Comparison of turbidometric immunoassay, refractometry, and gamma‐glutamyl transferase to radial immunodiffusion for assessment of transfer of passive immunity in high‐risk beef calves

Abstract

Background

Attainment of adequate transfer of passive immunity (TPI) is critical to health of calves; however, studies comparing available tools for measurement of TPI in individual beef animals are limited.

Objectives

To report agreement between 4 tests evaluating individual TPI status in beef calves.

Animals

One hundred ninety‐six beef calves born to cows and heifers presenting for calving management or dystocia.

Methods

Retrospective study to assess serum immunoglobulin (IgG) concentrations via turbidimetric immunoassay (TI), gamma‐glutamyl transferase (GGT), serum total protein (TP), and single radial immunodiffusion (RID; reference standard). Test agreement was evaluated using Passing‐Bablok regression, Bland‐Altman analysis, Cohen's kappa, and receiver operating characteristic (ROC) curves with and without covariate adjustment to determine optimal thresholds.

Results

Correlation between RID and test results varied: TI, ρ = 0.757; TP, ρ = 0.715; GGT: ρ = 0.413. For the TI compared to RID, regression analysis identified a constant (intercept = −0.51 [CI: −2.63, 3.05]) and proportional (slope = 1.87 [CI: 1.69, 2.08]) bias. Based on ROC, TI concentrations of ≤9.89 and ≤13.76 g/L, and TP concentrations of ≤5.5 and ≤6.0 g/dL, indicated IgG concentrations <18.0 and <25.0 g/L, respectively.

Conclusions and Clinical Importance

Within this cohort of calves, TI demonstrated the best correlation with RID; however, significant bias was identified which led to frequent underestimation of IgG concentration. Serum total protein demonstrated less correlation with RID but had less misclassification than TI. Both TI and TP demonstrated less correlation for calves that received colostrum replacement prompting clinical awareness of colostrum type when evaluating individual TPI in beef calves.

Abbreviations

AAUC

adjusted area under the curve

ALL

all calves in study

AROC

adjusted receiver operator characteristics

AUC

area under the curve

CV

coefficient of variation

GGT

gamma‐glutamyl transferase

IgG

immunoglobulin G

LoA

limits of agreement

MAT

calves receiving maternal colostrum only

POC

point‐of‐care

REP

calves receiving colostrum replacer or supplementation

RID

radial immunodiffusion

ROC

receiver operator characteristics

TI

turbidimetric immunoassay

TPI

transfer of passive immunity

1. INTRODUCTION

Transfer of passive immunity (TPI) via consumption of high‐quality colostrum containing adequate immunoglobulin is critical to neonatal calf health because of the lack of placental transfer of immunoglobulins. Serum IgG concentrations <24.0 g/L in the perinatal period are associated with 1.6 times higher morbidity and 2.7 times higher mortality preweaning when compared to calves with >24.0 g/L IgG. ¹ The detection of inadequate TPI is therefore critical to optimizing beef calf health on both the individual and herd level yet is rarely undertaken. Calves that experience dystocia might be predisposed to additional negative health effects; the ability to rapidly detect inadequate TPI and intervene through either IgG supplementation or increased monitoring could assist in decreasing morbidity and mortality in these animals.

In a veterinary teaching hospital setting, the ability to determine if adequate TPI has occurred before discharge after dystocia or calving management is especially critical to ensuring optimal calf health and building client confidence. Single radial immunodiffusion (RID) is considered the reference standard for measurement of bovine serum IgG levels; however, its routine use for individual animal assessment in clinical practice is not feasible because of the cost and time to run the assay. The use of refractometry for estimating TPI status has been extensively studied for use at the herd level in dairy cattle ² , ³ but has several limitations for use on an individual animal basis. ⁴ Serum gamma‐glutamyl transferase (GGT) is associated with TPI status in ruminants, ⁵ , ⁶ , ⁷ but studies involving this testing modality in comparison to other available methods are lacking in beef calves with only 1 study reporting a minimal association with IgG concentration. ⁸

Recently, a new point‐of‐care (POC) turbidimetric immunoassay (TI) has been developed and is currently marketed in the United States for measurement of bovine, camelid, and equine IgG. Validation of this test from the same manufacturer has been performed for assessment of TPI in foals, ⁹ but to our knowledge, no studies have been published to date comparing the use of this test to other currently available methods for assessment of TPI status in the calf. Previous studies utilizing a similar TI POC assay from a different manufacturer which is no longer commercially available demonstrated good agreement with RID. ¹⁰ , ¹¹ The objective of this study was thus to compare 3 different readily available methods (GGT, TP, and TI) for assessment of TPI status in beef calves born at a teaching hospital and determine which method provides the best estimate of individual TPI status compared to the reference standard (RID).

