The association of hypertension and diabetes pharmacotherapy with COVID-19 severity and immune signatures: an observational study

Abstract

Coronavirus disease 2019 (COVID-19) in patients with co-existing diabetes and hypertension is associated with an increased risk of severe infections and mortality.1 There is a paucity of data on the association of pharmacotherapy of these conditions with COVID-19 disease severity. We conducted a retrospective, observational cohort study of 717 patients with PCR-confirmed COVID-19 who were hospitalized at the National Centre of Infectious Diseases (NCID), Singapore up to 15 April 2020. Data collected included demographics, comorbidities, concomitant medications, and clinical outcomes. Primary outcomes were hypoxia (requirement for supplemental oxygen to maintain blood oxygen saturations >93%), intensive care unit (ICU) admission, mechanical ventilation, or death. The study was conducted in accordance with institutional guidelines; study protocols were reviewed and approved by the Ministry of Health, Singapore and the institutional ethics committee (reference 2012/00917). Plasma immune mediator concentrations were measured using multiplex microbead-based immunoassay: Cytokine/Chemokine/Growth Factor 45-plex Human ProcartaPlex™ panel 1 (ThermoFisher Scientific). We used the modified Poisson regression approach2 to calculate the relative risk (RR) for the association between requirement for supplementary oxygen, ICU admission, mechanical ventilation, and death with diabetes, hypertension, and pharmacotherapeutics. RR was adjusted for demographics, comorbidities, and co-medications. Multivariable linear regression models were used to assess the association between inflammatory markers and the use of medications. All statistical tests were two-sided, and statistical significance was taken as P < 0.05. All statistical analyses were performed using Stata version 15. Cytokine levels across groups were compared using two-tailed Mann–Whitney U-test. Statistical analyses and data plotting were performed using GraphPad Prism (Version 8.2.1) In this cohort, 139 (19.4%) had hypertension and 76 (10.6%) had type 2 diabetes mellitus (Table 1). Hypoxia was reported in 91 (12.7%), ICU admission in 47 patients (6.6%), mechanical ventilation in 25 (3.5%), and 12 patients (1.67%) died. Association of the use of antihypertensives and oral hypoglycaemic agents with clinical indicators of COVID-19 severity in patients with hypertension (n = 139) and diabetes (n = 76) Hypertension: Adjusted for age, gender, ethnicity, use of other antihypertensive medications, diabetes medications, and statins. Since no patients on an ACE-I was on an ARB, and vice versa, these were evaluated in two separate models including only ACE-I or ARB. Diabetes: Adjusted for age, gender, ethnicity, use of other antihypertensive medications, other diabetes medications (except metformin), and statins. Additionally adjusted for HbA1c, systolic and diastolic blood pressure, and body mass index. RR, relative risk; aRR, adjusted relative risk; CI, confidence interval. P < 0.05. Diabetes and hypertension were associated with hypoxia [diabetes, adjusted RR (aRR) 1.76, 95% confidence interval (CI) 1.18–2.63; hypertension, aRR 1.96, 95% CI 1.25–3.07] and ICU admission (diabetes, aRR 2.17, 95% CI 1.23–3.83; hypertension, aRR 2.23, 95% CI 1.17–4.24). In the hypertension subgroup, angiotensin-converting enzyme inhibitor (ACE-I) treatment was associated with a lower risk of ICU admission (aRR 0.26, 95% CI 0.10–0.68) and mechanical ventilation (aRR 0.09, 95% CI 0.02–0.36), whereas angiotensin receptor blockers (ARBs) were associated with a higher risk of ICU admission (aRR 2.19, 95% CI 1.08–4.43) (Table 1). In the diabetes subgroup, patients receiving a dipeptidyl peptidase 4 inhibitor (DPP-4i) were at higher risk of ICU admission (aRR 4.07, 95% CI 1.42–11.66) (Table 1). Sodium–glucose co-transporter 2 inhibitor (SGLT2i) use was associated with a marginally lower risk of mechanical ventilation (aRR 0.03, 95% CI 0.00–0.70). Sensitivity analysis adjusting for body mass index (BMI), glycated haemoglobin (HbA1c), and systolic and diastolic blood pressure at baseline showed similar results (Table 1). The results also remained materially unchanged when the analysis was confined to patients on metformin. Since 88% of patient with diabetes received metformin, it was not possible to analyse its effects. In the hypertension subgroup, the use of an ARB was associated with higher C-reactive protein concentrations (β = 19.0; P = 0.038), white blood cell counts (β = 1.14, P = 0.006), and neutrophil counts (β = 1.24, P = 0.002) when adjusted for age, gender, and ethnicity. Diabetes patients on a DPP-4i had significantly lower brain-derived neurotrophic factor (BDNF). Hypertension patients on ARB treatment had significantly higher monocyte chemoattractant protein 1 (MCP-1) and interferon-γ (IFN-γ)-induced protein 10 (IP-10) concentrations compared with non-ARB users (Figure 1). Immune signatures of COVID-19 patients with diabetes mellitus and hypertension treated with a DPP-4i, ACE-I, or ARB. Plasma fractions were isolated from the blood of COVID-19 patients without diabetes mellitus (DM) and hypertension (HTN) (n = 12), with DM (n = 17; 7 patients on DPP-4i treatment), or with HTN (n = 19; 2 patients on ACE-I treatment and 7 patients on ARB treatment). Concentrations of immune mediators were quantified using a 45-plex microbead-based immunoassay. Cytokine levels were measured on the first plasma samples collected upon hospitalization [median 8 days post-illness onset (PIO)]. Immune mediator profiles that are differently regulated in patients receiving antidiabetic (DPP-4i) or antihypertensive (ACE-I or ARB) treatment are illustrated as scatter plots. Statistical analyses were performed using Mann–Whitney U-test (*P < 0.05). Cytokine levels of