Optimizing SARS-CoV-2 Immunoassays for Specificity in Dengue-Co-Endemic Areas

Introduction The overlap in clinical presentation between COVID-19 and dengue poses challenges for diagnosis in co-endemic regions. Furthermore, there have been reports of antibody cross-reactivity between SARS-CoV-2 and dengue. Our research aims to evaluate SARS-CoV-2 antigens for serological testing while reducing the possibility of cross-reactivity with anti-dengue antibodies. Method Two hundred and ten serum samples were collected from 179 patients and divided into four panels. Panels 1 and 2 consisted of COVID-19-negative healthy donors (n=81) and pre-pandemic dengue patients (n=50), respectively. Alternatively, Panel 3 (n=19) was composed of reverse transcription-quantitative polymerase chain reaction (RT-qPCR)-positive samples collected within two weeks of COVID-19 symptom onset, while Panel 4 (n=60) was composed of positive samples collected after two weeks of symptom onset. Previously developed and characterized in-house SARS-CoV-2 spike-1 (S1), receptor binding domain (RBD), and nucleocapsid (N) immunoglobin G (IgG)-enzyme-linked immunosorbent assay (ELISA) assays were used for the study. Results Six dengue-positive sera cross-reacted with the RBD of SARS-CoV-2. However, only one dengue-positive sera cross-reacted with the S1 and N proteins of SARS-CoV-2. Co-immobilization of S1 and RBD in different ratios revealed an 80:20 (S1:RBD) ratio as optimal for achieving an overall 96.2% sensitivity with the least cross-reaction to anti-dengue antibodies. Conclusion Our findings indicated that SARS-CoV-2 RBD-based immunoassays present more cross-reactivity with anti-dengue antibodies than S1 and N proteins. Furthermore, co-immobilization of S1 and RBD reduces the cross-reactivity with anti-dengue antibodies compared to RBD, thereby increasing the immunoassay specificity without affecting overall sensitivity for the dengue-endemic areas.


Introduction
The co-endemicity of dengue fever and COVID-19, especially in tropical and subtropical areas of the world, has shaken the healthcare sector once again, which gradually started recovering by introducing COVID-19 vaccines [1][2][3].In 2023, more than 3.7 million dengue cases were reported globally, with the number still rising with the emergence of Omicron (BA.2.86), a SARS-CoV-2 variant [4,5].Unfortunately, despite the rollout of current vaccines, new COVID-19 variants are still emerging [6].These dual attacks by COVID-19 and dengue generate further economic downfall, especially in the dengue-affected regions.
The ubiquity of molecular testing platforms in developing and developed countries ensured early diagnosis of COVID-19, aiding in combating the pandemic.With the evolution of the disease, multiple low-cost, highsensitivity tests such as antigen tests, reverse transcription-quantitative polymerase chain reaction (RT-qPCR), enzyme-linked immunosorbent assay (ELISA), matrix-assisted laser disruption ionization time-offlight mass spectroscopy (MALDI-ToF-MS), loop-mediated isothermal amplification (LAMP), and others have evolved [14][15][16][17].On the other hand, dengue fever is primarily diagnosed virologically or serologically, the latter being more widely practiced [18,19].
Several serodiagnosis assays have been manufactured targeting N, S, or receptor binding domain (RBD) proteins for SARS-CoV-2 [27].These kits can detect pan-immunoglobin (Ig), IgG, IgM, or IgA following COVID-19 infection [28].However, several studies have reported co-endemicity and cross-reaction between COVID-19-positive sera with anti-dengue IgG/IgM kits and vice versa [10,11].Moreover, some dengue NS1 and IgM rapid diagnostic tests (RDTs) are also reported to cross-react with Zika and Chikungunya virusinfected sera [19].Thus, it is essential to properly evaluate RDTs and other serodiagnosis assays for their cross-reactivity with closely related pathogens with whom the clinical manifestations coincide before being introduced to the general population.This study aims to evaluate pre-COVID-19 pandemic dengue sera for their potential cross-reaction with SARS-CoV-2 S1, RBD, and N proteins using previously developed inhouse ELISAs.Moreover, immunoassays were optimized to minimize the cross-reaction without affecting sensitivity.
A portion of this paper was published earlier on the medRxiv preprint server on December 22, 2020.

