Introduction
The coronavirus disease 2019 (COVID-19) outbreak has spread across the world with more than 5.1 million confirmed cases [
1]. Early reports on clinical characteristics have identified hypertension as the most common comorbidity reported in COVID-19 patients [
2–
5]. Zhou
et al. documented pre-existing hypertension as a risk factor for in-hospital death of COVID-19 [
6]. This observation is of great interest since the renin-angiotensin system (RAS) is an important component in regulating blood pressure, and the transmembrane angiotensin-converting enzyme 2 (ACE2), which catalyzes the cleavage of angiotensin II (a vasoconstrictor) into vasodilatory peptides, is also the cellular entry receptor for SARS-CoV-2.
Given the common use of angiotensin-converting enzyme inhibitors (ACEIs) and angiotensin II receptor blockers (ARBs) in patients with cardiovascular diseases [
7], there has been wide discussion whether administration of ACEI/ARB might modulate COVID-19 disease severity. Concerns have been raised whether hypertensive patients should stop taking ACEI/ARB and switch to other anti-hypertensive drugs in the context of COVID-19 pandemic [
8,
9]. There have also been counter-arguments in favor of the use of ACEI/ARB in COVID-19 patients, based on previous studies showing that ACEI/ARB are associated with improvements in pneumonia-related outcomes [
10,
11], as well as a protective role of ARBs in SARS-CoV and influenza infected animals [
12,
13].
The renin-angiotensin system (RAS) plays a central role in regulating blood pressure and maintaining hemodynamic homeostasis. In hypovolemic conditions, angiotensinogen is serially cleaved by renin and angiotensin-converting enzyme (ACE) into angiotensin I (Ang I) and angiotensin II (Ang II), respectively. Ang II then acts on the cellular receptor angiotensin II receptor I (AT1) to constrict the blood vessels and increases blood pressure (Fig. 1). To counteract this vaso-constrictive effect of RAS, the human body has developed another system in an exquisite “
yin-yang” balance. The ACE2 is a homolog of ACE [
14,
15] that metabolizes Ang II into angiotensin-(1–7) (Ang
1–7), which dilates the blood vessels and counter-balance ACE/Ang II in blood pressure regulation [
16] (Fig. 1). Of note, in a rat model of lipopolysaccharide-induced acute respiratory distress syndrome (ARDS), reduced pulmonary levels of Ang
1–7 contribute to disease pathogenesis, and administration of this peptide or the ARB losartan reduces the development of ARDS [
17]. The non-angiotensinase function of ACE2 was recognized when it was identified to be cellular receptor for severe acute respiratory syndrome coronavirus (SARS-CoV) [
18], as well as more recently for SARS-CoV-2 [
19–
21]. ACE2 expression has been documented in various human organs, including oral and nasal mucosa, lung, intestine, kidney, heart, arterial/venous endothelium and others [
22–
25].
High-quality clinical evidence for ACEI/ARB usage in COVID-19 patients is insufficient to draw definitive conclusions whether chronic ACEI/ARB should be continued or stopped [
26]. Several recent retrospective, hospital-based studies have reported no adverse or marginal protective effect of ACEI/ARB on the clinical outcomes of COVID-19 patients [
27–
32]. To contribute to this discussion, we have performed a secondary analysis on a previously-reported single-center retrospective cohort of COVID-19 patients in Wuhan, China [
6].
Materials and methods
Study design and data collection
We previously collected data from 702 hospitalized adult COVID-19 patients admitted to Jinyintan Hospital (Wuhan, China), who had definite outcome (dead or discharged) between December 29, 2019 and February 15, 2020 [
6]. Data were extracted from medical records into an electronic data collection form by two independent researchers. Their electronic medical records were screened for medical history of hypertension in past medical history or admitting diagnosis and for oral anti-hypertensive treatment before and during hospitalization. The anti-hypertensive treatments we included in the analysis were ACEI, ARB,
b-blockers, calcium channel blockers (CCBs), diuretics, and others (
a-blockers and traditional Chinese medicine). Patients with definitive long-term anti-hypertensive treatments, as either recorded in the medical history or in the prescribed medication chart as standing order during hospitalization, were included in the final analysis. The comorbidities were either self-reported by the patients or evaluated by attending physicians and recorded in admitting diagnosis.
