Prevalence of SARS-CoV-2 Antibodies in Children and Adults with Type 1 Diabetes

Abstract

Objective: As diabetes is a risk factor for severe symptoms, hospitalization, and death with COVID-19 disease, we aimed to assess the prevalence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) antibodies in children and adults with and without type 1 diabetes in Colorado during 2020.

Research Design and Methods: We developed a highly sensitive and specific test for antibodies against SARS-CoV-2 and measured the antibodies in children and adults with new-onset (n = 129) and established type 1 diabetes (n = 94) seen for routine diabetes care at our center between January and October 2020. The antibodies were also measured in 562 children and 102 adults from the general population of Colorado.

Results: The prevalence of SARS-CoV-2 antibodies in persons with new-onset type 1 diabetes (0.8%; 95% confidence interval 0.1%–4.2%) or those with established disease (4.3%; 1.7%–10.4%) did not differ from that in the general population children (2.8%; 1.8%–4.6%) or adults (3.9%; 1.5%–9.7%). In a subset of individuals with positive antibodies (n = 31), antibodies remained positive for up to 9 months, although the levels decreased starting 3 months after the infection (P = 0.007).

Conclusions: From January to October 2020, the prevalence of SARS-CoV-2 antibodies were not different in children and adults with and without type 1 diabetes in Colorado. We found no evidence for increased prevalence of COVID-19 infections among youth with newly diagnosed type 1 diabetes. (COMIRB Protocol 20-1007)

Introduction

SARS-CoV-2 (severe acute respiratory syndrome coronavirus 2) causes COVID-19 syndrome and has resulted in a global pandemic. Molecular tests real time reverse transcription polymerase chain reaction (RT-PCR) are in place to diagnose the disease; however, serological tests are needed to determine who had a prior infection and potentially has immunity. People with diabetes are at a higher risk for more severe COVID-19 symptoms, hospitalization, and mortality compared with the general population.^1–4 Recent case reports have associated COVID-19 disease with the onset of type 1 diabetes^5–7; however, the evidence remains limited and inconsistent.^8–10 We developed a highly sensitive and specific test to measure COVID-19 antibodies and applied this assay to people with new-onset and established type 1 diabetes as well as children and adults from the underlying general population during the current pandemic.

Research Design and Methods

COVID-19 antibody assay

We developed a fluid-phase assay to detect antibodies in serum directed against the receptor-binding domain (RBD) in the spike protein of SARS-CoV-2 (Fig. 1A). The test measures total antibodies, including both IgM and IgG, independent of isotype. The format of the electrochemiluminescence (ECL) assay was adapted from those with islet autoantibodies.^11,12 In brief, serum is diluted five times in phosphate buffered saline (PBS) and then incubated with SULFO-TAG-labeled RBD protein 150 ng/mL (Creative Biomart, Shirley, NY) and biotin-labeled protein 150 ng/mL in PBS containing 5% bovine serum albumin overnight at 4°C. Next, the samples are incubated in a streptavidin-coated plate (Meso Scale Diagnostics [MSD], Rockville, MD) at room temperature for 1 h and gently agitated. The plate is washed three times with PBS with 0.05% Tween-20 buffer followed by the addition of reader buffer and then counted using a MESO QuickPlex SQ 120 instrument (MSD). Results are reported as index (index = [(Signal_sample − Signal_{negative control})/(Signal_{positive control} − Signal_{negative control})] × 100). The assay cutoff of index at 5 was set at the 99.9th percentile for 922 control sera samples obtained before September 2019 (pre-COVID-19 disease). The intra-assay coefficient of variance was 8.5% and interassay variance was 10.0% (n = 10). We subsequently adapted the assay to measure antibodies directed against the nucleocapsid protein of SARS-CoV-2, which had a similar performance to that for RBD antibodies and applied an identical assay cutoff.

FIG. 1.

COVID-19 antibody ECL test. (A) Assay diagram that uses fluid-phase binding of antibody in serum to two labeled RBD proteins. The SULFO-TAG-labeled RBD protein emits ECL when stimulated. (B) Results of validation testing with SARS-CoV-2 PCR+ convalescent sera (n = 58), SARS-CoV-2 PCR- but with other viral infections confirmed by PCR (n = 7), and pre-COVID-19 samples (n = 922). The mean value from duplicate wells are plotted for each sample and box plots showed the minimum, maximum, median, and 25th and 75th percentiles for each group. Dotted line at an index of 5 is the cutoff for positivity. ECL, electrochemiluminescence; PCR, polymerase chain reaction; RBD, receptor-binding domain; SARS-CoV-2, severe acute respiratory syndrome coronavirus 2. Color images are available online.

