Determination of an RT-qPCR viral load cutoff point for the etiologic diagnosis of rotavirus A diarrhea in neonate dairy calves

Abstract

Rotavirus A (RVA) is amongst the most widespread causes of neonatal calf diarrhea. Because subclinical infections are common, the diagnosis of RVA-induced diarrhea cannot rely solely on molecular viral detection. However, RT-qPCR allows for quantification of RVA shedding in feces, which can be correlated with clinical disease. Here, we determine an optimal cutoff of rotaviral load quantified by RT-qPCR to predict RVA causality in diarrheic neonate calves, using RVA antigen-capture ELISA as reference test. Feces from 328 diarrheic (n = 175) and non-diarrheic (n = 153), <30-day-old dairy calves that had been tested by ELISA and tested positive by RT-qPCR were included. Of 82/328 (25.0%) ELISA-positive calves, 53/175 (30.3%) were diarrheic, whereas 124/153 (81.0%) non-diarrheic calves tested negative by ELISA. The median log10 viral load was significantly higher in diarrheic vs. non-diarrheic and ELISA-positive vs. -negative calves, indicating a higher viral load in diarrheic and ELISA-positive calves. A receiver operating characteristic (ROC) analysis was conducted using the viral loads of the 175 diarrheic calves that had tested either positive (n = 53, cases) or negative (n = 122, controls) by ELISA. The optimal log10 viral load cutoff that predicted RVA causality in diarrheic calves was 9.171. A bootstrapping procedure was performed to assess the out-of-bag performance of this cutoff point, resulting in sensitivity = 0.812, specificity = 0.886, area under the curve = 0.922, and positive and negative diagnostic likelihood ratios of 11.184 and 0.142, respectively. The diagnostic accuracy of the cutoff was excellent to outstanding. This information will help in the interpretation of RVA RT-qPCR results in feces of diarrheic calves submitted for laboratory testing.