Clinical prognostic scale for the Pacific oyster Crassostrea gigas exposed to a pathogen: application to OsHV-1 (Oyster Herpesvirus type 1)

Abstract

When epizooty involving a pathogen affects the Pacific oyster Crassostrea gigas, the current diagnostic of this disease requires sacrificing animals to sample tissues before performing laboratory tests. This study highlights the interest of a clinical scale based on physical examinations of C. gigas during breeding to predict the outcomes of a disease involving a pathogen, with the example of an infection with OsHV-1 (Oyster Herpesvirus type 1). Established during preliminary tests of experimental transmissions of OsHV-1, this ordinal and nested scale on 6 stages has been the subject of several (i) development assays with the proposal of a univariate Cox model of the survival duration of C. gigas exposed to seawater contaminated by OsHV-1, and (ii) internal validation assays for its discrimination (c statistic, ROC curve) and its calibration (Hosmer-Lemeshow test, graphic representation of observed and predicted survival durations). The scale exhibits very good discrimination on the 3rd day (D2 post exposure) and on the 4th day (D3 post exposure), with c-statistics of 0.86 and 0.89 respectively. Concerning calibration, Hosmer-Lemeshow test shows the observed values were not significantly different from the predicted values at a threshold of 5% (ddl = 5; chi2 < 11.7), confirming the graphic representation of the relation between the observed and the predicted survival durations. Improvement assays of the clinical prognostic scale will be considered by adjusting the number of its stages and by adding non-destructive physiological biomarkers to move from an univariate to a multivariate model of survival duration.