Detection of SARS-CoV-2 Antibodies in Kidney Transplant Recipients.

Abstract

Kidney transplant recipients and other patient groups receiving immunosuppression have a poor prognosis following presentation with symptomatic severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection.1 The immune response to SARS-CoV-2 in an immunocompromised population has not been systematically reported. Recognition that humoral immune responses against common viral infections are blunted in such patients has led to their exclusion from validation studies of serologic assays for SARS-CoV-2.2,3 In this study, we analyze the seroprevalence of SARS-CoV-2 antibodies in a transplant population. In order to ensure the accuracy of the seroprevalence rate, we also evaluate the performance of different serologic assays within this patient cohort. We investigated 855 consecutive kidney transplant recipients who attended the phlebotomy service at the Imperial College Renal and Transplant Centre (ICTRC) London in June 2020 for SARS-CoV-2 antibodies. Patient demographics were obtained from the ICRTC transplant registry (Table 1). The study was approved by the Health Research Authority, Research Ethics Committee (reference: 20/WA/0123). Table 1. - Patient characteristics Variable Patients, n=855 (%) Sex Men 546 (63.9) Women 309 (36.1) Age, yr Median 57 (45–66) Ethnicity White 305 (35.7) Black, Asian, and minority ethnic 550 (64.3) Cause of ESKD ADPKD 82 (9.6) GN 235 (27.5) Diabetic nephropathy 199 (23.3) Urologic 50 (5.8) Unknown 213 (24.9) Other 76 (8.9) Time at ESKD Pre-emptive 172 (20.1) Dialysis dependence 683 (79.9) Median time, yr 1.7 (0.2–3.5) Timing post-transplant, yr ≤1 191 (22.3) >1 664 (77.7) Median time 3.7 (1.1–7.8) Type of graft Deceased donor transplant 516 (60.4) Living donor transplant 276 (32.3) Simultaneous pancreas-kidney transplant 42 (4.9) Antibody-incompatible transplant 21 (2.5) Immunosuppression at diagnosis FK monotherapy 512 (59.9) FK and MMF 187 (21.9) FK and steroids 44 (5.1) FK, MMF, and steroids 111 (13.0) Other 1 (0.1) Induction agent used Alemtuzumab 766 (89.6) IL-2 receptor blocker 89 (10.4) History of rejection Yes 115 (13.5) No 740 (86.5) Transplant number First 761 (89.0) Greater than or equal to second graft 94 (11.0) Diabetes Yes 312 (36.5) No 543 (63.5) ADPKD, autosomal dominant polycystic kidney disease; FK, tacrolimus; MMF, mycophenolate mofetil. Sera from all patients were tested for the presence of nucleocapsid protein (NP) antibodies using the Abbott SARS-CoV-2 IgG assay on the Abbott Architect system. Samples were interpreted as positive or negative according to the manufacturer’s instructions with a cutoff index of 1.4.4 All samples with an index value of >0.25 (Supplemental Material) were run on a second assay, the Fortress Diagnostics COVID-19 Total Antibody assay, which is a nonquantitative two-step antigen sandwich ELISA that detects total Ig against the receptor binding domain (RBD). Samples are interpreted as positive on the basis of a cutoff value from negative controls assayed on the sample microplate. Samples showing discordant results on the Abbott and Fortress assays were additionally tested using a commercially available lateral flow immunoassay (LFIA; Biomedomics Inc.), which detects both IgM and IgG to the recombinant antigen MK201027 of the RBD.5 The assay was used as per the manufacturer’s instructions, results were assessed by two independent blinded observers, and only IgG results were considered. Serologic samples taken from 85 health care workers (HCWs) with RT-PCR–confirmed infection were used to compare assay performance in an immunocompetent population. Statistical and graphical analyses were performed with MedCalc v19.2.1. The two-sided level of significance was set at P<0.05. The 95% confidence interval (95% CI) of the seroprevalence was calculated from binomial probabilities using Wilson methods. Concordance between assays was analyzed using Cohen κ coefficient of qualitative results. Sixty-nine of 855 patients tested positive for SARS-CoV-2 IgG using the Abbott assay, giving a seroprevalence of 8.1% (95% CI, 6.4 to 10.1). However, it was noted that 33 of 855 (3.9%) study patients had prior infection confirmed by RT-PCR, of whom 11 of 33 (33.3%) were serologically negative for IgG using the Abbott assay at a median time of testing of 36 (28–58) days postdiagnosis. To investigate the