Study population

Subjects were recruited from the Barbara Davis Center for Diabetes Clinics and >95% of participants were from or near great Denver area. All patients met the American Diabetes Association diagnostic criteria for type 1 diabetes mellitus. Serum samples collected between January 3, 2020 and October 16, 2020 were tested for COVID-19 antibodies. In addition, general population nondiabetic children participating in the Autoimmunity Screening for Kids (ASK) study¹³ and their parents were included if they were screened between July 8 and October 23, 2020. All participants provided written informed consent, and the Colorado Multiple Institutional Review Board approved the study.

Islet autoantibody measurements

Islet autoantibodies to glutamic decarboxylase (GADA), insulinoma antigen-2 (IA-2A), zinc transporter 8 (ZnT8) and insulin (IAA) were measured from serum using radio-binding assays as previously described.¹⁴

Statistical analyses

Statistical analyses were performed using GraphPad Prism 8.4 software (GraphPad Software). A two-tailed paired t-test was used to compare COVID-19 antibodies over time. Continuous variables (age and diabetes duration) were compared using a two-tailed t-test and a Fisher's exact test compared prevalence of COVID-19 antibody positivity between the cohorts. To analyze the changes in COVID-19 antibody levels over time, the index data were normalized using the Box–Cox transformation¹⁵; the optimal power parameter (lambda of 0.25) was obtained using PROC TRANSREG by using a maximum likelihood criterion (SAS 9.4, Cary, NC). Both log transformation and square root transformation failed to normalize the data. P-values <0.05 were considered significant.

Results

Using our COVID-19 antibody ECL test, we conducted validation testing with serum from individuals who were RT-PCR positive for the SARS-CoV-2 virus (n = 58) and previously collected serum samples before the pandemic from individuals aged 4.5–76.1 years (Fig. 1B). The sensitivity of the assay for RBD antibodies is 100% and specificity is 99.9%. Assuming a 5% prior infection rate, the negative predictive value is calculated to be 100% and the positive predictive value is 98.1%. Importantly, sera from seven individuals with PCR-confirmed human coronavirus infections NL63, OC43, and HKU1, rhinovirus, respiratory syncytial virus, and parainfluenza virus type 4 were negative in our assay,¹⁶ indicating biological specificity for COVID-19-related antibodies.

A subset of persons who were RT-PCR positive for the SARS-CoV-2 (n = 31) with repeat testing (up to five times) has demonstrated that COVID-19 antibodies remained positive for up to 9 months after the infection (Fig. 2). The antibody levels (index) increased during the initial 3 months after infection (P = 0.029) and decreased thereafter (P = 0.007).

FIG. 2.

COVID-19 antibody levels over time. Repeat testing of COVID-19 antibodies in 31 positive subjects. X-axis represents follow-up time in month since COVID-19 infection. An index of 5 is the cutoff for positivity. The antibody levels (index) were Box–Cox transformed to normalize the data. The levels increased during the initial 3 months after infection (P = 0.029) and decreased thereafter (P = 0.007).

Next, we measured antibodies to both the RBD and the viral nucleocapsid protein in new-onset type 1 diabetes patients and those with established type 1 diabetes that were seen at our center during the current pandemic (Table 1). Those with new-onset type 1 diabetes were predominantly children and adolescents, only had diabetes for a median of 2 days with all having diabetes <1 month, and 84% were positive for at least one islet autoantibody. The cohort with established type 1 diabetes was older than those with new-onset diabetes and had diabetes for a median of 866 days (Table 1). The prevalence of COVID-19 antibodies was 0.8% (95% confidence interval 0.1%–4.2%) among new-onset patients and 4.3% (1.7%–10.4%) among those with established disease (P = 0.17). All of the people with diabetes who had COVID-19 antibodies were positive for both antibodies directed against the RBD and nucleocapsid protein, and this was also the case among the 58 persons with PCR-confirmed COVID-19 infection (Fig. 1B). These data indicate that the addition of the nucleocapsid protein does not increase the sensitivity or specificity of our assay for detecting COVID-19 antibodies.

Table 1.

Characteristics of Study Participants and Prevalence of COVID-19 Antibodies

	Patients with type 1 diabetes		Controls without diabetes
	New-onset (n = 129)	Established (n = 94)	Children (n = 562)	Adults (n = 102)
Age, years
Mean (SD)	12.3 (8.2)	20.7 (14.4)	9.3 (4.5)	41.4 (7.7)
Median	11.4	16.2	9.0	41.7
Range	1.3–69.3	1.5–66.8	1.0–18	19–60.5
Diabetes duration, days
Mean (SD)	2.4 (3.9)	3208 (4348)	NA	NA
Median	2.0	866	NA	NA
Range	0–29	36–19,983	NA	NA
COVID-19 antibodies, % (n) 95% CI
Receptor-binding domain	0.8% (1) 0.1%–4.2%	4.3% (4) 1.7%–10.4%	2.8% (16) 1.8%–4.6%	3.9% (3) 1.5%–9.7%
Nucleocapsid	0.8% (1) 0.1%–4.2%	4.3% (4) 1.7%–10.4%	NA	NA

95% CI calculated using Wilson score interval.

