associated with influenza severity . IFITM3 encodes an immune effector protein critical to viral restriction and acts to restrict membrane fusion . It is currently unknown whether IFITM3 shows genetic association with the severity of COVID-19.
To investigate whether the IFITM3 variant was associated with disease severity, we recruited a cohort of hospitalized patients with laboratory-confirmed COVID-19. These comprised a consecutive series of 80 patients admitted to Beijing Youan Hospital during January to and 2020. We phenotyped the patients into mild and severe disease groups, according to clinical definitions from the National Health Commission of China .
Patients were recruited from Beijing Youan Hospital, Capital Medical University, Beijing, between January 2020 and February 2020. All the candidates were hospitalized patients with laboratory-confirmed COVID-19. Their clinical data and whole blood samples were collected. The study was approved by Beijing Youan Hospital’s institutional review board. Written informed consent was obtained from all patients.
COVID-19 was diagnosed according to the diagnosis and treatment of COVID-19 recommended by the National Health Commission of China . Laboratory confirmation was defined as a positive result at high-throughput sequencing or real-time reverse-transcription polymerase chain reaction assay of nasal and pharyngeal swab specimens . Mild disease was defined as COVID-19 with the presence of fever, respiratory symptoms, and pneumonia at imaging. Severe disease was defined as COVID-19 with the additional presence of significant respiratory distress (respirations, >30/min), blood oxygen saturation <93%, ratio of arterial oxygen pressure to fraction of inspired oxygen <300 mm Hg, respiratory failure with mechanical ventilation, shock, or other organ failure requiring intensive care in the intensive care unit .
Genomic DNA was extracted from peripheral blood mononuclear cells using the PureGene DNA Isolation Kit (Gentra System). The region encompassing the human IFITM3 rs12252 sequence was amplified, sequenced, and analyzed as described elsewhere [4, 8].
Statistical analysis was performed using χ 2 and Fisher exact tests. Student t tests were used to compare values between mild and severe disease groups where data were normally distributed (evaluated with Kolmogorov-Smirnov test), and nonparametric t tests (Mann-Whitney tests) were used where data were not normally distributed. Statistical test differences were considered significant when P values were < .05. Analyses were performed using GraphPad Prism, version 7 (GraphPad Software). The age effect was determined using a linear model approach with patient age as dependent variable and disease severity as independent variable. The associations between categorical variables (age groups or genotypes) and disease severity were assessed using logistic regression. The analyses were performed using R software (version 3.6.0). Differences were considered statistically significant at P < .05.
Demographic and phenotypic data are shown in Table 1. Mild disease (n = 56) was defined as the presence of fever, respiratory symptoms, and pneumonia seen with imaging. In patients with severe disease (n = 24), significant tachypnea, hypoxia, respiratory failure, or other organ failure also developed. The patient cohort was broadly representative of published clinical reports in the outbreak to date in terms of case mix and severity . We found that blood oxygen saturation was significantly lower in patients with severe disease (77% vs 96.7% in patients with mild disease; P = .001), and the ratio of arterial oxygen partial pressure to the fraction of inspired oxygen was significantly lower (223.5 ± 45.8 vs 466.7 ± 135.6 mm Hg, respectively; P < .001). Seven patients in the severe disease group received intensive care in the intensive care unit; 6 were mechanically ventilated, and 3 of the 6 died.
|Characteristic||Patients Hospitalized With COVID-10, No. (%)s||P Valueb|
|Total (N = 80)||Mild Disease (n = 56)||Severe Disease (n = 24)||Died (n = 3)|
|Age, median (IQR), y||49.5 (37.75–63.5)||43.5 (34–56.5)||67.5 (57.75–74.25)||86 (80.5–88.5)||<.001|
|Male sex||33 (41.25)||24 (42.86)||9 (37.5)||1 (33.3)||.66|
|Diabetes||9 (11.25)||5 (8.93)||4 (16.67)||1 (33.33)||.44|
|Hypertension||21 (26.25)||9 (16.07)||12 (50)||2 (66.67)||.002|
|Cancer||4 (5)||2 (3.57)||2 (8.33)||0 (0)||.58|
|Chronic liver disease||2 (2.5)||2 (3.57)||0 (0)||0 (0)||>.99|
|Fever||62 (77.50)||41 (73.21)||21 (87.50)||2 (66.67)||.16|
|Cough||51 (63.75)||33 (58.93)||18 (75.00)||2 (66.67)||.17|
|Expectoration||26 (32.50)||15 (26.79)||11 (45.83)||2 (66.67)||.10|
|Vomiting||1 (1.25)||0 (0)||1 (4.17)||0 (0)||.30|
|Diarrhea||1 (1.25)||1 (1.79)||0 (0)||0 (0)||>.99|
|Physiological variables, median (IQR)|
|Respirations/min||20 (20–21)||20 (20-20)||21 (20–24.75)||23 (20–25)||<.001|
|Sao2 (n = 56)||94.95 (88.125–97.625)||97.2 (95.5–98.1)||88.0 (79.6–90.9)||79.6 (77.6–80.3)||<.001|
|Pao2/Fio2 (n = 45)||386.5 (261.85–472.0)||449.5 (379.1–494.3)||211.35 (192–260)||193.8 (187–200)||<.001|
|ICU admission||7 (8.75)||0 (0)||7 (29.17)||3 (100)||<.001|
|Mechanical ventilation||6 (7.5)||0 (0)||6 (25.00)||3 (100)||<.001|
|Death within 28 d||3 (3.75)||0 (0)||3 (12.5)||3 (100)||.02|
We found evidence of a clear relationship between age and disease severity. The median age was 49.5 years (interquartile range, 37.75–63.5; Table 2), and comparing patients with mild or severe disease, the median age increased from 43.5 to 67.5 years (P < .001; β = 20.8; Supplementary Figure 1). We observed that 14 (58.3%) of the 24 patients with severe disease (including all patients who died) were >63.5 years old (third quantile), compared with only 6 (10.7%) of the 56 patients with mild disease (P < .001; odds ratio [OR], 11.7; Table 2). The age-related difference in disease severity was also profound when we restricted the analyses to the median age (≥49.5 or <49.5 years) of the cohort (P < .001; OR, 13.6).
