key: cord-0856569-68rpjh4y
authors: Molinari, Chiara; Laurenzi, Andrea; Caretto, Amelia; Rovere-Querini, Patrizia; Ciceri, Fabio; Lampasona, Vito; Scavini, Marina; Piemonti, Lorenzo
title: Dysglycemia after COVID-19 pneumonia: a six-month cohort study
date: 2021-06-04
journal: Acta Diabetol
DOI: 10.1007/s00592-021-01751-5
sha: 73e302f1f689298f065c4696058cc2b8f34f233f
doc_id: 856569
cord_uid: 68rpjh4y

AIM: The aim of this study was to understand whether the dysglycemia associated with SARS-CoV-2 infection persists or reverts when the viral infection resolves. METHODS: We analyzed fasting blood glucose (FBG) after hospital discharge in a cohort of 621 adult cases with suspected COVID-19 pneumonia. RESULTS: At admission, 18.8% of the patients in our cohort had pre-existing diabetes, 9.3% fasting glucose in the diabetes range without a prior diagnosis (DFG), 26% impaired fasting glucose (IFG), 44.9% normal fasting glucose (NFG), while 2% had no FBG available. FBG categories were similarly distributed in the 71 patients without confirmed COVID-19 pneumonia. During follow-up (median time 6 month) FBG was available for 321 out of the 453 (70.9%) surviving patients and showed a trend to a marginal increase [from 97 (87–116) to 100 (92–114) mg/dL; p = 0.071]. Transitions between FBG categories were analyzed in subjects without pre-existing diabetes (265 out of 321). We identified three groups: (i) patients who maintained or improved FBG during follow-up [Group A, n = 185; from 100 (86–109) to 94 (88–99) mg/dL; p < 0.001]; (ii) patients who moved from the NFG to IFG category [Group B, n = 66: from 89 (85–96) to 106 (102–113) mg/dl; p < 0.001]; (iii) patients who maintained or reached DFG during follow-up [Group C, n = 14: from 114 (94–138) to 134 (126–143) mg/dl; p = 0.035]. Male sex and ICU admission during the hospitalization were more prevalent in Group C compared to Group A or B. CONCLUSIONS: Six months after the SARS-CoV-2 infection DFG was evident in only few patients who experienced severe COVID-19 pneumonia.

Increasing evidences suggest a bidirectional link between COVID-19 pneumonia and diabetes [1] [2] [3] [4] . Many studies have confirmed that diabetes and hyperglycemia are risk factors for the progression and poor prognosis of COVID-19 [5] [6] [7] [8] , including a higher risk for extrapulmonary complications, like cardiac injury [9] , end-stage renal disease requiring replacement therapy [10] and thromboembolic events [11] . On the other hand, new-onset diabetes/hyperglycemia [3] and acute metabolic decompensation of preexisting diabetes [12] are now emerging as complications of SARS-CoV-2 infection and an infection-related diabetes was hypothesized as a result of virus-associated β-cell destruction [13] . In fact, SARS-CoV-2 can infect cells of the human exocrine and endocrine pancreas ex vivo and in vivo [13] since the viral entry receptor angiotensin-converting enzyme 2 (ACE2) is expressed in human pancreatic β-cells and in the human pancreas microvasculature [14, 15] . However, the evidence that the exocrine and endocrine compartments of the pancreas are susceptible to productive SARS-CoV-2 infection does not necessarily imply that in COVID-19 SARS-CoV-2 infection directly affects glucose homoeostasis or triggers diabetes mellitus. For instance, we recently Managed by Massimo Federici. showed that influenza viruses are able to replicate in human pancreatic islets and cause diabetes in animal models [16] , but a direct evidence of a correlation between influenza virus infection and diabetes onset in humans was inconsistent [17] . Moreover, the clinical entity of new-onset diabetes/hyperglycemia associated with SARS-CoV-2 infection still has not been adequately characterized and well discriminated from pre-existing diabetes. In fact, it can include previously unrecognized (pre)diabetes (either type 2 or, less likely, type 1 diabetes) [18] , since excluding pre-existing diabetes can be difficult in the context of COVID-19 admissions. Most physicians do not request hemoglobin A1c measurements in the absence of a clinical suspicion of diabetes and hemoglobin A1c measurements in the months preceding SARS-CoV-2 infection are almost invariably unavailable for most COVID-19 patients. Many patients may present with stress hyperglycemia or they may be close to the diagnostic criteria for diabetes and exceed the threshold only at the time of SARS-CoV-2 infection. Furthermore, whether SARS-CoV-2 infection affects glucose metabolism more than other infections, as the community-acquired pneumonia, has still to be clarified. To address this gap in our knowledge, we studied a cohort of 621 adult cases hospitalized with suspected COVID-19 pneumonia and assessed the presence of dysglycemia at the time of hospital admission and during post-discharge follow-up to document whether it persists or reverts when the viral infection resolves.