lack of seroconversion versus inadequate assay sensitivity in an immunocompromised population, we tested samples from 38 transplant recipients (including 33 from our screening cohort) with PCR-confirmed infection across three assays. Patients were tested at a median time of 35 (22–53) days postdiagnosis. All paired historical control samples, which had been taken and stored from our study patients prior to July 2019, were negative for IgG across all three assays. The numbers of patients with antibodies detected by the Abbott, Fortress, and LFIA assays were 26 of 38 (68.4%), 35 of 38 (92.1%), and 31 of 38 (81.6%), respectively (Table 2). Patient characteristics by antibody status are shown in Table 3. Three of 38 (7.9%) patients did not have detectable antibodies on any assay, and these patients may represent true failure to seroconvert. Table 2. - Test characteristics as determined by sampling historic and current sera from 38 patients who were SARS-CoV-2 RT-PCR positive Result Abbott Fortress Diagnostics Biomedomics (LFIA) Historic negative controls Positive 0 0 0 Negative 38 38 38 Positive controls Positive 26 35 32 Negative 12 3 6 Sensitivity 68.4 (51.3–82.5) 92.1 (78.6–98.3) 84.2 (68.7–94.0) Specificity 100.0 (92.3–100.0) 100.0 (92.3–100.0) 100.0 (92.3–100.0) Positive predictive value 100.0 100.0 100.0 Negative predictive value 79.3 (70.6–6.0) 93.9 (83.8–97.8) 88.5 (78.6–94.1) Accuracy 85.7 (76.4–92.4) 96.4 (89.9–99.3) 92.9 (85.1–97.3) Table 3. - Patient characteristics by antibody status and assay in patients who were RT-PCR positive Patient No. Abbott Fortress LFIA Timing of Test Postdiagnosis (d) Age, yr Sex Ethnicity First Year Post-Transplant Induction Agent Used Immunotherapy at Diagnosis 22 Negative Negative Negative 35 53 Man BAME Yes Alemtuzumab FK 29 Negative Negative Negative 28 28 Woman BAME Yes Alemtuzumab FK 36 Negative Negative Negative 27 72 Man White Yes Alemtuzumab FK, MMF 1 Negative Positive Positive 65 63 Man BAME No Unknown FK, MMF 6 Negative Positive Positive 70 48 Man BAME No Alemtuzumab FK 11 Negative Positive Positive 55 46 Woman BAME Yes Alemtuzumab FK, MMF 13 Negative Positive Positive 60 79 Woman BAME No Alemtuzumab FK 16 Negative Positive Positive 46 43 Man BAME Yes Alemtuzumab FK, MMF 18 Negative Positive Positive 37 67 Woman BAME No Alemtuzumab FK, MMF 31 Negative Positive Positive 29 76 Man White No Alemtuzumab FK, MMF 38 Negative Positive Positive 9 55 Man BAME No Alemtuzumab FK 33 Negative Positive Negative 21 66 Man BAME Yes Basiliximab FK, MMF 7 Positive Positive Negative 73 50 Man White No Alemtuzumab FK 26 Positive Positive Negative 28 49 Woman BAME Yes Alemtuzumab FK, Pred 24 patients were positive across all three assays 32 (22–41) (median) 52±13 (mean) 14 (58.3%) man 5 (20.8%) White 3 (12.5%) yes 19 (79.2%) alemtuzumab 9 (37.5%) (FK, MMF, Pred) Paired historic sera from all patients were negative across the three assays. BAME, Black, Asian, and minority ethnic; FK, tacrolimus; MMF, mycophenolate mofetil; Pred, prednisolone. To compare assay performance in an immunocompetent population, we tested 85 HCWs with RT-PCR–confirmed infection. At a median time of 31 (19–45) days postdiagnosis, three of 85 (3.5%) HCWs had no detectable antibodies by either the Abbott or Fortress assay, and an additional five of 82 (6.1%) HCWs had no antibodies detected by the Abbott assay. The sensitivity values of the Abbott and Fortress assays in HCWs were 90.6% (95% CI, 82.5 to 95.2) and 96.5% (95% CI, 90.1 to 98.8), respectively. Although there was no difference in the proportion of detectable antibody between the immunosuppressed patients and HCWs using the Fortress assay (P=0.30), immunosuppressed patients were less likely to have a positive serologic test using the Abbott assay compared with HCWs (P=0.002). To investigate potential missed cases of patients who were SARS-CoV-2 IgG positive in our overall cohort screened by the Abbott assay alone, we re-examined the 822 study patients without confirmed infection; 147 of 822 (17.9%) patients had an antibody index value of >0.25 by the Abbott assay, of which 100 had a value between 0.25 and 1.4 and 47 patients had a value ≥1.4. All but four patients were retested using the Fortress assay, and discordant results were seen in 18 of 143 (12.6%) patients. Twelve (12%) of 100 patients negative on the Abbott assay were positive on the Fortress assay, whereas six