Specimen selection and assortment
To characterize the serological cross-reaction between SARS-CoV-2 and DENV, serum samples (n=210) were collected from 179 individuals (Figure 1).The negative sera were categorized into two panels (P): P1 samples collected from individuals with no COVID-19 symptoms.They were N-IgG negative from April to June 2020 (n=81), and P2 samples were collected from pre-pandemic dengue-infected individuals (n=50).Seventy-nine positive serum samples were collected from confirmed RT-qPCR-positive SARS-CoV-2 individuals.These positive samples were categorized into two panels: P3 contained sera from SARS-CoV-2 qRT-PCR positive patients with <14 days of symptom appearance (n=19), and P4 contained sera from SARS-CoV-2 positive patients with >14 days from symptom onset (n=60).All specimens were aliquoted in small volumes and preserved at -80 °C for further analysis.

Stratification of COVID-19 and dengue samples for seropositivity and seronegativity
The samples were characterized based on four factors: clinical symptoms, RT-qPCR test outcomes, the absence of nucleocapsid protein, and samples collected before the pandemic.Following the manufacturer's instructions, seropositive dengue samples (P2) were re-characterized using the dengue IgG/IgM immunoassay Scan® dengue IgG+IgM card test (Bhat BioTech, Bangalore, India).Furthermore, P1, P3, and P4 were also tested with this same assay for anti-dengue antibodies and confirmed as seronegative.

Statistical analysis
Statistical analysis was performed with STATA 15 (StataCorp LLC, College Station, TX) and the graphical representation using Prism version 7.05 (GraphPad Software, La Jolla, CA).A p-value of <0.05 was considered significant.The precision of the in-house ELISA assay compared to that of the true positives was characterized by a 95% confidence interval in its sensitivity, specificity, receiver operating characteristic (ROC), positive predictive value (PPV), and negative predictive value (NPV).Test agreement estimation was carried out using Cohen's Kappa test.We used a univariate regression model to evaluate the mean difference of S1, RBD, and N-IgG titers among P1, P2, P3, and P4 panels.A non-parametric Wilcoxon signed-ranks test was used to see the difference of S1, RBD, and N-IgG titers separately in the overall population: <14 days (P3) and >14 days (P4).