The study population consisted of adult patients (≥ 18 years) with suspected COVID-19 pneumonia admitted between February 25 and May 2, 2020, to the Emergency or Clinical departments of the IRCCS San Raffaele Hospital (Milan, Italy) and for whom a serum sample was stored in our institution biobank. This series of patients is part of an institutional clinical-biological cohort (COVID-BioB; ClinicalTrials. gov Identifier: NCT04318366) of patients with COVID-19 pneumonia [19] . The Institutional Review Board (protocol number 34/int/2020) approved the study. Informed consent was obtained according to IRB guidelines. We defined as a confirmed infection case a patient with SARS-CoV-2 positive reverse-transcriptase polymerase chain reaction from a nasal/throat swab and signs, symptoms and radiological findings suggestive of COVID-19 pneumonia (n = 529). In case of multiple (at least two) SARS-CoV-2 negative reversetranscriptase polymerase chain reactions in the presence of radiological findings suggestive of COVID-19 pneumonia, subjects were classified with confirmed infection in the presence of positivity for IgM/IgG against SARS-CoV-2 spike protein [20] (n = 21). SARS-CoV-2 infection was excluded in subjects with multiple (at least two) SARS-CoV-2 negative reverse-transcriptase polymerase chain reactions and negativity for IgM/IgG against SARS-CoV-2 spike protein (n = 71). Data were collected through patient interview or medical chart review and entered in a case report form (CRF). Before analysis, CRF data were crosschecked with medical charts and verified by data managers and clinicians for accuracy (last data collection on March 17, 2021).

Routine blood tests encompassed serum biochemistry [including renal and liver function, lactate dehydrogenase (LDH)], complete blood count with differential and C-reactive protein (CRP) as inflammation marker. Specific antibodies to different SARS-CoV-2 antigens, interferon alpha-4 and glutamic acid decarboxylase (GAD) were measured by a luciferase immunoprecipitation system (LIPS) assay, as previously described [20] [21] [22] [23] .

Study participants were defined as having: a) pre-existing diabetes if they had a documented diagnosis of diabetes before the hospital admission for COVID-19 pneumonia [fasting plasma glucose (FPG) ≥ 126 mg/dl or HbA1c ≥ 6.5% (48 mmol/mol), or they were prescribed diabetes medications]; b) new-onset hyperglycemia if they had a mean fasting plasma glucose ≥ 126 mg/dl during the hospitalization for COVID-19 pneumonia in the presence of a negative history for diabetes and/or normal glycated hemoglobin level in the last year when available. We computed mean fasting glucose and glucose variability (standard deviation) from all laboratory fasting glucose values measured during hospitalization. Normal Fasting Glucose (NFG), impaired fasting glucose (IFG) and fasting glucose in the diabetes range (DFG) were defined according to ADA criteria.