positive patients by the Abbott assay were negative on the Fortress assay. When these 18 samples were tested by the LFIA, agreement was seen with the Fortress assay in 14 of 18 (77.8%) patients (Supplemental Material). Analyzing the concordance of the assays, we found only a moderate agreement between the Abbott and Fortress assays (κ=0.73 [0.64–0.82]) and between the Abbott and LFIA assays (κ=0.60 [0.46–0.74]), whereas concordance between the Fortress and LFIA assays was strong (κ=0.86 [0.77–0.95]). On amalgamating the results of the Fortress and Abbott assays, the overall seroprevalence in our transplant cohort increased to 10.4% (95% CI, 8.5 to 12.6). The finding of a seroprevalence of 10.4% (95% CI, 8.5 to 12.6) in a cohort of shielded patients with kidney transplants was higher than expected, albeit in patients from a region with a community seroprevalence rate of 13% (Ward H, Atchison CJ, Whitaker M, Ainslie KCE, Elliott J, Okell LC, et al.: Antibody prevalence for SARS-CoV-2 in England following first peak of the pandemic: REACT2 study in 100,000 adults. medRxiv, 2020 10.1101/2020.08.12.20173690). Notably, our study demonstrates the influence of the assay utilized to detect SARS-CoV-2 antibodies and hence, estimate seroprevalence in an immunosuppressed cohort. Our results indicate that the Fortress ELISA and LFIA are more sensitive than the Abbott test at detecting SARS-CoV-2 antibodies in kidney transplant recipients. We also showed that there was better concordance between the Fortress and LFIA assays compared with the Fortress and Abbott assays, which also suggests the potential importance of the target antigen in the serologic assays. The LFIA and Fortress assays share the RBD as their target antigen, whereas the Abbott assay utilizes the NP. It may be proposed that the RBD is more immunogenic than the NP, making it a better stimulus for an immune response in patients who are immunosuppressed. Recently published evaluation studies from Public Health England support this, suggesting that assays targeting NP are less sensitive in immunocompetent populations too.6 Further, we have demonstrated that the Abbott assay was significantly less likely to detect antibody in the immunosuppressed population compared with HCWs. In addition to the greater sensitivity of RBD assays, there is evidence that RBD antibodies may provide information on functional immunity given reported correlations between RBD antibodies and neutralizing antibodies.7,8 It, therefore, follows that assays utilizing the RBD, rather than the NP, may be clinically more relevant for immunosuppressed patients. Our study would have been strengthened by analyzing larger numbers of patients who were RT-PCR positive, incorporating serial sampling, and including demographic data on our HCWs, and we acknowledge that we have not been able to exclude discordance related to the detection of IgM or IgA by the Fortress assay. However, to our knowledge, this is the first SARS-CoV-2 seroprevalence study in patients with transplants, and we have shown that immunoassays that incorporate the RBD as their antigenic target may be superior in testing for SARS-CoV-2 antibodies, without compromising specificity (Table 2). This finding may be seen in immunocompetent people but seems to have a greater effect in an immunosuppressed transplant population.6 Disclosures P. Kelleher reports scientific advisor or membership as an editorial board member for HIV medicine. S. McAdoo reports consultancy agreements with GSK; and honoraria from Rigel Pharmaceuticals, ThermoFisher Scientific, and Celltrion. L. Moore reports research funding from the Chelsea and Westminster Hospital Charity and National Institute of Health Research; honoraria from Eumedica and Profile Pharma; scientific advisor or membership with bioMerieux, Pfizer, and Umovis Lab; and speakers bureau for bioMerieux, Pfizer, and Umovis Lab. N. Mughal reports honoraria from Baxter, Eudmedica, and Pfizer. M. Willicombe reports research funding from Chiesi Pharmaceuticals. All remaining authors have nothing to disclose. Funding This research is supported by the National Institute for Health Research (NIHR) NIHR Imperial Biomedical Research Centre based at Imperial College Healthcare National Health Service Trust and Imperial College London. Data Sharing Statement Data are available from the corresponding author upon reasonable request.