2. MATERIALS AND METHODS

2.1. Animal information and sample collection

A retrospective study of clinical practice was conducted to evaluate the agreement of currently available stall‐side tests for assessing TPI in beef calves. The study cohort consisted of beef calves born to cows and heifers admitted to the Iowa State University Food Animal and Camelid Hospital at the Lloyd Veterinary Medical Center for calving management or dystocia. Immediately after calving, colostrum samples were collected for analysis of IgG quantity via Brix refractometry; if colostrum quality was deemed unacceptable by the attending clinician (criteria varied based on clinician discretion), colostrum supplementation (60 or 120 g IgG) or replacement (≥180 g IgG) was administered using a commercial colostrum replacer product (Calf's Choice Total Gold, Saskatoon Colostrum Company, Saskatoon, Saskatchewan). The amount of colostrum supplementation or replacement provided along with the time to first colostrum consumption was recorded. Calves that were not observed to nurse dam on their own by 4 hours were given maternal colostrum either by bottle or esophageal feeder if quality was deemed acceptable and it could be safely collected.

Convenience serum samples collected as part of routine clinical management of calves after birth and surviving until sample collection from November 1, 2016, to December 1, 2018, were then utilized for retrospective analysis of methods for TPI assessment in this study. As per clinical protocol, serum was collected via jugular venipuncture on all calves born in hospital to determine TPI status before hospital discharge to intervene if necessary; timing of blood draw ranged from 12 to 40 hours after birth based on discharge timing and patient status. Whole blood was placed into red top tubes and allowed to clot at room temperature before centrifugation at 4500×g for 12 minutes. After centrifugation, the serum was removed from the clot and immediately processed for TP and TI as per clinical protocol for stall‐side evaluation of TPI status; residual serum sample if available was then divided into 2 aliquots and stored at −20°C for further testing via RID and GGT. A subset of 6 serum samples were saved in 4 separate aliquots for analysis of the stability of GGT levels over time to ensure accuracy in batch reporting of results.

2.2. TP and TI

Serum total protein was estimated using a digital refractometer (MISCO Digital‐Dairy, Solon, Ohio). Before each calving season (in November), the refractometer was calibrated based on manufacturer's instructions; calibration of testing unit was also periodically confirmed during the calving season. For each sample, approximately 200 μL of serum was placed on the refractometer, and the digital reading was recorded on a paper record and in the electronic medical record. A quantitative estimate of immunoglobulin levels was then performed using the DVM Rapid Test II for Bovine Serum (MAI Animal Health, Elmwood, Wisconsin). Before each calving season, the testing unit was calibrated based on manufacturer's instructions and standards provided with the test; calibration of testing unit was also periodically confirmed during the calving season using the standard provided with each lot of test kits (10/kit). All samples were tested at the time of blood draw according to the manufacturer's instructions. Briefly, 10 μL of serum (measured using the manufacturer's provided pipettor) was added to both the manufacturer provided reagent and control tubes, mixed via inversion, and allowed to incubate for 10 minutes before analysis using the bovine serum IgG (BSIG) setting on the testing unit. Test results were recorded identical to TP.

2.3. GGT and RID

For analysis of GGT levels, serum aliquots were submitted to the Iowa State University Lloyd Veterinary Medical Center Clinical Pathology Laboratory for analysis using the VITROS Chemistry Products GGT Slides on the VITROS 5,1 FS Chemistry System (Ortho Clinical Diagnostics, Raritan, NJ, USA). Samples were stored for up to 2 months at −20°C before analysis. When the longest stored sample had been stored for approximately 2 months, all samples stored for the previous 60 days were submitted together for analysis. For the 6 samples utilized for GGT freezer stability analysis, aliquots were submitted on Days 1, 8, 31, and 66 of storage at −20°C.

For analysis of quantitative IgG levels via RID, the second serum aliquot was thawed and processed at the end of the sample collection period in December 2018 and tested using commercial bovine serum IgG RID test kits (Triple J Farms, Kent Labs, Bellingham, Washington). Initial analysis revealed that most samples when tested undiluted failed to achieve a reading within the range of accuracy based on the test standards (approximately 2.0‐30.0 g/L). Therefore, to achieve the appropriate testing window, serum samples were first diluted 1:1 in sterile Dulbecco's phosphate buffered saline (1× DPBS, Corning, Ref 21‐030‐CM) at a ratio 150 μL serum to 150 μL DPBS. Samples were mixed well by pipetting, and 5 μL was utilized for sample analysis as per the manufacturer's instructions. Results were read at completion of the reaction using a digital caliper, and sample results were calculated according to the manufacturer's instructions using the standards provided with the test kits.

2.4. Testing agreement and standards assessment

A subset of 20 serum samples were submitted to an independent testing laboratory (Saskatoon Colostrum Company [SCC], Saskatoon, Saskatchewan) and subjected to RID testing to assess the repeatability of the RID test results as a reference standard. Further analysis of the agreement of testing standards between the TI and RID was also undertaken as previously reported. ¹²

