1. Introduction
“New-onset” hyperglycaemia and acute metabolic decompensation of pre-existing diabetes are commonly recognized phenomena associated with past coronavirus infections [
1,
2]. During the Middle East Respiratory Syndrome Coronavirus (MERS-CoV) outbreak in 2012 and the Severe Acute Respiratory Syndrome Coronavirus-1 (SARS-COV-1) outbreak in 2003, hyperglycaemia was a commonly observed finding, even in patients without a prior history of diabetes and who did not use glucocorticoids and was an independent predictor for mortality [
1,
2]. Not unexpectedly, pre-existing diabetes also resulted in an association with poor outcomes and death after SARS-CoV-2 infection [
3,
4,
5], and “new-onset” hyperglycaemia [
6] and acute metabolic decompensation of pre-existing diabetes [
7] are not uncommonly recognized phenomena associated with COVID-19 pneumonia, especially in patients requiring hospitalization. Moreover, isolated cases of true new-onset diabetes have been reported during COVID-19 [
8,
9,
10,
11,
12,
13,
14]. These finding suggest the existence of a bidirectional link between COVID-19 and diabetes [
15]. Both reduced insulin secretion and increased insulin resistance have been proposed as possible mechanisms [
16,
17]. SARS-CoV-2 might exert direct cytotoxicity against beta cells with a diabetogenic effect [
18,
19]. Whether beta cells express ACE2 and TMPRSS2 (two factors associated with SARS-CoV-2 infection [
20]) and are permissive to SARS-CoV-2 infection is debated [
21,
22,
23,
24,
25,
26]. However, there is a common agreement that SARS-CoV-2 can be detected in the pancreas where it may cause inflammation that secondarily affects beta cells, and pancreatic enlargement, abnormal amylase or lipase levels, and pancreatitis were described in critically ill COVID-19 patients [
27,
28,
29]. On the other hand, the surge in cytokine production associated with viral infection may be responsible for an increase in insulin resistance [
30]. Regardless of the mechanism responsible for the metabolic dysregulation observed, the clinical entity of COVID-associated hyperglycaemia still has not been adequately characterized and separated from the entity of pre-existing diabetes [
6]. Few studies have comparatively characterized the two conditions for the presence of comorbidities, pre-hospitalization treatments, symptoms at admission, and laboratory variables associated with the severity of infection. Many studies have recently reported that COVID-associated hyperglycaemia is associated with a poorer outcome compared to that of the normoglycaemic individuals [
31,
32,
33,
34,
35,
36]. However, whether COVID-associated hyperglycaemia is associated with a poorer clinical outcome compared to pre-existing diabetes is still an open question, and there have been conflicting findings [
32,
34,
37,
38,
39]. Moreover, data on islet autoimmunity prevalence, anti-SARS-CoV-2 antibody responses, and timing for viral clearance are still missing. Finally, it is unclear whether COVID-associated hyperglycaemia persists or reverts when the viral infection resolves. To address this gap in our knowledge, we studied a cohort of 176 adult patients with confirmed COVID-19 pneumonia who also had a diagnosis of pre-existing diabetes or COVID-associated hyperglycaemia.
4. Discussion
Hyperglycaemia [
6] is emerging as a common feature among patients hospitalized for COVID-19 pneumonia [
40]. This COVID-associated hyperglycaemia still needs to be adequately characterized since there are only few studies documenting how it differs from pre-existing diabetes in COVID-19 patients. To address this issue, we studied a cohort of 176 adult patients with COVID-19 pneumonia with a diagnosis of pre-existing diabetes or hyperglycaemia. Our study generated several interesting findings. First, patients with COVID-associated hyperglycaemia had significantly less comorbidities related to the metabolic syndrome than those with pre-existing diabetes. The main reason for this difference probably lies in the lower prevalence of obesity and a shorter hyperglycaemia duration. Consistent with this, prior to admission, only a low proportion of patients had been prescribed antihypertensive drugs, lipid-lowering agents, or antiplatelet therapy. However, whether the lower prevalence is an effect related to an underdiagnosis driven by the absence of diabetes diagnosis cannot be excluded. Similarly, the higher prevalence of neurodegenerative disease in subjects with COVID-associated hyperglycaemia could be related to the lower use of health care services and therefore lower rates of a diabetes diagnosis pre-admission in patients with a cognitive impairment. Beyond the potentially confounding factors, the lower prevalence of a metabolic signature suggests a specific pathogenesis for hyperglycaemia associated with COVID-19.
Second, patients with COVID-associated hyperglycaemia showed increased levels of inflammatory markers and indicators of multi-organ injury compared to patients with pre-existing diabetes. The correlation of hyperlgycaemia with high levels of inflammation is consistent with other conditions [
41] and it is not unexpected in this context. In fact, during an overwhelming SARS-CoV-2 infection, the massive release of glucocorticoid [
42] and of cytokines [
43] might stimulate the gluconeogenesis and increase insulin resistance, which can contribute to the elevation in blood glucose levels. Based on this, it can be speculated that in the absence of a pre-existing diabetes, hyperglycaemia pre-existing could represent a proxy of illness severity, as also suggested by the association with and indicators of multi-organ injury.
Third, the prevalence of autoimmune biomarkers such as anti-GAD (a marker of islet autoimmunity) or interferon alpha-4 antibody (an autoimmune marker recently associated with COVID-19 severity) [
44] was generally very low and did not differ between COVID-associated hyperglycaemia and pre-existing diabetes. Viral infections were described to be associated with the development of pancreatic autoantibodies leading to T1D in genetically predisposed individuals in the TEDDY study [
45], and coronaviruses were identified as one of the incriminating pathogens. Our data tend to exclude that this is the case for SARS-CoV-2. The extremely low prevalence of GAD autoantibodies excludes a previously unrecognized autoimmune (pre)diabetes (i.e., LADA) as the possible cause of new onset diabetes. Moreover, the data are in agreement with our previous results on children aged 1 to 18 years participating in a public health type 1 diabetes screening program in Bavaria, where the prevalence of antibodies against SARS-CoV-2 was not associated with type 1 diabetes autoimmunity [
46].
