Interpretation of Pooled Metabolic Profiles for Evaluating Transition Cow Health Status
For metabolic profiling, typically between 8 and 12 healthy individual cows within defined physiologic state categories are sampled and a battery of blood analytes determined. Metabolic profile results are interpreted as either individual mean values or percent of individuals (8 to 12) deviating from some defined criteria. Use of pooled samples was evaluated as a method to collect usable information on herd metabolic status without the high cost of individual sampling. Objective of this study was to determine if diagnostic interpretation guidelines can be established for pooled metabolic profile samples.
Methods and Materials
Blood samples collected from 113 cows on 15 different farms for three defined time periods relative to calving (Early Dry, Close-up Dry, Fresh). Pooled samples (n=48, 16 per time period) containing between 5 and 12 individuals were randomly composited by blending equal volumes (100 to 500 μl) of individual serum. Metabolic profiles were performed on individual and pooled samples and included urea nitrogen (SUN), glucose (Glu), albumin (Alb), aspartate amino-transferase (AST), sodium (Na), potassium (K), chloride (Cl), calcium (Ca), magnesium (Mg), total cholesterol (Chol), β- hydroxybutyrate (BHB), and nonesterified fatty acids (NEFA). Defined criteria were used to determine normal or abnormal analyte status on individuals and percent abnormal values within a pool. Using population statistics (median, standard deviation [SD]) from individual healthy cows (n=49), relationship between percent abnormal values and standard deviation difference from median were tested by ANOVA and regression modeling.
Results and Discussion
Pooled samples cannot be interpreted in the same manner as individual animal samples; comparing the measured value to a defined laboratory reference range. Expected values for pooled samples were derived from median blood analyte concentrations from a defined population of healthy fresh dairy cows (Table 1). For each analyte measured, arithmetic mean or pooled sample value were subtracted from healthy population median and divided by the analyte’s SD for the healthy population to quantify sample deviation from expected. Number of SD the mean individual and pooled sample values deviated from population median was related to percent of abnormal values within the sample population, based on individual animal evaluations.
Table 1. Populations statistics (median and standard deviation) for selected analytes in healthy fresh cows and reference values, used on individual cow samples within a sampled group, to determine the percent abnormal values within a group.
|BUN||mmol/L||10.7||3.4||< 7.1, >12.8|
|Calcium||mmol/L||2.3||0.25||< 2, > 3|
A linear model (Figure 1) was determined to be the best association between percent abnormal values within a pool, independent of pool size, and SD difference from reference healthy population median. All models within Fresh period were significant (r2=.26 to .81, P<.04 - .0001), except Cl. Close-up period models were significant for Alb, Chol, Cl, Mg, Na, K and NEFA (r2=.36 to .79, P<.01 - .0001).
Figure 1. Relationship between percent abnormal values and number of standard deviations the sample value (mean of individuals or pooled sample) deviates from healthy population median NEFA (A), BHB (B), albumin (C) and glucose (D) concentrations in sampled fresh cows.
For every increase of 10% abnormal values within a pooled sample, analyte concentration deviated from the healthy population median by 0.23 SD (range: 0.105 - 0.5) with deviations being analyte-specific. Pooled sample reference ranges, relative to percent abnormal values within the pool, can be determined from healthy population SD multiplied by predicted analyte deviations (Table 2).
Table 2. Preliminary reference criteria for pooled samples of selected analytes based on various sigma levels around the healthy population median.1
|Analyte||Slope2||Sigma Level Range: 0.25
||Sigma Level Range: 0.50
||Sigma Level Range: 0.75
||Sigma Level Range: 1.0
1Yellow (italicized) and red (bold and italicized) shaded cells represent >20% and >40% abnormal values, respectively, if a pooled sample value is outside previous range and within highlighted ranges.
2Slope value was obtained from linear models regressing number of SD per 10% of abnormal values within a sampled population.
- Though some variation may be masked, pooled sampling may be used as an economic herd metabolic status screening tool.
- Statistical process control may be a reasonable approach to using pooled samples for on-going herd monitoring.
- This interpretation approach to pooled samples requires further refinement and validation, but is a reasonable starting point.
Research funded by Pennsylvania Department of Agriculture.
Prepared by Robert J. Van Saun, Department of Veterinary and Biomedical Sciences Pennsylvania State University; University Park, Pennsylvania USA 16802-3500
table 2. as colored graphic
TitleInterpretation of Pooled Metabolic Profiles for Evaluating Transition Cow Health Status
This publication is available in alternative media on request.