2.5. Statistical analysis

Descriptive statistics were used initially to describe the results of the study. Serum samples that had all tests completed were then analyzed for agreement between the methods. When necessary, samples were evaluated first for all samples (ALL), and then separately for calves that only received maternal colostrum (MAT), and samples from calves that received any level of commercial colostrum replacer (REP). Association was initially evaluated via linear regression analysis of TP, TI, and GGT compared to RID IgG levels as the reference standard. A correlation of <0.975 was observed for all methods, therefore a Passing‐Bablok regression analysis with calculation of a nonparametric Spearman's correlation coefficient was performed to allow for inclusion of extreme values and the assumption that imprecision exists within both the RID and other method compared which might not be normally distributed. ¹³ Agreement between the RID and TI including bias and limits of agreement (LoA) was then further evaluated by use of the Bland‐Altman method ¹⁴ ; any sample above the limit of detection for the TI assay (result ≥30.0 g/L) was not included in this analysis. Because a proportional bias was identified, an extended Bland‐Altman plot was then generated using the Taffé method on the subset of samples that had the RID performed twice. ¹⁵ Receiver operating characteristic (ROC) curves were used to identify the optimal cutoff point for each indirect method to correlate with RID IgG levels of 10.0, 18.0, and 25.0 g/L as recently proposed for fair, good, and excellent TPI in dairy calves ¹⁶ and consistent with a cutoff of 24.0 g/L for risk of pre‐weaning morbidity and mortality in beef calves. ¹ As previous work has demonstrated that cutoff values for TP might differ between calves that have received maternal colostrum vs commercial colostrum replacement, ¹⁶ a covariate‐adjusted ROC curve (AROC) was generated first and the covariate‐adjusted area under the curve (AAUC), Youden's index, and false positive fraction (FPF) were calculated. The pooled ROC curve (ALL), ROC curve for samples from calves that only received maternal colostrum (MAT), and samples from calves that received any level of commercial colostrum replacer (REP) were then generated and the AUC, and Youden's J statistic were calculated and reported. DeLong's test for 2 correlated ROC curves was also calculated to further evaluate the discriminating ability of these tests.

The repeatability of the RID and TP assays was assessed by sending a subset of 20 samples (range, 5.0‐75.0 g/L) to an outside laboratory for analysis; intra‐assay variation was assessed by calculating the coefficient of variation (CV) as previously described, ⁹ and bias and LoA were determined by use of the Bland‐Altman method. ¹⁴ The stability of GGT in storage was analyzed using a mixed model for repeated measure. In addition, classification agreement of samples between testing modalities (RID, TI, and TP) based on the classification scheme described above from Godden et al. ¹⁶ was evaluated using Cohen's kappa statistic with linear weights. Statistical analyses were performed either in Microsoft Excel (version 16.52) for descriptive statistics, or for all other analyses, in R software (version 3.6.2) with ROC curves fitted with the MethodCompare, ¹⁵ pROC, ¹⁷ and ROCnReg ¹⁸ packages and mixed model for repeated measures using the Imer function of the Ime4 package.

3. RESULTS

Data were collected on 233 live calves; the average time from birth to collection of serum sample was 20.4 hours (SD ±4.8 hours) which was primarily associated with discharge from the hospital typically occurring within 24 hours of birth. A total of 196 calves had data collected and serum saved for further analysis via all test types and were included in this study (ALL). Of the 196 calves, 134 calves (MAT) only received maternal colostrum either via nursing directly from the dam, or via collection and administration via bottle or esophageal feeder. The remaining 62 of the 196 calves (REP) received either colostrum supplementation or complete replacement in addition to access to nursing from dam (average of 2.1 bags, or 126 g, IgG, ±1.0 bag or 60 g; min = 1.0, max = 6.0).

Table 1 describes the results of each individual test for TPI status, with the distribution of results of RID testing stratified by MAT or REP provided in Figure 1. Based on the RID reference standard testing, only 1.5% (3/196) of the calves were considered to have poor TPI (<10.0 g/L), whereas 8.7% (17/196) were classified as achieving fair (10.0‐17.9 g/L), 25.0% (49/196) good (18.0‐24.9 g/L), and 64.8% (127/196) excellent (>25.0 g/L) TPI. For the calves that received maternal colostrum only (MAT, n = 134), 0.7% (1/134) were poor, 6.0% (8/134) were fair, 17.2% (23/134) were good, and 76.1% (102/134) were excellent. For the calves that received supplementation or replacement (REP, n = 62), 3.2% (2/62) were poor, 14.5% (9/62) were fair, 41.9% (26/62) good, and 25/62 (40.4%) were excellent.

TABLE 1.

Descriptive statistics of serum samples tested via turbidimetric immunoassay (TI), serum total protein (TP), gamma‐glutamyl transferase (GGT), and radial immunodiffusion (RID) from 196 calves born at a teaching hospital.

Test	N	Mean	Median	SD	Min	Max
RID IgG (g/L)	196	34.39	31.51	±15.39	8.91	79.99
Maternal only	134	38.92	38.86	±15.85	8.91	79.99
Supplement/replacer	62	24.60	23.51	±8.39	9.25	55.46
TP (g/dL)	196	6.3	6.1	±1.0	4.5	9.6
Maternal only	134	6.6	6.6	±1.0	4.7	9.6
Supplement/replacer	62	5.7	5.7	±0.6	4.5	7.3
GGT (IU/L)	196	4051	3239	±2950	19	13 798
Maternal only	134	4110	3082	±3208	166	13 798
Supplement/replacer	62	3923	3538	±2353	19	10 604
TI‐IgG (g/L)	196	18.92	17.91	±8.00	4.31	>30.0
Maternal only	134	21.43	23.08	±7.83	4.31	>30.0
Supplement/replacer	62	13.50	12.59	±5.29	5.26	29.53

FIGURE 1.

Distribution of the quantity of IgG (g/L) based on single radial immunodiffusion for serum samples collected from beef calves receiving maternal colostrum only vs receiving either a supplemental or complete colostrum replacement.

