Materials and methods

The study was conducted on clinically and gynecologically healthy lactating Holstein-Friesiancows at two dairy farms in Eastern Germany between August 2009 and De-cember 2010. These cows were housed in freestall barns, fed a total mixed ration (primarily corn and grass silage plus concentrate) ad libitum and had free access to water. The study was approved and in accordance with the German legislation on animal welfare (Saxony: A18/09; Brandenburg, 23-2347-A-19-2-2010).

4.3.2 Study design

Beginning on Day 50 post partum (p.p.), 180 cows that were clinically healthy, with no apparent reproductive abnormalities were selected and a transrectal ultrasound examination was performed. All cows with a corpus luteum (CL) > 20 mm in diameter (n = 150) on the basis of the ultrasound examination were incorporated in this study.

Those cows received 25 mg prostaglandin F2α (PGF2α; Dinoprost, Enzaprost T ^®, CEVA, Libourne, France) and were observed twice daily for evidence of estrus. Ten percent of the cows, randomly selected, were not inseminated (n = 15) and served as a cycling control group.Allother cows (n=135)wereartificially inseminated (AI) ap-proximately 12 hours after detection of standing estrus by one experienced techni-cian. In all cows ovulation was determined by examining both ovaries directly before and every 12 ± 2 hours after AI (or rather estrus if not inseminated) with transrectal ultrasonography. Cows without signs of estrus within the first 96 + 4 hours after PGF2α treatment were excluded from the study. Cows not ovulating within 30 + 2 h after insemination were either inseminated again or excluded from the study as well.

Two inseminations per cow were performed at most. The day of ovulation was defined as Day 1. In retrospect, the day of the last AI was defined as Day 0.

On Day 5 the cows ultimately enrolled in the study were blood sampled and ultra-sonic examinations (B-mode and Power-mode) were performed. Further blood samples were taken on Days 9, 14, 19 and 20.

Chapter I: Materials and methods

To detect non-pregnant cows, all cows were observed twice daily for estrus behavior beginning on Day 19. Accessorily milk samples were taken to determine whole milk progesterone on Days 22 and 25. Milk progesterone was analyzed directly in the barn by an automatic, semiquantitative rapid test (eProCheck^®, Minitüb, Tiefenbach, Germany). Cows detected in estrus and/or showing a P4 concentration in milk

≤ 3 ng/ml by Day 25 were examined using transrectal ultrasonography. In case of a preovulatory follicle on at least one ovary and a CL < 20 mm, they were inseminated and recorded as non-pregnant from the former AI. The threshold for milk P4

concentration in order to detect cows in heat was determined in preliminary investigations for this project, based on the milk P4 concentrations of 25 cows in heat.

As for all cows not observed in estrus and not showing milk P4 < 3 ng/ml until Day 30 after artificial insemination, transrectal ultrasonographic pregnancy diagnosis (Tringa L^®; Esaote Pie Medical, Cologne, Germany) was conducted. In addition, on Day 42, maintenance of pregnancy was confirmed by manual transrectal palpation.

Moreover, the following reproductive and cow-associated parameters were assessed using entries of the farm´s veterinarian in the herd management program (Herde^®; dsp-Agrosoft GmbH, Ketzin, Germany) or our own recordings : age of the cow (age), lactation number (LAC), milk yield at AI (MYAI), 305 days lactation records (MY305), calving ease, start of first cycle after parturition (SCP), uterine involution (UI), quality of estrus signs (QE), size of preovulatory follicle (SPF) as well as body condition score (BCS) and lameness score (LNS) by Sprecher et al. [121] at AI.

