key: cord-0910394-6qqolhqh
authors: Anaya, Juan-Manuel; Monsalve, Diana M.; Rojas, Manuel; Rodríguez, Yhojan; Montoya-García, Norma; Mancera-Navarro, Laura Milena; Villadiego-Santana, Ana María; Rodríguez-Leguizamón, Giovanni; Acosta-Ampudia, Yeny; Ramírez-Santana, Carolina
title: Latent Rheumatic, Thyroid and Phospholipid Autoimmunity in Hospitalized Patients with COVID-19
date: 2021-03-02
journal: J Transl Autoimmun
DOI: 10.1016/j.jtauto.2021.100091
sha: ab33f32e98f9d718f3ac5c4144a3678be7e467e0
doc_id: 910394
cord_uid: 6qqolhqh

Autoimmune responses mediated by autoantibodies have been previously observed in SARS-CoV-2 infection. Herein, we evaluated the presence of rheumatic, thyroid and phospholipid autoantibodies in sera samples from 120 hospitalized patients with COVID-19 in comparison to pre-pandemic samples from 100 healthy individuals. In addition, to estimate the frequency of these autoantibodies in COVID-19, a meta-analysis of selected articles was conducted. Hospitalized patients with COVID-19 displayed latent autoimmunity mainly mediated by a high frequency of anti-thyroid peroxidase antibodies, rheumatoid factor (RF), anti-cyclic citrullinated peptide third generation antibodies, antinuclear antibodies (ANAs), IgM β2-glycoprotein I (β2GP1) and IgM anti-cardiolipin antibodies. The meta-analysis confirmed our results, being RF and ANAs the most common autoantibodies. In addition, cluster analysis revealed that those patients with high positivity for RF, IgM β2GP1 antibodies and ANAs had a longer hospital stay, required more vasopressors during hospitalization, and were more likely to develop critical disease. These data suggest that latent autoimmunity influences the severity of COVID-19, and support further post-COVID studies in order to evaluate the development of overt autoimmunity.

The natural history of COVID-19, the disease caused by SARS-CoV-2, is 26 beginning to be deciphered thanks to research and scientific collaboration. One of 27 the most intriguing phenomena of COVID-19 is the presence of autoimmunity. [19, 20] . In addition, antinuclear antibodies (ANAs) were evaluated by using an 79 indirect immunofluorescence assay. Positive results were considered from dilution 80 1/80. In case of ANA positivity, anti-SSA/Ro, anti-SSB/La, anti-ribonucleoprotein 81 (RNP) and anti-smith (Sm) antibodies were further evaluated by a commercial 82 ELISA. All the assay kits were from Inova Diagnostics, Inc (San Diego, CA, USA).

Univariate descriptive statistics were performed. Categorical variables were 85 analyzed using frequencies, and quantitative continuous variables were expressed 86 as the mean and standard deviation (SD) or the median and interquartile range 87 (IQR). The Kruskal-Wallis, Mann-Whitney U-test, or Fisher exact tests were used 88 base on the results. Next, we tested the association between antibody levels and 89 critical disease (i.e., MV or died) using multivariable logistic regression. To account 90 for confounding factors, we included age and sex in the regression analysis. We 91 J o u r n a l P r e -p r o o f then used a margins analysis to graphically display mortality risk at a range of 92 antibody levels.

To summarize the diverse information of frequencies of autoantibodies in COVID-94 19, a meta-analysis approach for selected articles was employed. The logit 95 transformed proportion was used to derive the weighted proportion. The overall 96 pooled prevalence and 95% CI were obtained using a random effect model for 97 latent autoantibodies. Statistical heterogeneity between studies was evaluated by 98 Cochran's Q-statistic, as well as Tau 2 and I 2 statistics. A P value > 0.10 in Q-99 statistics or <50% in I 2 statistic indicated a lack of heterogeneity [21] .

To determine clusters of patients of COVID-19 with similar characteristics based on 101 autoantibodies positivity, we used the mixed-cluster methodology proposed by 102 Lebart et al. [22] . Briefly, a multiple correspondence analysis was done to obtain 103 the representation of data based on principal components. Following, we 104 determined the number of clusters by a hierarchical cluster analysis, and finally we 105 performed a consolidation step by k-means clustering. Autoantibodies with 106 frequencies < 5% were excluded since these variables with low frequencies tend to 107 generate clusters that include only those atypical values. Then, to evaluate the 108 clinical relevance of clusters obtained, we tested the risk for critical disease (i.e., 109 MV or died) using a multivariable logistic regression adjusted for age and sex. A P 110 value of <0.05 was set as significant for all type of comparisons. All analyses were 111 done using R version 4.0.1. analysis disclosed a heterogeneous autoimmune phenomenon (i.e., latent 149 autoimmunity). ANAs and RF were the most common, whereas other 150 autoantibodies exhibited frequencies lower than 11% (Table 3) . test, P= 0.0314), and were more likely to develop a critical disease (AOR, 2.75; 161 95% CI, 1.08 to 7.02; P=0. 0339) ( Table 5) . were associated with mortality [32]. In our study, it was found that IgG ACA levels 221 were associated with prediction of a critical disease, suggesting that levels of 222 phospholipid antibodies may help monitoring the disease and guide the treatment 223 (e.g., appropriate anticoagulation regimens).

Our study has several strengths. We included hospitalized patients with COVID-19 225 that did not have prior history of autoimmunity, and those patients with overt 226 autoimmunity were excluded from the study. This guaranteed that evaluation of 227 latency was accurate allowing a precise estimation of the real clinical effect of 228 J o u r n a l P r e -p r o o f latencies in COVID-19 from real-world data. In addition, loss of data was lower 229 than 1%.

Limitations must be also acknowledged. This was a retrospective study that could 231 have been susceptible for selection bias. However, grouping for this study was 232 researcher independent given by the unsupervised machine learning approach 233 implemented. In addition, given the significant results after adjustments and 234 corrections, it is highly unlikely that our results might be influenced by chance 235 alone or the moderate sample size. 

The presence of autoantibodies or biomarkers of autoimmunity without clinical symptoms and signs of autoimmune disease (AD). Also known as pre-clinical AD.

The presence of clinical symptoms or signs as a consequence of T cell activation (i.e., positive tissue biopsy) or B cell activation (i.e., presence of speci c autoantibodies) con guring a disease.

The presence of autoantibodies or biomarkers of autoimmunity unrelated (non-speci c) to an undergoing AD, with no clinical symptoms and signs of such an additional autoimmune condition. Expanded polyautoimmunity, when there are autoantibodies or biomarkers associated with at least two ADs.

The presence of two or more well-characterized and clinically manifested ADs in a single patient. Multiple autoimmune syndrome, if three or more ADs. The simultaneous presence of both overt and latent polyautoimmunity is possible as well as the presence of both systemic and organ-speci c autoimmunity.

ACE: Angiotensin-converting enzyme; ARB II: Angiotensin receptor blockers 2; COPD: Chronic 445 pulmonary obstructive disease

• ANAs, RF, CCP3, antiphospholipid and anti-TPO antibodies are the most common autoantibodies in patients with COVID-19.• Levels of IgG ACA are associated with critical illness.• The presence of ANAs, RF, and IgM β2GP1 antibodies is a risk factor for critical disease.J o u r n a l P r e -p r o o f

☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:J o u r n a l P r e -p r o o f