The study was approved by the Research Ethics Commission of Jinyintan Hospital (KY-2020-01.01) and the informed consent was waived by the Research Ethics Commission.
Definitions
Fever was defined as axillary temperature of at least 37.3 °C. Sepsis and septic shock were evaluated based on Sepsis-3 International Consensus [
33]. ARDS was diagnosed according to the Berlin Definition [
34]. Acute kidney injury was defined according to the KDIGO clinical practice guidelines [
35]. Diagnosis of acute cardiac injury was made if serum levels of high-sensitive cardiac troponin I was above the 99th percentile upper reference limit [
36]. Secondary infection, hypoproteinemia, and coagulopathy were defined as previously [
6].
Statistical analysis
Continuous variables were expressed as median (interquartile range (IQR)) and categorical variables as number (proportion). Two-group comparisons (ACEI/ARB vs. non-ACEI/ARB) were conducted with Mann–Whitney U test or c2/Fisher exact test where appropriate. The risk of death, ICU admission, invasive mechanical ventilation and corresponding odds ratio both with and without adjustment for age and gender were calculated by Logistic model, comparing patients in ACEI/ARB group with those in non-ACEI/ARB group. Four previously-reported risk factors for death (age, history of diabetes, qSOFA score, and D-dimer) were included in univariable and multivariable analysis with Logistic model. A 2-sided a less than 0.05 was considered statistically significant. Statistical analyses were conducted using SAS software (version 9.4, SAS Institute) and SPSS software (version 25, IBM).
Results
Patients and clinical characteristics
A total of 702 adult laboratory-confirmed COVID-19 patients admitted to Jinyintan Hospital, who had definite clinical outcome by February 15, 2020, were screened in this study, of whom 188 patients had a medical history of hypertension. Among them, 101 patients had anti-hypertensive treatment medications recorded in their medical records and were included in the final analysis. The schematic of patient selection is shown in Fig. 2. Patients who took ACEI or ARB as part of their anti-hypertensive therapy were included in the ACEI/ARB cohort (n = 40), and those without were categorized in the non-ACEI/ARB cohort (n = 61).
Patients in ACEI/ARB group and non-ACEI/ARB group were comparable in age (median, 66.5 vs. 65.0 years, P = 0.80) and gender distribution (male, 48% vs. 56%, P = 0.42), as shown in Table 1. They were also in similar phase of the disease with median time from symptom onset to admission being 12.5 days (IQR 8.5–16.0 days) and 13.0 days (IQR 8.5–15.5 days), respectively. In addition to hypertension, patients in two groups also shared similar proportions of comorbidities and smoking history.
The vital signs and laboratory test results on admission were generally similar between the two groups (Table 1). Comparable proportions of patients in two groups had lymphocyte count less than 0.8 × 109/L (55% vs. 48%, P = 0.51), increased serum cardiac troponin I (18% vs. 19%, P = 0.88), and increased serum D-dimer levels (59% vs. 58%, P = 0.95). The percentages of heart failure, defined as brain natriuretic peptide (BNP) value≥100 pg/mL on admission, were also similar among the two groups (28% vs. 27%, P = 0.93). Values of the inflammatory markers, including serum ferritin, procalcitonin, and interleukin-6 were not different between the two groups, nor were chest imaging features distinguishable. The sequential organ failure assessment (SOFA) scores in ACEI/ARB group (median 2.0, IQR 1.0–4.0) were numerically smaller than those in non-ACEI/ARB group (median 4.0, IQR 2.0–4.5), although the difference is not statistically significant (P = 0.29).