Discussion
The limitations of molecular diagnostics, specifically the detection of viral ribonucleic acid (RNA) in the later stages of COVID-19 diagnosis, make it necessary to implement serological tests.These tests help provide a better understanding of the disease's progression and enable the identification of asymptomatic infections [32].Scientists worldwide are exploring the development of various immunoassays, leading to the availability of tests based on ELISA, lateral flow immunoassays (LFIA), chemiluminescent assays, flowthrough dot-blot assays, and more [29].However, the available serological assays have indicated lower sensitivities, particularly for LFIA, even when conducted more than 21 days after the onset of symptoms [33].
Developing immunoassays with higher sensitivity and specificity is essential to avoid false-positive and false-negative outcomes.
Viral entry into cells is facilitated by binding the RBD part of the S1 protein and human (part of S1) angiotensin-converting enzyme-2 (hACE-2).445-456, 473-477, and 484-505 residues of the RBD part are responsible for the interaction, ultimately leading to a cascade of events [34,35].Upon infection with the SARS-CoV-2 virus, the anti-RBD IgG thus produced lingers within the body for up to 75 days.These antibodies do not cross-react with other known circulating human coronaviruses [36,37].Henceforth, the developed ELISA assay can detect and study seroconversion and vaccine efficacy [31].
This study aims to develop local ELISA assays targeting antibodies against a combination of spike S1 protein and RBD.When S1, RBD, and N were immobilized individually, surprisingly, S1 showed the lowest sensitivity (68.4%) among the three antigens when challenged with samples collected <14 days after symptom onset (Table 1).A similar observation was found by Brochot et al., where RBD and N showed better sensitivity than S1 in the earlier phases of infection [38].Despite RBD being a part of S1, the immobilized proteins' equimolar concentration can explain the early sensitivity phenomenon, where RBD is the dominant immune-reactive site [39].However, all three antigens were good seroconversion markers for convalescent sera characterization, reaching a sensitivity close to 100% (Table 1).However, RBD was found to cross-react more (6/50) with dengue sera than the rest of the two (1/50) (Figure 4), an observation previously reported by Nath et al. [40].
Comparing our in-house N-specific IgG ELISA and commercially available chemiluminescence immunoassay (Roche) with P3 samples, we obtained equal sensitivity for N-specific IgG ELISA and RBD-ELISA while being higher when compared with that of S1-ELISA (Table 1).However, in P4 samples, N-IgG ELISA misidentified ~2% of the positive samples as negative at a rate similar to previously reported (Table 1) [41].The presentation of the early inception of IgG against N can be attributed to the 90% amino acid homology of SARS-CoV.Exposure to SARS-CoV or other human coronaviruses may cause cross-reaction while declining the specificity [42,43].Our observation of a higher titer of S1 antibodies than RBD and N in COVID-19 patients (Figure 3) is in congruence with the dynamic and kinetic heterogeneity of the antibody titers reported previously [31,44].A similar observation was also reported by studies performed with children with multisystem inflammatory syndrome and other diseases [27,45].
In the dengue-endemic region, especially in the lower and lower-middle-income countries, dengue is primarily diagnosed using serological assays.In the COVID-19 pandemic era, cross-reaction of dengue samples in COVID-19 serological assays or vice versa creates misdiagnosis and generates false epidemiological data [10][11][12][13].Yan et al. have reported cases where COVID-19 patients' sera cross-reacted with dengue IgG/IgM assays [11].Additionally, others have said that the dengue-infected patient's sera cross-reacted with SARS-CoV-2 antigens [10,46,47].On the other hand, Spinicci et al. reported low chances of a cross-reaction between these two diseases [48].The target antigens and their quality may contribute to the variable outcomes of the assay kits.
Nath et al., using computational modeling with DENV antibody crystal structures, predicted 19 antigenic cross-reaction sites in the SARS-CoV-2 RBD domain, among which seven fall in the ACE-2 binding region (438-506) [40].Similar cross-reactions can be found in the Masyeni et al. study, where DENV-positive sera cross-reacted in COVID-19 serological assay kits with RBD as the target antigens [10].Lustig et al., on the other hand, using an in silico model, predicted the major cross-reacting site of spike protein to be the HR2 domain [46].The latter reported a 22% cross-reaction of DENV sera with the COVID-19 assay targeting S protein (EUROIMMUN ELISA).However, their claims contradict their in silico result, as the assay kit used the S1 domain, in which the HR2 domain is absent [49,50].Moreover, dengue pre-pandemic infection sera reacted with RBD (3/37), whereas S1 showed less cross-reaction (1/47) when they checked the IgG antibodies.Because RBD is a part of the S1 domain, a higher molar concentration of RBD can justify such a cross-reaction when the equimolar amounts of these antigens are compared.Furthermore, a higher crossreaction was observed in S1 (10/47) compared to RBD (1/37) when they carried out the assay for IgA antibodies, which can be further explained by the polymeric forms of IgA, which can recognize a more comprehensive range of antigenic sites [51].
Co-immobilization of antigens, such as N and RBD/S, is found to augment the sensitivity of the immunoassays [52,53].We aimed to improve the overall sensitivity of the assay as well as reduce the crossreaction.When combinations of S1 and RBD were challenged with dengue sera, minimal cross-reaction was observed at an S1:RBD ratio of 80:200 (Table 2).Interestingly, the early sensitivity improved at this ratio compared to S1-ELISA and increased from 68.4 to 84.2 (Tables 1 and 3).Similarly, the specificity was found to be 97.7%, which improved when compared to RBD (95.4%) but reduced slightly concerning S1 (99.2%) (Tables 1 and 3).This observation indicates the RBD domain of S as the dengue cross-reacting part, but again, it is an essential one for overall sensitivity.Despite RBD being a part of S1, a lower molar ratio reduces cross-reaction with dengue-positive sera.Still, careful tweaking of the co-immobilized balance of RBD and S1 can improve sensitivity with a minimum compromise of specificity.
The rigorous and spontaneous mutation of SARS-CoV-2 has created chaos worldwide, occurring in multiple infection waves with several deadly variants [54][55][56].Worldwide, proper vaccine distribution and heterologous doses have become a critical concern under these circumstances [57][58][59].Besides, serological testing is required to assess the efficacy of different vaccines and understand booster doses' necessity [60].Therefore, it is essential to reduce the plausible cross-reaction of serological tests to critically investigate the requirement of a booster dose along with the minimization of misdiagnosis, accurate prognosis, and sero-surveillance in suspected confirmed hospitalized, recovered, and vaccinated patients [60,61].
While our findings are noteworthy, as they reveal cross-reactivity between anti-dengue antibodies against the SARS-CoV-2 antigens, it's important to acknowledge certain limitations.Unfortunately, we could not evaluate cross-reactivity with SARS-CoV and other prevalent coronaviruses due to budgetary constraints and limited facilities.Additionally, the relatively small sample size in the P3 group adds to the limitations of this study.Nevertheless, the minimal observed cross-reactivity suggests that the developed ELISA assay can be effectively utilized in regions where dengue is endemic.Moreover, spectrum biases do not influence these assays, further enhancing their applicability and reliability.

