Surrogate test performance for SARS-CoV-2 neutralizing antibodies (nAbs) for convalescent plasma (CCP): How useful could they be?

Abstract

Background: COVID-19 high-titer CCP selection is a concern, because neutralizing antibody (nAb) testing requires sophisticated labs and methods. Surrogate tests are an alternative for measuring nAb levels in plasma bags, including those that are pathogen-reduced.

Study design/methods: We studied a panel consisting of 191 samples from convalescent donors tested by nAb (CPE-VNT), obtained from 180 CCP donations (collection: March 20-January 21) and 11 negative controls, with a total of 80 and 111 serum and plasma samples (71 amotosalen/UV treated), with nAb titers ranging from negative to 10,240. Samples were blindly tested for several surrogates: one anti-RBD, two anti-spike, and four anti-nucleocapsid tests, either isolated or combined to improve their positive predictive values as predictors of the presence of high-titer nAbs, defined as those with titers ≥160.

Results: Except for combined and anti-IgA/M tests, all isolated surrogate tests showed excellent performance for nAb detection: sensitivity (98.3%-100%), specificity (85.7%-100%), PPV (98.9%-100%), NPV (81.3%-100%), and AUC (0.93-0.96), with a variable decrease in sensitivity and considerably lower specificity when using FDA authorization and concomitant nAb titers ≥160. All surrogates had AUCs that were statistically different from CPE-VNT if nAb≥160, including when using combined, orthogonal approaches.

Conclusions: Surrogate tests (isolated or in combination) have an indirect good performance in detecting the presence of nAb, with lower sensitivity and specificity when high nAb titer samples are used, possibly accepting a considerable number of donors whose nAb titers are actually low, which should be evaluated by each laboratory responsible for CCP collection.

Keywords: COVID-19; SARS-CoV-2; coronavirus; convalescent plasma therapy; passive immune therapy; surrogate tests.

FIGURE 1

Diagram of samples distributed according to serum/plasma or pathogen reduction treatment. All samples were tested by eight isolated tests. anti‐RDB, competitive anti‐RBD inhibition test; nAb, neutralizing antibodies; NP, nucleoprotein

FIGURE 2

Left: Box‐plot from nAb titer (natural log) from CCP donor samples, based on plasma or sera (top left) or pathogen reduction treatment (bottom left); with no statistical difference between groups (Mann–Whitney test); negative controls (NC; n = 11) were excluded. Right: Histogram density distribution of nAb titer (natural log), based on material (plasma or sera, top right), or pathogen reduction treatment (bottom right); both had a statistical difference in nAb titer distribution, (p = .048 and < .001, respectively, by the chi‐square test Χ²). The vertical dashed lines represent nAb titer ≥160 [Color figure can be viewed at wileyonlinelibrary.com]

FIGURE 3

Median reactivity of anti‐SARS‐CoV‐2 tests from 180 samples, based on material (plasma and sera, upper quadrants) or pathogen reduction treatment (lower quadrants). Negative controls were not included, as they did not bear any SARS‐CoV‐2 Abs. There is no statistical difference in reactivity for each test based on plasma/sera or non‐treated/treated samples. Except for anti‐RBD test (given in % inhibition – left), all signal/cut‐off (S/CO) ratio were normalized into natural logarithm. ICB‐G, anti‐IgG NP; ICB‐A, anti‐IgA NP; ICB‐M, anti‐IgM NP; R, Roche; O, Ortho; Spk, Spike; NP, nucleoprotein [Color figure can be viewed at wileyonlinelibrary.com]

FIGURE 4

Quadratic regression between lnAb and anti‐RBD (% inhibition, upper left); ICB anti‐IgG NP (Ln S/CO, upper middle); Roche NP (Ln COI, upper right); Ortho Spike total (Ln Index, bottom left) and Roche Spike (Ln U/ml, bottom right). Horizontal red and black dashed lines represent reactive reactions based on manufacturer's instructions and according to the FDA authorization, respectively (except for ICB‐IgG, which was defined as S/CO ≥5.0); vertical black dashed line marks nAb titer ≥160 [Color figure can be viewed at wileyonlinelibrary.com]

FIGURE 5

ROC curve for surrogate tests compared against samples with nAb titers ≥20 (gold standard, left) or nAb titers ≥160 and FDA authorization for high‐titer samples (right). Except for ICB‐A, ICB‐M, and all combined tests (O‐Spk+anti‐RBD68 or R‐Spk+anti‐RBD68), all methods were statistically non‐significant whenever nAb titers ≥20 and manufacturer's instructions were followed (left); however, all surrogate methods following FDA authorization were statistically inferior (p < .0001) when compared to samples with nAb titers ≥160 (right). ICB‐A and ‐M were not shown in the right quadrant, because they already were statistically different by the former analysis. nAb20, neutralizing antibody titer ≥20; anti‐RBD, competitive anti‐RBD inhibition test; anti‐RBD68, ≥68% competitive anti‐RBD inhibition test; ICB‐G, anti‐IgG NP; ICB‐A, anti‐IgA NP; ICB‐M, anti‐IgM NP; R‐NP, Roche NP; O‐SPk, Ortho Spike; R‐Spk, Roche Spike; Spk‐0.8 and Spk1.0, anti‐Spike tests based on the manufacturer's cut‐off; SPk109 and SPk132, anti‐Spike tests based on FDA guidance for high‐titer; NP, nucleoprotein. *Both combinations for O‐Spk+ anti‐RBD68 presented the same result, being merged into a single line. **p < .0001 against reference (nAb20) [Color figure can be viewed at wileyonlinelibrary.com]