3.1. Regression analysis and Bland‐Altman

When all serum samples were considered together, the Spearman's correlation coefficients as calculated using Passing‐Bablok regression analysis and compared to RID were: TI, ρ = 0.757 (Figure 2); TP, ρ = 0.715 (Figure 3); and GGT, ρ = 0.413 (Figure 4). For the TI compared to RID, regression analysis also identified both a constant (intercept = −0.51 [CI: −2.63, 3.05]) and proportional (slope = 1.87 [CI: 1.69, 2.08]) bias between the 2 methods. For samples from calves that received maternal colostrum only (MAT, n = 134), the Spearman's correlation as determined by a Passing‐Bablok regression and compared to RID were: TI, ρ = 0.72 (Figure S1); TP, ρ = 0.695 (Figure S2); and GGT, ρ = 0.53 (Figure S3). For the TI compared to RID, regression analysis again identified both a constant (intercept = −1.15 [CI: −5.55, 3.50]) and proportional (slope = 1.97 [CI: 1.7, 2.23]) bias between the 2 methods. For samples from calves that received either partial supplementation or complete colostrum replacement (REP, n = 62), the Spearman's correlation as determined by a Passing‐Bablok regression and compared to RID were: TI, ρ = 0.678 (Figure S4); TP, ρ = 0.529 (Figure S5); and GGT, ρ = 0.379 (Figure S6). For the TI compared to RID, regression analysis again identified both a constant (intercept = 2.65 [CI: −1.94, 6.63]) and proportional (slope = 1.65 [CI: 1.29, 2.05]) bias between the 2 methods.

FIGURE 2.

Passing‐Bablok regression analysis of turbidimetric immunoassay (TI) vs single radial immunodiffusion (RID) values obtained for all calf serum samples. Blue line represents the regression line, whereas dashed red line represents the identity line for perfect correlation.

FIGURE 3.

Passing‐Bablok regression analysis of serum total protein (TP) vs single radial immunodiffusion (RID) values obtained for all calf serum samples. Blue line represents the regression line.

FIGURE 4.

Passing‐Bablok regression analysis of gamma‐glutamyl transferase (GGT) vs single radial immunodiffusion (RID) values obtained for all calf serum samples. Blue line represents the regression line.

Agreement between the RID and TI was further analyzed via a Bland‐Altman LoA plot for all samples collected (Figure 5A). Because a proportional bias was identified, an extended Bland‐Altman LoA plot was also generated to further explore this for a subset of the data (n = 20) for which the reference method (RID) was repeated (Figure 5B); using this method, a differential bias of 4.05 (CI: 0.91, 7.20) and a proportional bias of 0.38 (CI: 0.29, 0.46) were identified within this subset of data. Therefore, for all analyses, a significant proportional bias was evident with the TI underestimating IgG concentrations as measured by RID assay, with the bias increasing as the IgG concentration increased.

FIGURE 5.

Agreement between the turbidimetric immunoassay vs radial immunodiffusion (RID) via traditional Bland‐Altman analyses for all tested samples (A) and an extended Bland‐Altman plot for a subset of samples (n = 20) in which the reference test (RID) was repeated (B). The differential bias between these 2 methods was 4.05 (CI: 0.91, 7.20) and proportional bias was 0.38 (CI: 0.29, 0.46). LoA, limits of agreement.

3.2. ROC curves

Covariate‐adjusted ROC curves were then performed to assess the effect of maternal colostrum vs colostrum replacer on the overall performance of the tests; standard ROC curves were also completed to determine optimal cutoffs for TI, TP, and GGT based on RID target values of <10.0, <18.0, and <25.0 g/L. For all ROC analyses, data for <10.0 g/L are not shown because of an insufficient number of samples for accurate analysis. Figure 6A‐C shows the covariate‐adjusted ROC curves compared to the standard pooled ROC curves (ALL) for TI, TP, and GGT, respectively, based on a RID target value of <18.0 g/L. Figure 7A‐C shows the covariate‐adjusted ROC curves compared to the standard pooled ROC curves (ALL) for TI, TP, and GGT, respectively, based on a RID target value of <25.0 g/L. The AAUC, Youden's index, and FPF for each covariate‐adjusted curve are demonstrated in Table 2. For both 18.0 and 25.0 g/L, the TI demonstrated the highest AAUC and Youden's index, and lowest FPF, indicating the best overall performance when adjusted for the effect of maternal colostrum vs colostrum replacer. The standard pooled ROC curves for all calves (ALL), calves that received maternal colostrum only (MAT), and calves that received any amount of colostrum supplementation or replacement (REP) are shown in Figures S7A‐C and S8A‐C for RID target values of <18.0 and 25.0 g/L, respectively. The AUC, cutoff concentrations for detecting RID IgG concentrations with highest sensitivity and specificity based on Youden's index, and prevalence of samples below the cutoff are given in Table 3 and separated into categories (ALL, MAT, REP). When all samples were considered together, the Youden index for detecting RID IgG concentrations of <18.0 and <25.0 g/L was associated with TI cutoff concentrations of 9.89 and 13.76 g/L, respectively, with TP cutoff concentrations of 5.5 and 6.0 g/dL, respectively, and with GGT cutoff concentrations of 2303 and 1831 IU/L, respectively. DeLong's test detected a significant difference (P < .05) between the AUC of the ROC curve of the TI compared to both TP and GGT at <18.0 and <25.0 g/L for all samples tested; TP demonstrated a significant difference from GGT only at <25.0 g/L.