Conclusions
The S1 and N proteins demonstrated a better seroprevalence marker for the analysis of COVID-19.Although RBD was found to be slightly cross-reactive with anti-dengue antibodies, it is crucial for the early diagnosis of COVID-19 antibodies in serum.Therefore, co-immobilization of S1 and RBD with an appropriate and stable ratio is suggested to minimize the plausible cross-reaction.
interest in the submitted work.Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.

FIGURE 2 :
FIGURE 2: Detection of SARS-CoV-2 S1-IgG, RBD-IgG, and N-IgG among the overall participants; the area under the ROC curve was considered both true and test cases S1: spike-1; IgG: immunoglobin G; RBD: receptor binding domain; N: nucleocapsid; ROC: receiver operating characteristic

FIGURE 3 :
FIGURE 3: Detection of SARS-CoV-2 N-IgG, RBD-IgG, and S1-IgG among the SARS-CoV-2 confirmed patients with an overall period (A), <14 days (B), and >14 days (C).The ratio of OD/cut-off of N-IgG, RBD-IgG, and S1-IgG of the confirmed positives with SARS-CoV-2 was shown.Data are presented as mean with range (minimum and maximum).A nonparametric Wilcoxon signed-rank test was used to estimate the p-value.

FIGURE 5 :
FIGURE 5: Detection of SARS-CoV-2 IgG against S1 and RBD combinations among the overall participants; the area under the ROC curve was considered both true and test cases S1: spike-1; IgG: immunoglobulin G; RBD: receptor binding domain; ROC: receiver operating characteristic

TABLE 3 : Sensitivity, specificity, and ROC analysis for S1 and RBD combination (0.8:0.2) IgG in RT-qPCR positive against SARS-CoV-2 and negative controls
Kohen's kappa test was used to estimate the test agreement.