|Characteristics||All patients (n = 80)||Mild (n = 56)||Severe (n = 24)||Died (n = 3)||P valuea||ORa||P valueb||ORb|
|Age, median (IQR), y||49.5 (37.75–63.5)||43.5 (34–56.5)||67.5 (57.75–74.25)||86 (80.5–88.5)||4.04E–05||11.67|
|≥63.5 (third quantile)||20 (25)||6 (10.7)||14 (58.3)||3 (100)|
|<63.5||60 (75)||50 (89.3)||10 (41.7)||0 (0)|
|Male||33 (41.25)||24 (42.86)||9 (37.5)||1 (33.3)|
|Female||47 (58.75)||32 (57.14)||15 (62.5)||2 (66.7)|
|CC||28 (35)||16 (28.57)||12 (50)||2 (66.7)|
|CT||37 (46.25)||30 (53.57)||7 (29.17)||1 (33.3)|
|TT||15 (18.75)||10 (17.86)||5 (20.83)||0 (0)|
To determine whether the homozygous C-allele carriers associate with the severity of COVID-19, we genotyped the cohort by sequencing a 300–base pair locus spanning rs12252. Among all hospitalized patients, we found that 35% were homozygous for the CC allele (46.25% CT heterozygotes and 18.75% TT homozygotes). Adjusting for age on regression analysis, we found a significant difference between mild and severe disease groups, with an association between homozygosity for the C allele (CC vs CT/TT) and disease severity (P <.0001; OR, 6.37; Table 2). In addition, 2 of the 3 patients who died carried the CC genotype. However, the frequency of CC genotype (28.6%) observed in our mild patient groups is similar to that in general Beijing population (26.2%), according to the International Genome Sample Resource (IGSR) (https://www.internationalgenome.org/1000-genomes-browsers).
These findings require further validation but represent important early evidence in a carefully phenotyped cohort for a role of IFITM3 in the severity of COVID-19. Our findings support a hypothesis that host defects in control of intracellular viral replication may result in more prolonged and severe disease. The mechanism of action of this synonymous variant of IFITM3 is still unknown; a truncated version of IFITM3 suggested by earlier work on influenza is unlikely to lead to the differences in viral restriction previously seen . Larger cohort studies are needed to confirm this genetic association with severity of COVID-19, but the data presented here support the need for such genetic analyses in different global populations, with the opportunity for early targeted intervention in persons at risk if robust genetic biomarkers can be established and potential therapeutic targets highlighted.
Author contributions. Study conception and design: Y. Zhang, T. D., and R. J. Performance of experiments and data analysis: Y. Zhang, L. Q., Y. Zhao, P. Z., and J. C. K. Clinical sampling and data collection: B. X., K. L., L. L., C. Z., Y. D., Y. F., J. S., Z. H., H. X., and R. J. Writing, review, and/or revision of the manuscript: Y. Zhang, J. C. K., and T. D.
Financial support. This work was supported by the Beijing Natural Science Foundation (grants 7191004 and 7202069); the Beijing Municipal Science & Technology Commission (grant Z171100001017078); the Beijing Municipal Administration of Hospitals (grants DFL20181701 and ZYLX201711); the Beijing Key Laboratory (grant BZ0373); the National Key Research and Development Project (grant 2020YFC0841700); the Chinese Academy of Medical Sciences Innovation Fund for Medical Sciences (grant 2018-I2M-2-002); the Medical Research Council (MRC), United Kingdom (grant MR/L018942/1 to T. D. and the MRC Human Immunology Unit Core); the Nuffield Department of Medicine, Oxford University (T. D.); the Wellcome Trust (investigator award (204969/Z/16/Z to J. C. K.); and the NIHR Oxford Biomedical Research Centre with Wellcome Trust (grants 090532/Z/09/Z and 203141/Z/16/Z to J. C. K. and core facilities of the Wellcome Centre for Human Genetics).
Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.