Standardized evaluation of Zika nucleic acid tests used in clinical settings and blood screening

Abstract

Early detection of Zika virus (ZIKV) transmission within geographic regions informs implementation of community mitigation measures such as vector reduction strategies, travel advisories, enhanced surveillance among pregnant women, and possible implementation of blood and organ donor screening or deferral. Standardized, comparative assessments of ZIKV assay and testing lab performance are important to develop optimal approaches to ZIKV diagnostic testing and surveillance. We conducted an expanded blinded panel study to characterize and compare the analytical performance of fifteen diagnostic and blood screening ZIKV NAT assays, including detection among single- and multiplex assays detecting ZIKV, dengue virus (DENV) and chikungunya virus (CHIKV). A 300 member blinded panel was constructed, consisting of 11 serial half-log dilutions ranging from ~104 to 10-1 genome equivalents/mL in 25 replicates each of the Tahitian Asian ZIKV isolate in ZIKV-negative human serum. Additionally, clinical samples from individuals with DENV-like syndrome or suspected ZIKV infection in Brazil were evaluated. The majority of assays demonstrated good specificity. Analytical sensitivities varied 1-2 logs, with a substantially higher limit of detection (LOD) in one outlier. Similar analytical sensitivity for ZIKV RNA detection in singleplex and multiplex assays of the Grifols and ThermoFisher tests were observed. Coefficient of Assay Efficiency (CE), calculated to characterize assays' RNA extraction and amplification efficiency, ranged from 0.13 for the Certest VIASURE multiplex and 0.75 for the Grifols multiplex assays. In general, assays using transcription mediated amplification (TMA) technology had greater CE compared to assays using conventional PCR technology. Donor screening NAT assays were significantly more sensitive than diagnostic RT-qPCR assays, primarily attributable to higher sample input volumes. However, ideal assays to maximize sensitivity and throughput may not be a viable option in all contexts, with other factors such as cost, instrumentation, and regulatory approval status influencing assay availability and selection, particularly in resource constrained settings.

Fig 1. Analytic performance of plasma blood screening and diagnostic assays.

Probit curves comparing analytical sensitivity of assays evaluated in this study by replicate testing of Polynesia isolate blinded panels, grouped by assay category or laboratories. The fitted probit regression model curve (solid lines) and standard errors were used to estimate LOD₅₀ and LOD₉₅ and confidence intervals presented in Table 2.