FIGURE 6.

Covariate‐adjusted receiver operating characteristic (AROC) curves (in black) compared to the standard pooled ROC curves (in red) showing the adjusted area under the curve (AAUC) and pooled AUC of the (A) turbidimetric immunoassay (g/L), (B) serum total protein (g/dL), and (C) gamma‐glutamyl transferase (IU/L) for the detection serum IgG concentrations of <18.0 g/L when adjusted for colostrum type (maternal vs replacer). Results are shown for all calves sampled. Black solid line is AAUC, dashed black represents confidence intervals; red solid line is pooled AUC, dashed red represents confidence intervals. X‐axis represents the false positive fraction (FPF), Y‐axis the true positive fraction (TPF).

FIGURE 7.

Covariate‐adjusted receiver operating characteristic (AROC) curves (in black) compared to the standard pooled ROC curves (in red) showing the adjusted area under the curve (AAUC) and pooled AUC of the (A) turbidimetric immunoassay (g/L), (B) serum total protein (g/dL), and (C) gamma‐glutamyl transferase (IU/L) for the detection serum IgG concentrations of <25.0 g/L when adjusted for colostrum type (maternal vs replacer). Results are shown for all calves sampled. Black solid line is AAUC, dashed black represents confidence intervals; red solid line is pooled AUC, dashed red represents confidence intervals. X‐axis represents the false positive fraction (FPF), Y‐axis the true positive fraction (TPF).

TABLE 2.

Covariate‐adjusted receiver operator characteristics of point‐of‐care assays turbidimetric immunoassay (TI), serum total protein (TP), and gamma‐glutamyl transferase (GGT) as compared to reference standard radial immunodiffusion (RID) for assessment of well‐managed calves for transfer of passive immunity status (n = 196, all calves in study).

Assay	RID (prevalence below threshold, g/L)	AAUC (95% CI)	Youden index (95% CI)	FPF (95% CI)
TI (g/L)	<18.0 (0.102)	0.927 (0.854‐0.971)	0.735 (0.578‐0.846)	0.131 (0.038‐0.288)
	<25.0 (0.352)	0.827 (0.754‐0.894)	0.590 (0.461‐0.707)	0.259 (0.135‐0.391)
TP (g/dL)	<18.0 (0.102)	0.759 (0.621‐0.878)	0.481 (0.273‐0.666)	0.342 (0.143‐0.603)
	<25.0 (0.352)	0.779 (0.695‐0.852)	0.503 (0.393‐0.619)	0.288 (0.154‐0.432)
GGT (IU/L)	<18.0 (0.102)	0.777 (0.683‐0.856)	0.461 (0.30‐0.628)	0.249 (0.071‐0.457)
	<25.0 (0.352)	0.741 (0.662‐0.811)	0.398 (0.283‐0.528)	0.391 (0.156‐0.555)

Abbreviations: AAUC, adjusted area under the curve of ROC curve analysis; CI, confidence interval; FPF, false positive fraction.

TABLE 3.

Receiver operator characteristics and optimal diagnostic cutoffs of point‐of‐care assays turbidimetric immunoassay (TI), serum total protein (TP), and gamma‐glutamyl transferase (GGT) as compared to reference standard radial immunodiffusion (RID) for assessment of individual beef calves for transfer of passive immunity status (ALL, n = 196; MAT, n = 134; REP, n = 62).

Assay	RID (prevalence below threshold, g/L)	Samples	Threshold a	AUC	Sensitivity (95% CI)	Specificity (95% CI)
TI (g/L)	<18.0 (0.102)	ALL	9.89	0.937	0.910 (0.861‐0.951)	0.888 (0.722‐1)
		MAT	11.4	0.960	0.882 (0.801‐0.946)	1 (1)
		REP	9.91	0.898	0.882 (0.784‐0.961)	0.909 (0.727‐1)
	<25.0 (0.352)	ALL	13.76	0.861	0.813 (0.729‐0.885)	0.818 (0.712‐0.909)
		MAT	13.16	0.853	0.930 (0.873‐0.986)	0.793 (0.621‐0.931)
		REP	10.06	0.795	1 (1)	0.459 (0.297‐0.622)
TP (g/dL)	<18.0 (0.102)	ALL	5.5	0.812	0.818 (0.761‐0.869)	0.75 (0.55‐0.9)
		MAT	5.5	0.768	0.912 (0.864‐0.96)	0.667 (0.333‐1)
		REP	5.6	0.793	0.686 (0.569‐0.804)	0.818 (0.546‐1)
	<25.0 (0.352)	ALL	6.0	0.818	0.756 (0.677‐0.827)	0.754 (0.652‐0.855)
		MAT	6.2	0.781	0.725 (0.628‐0.814)	0.781 (0.625‐0.906)
		REP	5.4	0.762	0.96 (0.88‐1)	0.514 (0.351‐0.676)
GGT (IU/L)	<18.0 (0.102)	ALL	2303	0.763	0.737 (0.669‐0.8)	0.7 (0.5‐0.9)
		MAT	1832	0.800	0.823 (0.758‐0.887)	0.778 (0.444‐1)
		REP	3490	0.786	0.647 (0.510‐0.784)	0.909 (0.727‐1)
	<25.0 (0.352)	ALL	1831	0.695	0.905 (0.849‐0.952)	0.406 (0.290‐0.522)
		MAT	1832	0.791	0.911 (0.852‐0.960)	0.625 (0.469‐0.781)
		REP	4501	0.653	0.52 (0.32‐0.72)	0.757 (0.622‐0.892)

Abbreviations: ALL, all calves in study; AUC, area under the curve of ROC curve analysis; CI, confidence interval; MAT, calves receiving maternal colostrum only; REP, calves receiving colostrum replacer or supplementation.