CI, confidence interval; SD, standard deviation; NA, not applicable.

Finally, we compared the prevalence of antibodies in persons with type 1 diabetes with that in the underlying general population. Since 2017, the ASK study has screened >25,000 Colorado children for preclinical type 1 diabetes and celiac disease. COVID-19 antibodies directed against the RBD were added to the screening in July 2020 and were measured in 562 children 1–17 years of age and in 102 of their parents (Table 1). Race/ethnicity reflected the underlying population of the Denver metropolitan area (66% non-Hispanic whites) and 51% were female. The prevalence of COVID-19 antibodies in the general population children (2.8%; 1.8%–4.6%) or adults (3.9%; 1.5%–9.7%) did not differ significantly from the prevalence observed in persons with type 1 diabetes.

Conclusions

We have successfully used the ECL assay format to measure antibodies for diagnostic purposes in type 1 diabetes and other autoimmune disorders.¹⁷ The ECL assay, as opposed to prototypical solid-phase enzyme-linked immunosorbent assay (ELISA), allows for a large dynamic range in terms of signal to noise and excellent sensitivity and specificity for very low concentration analytes such as antibodies. Furthermore, the test for COVID-19 antibodies is independent of antibody isotype (e.g., IgM, IgG, IgA, IgD, or IgE) making it useful for an initial screen for antibody positivity in large populations. In contrast to several commercial ELISA assays,¹⁸ the addition of antibodies targeting the nucleocapsid antigen on SARS-CoV-2 did not appear necessary to achieve high sensitivity and specificity in our assay measuring antibodies to RBD alone. RBD antibodies are also less likely to quickly wane postinfection compared with nucleocapsid antibodies¹⁹—a desirable feature for seroepidemiological studies. In fact, all 31 subjects who were positive for RBD antibodies and retested repeatedly for up to 9 months remained positive, although the levels decreased, as expected.²⁰

Although our data are preliminary, we did not find statistical differences in the prevalence of COVID-19 antibodies in people with type 1 diabetes and the general population. This observation needs to be confirmed in a larger sample with multivariate adjustment for age, month of blood draw, and potentially other confounders, for example, occupation and ethnicity. Although the prevalence of COVID-19 antibodies in persons with established type 1 diabetes was not significantly higher than in those newly diagnosed, a small difference (4.3% vs. 0.8%) could be due to the older age of these individuals or diabetes being a potential risk factor for COVID-19 disease. Future larger studies are warranted to determine the seroprevalence of COVID-19 antibodies in diabetes populations, and verify whether these antibodies provide immunity to repeat exposure in people with diabetes.

SARS-CoV-2 is known to bind angiotensin-converting enzyme 2 (ACE2) receptors, thereby allowing entry into cells. As ACE2 receptors have been identified on human beta cells,^21,22 a potential mechanism exists for direct viral cytotoxicity of IAA producing cells with a SARS-CoV-2 infection. However, there is a conflicting report that analyzed pancreata from organ donors with COVID-19 and predominantly found ACE2 expression in the pancreatic ductal epithelium and microvasculature but rare ACE2 mRNA levels within endocrine cells; there were thrombotic lesions containing SARS-CoV-2 nucleocapsid protein within ducts indicating pancreatic infection with the virus.²³

Interestingly, the only new-onset patient that tested positive for COVID-19 antibodies in our cohort did not have any of the four prototypical islet autoantibodies and could have type 2 diabetes. She was a 14-year-old Hispanic with a history of overweight and a 30 pound weight loss before diagnosis. Although she presented in moderately severe diabetic ketoacidosis, this is not unusual in teens with type 2 diabetes.²⁴ Therefore, we found no evidence for an association between COVID-19 infection and newly diagnosed type 1 diabetes in this patient.

There is another case report of new-onset type 1 diabetes in a patient that lacked islet autoantibodies, had a prior COVID-19 infection, and presented with COVID-19 antibodies.⁷ The mechanisms of new-onset diabetes during the COVID-19 pandemic warrant further investigation, and a global registry is in place to collect information on patients with new-onset diabetes.²⁵

Authors' Contributions

M.J.R., A.W.M., and L.Y. designed the studies and wrote the article. X.J., P.G., C.G.R., L.H., and A.A.A. performed the research and reviewed the article. F.D. assisted with the statistical analyses. L.Y. is the guarantor of this study, had full access to all the data in the study, and takes responsibility for the data integrity and accuracy of analysis.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This study was supported by NIH grants (DK108868, DK032083, DK099317, and DK116073), JDRF grant 3-SRA-2018-564-M-N (Autoimmunity Screening for Kids, funded jointly by JDRF, The Leona M. and Harry B. Helmsley Charitable Trust, and Janssen Pharmaceutical) and the Patten-Davis Foundation.