Categorical variables are reported as frequency or percent, continuous variables as median with interquartile range (IQR) in parenthesis. Categorical variables were compared using Chi-square or Fischer's exact test, as appropriate; continuous variables using the Mann-Whitney test, Wilcoxon signed rank test or Mann-Whitney U test, as appropriate. Survival was estimated according to Kaplan-Meier. The time-to-event was calculated from the date of symptom onset to the date of the event, or of last follow-up visit, whichever occurred first. Two-tailed P values are reported, with P value < 0.05 indicating statistical significance. All confidence intervals are two-sided and not adjusted for multiple testing. Statistical analyses were performed with the SPSS 24 (SPSS Inc. /IBM). Sankey diagram of transitions between glucose tolerance categories was made by Sankey Diagram Generator by Dénes Csala, based on the Sankey plugin for D3 by Mike Bostock; https:// sankey. csala den. es; 2014.

We evaluated a series of 621 adult cases with suspected COVID-19 pneumonia enrolled from February 25 to May 2, 2020, in our institutional clinical-biological cohort (COVID-BioB). A confirmed COVID-19 pneumonia was present in 550 out of 621 (88.6%) of cases (COVID cohort), while the SARS-CoV-2 infection was excluded in the remaining 71 cases (No-COVID cohort). The characteristics of study participants are reported in Table 1 

As of March 18, 2021, the median follow-up time after symptoms onset was 213 (95% CI: 205-220) and 195 (100-289) days for the COVID and No-COVID cohorts, respectively.

We recorded fasting blood glucose during the postdischarge follow with outpatient visits at 1, 3, 6, and 9 months. In the COVID cohort, 97 patients died during follow-up (17.6%; 96 during hospitalization, 1 after hospital discharge), 133 (24.2%) had no glucose measurements either during follow-up (n = 122) or at admission (n = 11). Considering the remaining 321 subjects with available data post-hospital discharge, the FBG showed a trend to a marginal increase during follow-up from 97 (87-116) to 100 mg/dl (92-114) (p = 0.071) (median length of follow-up: 6 months). In the No-COVID cohort five (7%) patients died during and after the hospitalization and 50 (70.4%) had no glucose measurements during follow-up. The remaining 16 subjects with available data post-hospital discharge, showed a marginal not significant decrease in FBG during follow-up from 105 (94-134) to 97 mg/dl (88-142) (p = 0.48).

Since the study population was heterogeneous in terms of glycemic glucose levels at baseline, we conducted a sub analysis in the COVID cohort taking into account the dysglycemia state at the time of admission (Fig. 1) . Among 98 patients with pre-existing diabetes, 29 died (29.6%) and 13 (13.3%) had no glucose measurements during follow-up. The remaining 56 subjects showed a non-significant decrease in FBG: from 141 (111-172) to 129 mg/dl (113-163) (p = 0.937). Among patients with new-onset hyperglycemia, 20 (39.2%) died and 10 (19.6%) had no glucose measurements during follow-up. Of the remaining 21 subjects, none was prescribed a diabetes treatment during follow-up and they showed a significant decreased of FBG: from 138 (134-145) to 101 mg/dl (91-126) (p = 0.001; Fig. 1 ). Among patients with IFG/NFG, 47 (12.1%) died and 99 (25.4%) had no glucose measurements during follow-up. The remaining 244 subjects showed a marginal, but significant increase in FBG: from 94 (80-103) to 97 mg/dl (91-106), p < 0.001 (Fig. 1) .

To identify whether SARS-CoV-2 infection was able to induce dysglycemia in susceptible individuals in the COVID cohort, a Sankey diagram of transitions between FBG categories was drawn for patients without pre-existing diabetes (Fig. 2 

According to the transitions between FBG categories during follow-up, patients in the COVID cohort were divided into three groups (Fig. 3) . Group A (n = 185) included patients who maintained or improved their fasting blood glucose category during follow-up (NFG to NFG, n = 92; IFG to IFG/ NFG, n = 78; DFG to IFG/NFG, n = 15) with FBG going from 100 (86-109) at admission to 94 (88-99) mg/dl at the last follow-up (p < 0.001). Group B (n = 66) included patients who shifted from the NFG to IFG category in which FBG went from 89 (85-96) at admission to 106 (102-113) mg/dl and at the last follow-up (p < 0.001). Group C (n = 14) included subjects who maintained or reached DFG during follow-up (NFG to DFG, n = 4; IFG to DFG, n = 4; DFG to DFG, n = 6), with FBG going from 114 (94-138) at admission to 134 (126-143) mg/dl at the last follow-up (p = 0.035). The characteristics of the study participants according to these three subgroups are reported in Tables 2 and 3 . Male sex and the need of ICU admission during hospitalization showed a significantly higher prevalence in Group C compared to Group A or B.