^a Optimal cutoff as determined by the Youden index.

3.3. Agreement between assays

To evaluate the repeatability of the RID assay, a subset of 20 samples were sent to an outside lab and the CV was calculated to evaluate intra‐assay variation for the RID and determined to be 7.3%. Bland‐Altman analysis revealed a mean bias of −0.44 (CI: −7.76, 6.87) g/L between labs for the RID and −0.015 (CI: −0.724, 0.694) g/dL for TP. To ensure that storage time did not affect GGT results (Table S1), a mixed model for repeated measures was used to demonstrate in a subset of samples that no significant difference (P = .1) was found between sampling timepoints, therefore the measurement of GGT in batch samples every 2 months was found to be acceptable for comparisons.

The weighted kappa value characterizing agreement between TI and RID assay for the classification of ALL samples within the categories of IgG <10.0 (poor), 10.0‐17.9 (fair), 18.0‐24.9 (good), and >25.0 (excellent) g/L was 0.127 (CI: −0.138, 0.390); for calves that received maternal colostrum only (MAT), the weighted kappa was 0.061 (CI: −0.438, 0.560), and for REP, the weighted kappa was 0.108 (CI: 0.069, 0.147). For TP, using the previously recommended cutoffs for TP by Godden et al., ¹⁶ the weighted kappa was 0.33 (CI: 0.052, 0.61) for ALL samples, 0.31 (CI: −0.069, 0.69) for MAT samples, and 0.014 (CI: −0.256, 0.280) for REP samples. Table S2 demonstrates the distribution of these results for all samples. Underestimation of IgG leading to misclassification occurred in 125/196 (63.8%) of samples by TI, and in 80/196 (40.8%) of samples by TP. Overestimation of IgG leading to misclassification occurred in 6/196 (3.0%) of samples by TI, and in 15/196 (7.7%) of samples by TP.

4. DISCUSSION

Overall, the mean concentration of IgG in the samples collected as measured by RID was higher than anticipated based on the TI samples obtained stall‐side, but within the range previously reported for beef cattle ¹ , ¹⁹ and indicates a successful program for ensuring optimal TPI within our cohort of beef calves after calving management or dystocia at a veterinary teaching hospital. Extensive clinical monitoring with intervention by 4 hours, if nursing was not noted, in a teaching hospital setting is more likely to lead to optimal outcomes with respect to TPI which could limit the external validity of this work. While our results indicate that the TI demonstrated the best correlation with RID as assessed via Spearman's ρ and displayed the highest AAUC of the AROC for all calves regardless of colostrum source, the TI showed a significant constant and proportional bias which led to poor classification agreement as measured by kappa analysis. The AAUC and Spearman's ρ for TP were less than for the TI, but TP demonstrated a higher kappa value and less misclassification than TI. Both tests demonstrated less correlation for calves that received any amount of colostrum replacement product as compared to those that received maternal colostrum only, particularly when assessing higher IgG levels (>25.0 g/L). Gamma‐glutamyl transferase levels demonstrated poor correlation with the RID regardless of the statistical measure.

Turbidimetric immunoassays have several potential advantages regarding measurement of TPI in calves at the individual as opposed to herd level. The results of these tests are quantitative and directly estimate the actual IgG concentration; TP or Brix refractometry are indirect measures of IgG. ²⁰ Previously marketed TI kits that are no longer commercially available in the United States have been documented in the literature to have excellent (R ² = 0.98) correlation with RID test results in dairy calves with serum IgG levels ranging from 4.60 to 36.40 g/L ¹⁰ ; in dairy and beef calves <2 weeks of age that test was also shown to have good sensitivity and specificity for detection of IgG <10.0 g/L. ¹¹ Ujvari et al. ⁹ demonstrated that the use of a similar product by the same manufacturer to assess TPI in foals had sufficient sensitivity, specificity, accuracy, and precision with acceptable intra‐ and interassay variability to be used as an alternative to other POC tests. ⁹ These results suggested that the currently marketed bovine serum TI should perform similarly well and represent a useful measure of TPI for beef calves born in our teaching hospital.