Anti-hypertensive treatment
Among the 101 patients recorded with anti-hypertensive medication, 40 cases had therapies based on ACEI/ARB, either as mono-therapy or combined with b-blockers, CCB, thiazide diuretics or spironolactone (Tables 2 and 3). No usage of sacubitril-valsartan was reported in this cohort. The non-ACEI/ARB therapy was mostly based on calcium channel blockers (Tables 2 and 3). The usage of other anti-hypertensive medications (a-blocker, b-blocker, and diuretics) was comparable between the two groups (Table 2). Similar percentages of patients were under optimal blood pressure control based on the vital signs taken on admission (Table 1). Chronic treatment on RAS antagonists was not associated with statistically significant changes in serum electrolyte (sodium and potassium) or serum creatinine levels (Table 1).
Association between ACEI/ARB usage with severe clinical outcome of COVID-19
Compared with the hypertensive patients who were not given ACEI/ARB, we observed no significant differences in percentages of in-hospital mortality (28% vs. 34%,
P = 0.46), ICU admission (20% vs. 28%,
P = 0.37), or mechanical ventilation (18% vs. 26%,
P = 0.31) in those receiving a regimen that included an ACEI or ARB. Of note, the percentages with these major outcomes were uniformly smaller in ACEI/ARB group. Similarly, numerically fewer patients in ACEI/ARB group developed respiratory failure (43% vs. 59%,
P = 0.10), sepsis (48% vs. 59%,
P = 0.26) or ARDS (28% vs. 36%,
P = 0.37) during hospitalization. Hospital length of stay and duration of viral shedding for discharged patients were similar between ACEI/ARB and non-ACEI/ARB groups (Table 4). We did not include the non-survivors in the analysis for viral-shedding duration because it has been previously reported that non-survivors had prolonged viral shedding until death [
6].
Further multivariable analysis adjusted for age and gender did not provide statistical evidence for association between ACEI/ARB treatment and severe clinical outcome in COVID-19 patients (Table 5). As previously reported in the total population [
6], older age, increased D-dimer values (>1µg/mL) and higher qSOFA scores were associated with higher risks of in-hospital mortality in hypertensive COVID-19 patients in the Logistic model (Table 6).
Discussion
The hypothesis that ACEI or ARB treatment changes COVID-19 disease progression was based in part on potential impact of RAS antagonists on ACE2 expression. Theoretically, does administration of an ACEI or ARB have an impact on ACE2 expression or activity? First of all, it has been shown that ACEIs do not affect ACE2 activity [
15]. In terms of ACE2 expression, olmesartan, but not other ACEIs, ARBs or calcium channel blockers, has been found to increase urinary ACE2 levels in hypertensive patients [
37,
38]. Also, ACEI treatment has been reported to increase serum ACE2 activity in diabetic patients [
39]. Yet the soluble ACE2 (sACE2) levels do not always reflect the tissue ACE2 expression in lung or other organs. In fact, it is reasonable to hypothesize that sACE2 may even negatively correlate with tissue ACE2 since sACE2 is cleaved from the membrane-bound enzyme before being released into the blood stream. Also, it has to be noted that increase in ACE2 enzymatic activity does not always match the upregulation of ACE2 expression [
40,
41]. Therefore, change in pulmonary tissue ACE2 after ACEI or ARB treatment has to be directly studied to provide answers for this question, although studies of upper respiratory tract ACE2 expression or ACE2 enzyme activity in bronchoalveolar lavage fluid might provide indirect evidence [
42,
43].
Hypertensive animal models might be a surrogate before clinical evidence is available. In rat cardiac myocytes, Ang II significantly reduces ACE2 activity and downregulates ACE2 mRNA expression; these effects can be blocked by the ARB losartan, indicating that Ang II regulates ACE2 [
24]. Mice deficient for ACE show markedly improved disease in an ARDS model induced by acid aspiration or sepsis, and recombinant ACE2 can protect mice from severe acute lung injury [
44]. In mice and rat models, ARB treatment consistently increases ACE2 mRNA and protein levels in heart, kidney, and aorta tissue, although its effect on lung ACE2 expression has not been studied. Also, the effects of ACEI administration on ACE2 expression differs among experiment models and tissues [
45]. Animal models of SARS-CoV-2 infection that recapitulate the pathogenesis of severe COVID-19 are needed to better understand the effects of ACE inhibitors and ARBs.
