Remote Fingerstick Blood Collection for Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Antibody Testing

Contrived whole blood was generated by mixing serum with an equal volume of type O–negative packed red blood cells (hematocrit of approximately 80%). Whole blood was obtained from discarded segments (tubing) from whole-blood donations. Fingerprick blood was expressed using a spring-loaded lancet (Becton Dickinson). There were 20-μL Mitra devices (Neoteryx LLC) used to collect fingerprick and blood sample specimens, followed by dry storage for at least 24 hours. Mitra is a US Food and Drug Administration class I exempt medical device registered with the Food and Drug Administration.

Mitra devices containing dried blood samples were placed in a deep (2-mL) 96-well plate with a fitted holder (Neoteryx LLC) that allows for the devices to be slightly suspended. The deep 96-well plates contained 200 μL per well of extraction buffer, which consisted of 1% bovine serum albumin, 50 mM Tris (pH 8.0), 140 mM NaCl, and 0.05% Tween-20. Our extraction buffer contained bovine serum albumin to help stabilize antibodies as well as 0.02% Tween-20 to help facilitate extraction and inactivate any potentially infectious virus.¹⁰  Plates were then placed on a shaker for 2 hours at 600 RPM. Devices were subsequently removed, leaving 200 μL of dried blood extract in each well. Of note, extracts exhibited significant hemolysis (dark red in color).

We used the electrochemiluminescence-based Roche Elecsys anti-SARS-CoV-2 immunoassay, which detects IgG, IgM, and/or IgA antibodies to the SARS-CoV-2 nucleocapsid protein. The assay was run on a Roche Cobas 8000 e801 Immunoassay Analyzer, which also measured hemolysis and lipemia indices. A minimum volume of approximately 200 μL of sample per run was required when using a tube insert.

Use of patient samples for the development and validation of SARS-CoV-2 diagnostic tests was approved by the Mass General Brigham Healthcare (Boston, Massachusetts) Institutional Review Board (protocol 2020P000895).

In order to determine the feasibility of running dried blood extracts from 20-μL Mitra devices on the Roche Elecsys SARS-CoV-2 total antibody assay, we first tested serum samples from COVID-19 patients (confirmed by nasopharyngeal swab polymerase chain reaction) diluted 1:20 in extraction buffer. A dilution of 1:20 was chosen based on a conservative estimate that assumes that of the 20 μL of whole blood that dries onto 20-μL Mitra device, there will be at least 10 μL of dried serum extracted into 200 μL of extraction buffer. Analysis of neat serum versus 1:20-diluted serum from 72 COVID-19 patients showed a positive, nonlinear correlation of cutoff index (COI) values measured by the Roche Cobas 8000 e801 Immunoassay Analyzer as well as a high dynamic range of detection with COI ranging from approximately 0.08 to 140 (Figure 1, A). Using the threshold of 1.0 established by the manufacturer, neat serum samples showed 86% (62 of 72) seropositivity, whereas diluted serum samples showed 46% (33 of 72) seropositivity. However, adjusting the threshold to 0.15, the seropositivity of diluted serum samples increased to 88% (63 of 72). Previous studies have demonstrated that this adjustment in the COI threshold for the Roche assay can increase the sensitivity of SARS-CoV-2 antibody detection without sacrificing specificity.¹¹ 