While the TI and RID demonstrated good correlation, a very large and clinically relevant bias was noted between the 2 testing modalities. Radial immunodiffusion has long been considered the reference standard for assessment of serum IgG levels in calves (recently reviewed in reference 20). However, both the RID and TI tested in our study rely heavily on the standards provided with the test kits to determine the result for each individual sample. Each TI kit is provided with a single control standard within the testing window; the commercial RID is provided with 3 standards to allow generation of a standard curve. In this study, a 1:1 dilution of the serum samples was necessary for the RID as over half of the samples tested were >30.0 g/L and out of the dynamic range of the RID without dilution; even with a 1:1 dilution, some samples (30/196, 15.3%) would have benefited from further dilution. For the TI, 34/196 (17.3%) of samples tested on the TI unit as >30.0 g/L and outside the dynamic range of the test; ideally these samples would have been diluted and retested, but as that was not clinically relevant at the time, this was not performed. The dynamic range of both tests without dilution most closely represents commonly reported TPI results for dairy neonates (approximately 2.0‐30.0 g/L). However, based on our data and others, ¹⁹ this does not appear to represent the reference range for well‐managed beef calves. Future evaluation of this TI testing unit for POC use in beef calf serum samples from calves receiving maternal colostrum could benefit from a 1:1 dilution which might place the sample closer to the testing window.

Statistical analysis revealed minimal bias (−0.44 g/L) and acceptable CV (7.3%) (as previously defined ⁹ ) between our RID results and those of an outside lab, suggesting that our RID results are consistent with what might be expected from other recently published studies utilizing the same laboratory. ¹⁹ , ²¹ As our recent comparison ¹² suggests that the standards provided with the RID provide a more accurate representation of the actual IgG concentrations than the TI, the large bias identified between the TI and RID results might be because of differences in standards and testing calibration. Future users of the TI unit could consider utilization of the cutoffs suggested by our work for individual classification of TPI status in beef calves, however, recalibration of the TI testing unit in the future to match the currently available USDA standard could also help to correct the observed bias. Potential users are encouraged to carefully consider these factors in the future when making individual clinical decisions based on TI results.

Serum TP as measured by refractometry is a portable, inexpensive, and practical tool for estimating herd TPI status in calves. New targets for serum TP levels in dairy herds have recently been proposed for calves fed maternal colostrum and were used in our study to stratify the results, however, it has been suggested that the cutoffs are inaccurate for calves fed colostrum replacement products. ¹⁶ Our data support this suggestion as we observed good differentiation of cutoff values for TP in maternal colostrum‐fed calves that are similar to those previously reported, but poor differentiation in calves that received any amount of colostrum replacement product. In beef calves, Gamsjäger et al. ¹⁹ and Akkose et al. ²² recently found ≤5.1 and 5.2 g/dL, respectively as the optimal indicators of both <10.0 and <16.0 g/L IgG, and ≤5.7 and < 6.4 g/dL, respectively for <24 g/L IgG; only a small percentage of animals in both studies were demonstrated to have IgG levels below 16.0 g/L which the authors acknowledge might have interfered with generation of adequate thresholds at those levels. The recommended optimal cutoffs identified in our study for TP from calves receiving maternal colostrum only (≤5.5 g/dL for <18 g/L IgG and ≤6.2 g/dL for <25 g/L IgG) are more similar to the consensus recommendations in dairy calves than the results reported by either study in beef calves. ¹⁹ Therefore, while more studied than most other practical POC methods, serum TP analysis still warrants future work, particularly in beef calves and those receiving colostrum replacement products.

The use of serum GGT as an indicator of colostrum ingestion in beef calves is utilized in clinical practice when blood chemistry results are available and colostrum consumption is unknown. Bovine colostrum contains exceptionally high levels of GGT, and this maternal GGT is absorbed from colostrum and present in the serum of neonatal calves that have ingested maternal colostrum. ⁶ , ⁷ Calves that fail to nurse colostrum have serum GGT levels equivalent to those of adults, whereas those that have ingested maternal colostrum display significantly elevated serum GGT. ⁷ However, conflicting results have been shown in studies assessing the utility of GGT as a predictor of TPI status beyond simple detection of colostrum ingestion. ⁵ , ²³ In otherwise healthy beef calves less than 18 days of age with a mean serum IgG of 33.16 g/L (range, 5.20‐49.50 g/L), serum GGT demonstrated a poor correlation with IgG results. ⁸ Interestingly, no studies to date have evaluated the effect of the use of colostrum replacement products on absorption of GGT despite review articles stating that GGT absorption from colostrum is interrupted by pasteurization or use of colostrum replacers. ²⁰ In this review, the article referenced for this statement only tested synthetic milk (the production of which was not described) and boiled cow's milk, not pasteurized colostrum. ²⁴ Our results indicate minimal difference in GGT levels in calves that did or did not receive supplementation with a colostrum replacement product derived from colostrum. Twenty of the calves included in this study received a full replacement volume of colostrum replacer and had GGT levels averaging 3846 IU/L, very similar to those receiving maternal colostrum only. Communication with the manufacturer of the colostrum replacement product utilized revealed unpublished data supporting high levels of GGT present in their colostrum replacement products (R. Sargent, SCC, personal communication). The inactivation of GGT has been investigated as a marker for successful pasteurization of milk and occurs within 1 minute at 80°C and within 10 minutes at 70°C, however, over 30% of initial enzyme activity was still present after 30 minutes at 60°C. ²⁵ Thus, it is likely that GGT is present in and actively absorbed from at least some colostrum‐based replacement products depending on heat treatment settings used. Regardless, based on our results and consistent with previously published reviews, ²⁰ , ²⁶ the use of GGT to differentiate the level of TPI achieved in beef calves does not appear to be justified.