Whether diabetes/hyperglycemia associated with SARS-CoV-2 infection should be considered a specific clinical entity is a matter of discussion. To address this issue, we studied a cohort of 621 adult cases with suspected COVID-19 pneumonia, assessing the presence of dysglycemia Our study generated several interesting findings. First, the prevalence of different FBG categories was similar between patients in the COVID and No-COVID cohorts. The No-COVID cohort patients were admitted to the hospital because of clinical signs of infection and respiratory insufficiency, but the diagnosis of COVID-19 was excluded by both molecular and serological testing. Thus, the No-COVID cohort represents a suitable sex and age matched control population to verify whether precipitating factors other than SARS-CoV-2 infection can induce dysglycemia. The results confirmed that dysglycemia per se is not unique to COVID-19, an expected finding as acute intercurrent illness of any kind are associated with metabolic abnormalities including impaired glucose use as well as decreased insulin secretion or increased counter-regulation. Second, new-onset hyperglycemia associated with COVID-19 pneumonia reversed in most patients after the viral infection resolved. A similar behavior was also evident in the No-COVID cohort and it is reasonable to speculate that reversible transient factors, such as inflammation-induced insulin resistance, may be causing hyperglycemia in those patients [24] . Third, a small group of patients without preexisting diabetes in COVID cohort maintained or achieved DFG during follow-up. These subjects corresponded to 4.3% of patients and showed a high prevalence of male sex and admission to intensive care, suggesting an association of DFG with a worse respiratory function during the acute phase of the disease. Fourth, a larger group of patients (about 20% of the COVID cohort) showed a modest increase in fasting blood glucose during follow-up, resulting in their shift to the IFG category. Of note, group B and group C (subjects whose glycemia increases) show a tendency toward a higher BMI. Even if this difference is not statistically significant, the presence of a higher insulin resistance linked to the weight could be one of the predisposing factors for the development or maintenance of dysglycemia during follow-up. On the other hand, BMI was associated with greater severity of COVID [25] , and could be indirectly linked as a proxy for severe infection [26] . This second hypothesis is less likely. In fact, these patients did not differ from the remaining COVID cohort in terms of IgG, IgM and IgA responses to the SARS-CoV-2 spike protein (RBD or S1 + S2), IgG response to NP, autoimmune antibodies anti-GAD, anti-interferon alpha-4, virus clearance, laboratory variables associated with COVID-19 pneumonia severity (C reactive protein, white blood cells, lymphocytes, lactate dehydrogenase), markers of liver and kidney function, comorbidities, preadmission treatments Tables 2, 3 ). The interpretation of these results is not simple but they could support the hypothesis that SARS-CoV-2 cannot directly cause dysglycemia and an attribution beyond a random variability of FBG remains unlikely. Our study has some limitations. First, FBG during followup was unavailable for 24% of patients in the COVID cohort. Even if FBG categories were similarly distributed at admission in the patients with or without glucose measurements, we cannot exclude a selection bias. Second, our COVID cohort did not include asymptomatic or pauci-symptomatic patients, as it only includes patients with COVID-19 pneumonia requiring hospital admission. Third, we only assessed fasting blood glucose. Studies on insulin secretion and resistance would have provided relevant information, but conducting such studies during the first wave of the COVID-19 pandemic was essentially impossible. Fourth, the lack of HbA1c measurements in our cohort is an additional limitation. The evidence of normal HbA1c at the time of admission would indicate no history of recent hyperglycemia and confirm the diagnosis of new-onset hyperglycemia. Fifth, the number of subjects included in No-COVID cohort is unbalanced respect to COVID cohort (71 vs 550) and this may prevent you from detecting small differences.

