Table 1.
Candidate biomarkers for ST abuse and their expected response upon ST treatment in human and cows.
Figure 1.
Work flow for serum preparation, generic serum pre-treatment and 4-plex FCIA for serum candidate biomarkers.
A detailed description can be found in Materials S1. Abbreviations: h – hour, IGF-1 – insulin-like growth factor 1, IGFBP2 – IGF binding protein 2, GS I – glycine solution I, GS II glycine solution II, min – minutes, PBST – phosphate-buffered saline with 0.05% (v/v) Tween-20, PBSTB – 0.1% (m/v) BSA in PBST, PE – phycoerythrin fluorescent label, rbST – recombinant bovine somatotropin, RT – room temperature, sec – seconds.
Table 2.
4-plex FCIA assay performance characteristics for the single candidate biomarkers.
Figure 2.
Biomarker profiles of rbST-treated (left) and untreated (right) dairy cows.
Profiles from animal study I (dotted lines) and animal study II (solid lines) are shown. Sera from adaption period (3 sera from every cow), treatment period (13 sera per cow from animal study I and 9 sera per cow from animal study II) and withdrawal period (5 sera per cow from animal study I and 6 sera per cow from animal study II) were measured in duplicate. Biomarkers shown are concentrations of IGF-1 (A), B/B0 levels of IGFBP2 (B), B/Bd levels of antibodies against rbST (C) and concentrations of osteocalcin (D). The rbST treatment schedules for both animal studies are indicated by two black horizontal bars and decision limit per biomarker by the green horizontal line. Note that cows from animal study II received two additional rbST injections after the biweekly treatment period.
Figure 3.
Predictive power of each single candidate biomarker for indicating rbST abuse.
True-positive rates were calculated for all samples from rbST-treated cows in their treatment and withdrawal periods of study I and II. False-positive rates were calculated for all samples from untreated cows from the two animal studies (adaptation period samples from all cows and all the samples from untreated cows). Samples were considered positive if their biomarker value exceeded the respective decision limit. The treatment schedules of the two controlled animal studies are indicated by black horizontal bars on top of the graph. The targeted 95% true-positive (<5% false-compliant) rate according to 2002/657/EC is indicated by the dotted horizontal line.
Figure 4.
Number of biomarkers reacting above the respective decision limit.
Results shown per cow (in animal studies I and II) and day of the controlled animal studies. Each row represents one individual cow. Vertical dotted lines indicate the treatment time points in both animal studies.
Figure 5.
Predictive power (shown as true-positive and false-positive rates) of the additive biomarker analysis.
True-positive rates were calculated for all samples from rbST-treated cows in their treatment and withdrawal periods of study I and II. False-positive rates were calculated for all samples from untreated cows from the two animal studies (adaptation period samples from all cows and all the samples from untreated cows). Samples were considered positive if one of the candidate biomarkers exceeded its respective decision limit. The treatment schedules of the two animal studies are indicated by black horizontal bars on top of the graph. The targeted 95% true-positive (<5% false-compliant) rate according to 2002/657/EC is indicated by the dotted horizontal line.
Figure 6.
True-positive rates following statistical multiple biomarker analysis.
True-positive rates, obtained with the prediction models based on the eleven different biomarker combinations, were calculated for rbST-treated cows from animal study I in their treatment and withdrawal period. The treatment schedules of animal study I is indicated by black horizontal bars on top of the graphs. The targeted 95% true-positive rate according to 2002/657/EC is indicated by the dotted horizontal lines.