4.3.3 Blood sampling and analysis of serum P4 concentration

Blood samples (approximately 10 ml) were collected from the coccygeal vessels into evacuated serum tubes (Monovette®; Sarstedt, Nürnbrecht, Germany) immediately after each sonographic examination on Day 5. Tubes were directly placed on ice and serum was separated by centrifugation (2.050 g, 15 min) and frozen at -20 °C within one hour after collection. Serum progesterone concentrations were determined using acommerciallyavailable radioimmunoassay (RIA; Coat-a-Count^® Radioimmuneassay Progesterone PITKPG-9,Siemens HealthcareDiagnostics GmbH, Los Angeles, CA, USA) previously validated for bovine serum. The SN of the assay was 0.1 ng

Chapter I: Materials and methods

progesterone. The specificity of the antibody was: 100% progesterone, 9.0 % 5 α-pregnan-3,20 dion, 3.2 % 5 β-pregnan-3.20 dion, 3,4% 17α dihydroprogesteron, 2.2

% 11-deoxycorticosteron and 0.9% corticosterone. The intra- and inter-assay coeffi-cients of variation (CVs) were 3.5 and 3.9%, respectively. The lower reference range was 0.02 ng/ml.

4.3.4 Milk sampling and analysis of milk P4 concentration

Milk samples were collected from each cow on Days 22 and 25 from the right hindquarter of the udder in non-sterile polyethylene centrifuge tubes (10 ml; Sarstedt, Nürnbrecht, Germany) and directly placed on ice. Within two hours after sampling, the whole milk was tested directly in the barn by an automatic semi-quantitative progesterone rapid test (eProCheck^®, Minitüb, Tiefenbach, Germany) on the basis of an EIA. For this study a testing option with two standards was chosen, at which standard 1 corresponded to 0 ng/ml and standard 2 to 16.8 ng/ml. The slide containing the samples comprised 8 cavities, one for each standard and 6 for the testing milk. To have an analysis made, 30 µl of homogenized whole milk were pipetted into each cavity and the testing block filled with all reagents was inserted.

Every sample was tested three times and the mean was taken for further analyses.

All results were displayed in ng/ml. The lower reference range was 0.1 ng/ml.

4.3.5 B-mode ultrasound examination for pregnancy diagnosis

Pregnancy diagnosis on Day 30 via B-mode ultrasound examination was carried out using a Tringa L^® ultrasound device (Esaote Pie Medical, Cologne, Germany), equipped with a 7.0 MHz linear probe. The entire uterus was examined systemati-cally. All cows in which amniotic fluid and an embryo with heartbeat was visible were regarded as pregnant. In doubt cows were again examined three days later. For pregnant cows, maintenance of pregnancy was confirmed by manual transrectal palpation on Day 42.

Chapter I: Materials and methods

4.3.6 B-Mode and colour Doppler ultrasonography, data recording and analysis

All ultrasonographic examinations were carried out 96 + 4 h following detected ovulation (Day 5 p.i.) by the same operator with a colour Doppler ultrasound device (Titan^®, 7.5 MHz linear probe, Sonosite, Bothell, USA) and took about 20 min for each cow. In brief, the CL was identified and three cross-sectional images with maxi-mal areas of the respective CL were frozen, recorded and digitized using B-mode ultrasonography (US). Measurements of the size of the CL and Doppler analysis were performed off-line on a personal computer using the image analysis software PixelFlux^® (Version 1.0; Chameleon Software, Leipzig, Germany). From each image the maximum diameter (DCL) as well as the maximum cross-sectional area of the CL (corpus luteum size / CS) was determined. Without a cavity within the CL, CS was in accordance with luteal size (LS). When a cavity was present within a CL, the size of the cavity area (CVS) was assessed separately in three cross-sectional images and subtracted from the total area of the CL in order to obtain LS (Figure 1). For further processing, the mean of the cross-sectional areas of the three images was calculated and used for data analysis.

The second part of the sonographic investigation was performed to assess para-meters representing luteal blood flow (PLBF) using colour Doppler US. Three cross-sectional images in Power-mode were recorded at the maximal DCL. Small meande-rings in position were allowed in order to achieve a maximum number of colour pixels in the luteal tissue. Care was taken to locate the whole section of the CL within the Doppler sample box, to avoid flash artefacts and to evaluate the maximum blood flow area (LBF) in the CL (Figure 1). In each image the absolute number of colour pixels was computed via PixelFlux^® Software and determined as a semi-quantitative para-meter of LBF. For each CL, average values of the three recorded images were calcu-lated and used for further data analysis. The parameter LBF was then dividedbyCS and accordingly LS to calculate the relative (rLBF) and absolute LBF(aLBF).