patient samples that are not detectable are presented as the value of logarithmic transformation of limit of quantification (LOQ). The cytokine level for healthy controls (HC) (n = 23) is indicated by the blue dotted line. Longitudinal profiling of these cytokines showed that IFN-γ, interleukin (IL)-18, IL-15, and BDNF trends were mixed during disease progression in diabetes patients with no differences with treatment, whilst MCP-1 and IP-10 concentrations continued to remain significantly higher in hypertensive ARB users vs. non-users. We found that patients with COVID-19 and diabetes or hypertension had more severe infections with a higher requirement for oxygen and ICU admission. These results are similar to those of other larger observational studies reported globally.1 As severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) uses angiotensin-converting enzyme-2 (ACE2) for entry into alveolar cells,3 there has been concern about the effect of renin–angiotensin system (RAS) modulators on COVID-19 severity. The concerns about possible increased viral transmission due to up-regulation of ACE2 enzyme expression has been recently debunked as a large population-based study showed no correlation with viral transmission.4,5 In the hypertension cohort, ACE-I use was associated with better outcomes. In a recent meta-analysis, which included 16 studies involving 24 676 patients, the authors similarly found that the use of an ACE-I was associated with better disease outcomes.6 Conflicting results have been reported from various centres, as highlighted in another meta-analysis.7 Many studies have combined ACE-inhibitors and ARBs into one group as they have patients on a combination and thus they are unable to study the effects individually. Our cohort did not have a single patient who was taking both an ACE-I and an ARB, allowing us to individually assess these medications. Since ACE-inhibitors block the formation of angiotensin II completely, it can lead to a counter up-regulation of the ACE2 pathway which can protect against respiratory failure.8 On the other hand, ARBs block the AT1R with a concomitant increase in angiotensin II. Angiotensin II can act through other receptors such as AT3R and AT4R which are not completely blocked by ARBs.9 Serum angiotensin I concentrations have been seen to correlate with viral load and lung injury in COVID-19.10 Moreover, angiotensin II can induce endothelial cell synthesis of MCP-1 and IP-10, proinflammatory cytokines which have been identified to be strongly associated with severe COVID-19.10,11 This is consistent with our observations that the patients receiving ARB treatment have higher levels of plasma MCP-1 and IP-10 compared with patients receiving ACE-I treatment or neither treatment. In the diabetes cohort, we found that patients on DPP-4 inhibitors were more likely to require ICU admission. In COVID-19, the adaptive T-cell immune response is thought to play a significant role in determining disease severity. If the T-cell and B-cell response are not effective, immune cells accumulate in the lungs, causing overproduction of proinflammatory cytokines and damaging inflammation.12DPP-4 inhibitors have been associated with the suppression of T-cell proliferation response which may result in a less effective immune response to the virus.13 In support of this hypothesis, we did observe a significant decrease in BDNF, which is important for T-cell maturation (Figure 1). In BDNF knockout mice and in conditional knockout mice lacking BDNF (specifically in this lymphoid subset), diminished T-cell cellularity in peripheral lymphoid organs has been demonstrated suggestive of a critical paracrine and autocrine role in thymocyte development.14 This study is limited by its single-centre, observational study design, and relatively small sample size. We found a beneficial signal for ACE-Is and a deleterious signal for ARBs and DPP-4-inhibitors with evidence of deleterious immune signatures. Data from larger observational cohorts and ongoing randomized controlled trials are urgently needed to confirm or refute these findings. Author contributions: R.D. conceptualized the study, performed data analysis, data interpretation, and literature review, and wrote the manuscript. W.T. and B.E.Y. conceptualized the study, performed data collection, and critically reviewed the manuscript. S.W.F., W.C.T., Y.H.C., L.R., and L.N. performed the microbead immunoassay, and analysed, interpreted, and reported the results for the T-cell response. L.W.A. performed data analysis and interpretation. D.E.K.C. performed data interpretation and critical review of the manuscript. D.C.L. conceptualized the study, and was involved in data interpretation and critical review of the manuscript. All authors reviewed the final manuscript. We thank Professor Olaf Rötzschke, Dr Bernett Lee, Wilson How, and Norman Leo Fernandez from the Singapore Immunology Network (SIgN) for their help in running the multiplex microbead immunoassay. We are also grateful to Dr Danielle Anderson and her team at Duke-NUS, for their technical assistance in virus inactivation procedures with Triton™ X-100 (ThermoFisher Scientific). The study is funded through the NMRC COVID-19 research fund (COVID19RF-001). The multiplex microbead-based immunoassays were supported by A*STAR COVID-19 Research funding (H/20/04/g1/006) provided to the Singapore Immunology Network (SIgN) by the Biomedical Research Council (BMRC), A*STAR, and by a National Research Foundation grant (#NRF2017_SISFP09) to the SIgN Immunomonitoring platform. R.D. is supported in part by a Ministry of Health, Clinician Scientist Award [MOH-000014] and National Medical Research Council Centre Grant [NMRC/CG/017/2013]. Conflict of interest: none declared. Data availability: The datasets are available from the corresponding author on reasonable request.

Topics & Concepts