Use of a multiantigen detection algorithm for diagnosis of Kaposi's sarcoma-associated herpesvirus infection

Abstract

The ability to readily and accurately diagnose Kaposi's sarcoma-associated herpesvirus (KSHV, or human herpesvirus 8) infection in individuals remains a demanding task. Among the available diagnostic methods, sensitivities and specificities range widely, and many are inadequate for large-scale screening studies. We examined a serological algorithm for detecting KSHV in human sera having high sensitivity and specificity. This method uses previously described open reading frame (ORF) K8.1 and ORF65 peptide-based enzyme-linked immunosorbent assays and a novel purified recombinant full-length LANA1 protein. We generated two multiantigen algorithms: one that maximized sensitivity and one that maximized specificity. These serological algorithms were then used to evaluate seroprevalence rates among populations of clinical and epidemiological importance. The serological algorithms yielded sensitivities of 96% and 93% and specificities of 94% and 98% for the more sensitive and specific algorithms, respectively. Among kidney donors, seroprevalence was low, 4.0% (2/50), and similar to that of blood donors (P = 0.46; odds ratio [OR], 1.4; confidence interval [CI], 0.14 to 7.9) using the highly specific algorithm. Using the sensitive algorithm, 8.0% (4/50) were infected compared to 6.4% (16/250) observed among blood donors (OR, 1.3; CI, 0.41 to 4.0; P = 0.43). Among subjects requiring bone marrow transplantation, seroprevalence rates were not elevated compared to those of blood donors (OR, 2.0; 95% CI, 0.10 to 122.9; P = 0.50). Because the need for high-quality KSHV detection methods are warranted and because questions remain about the optimal methods for assessing KSHV infection in individuals, we propose a systematic approach to standardize and optimize the assessment of KSHV infection rates using a combination of established and novel serological assays and methods.

FIG. 1.

ELISA plating scheme. Each sample tested has two wells coated with antigen (indicated by gray shading in columns 1, 2, 5, 6, 9, and 10) and two wells that are not coated with antigen (unshaded circles of columns 3, 4, 7, 8, 11, and 12). The average of the noncoated wells, used as a background correction, is subtracted from the average of the peptide-coated wells to give the adjusted optical density.

FIG. 2.

Effect of antigen concentration on rLANA-G reactivity for case (left) and a control (right) sera. Increasing amounts of rLANA-G antigen increase case serum reactivity until a plateau is reached at about 0.125 μg per plate, without increasing nonspecific reactivity for the control serum. Note that the control serum reactivities are shown in a rescaled graph to allow comparison.

FIG. 3.

Distribution of optical density values among blood donors (closed circles) and subjects with Kaposi's sarcoma (open circles) for K8.1 (A), ORF65 (B), and LANA1 (C) ELISAs.

FIG. 4.

Receiver operator characteristic curve of individual assay performance. Symbols indicate sensitivity and specificity of assays at differing optical density values.

FIG. 5.

Selection method for determination of high-sensitivity and high-specificity assay algorithms.

FIG. 6.

The effects of multiple freeze-thaws on a human seroreactive serum sample. (A) One patient serum seroreactive for ORF65, LANA1, and K8.1 was frozen and thawed a total of 22 times and tested in each assay. (B) Linear regression of samples from panel A.

FIG. 7.

Representative IFA examples from a KSHV-infected subject with KS and from two SLE patients with seroreactive ELISA results. Note that the characteristic speckled nuclear staining pattern seen in the KS patient serum is absent from the SLE patient sera, suggesting nonspecific cross-reactivity.