In this single-center retrospective study, we did not observe difference in clinical presentations of patients with or without ACEI/ARB treatment on admission. Nor did we find any important differences in severe clinical outcomes, namely in-hospital death, ICU admission, and mechanical ventilation, in these two groups. However, our preliminary analysis was limited by the retrospective nature of the cohort, its small sample size, and possibly nonrecorded data as the prescription records for drugs not directly relevant to COVID-19 treatment may be incomplete. Only patients with complete anti-hypertensive treatment information were included in the analysis, but this strategy, by excluding about half the hypertensive cases in the original hypertensive cohort, reduced statistical power of this study. Based on the percentages of in-hospital mortality and ICU admission reported in this cohort, around 500–1000 patients for each group would be necessary to provide statistical difference given two-tailed a = 0.05 and b = 0.2. Although this cohort had enough power to detect the association between older age and increased COVID-19 mortality (Table 6), it is under-powered for the marginal effect of ACEI/ARB on altered clinical outcome if there was association.
Other confounding factors including age should be carefully evaluated before drawing the final conclusion from retrospective analysis. It has to be noted that while ACEI/ARB are the most popular anti-hypertensive drug in the US [
46], CCBs are more widely used in China [
7,
47]. This partially explains the high CCB usage in the non-ACEI/ARB group (Tables 2 and 3). After adjustment for the difference of CCB prescription, ACEI/ARB still does not show association with COVID-19 mortality (OR 0.70, 95% CI 0.24−1.99,
P = 0.5011). We did not observe association between CCB usage and COVID-19 mortality, either, from this cohort (data not shown). Pre-existing heart failure should also be assessed because RAS antagonists including ACEI, ARB, and aldosterone inhibitors were basic treatment options for chronic heart failure. Proper multivariable adjustment on confounding factors should be implemented in retrospective analysis. We did not adjust for history of heart failure concerning the small sample size and the self-reporting nature of this factor. We did not have information for viral load of patients on admission, because only qualitative RT-PCR test results were available in the medical records. It would be interesting to examine whether chronic ACEI or ARB treatment would affect virus replication in future studies.
Similar to our findings, no association was observed between ACEI/ARB prescription and poor clinical prognosis or severe disease of COVID-19 in several other retrospective cohorts in China [
27,
28,
48,
49] or globally [
29–
32]. Compared with these reports, patients in this cohort were sicker on admission and the in-hospital mortality was higher (32%). This likely related to the study hospital being a referral center with severe cases during early epidemic in Wuhan.
With accumulating information from multiple large cohort studies, it seems clear that prior receipt of ACEI or ARB does not bring additional risks of hospital admission or poor prognosis of COVID-19. The discussion has therefore been gradually shifted to whether chronic treatment of ACEI or ARB before COVID-19 onset is associated with less severe illness. ACEI and ARB have been reported with protective effect in a previous meta-analysis for pneumonia-related mortality in general [
10]. In terms of COVID-19, de Abajo
et al. found a decreased risk of COVID-19 hospital admission in the subgroup analysis of diabetic patients with RAS inhibitors [
31]. Information from other large COVID-19 case series and preferably population-based cohort studies, optimally linked to serologic testing, are necessary to determine whether receipt of ACEI or ARBs are associated with reduced risk of symptomatic SARS-CoV-2 infection and especially of severe COVID-19 illness. Systematic analysis combining data from multiple studies would be helpful in answering this question. There have also been randomized clinical trials in progress studying the efficacy of acute initiation of ACEI for treatment of COVID-19 (e.g., NCT04366050 and NCT04355429).
In conclusion, there has been accumulating evidence for no association between chronic ACEI/ARB treatment and severe outcome of COVID-19. Meanwhile, current studies are pointing toward a putative protective role of prior ACEI or ARB receipt in COVID-19 illness, which awaits further evidence for confirmation. Several professional societies have released statements on continuing current anti-hypertensive treatment during the COVID-19 pandemic [
50], considering the risk of destabilizing blood pressure after changing medications. We agree that the best option is to continue previous therapy and wait for further evidence.
PDF
(552KB)