Chapter I: Materials and methods

4.3.7 Assignment to groups

Cows were assigned to one of four groups after pregnancy diagnosis was performed:

(1) pregnant [P]: amniotic fluid and an embryo with heartbeat visible on Day 30; (2) early embryonic loss [EEL]: no amniotic fluid and no embryo with heartbeat visible on Day 30, no sonographically confirmed estrus and/or AI until Day 25; (3) non-pregnant [NP]: sonographically confirmed estrus and/or AI until Day 25; (4) non-inseminated [NI].

4.3.8 Statistical analysis

All statistical analyses were carried out using the statistical analysis software SPSS (version 17.0.; SPSS Inc., Chicago, USA). The distribution of the data was tested visually using histogram analysis and by means of the Kolmogorov-Smirnov test modified according to Lilliefors (n >90) or Shapiro-Wilk test (n < 90). Data sets not differing significantly from normal distribution (< 0.05) as indicated by the tests or by graphic account were presented as mean ± standard error of mean (SEM), quartiles and ranges. In order to achieve a normal distribution, before parametric statistical methods were applied; a logarithmic transformation was performed on P4 values (Days 5, 9, 13, 19 and 20) and the size of the cavity (CVS), as well as the relative (rLBF) and absolute luteal blood flow (aLBF). Relationships were viewed pairwise using the Cramer´s V (nominal variables) or alternatively Spearman`s rank correlation coefficient (rs; ordinalor metric variables). A rs > 0 represented a positive relationship, whereas a negative relationship was indicated by rs < 0. For both correlation coefficients (cc) a relationship was assessed as low if 0.2 < cc < 0.4, as moderate if 0.4 < cc < 0.8 or high if 0.8 < cc < 1.0. Moreover, boxplots were charted in order to compare different groups of cows. For a comparison of the P4, PCS and PLBF means among groups with different pregnancy status on different days a t-test for independent samples was used for each separate day. In order to determine the influence of P4 concentration during the early luteal phase on pregnancy outcome more precisely, cows were additionally grouped in three groups (low:P4 <1.0 ng/ml;

intermediate:P4 >1.0 and<1.7ng/ml;high: P4 > 1.7 ng/ml) by P4 on Day 5.

Estima-Chapter I: Materials and methods

tion of the variance component was used to compare the effect of the factor cow on the variables P4, PCS and PLBF. Differences with P < 0.05 were considered to be significant. Significant P values were further classified as significant: P < 0.05 (*), highly significant: P < 0.01 (**) or extremely significant: P < 0.001 (***). A tendancy was assumed with P > 0.05 and < 0.10. Intra-observer reproducibility of sonographic measurements was expressed as intra-class correlation coefficients. Regression analysis was done to assess the predictability of P4 by analyzing PCS and PLBF.

Determination of the P4 concentration was considered as the gold standard for the distinction between a functional and subfunctional corpus luteum in this experiment.

Cows having a P4 concentration >1.0 ng/ml were assumed to have a physiological hormone concentration and consequently a functional CL, while cows having a P4

concentration <1.0 ng/ml were assumed to have a reduced hormone concentration and consequently a subfunctional CL. A logistic regression model was used for calculating the cut-off value at which likelihood of P4 > 1.0 ng/ml was more than 0.5.

Sensitivity [SN], specificity [SP], positive predictive value (PPV), and negative predictive value (NPV), all denoted in %, were estimated as described by Herzog et al. (2009) in order to calculate the optimized cut-off for parameters of luteal size and blood flow. The SN was defined as the probability of having a CL above the de-termining LS/LBF cut-off when P4 concentration was > 1.0 ng/ml. The SP was de-fined as the probability of having a CL below the determining LS/LBF cut-off when P4

concentration was < 1.0 ng/ml. The PPV was defined as the probability of having a P4 level>1.0 ng/ml if parameters of LS and LBF were above cut-off. The NPP was defined as the probability of having a P4 level < 1.0 ng/ml if parameter of LS and LBF were below selected cut-off. The cut-offs for LS and LBF were evaluated using receiver operating characteristic (ROC) analysis in order to detect the optimum combination of SN and SP for the detection of a P4 concentration > 1.0 ng/ml.

Chapter I: Results