Fourth, COVID-associated hyperglycaemia was associated with a poorer clinical outcome and a delayed time to viral clearance compared to pre-existing diabetes despite the presence of superimposable humoral responses to SARS-CoV-2 and similar glucose levels. Many studies have previously reported a worse outcome in patients with new-onset hyperglycaemia without diabetes versus normoglycaemic COVID-19 patients [
35,
38,
47], while only a few studies have reported the outcomes in patients with new-onset hyperglycaemia without diabetes versus diabetes with discordant results [
33,
35,
38]. Of great interest is the finding of a slow viral clearance in patients with COVID-associated hyperglycaemia. Discordant data exist about differences in viral clearance or shedding in people with diabetes [
48]. As increased glucose levels and glycolysis promoted Sars-CoV-2 replication in monocytes via ROS/HIFα pathway activation lead to secondary T-cell dysfunction [
49], this could explain the delayed viral clearance associated with the acute hypeglycaemia in patients with COVID-associated hyperglycaemia.
Fifth, COVID-associated hyperglycaemia reversed in most patients when the viral infection resolved. As hyperglycaemia reverted in most patients when the infection resolved, it is reasonable to hypothesize that reversible transient factors, such as inflammation-induced insulin resistance, may play a role in causing hyperglycaemia in those patients [
50]. Whether COVID-associated hyperglycaemia should be considered a specific clinical entity remains a matter of discussion. The use of glucocorticoids in the first wave of the pandemic was very limited and, therefore, unlikely to have played a role in the pathogenesis of hyperglycaemia. On the other hand, it is very likely that more than one cause may contribute to the hyperglycaemia associated with COVID-19. In our study, the absence of islet autoimmunity markers and the lower prevalence of comorbidities related to metabolic syndrome do suggest specific pathophysiological mechanisms responsible for hyperglycaemia. Regardless of the pathophysiological mechanism, our study documented that COVID-associated hyperglycaemia was associated with an adverse clinical outcome of SARS-CoV-2 infection and that this association was independent from the major risk factors for disease severity. As good glycemic control was indeed shown to reduce disease severity and COVID-19 mortality patients with hyperglycaemia [
51], early recognition and treatment of COVID-associated hyperglycaemia may greatly benefit these patients. Therefore, our findings do strongly support the need to screen all patients with COVID-19 pneumonia for hyperglycaemia (i.e., blood glucose and/or HbA1c) at the time of admission despite a mute personal or family history of diabetes. Since blood glucose levels in our study were not different in patients with either COVID-associated hyperglycaemia or pre-existing diabetes, factors other than hyperglycaemia should be considered to explain the differences in clinical outcome. A different drug treatment prior to admission could play a role. Of note, in our study, sex-and age-adjusted Cox proportional hazards models indicated that angiotensin-converting enzyme inhibitors (ACEi) and/or angiotensin receptor blockers (ARB) treatment was significantly associated with a better clinical outcome in pre-existing diabetes. Most studies, including several meta-analyses, found no substantial differences in the risk for severe COVID-19 pneumonia associated with the prescription of common classes of antihypertensive medications [
52]. On the other hand, evidence of the beneficial effects of chronic ACEi/ARB use, especially in hypertensive cohort of patients with COVID-19 pneumonia, was reported [
53,
54,
55,
56]. As the prevalence of hypertension is higher among patients with pre-existing diabetes, ACEi/ARB use may mediate a selective beneficial effect. Similarly, a role for antidiabetics could be speculated. In our study, insulin and metformin had different associations with clinical outcomes: around 60% of patients with pre-existing diabetes were taking metformin, and the treatment was associated with a better clinical outcome. Conversely, insulin treatment was associated with a worse outcome. This is in agreement with previous small studies [
57,
58,
59,
60,
61,
62] and was recently confirmed by a large observational nationwide study in England [
63] where metformin, SGLT2 inhibitors, and sulfonylureas resulted in an association of reduced risks of the COVID-19-related mortality, whereas insulin and DPP-4 inhibitors were associated with an increased risk, and neutral results were found for GLP-1 receptor agonists and thiazolidinediones. However, because of the limitations of such real-world studies, these findings should be considered with caution and are likely due to be confounding by indication, in view of the use of different drug classes in the early and late stages of the type 2 diabetes disease trajectory (metformin is used early in the disease trajectory of type 2 diabetes, whereas insulin is typically initiated later). Few randomized clinical trials assessing the role of glucose-lowering therapies on COVID-19 outcomes in patients with type 2 diabetes are ongoing. The DARE-19 study [
64] investigated the effect of the SGLT2 inhibitor dapagliflozin versus a placebo on the risk of death or organ dysfunction in patients admitted to hospital with COVID-19 and failed to demonstrate any advantages.
Our study has some limitations. First, the analysis was performed on a subcohort of 176 patients selected for having hyperglycaemia or pre-existing diabetes out of 584 subjects of our original cohort. All patients with diabetes/hyperglycaemia were included, and the age, sex, hyperglycaemia prevalence, and the clinical outcome of our cohort appears superimposable to those reported by many authors. Despite this, we cannot exclude selection bias. Second, our cohort is limited to hospitalized patients with COVID-19 pneumonia. Third, we only assessed fasting blood glucose, and we acknowledge that more specific indicators of beta cell function, such as serum insulin and or C peptide levels, should have been measured.