Having demonstrated that the Mitra-Roche assay can accurately detect SARS-CoV-2 antibodies in diluted serum samples, we proceeded to generate extracts from dried blood collected on Mitra devices using contrived whole blood with known serostatus. For this, we mixed serum samples from 48 COVID-19 patients and 48 prepandemic individuals with an equal volume of O-negative donor packed red blood cells to mimic whole blood, which was subsequently collected onto 20-μL Mitra devices. After 24 hours of dry storage, samples were extracted and run on the Roche Elecsys assay. Quality control measurements made by the Roche Cobas instrument of hemolysis and lipemia were included, and each extracted sample, which was dark red in color, unsurprisingly contained high hemolysis and lipemia indices due to the nature of the sample (data not shown). Nevertheless, this did not affect the SARS-CoV-2 antibody assay results, and using a threshold of 1.0 for neat serum and 0.15 for extracted samples, a concordance of 98% (94 of 96) was achieved, with the only 2 discordant specimens being COVID-19 patient samples (Figure 1, B). Of the 2 discrepant COVID-19 patient samples, one was positive by the neat serum specimen but just below the threshold by the dried blood specimen, and the other was positive by the dried blood specimen but just below the threshold by the neat serum specimen. In addition, a false-positive prepandemic specimen—which was selected as 1 of 2 prepandemic specimens that were false positive in the Roche assay in a cohort of 1200 prepandemic specimens—was still detected in the extract sample, indicating that false-positive signals indeed dilute with sample dilution, and could reflect truly cross-reactive antibodies. We also generated duplicate Mitra collections devices from contrived whole blood and performed extractions in 2 independent facilities and found high precision (R² = 0.97; Figure 1, C), highlighting the reproducibility of this testing strategy.

To test the performance of the combined Mitra-Roche assay on low- and high-seroprevalence cohorts during the pandemic, we collected specimens from prescreened blood donors at the Massachusetts General Hospital Blood Donor Center as well as in a community seroprevalence study conducted in Chelsea, Massachusetts, the densest and most affected area for SARS-CoV-2 infection.¹²  For blood donors, we collected serum and dried venous blood on Mitra devices and found that the assay showed a seroprevalence of 1% (1 of 143) with 100% concordance between serum and extracted samples (Figure 2, A). For community samples, we collected dried fingerprick blood on Mitra devices and serum from antecubital vein phlebotomy in 402 participants. Dried fingerprick blood extracts and sera were run on the Roche Cobas instrument, which revealed a seroprevalence of 24% (98 of 402) in serum samples, 24% (97 of 402) in extracted samples, and a concordance of 99% (397 of 402; Figure 2, B). Interestingly, we saw a threshold effect in neat serum samples, with extracted samples that showed the highest COI demonstrating slightly decreased COI in the neat serum, most likely representing a “hook effect” (Figure 2, C). This effect, however, did not decrease COI sufficiently to approximate the COI threshold, and thus is unlikely to affect Roche results as a qualitative readout.

The need for remote and large-scale testing for SARS-CoV-2 antibodies is increasing as the population of convalescent individuals continues to increase and vaccines are soon to be implemented. In this study, we validate the use of a volumetric absorptive microsampling device (Mitra) for fingerprick blood on a widely used, EUA-approved SARS-CoV-2 antibody test (Roche Elecsys). Currently, the Roche Elecsys test detects IgG, IgM, and/or IgA antibodies to SARS-CoV-2 nucleocapsid protein, an antigen that is not present in most SARS-CoV-2 vaccines. However, a Roche Elecsys assay that detects antibodies to SARS-CoV-2 spike is currently being developed¹³  and will be useful to detect immune responses to vaccines, all of which contain all or a portion of the spike antigen. Both tests together can even help discriminate individuals who were infected with circulating virus (ie, positive for both nucleocapsid and spike antibodies) or administered one of the vaccines being developed (ie, negative for nucleocapsid antibodies but positive for spike antibodies). As such, this protocol and collection method, which is readily adaptable to the upcoming version of Roche Elecsys assays, can enable remote confirmation of vaccination responses. Currently, the National Institutes of Health is using this collection method in conjunction with an in-house developed serologic assay, but our study validates its use in facilities already deploying the Roche Elecsys SARS-CoV-2 total antibody assay. In addition, although performance may not be guaranteed because of potential interferences from extracted dried blood, this study provides a straightforward framework for validating other SARS-CoV-2 serologic testing platforms, such as those from Euroimmun, Bio-Rad, Diasorin, Abbott, and Siemens. Overall, we believe these findings to be of significance for clinical laboratories around the world as we combat this pandemic.

We wish to thank Robert S. Makar, MD, PhD, for providing whole-blood samples from blood donors. We also thank Edward T. Ryan, MD, for providing prepandemic serum samples.