In conclusion, new-onset diabetes/hyperglycemia was documented as a complication of COVID-19 pneumonia. It is clear from our study that only a small proportion of patients without pre-existing diabetes maintained or shifted to DFG during follow-up and this finding may not be related to the SARS-CoV-2 infection. We cannot dismiss the possibility that secondary diabetes can be a distinct clinical entity within 'long COVID' or PARS (post-acute sequelae SARS-CoV-2 infection); however, our data suggest that the frequency of this event would be low. Large epidemiological studies in the next years will be required to clarify whether COVID-19 induce permanent diabetes and the alarmistic claims regarding the diabetes risk associated with COVID-19 should be interpreted with caution.

Diabetes mellitus is associated with severe infection and mortality in patients with COVID-19: a systematic review and meta-analysis

Extrapulmonary manifestations of COVID-19

New-onset diabetes in Covid-19

Covid-19 and diabetes: a complex bidirectional relationship

Diabetes as a risk factor for greater COVID-19 severity and in-hospital death: a meta-analysis of observational studies

Burden of diabetes mellitus and its impact on COVID-19 patients: a meta-analysis of real-world evidence

Predictors of adverse prognosis in COVID-19: a systematic review and metaanalysis

Sanchis-Gomar F (2020) Diabetes mellitus association with coronavirus disease 2019 (COVID-19) severity and mortality: a pooled analysis

Effects of hyperglycaemia on complications of COVID-19: a meta-analysis of observational studies

Acute kidney injury and kidney replacement therapy in COVID-19: a systematic review and meta-analysis

Registry of arterial and venous thromboembolic complications in patients with COVID-19

COVID-19 infection may cause ketosis and ketoacidosis

SARS-CoV-2 infects and replicates in cells of the human endocrine and exocrine pancreas

ACE2 expression in pancreas may cause pancreatic damage after SARS-CoV-2 infection

SARS-CoV-2 receptor angiotensin I-converting enzyme type 2 (ACE2) is expressed in human pancreatic β-cells and in the human pancreas microvasculature

Influenza A viruses grow in human pancreatic cells and cause pancreatitis and diabetes in an animal model

Study of 2009 H1N1 pandemic influenza virus as a possible causative agent of diabetes

Can COVID-19 cause diabetes?

Early predictors of clinical outcomes of COVID-19 outbreak in

COVID-19 survival associates with the immunoglobulin response to the SARS-CoV-2 spike receptor binding domain

Autoantibodies against type I IFNs in patients with life-threatening COVID-19

Antibody response to multiple antigens of SARS-CoV-2 in patients with diabetes: an observational cohort study

Robust neutralizing antibodies to SARS-CoV-2 develop and persist in subjects with diabetes and COVID-19 pneumonia

Potential metabolic and inflammatory pathways between COVID-19 and newonset diabetes

Obesity could shift severe COVID-19 disease to younger ages

High body mass index and night shift work are associated with COVID-19 in health care workers

Acknowledgement Professor Lorenzo Piemonti is the guarantor of this work and, as such, had full access to all the data presented in the study and takes responsibility for the integrity of data and the accuracy of data analysis.

Authors contribution LP, VL and MS contributed to the conception of the study, wrote the manuscript, researched data and contributed to the discussion. AL, AC and CM, contributed to the acquisition, analysis and interpretation of clinical data, contributed to the design of the study and critically reviewed/edited the manuscript. PRQ and GL contributed to the collection of biological samples, managed the biobanking activities, critically revised the manuscript, recruited patients and managed sample biobanking.Funding This work was funded by the IRCCS Ospedale San Raffaele (Program Project COVID-19) and the Italian Ministry of Health (Ministero della Salute COVID-2020-12371617).

The datasets generated during and/or analyzed during the current study are not publicly available due to concerns about data confidentiality, but are available from the corresponding author on reasonable request.

The authors have no conflict of interest to disclose in relation to the topic of this manuscript. The authors declare that there are no relationships or activities that might bias, or be perceived to bias, their work.Ethics approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. The study was approved by the Ethics Committee of the IRCCS Ospedale San Raffaele (protocol number 34/int/2020).Informed consent Informed consent was obtained from all individual participants included in the study.