Our results are not without limitations; because this was a retrospective study of clinical practice whereby TP and TI data were generated in real‐time and performed by multiple operators on a clinical service, overall test agreement was less than previously reported in other similar studies in which the data were collected prospectively by a dedicated research team. In addition, it is important to note that many of the calves were tested before 24 hours of age as the clinical goal of this testing was to identify TPI status before hospital discharge; therefore, these results likely do not represent peak IgG concentrations in this group of calves. However, it does demonstrate that for the purposes of individual animal assessment, TPI measurements can successfully be performed before 24 hours of age. Our study also lacked a cohort of calves with IgG <10.0 g/L; as a larger emphasis is placed on increasing IgG targets to higher levels, the ability to differentiate between fair and good vs excellent TPI becomes more critical. Therefore, the need to monitor TPI even in well‐managed herds will remain, and our results offer insights into the best methodology for monitoring in these situations.

In conclusion, both TI and TP results demonstrated moderate agreement with RID and represent potentially viable yet less than perfect options for estimating individual TPI status in beef calves; GGT was not well correlated with IgG level as measured by RID and should not be used other than to indicate colostrum consumption. In this group of calves, the TI demonstrated the best overall agreement with RID and did not require adjustment for colostrum source, however, a large bias was detected between the TI and RID results which currently limits the usefulness of the TI. Because of varying amounts of colostrum replacer clinically used, TP was not as predictive of individual TPI status in this cohort of animals.

CONFLICT OF INTEREST DECLARATION

The testing unit (DVM Rapid Test II) for the turbidometric immunoassay was donated to Iowa State University in 2015 by MAI Animal Health. No additional compensation or provision of testing supplies was provided for the work that was performed in this manuscript.

OFF‐LABEL ANTIMICROBIAL DECLARATION

Authors declare no off‐label use of antimicrobials.

INSTITUTIONAL ANIMAL CARE AND USE COMMITTEE (IACUC) OR OTHER APPROVAL DECLARATION

Authors declare no IACUC or other approval was needed.

HUMAN ETHICS APPROVAL DECLARATION

Authors declare human ethics approval was not needed for this study.

Supporting information

Figure S1. Passing‐Bablok regression analysis of turbidimetric immunoassay (TI) vs single radial immunodiffusion (RID) values obtained from serum samples of calves receiving maternal colostrum only. Blue line = regression line; red dashed line = identity line.

Figure S2. Passing‐Bablok regression analysis of serum total protein (TP) vs single radial immunodiffusion (RID) values obtained from serum samples of calves receiving maternal colostrum only.

Figure S3. Passing‐Bablok regression analysis of gamma glutamyl transferase (GGT) vs single radial immunodiffusion (RID) values obtained from serum samples of calves receiving maternal colostrum only.

Figure S4. Passing‐Bablok regression analysis of turbidimetric immunoassay (TI) vs single radial immunodiffusion (RID) values obtained from serum samples of calves receiving partial or complete colostrum replacement in addition to access to dam.

Figure S5. Passing‐Bablok regression analysis of serum total protein (TP) vs single radial immunodiffusion (RID) values obtained from serum samples of calves receiving partial or complete colostrum replacement in addition to access to dam.

Figure S6. Passing‐Bablok regression analysis of gamma glutamyl transferase (GGT) vs single radial immunodiffusion (RID) values obtained from serum samples of calves receiving partial or complete colostrum replacement in addition to access to dam.

Figure S7. (A‐C) Receiver operating characteristic (ROC) curves showing the sensitivity and specificity at a cutoff of 18 g/L for the turbidimetric immunoassay (TI, g/L) (green line), serum total protein (TP, g/dL) (blue line) and gamma glutamyl transferase (GGT, IU/L) (red line) for (A) the detection serum IgG concentrations of <1000 mg/dL measured by RID all serum samples; (B) serum samples from calves receiving maternal colostrum only; and (C) serum samples from calves receiving some amount of colostrum replacement product.

Figure S8. (A‐C) Receiver operating characteristic (ROC) curves showing the sensitivity and specificity at a cutoff of 25 g/L for the turbidimetric immunoassay (TI, g/L) (green line), serum total protein (TP, g/dL) (blue line) and gamma glutamyl transferase (GGT, IU/L) (red line) for (A) all serum samples; (B) serum samples from calves receiving maternal colostrum only; and (C) serum samples from calves receiving some amount of colostrum replacement product.

Table S1. GGT freezer stability analysis on a subset of serum samples following 1, 8, 31, and 66 days of storage at −20°C (N = 6); a mixed model for repeated measures was used to demonstrate that no significant difference (P = .1) was found between sampling timepoints.

Table S2. Classification between point‐of‐care assays turbidimetric immunoassay (TI) and serum total protein (TP) for assessment of TPI status based on criteria established by Godden et al. ¹⁶ [25.0 g/L (excellent)].

Kreuder AJ, Breuer RM, Wiley C, Dohlman T, Smith JS, McKeen L. Comparison of turbidometric immunoassay, refractometry, and gamma‐glutamyl transferase to radial immunodiffusion for assessment of transfer of passive immunity in high‐risk beef calves. J Vet Intern Med. 2023;37(5):1923‐1933. doi: 10.1111/jvim.16831