Accepted Manuscript Accuracy of diagnosing double corpora lutea and twin pregnancy by measuring serum progesterone and bovine pregnancy-associated glycoprotein-1 in the first trimester of gestation in dairy cows Zoltán Szelényi, Attila Répási, Noelita Melo de Sousa, Jean Francois Beckers, Otto Szenci PII:
S0093-691X(15)00087-4
DOI:
10.1016/j.theriogenology.2015.02.014
Reference:
THE 13101
To appear in:
Theriogenology
Received Date: 2 November 2014 Revised Date:
8 February 2015
Accepted Date: 11 February 2015
Please cite this article as: Szelényi Z, Répási A, Melo de Sousa N, Beckers JF, Szenci O, Accuracy of diagnosing double corpora lutea and twin pregnancy by measuring serum progesterone and bovine pregnancy-associated glycoprotein-1 in the first trimester of gestation in dairy cows, Theriogenology (2015), doi: 10.1016/j.theriogenology.2015.02.014. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Revised Highlighted
1 2
Accuracy of diagnosing double corpora lutea and twin pregnancy by measuring serum
4
progesterone and bovine pregnancy-associated glycoprotein-1 in the first trimester of
5
gestation in dairy cows
6
RI PT
3
Zoltán Szelényi1,2*, Attila Répási2, Noelita Melo de Sousa3, Jean Francois Beckers3, Otto
8
Szenci1,2
SC
7
M AN U
9 10
1
11
Animals, H-2225 Üllő-Dóra major, Hungary
12
2
MTA-SZIE Large Animal Clinical Research Group, H-2225 Üllő-Dóra major, Hungary
13
3
University of Liege, Faculty of Veterinary Medicine, Fundamental and Applied Research for
14
Animals & Health (FARAH), Laboratory of Animal Endocrinology and Reproduction, Liege,
15
Belgium
EP
16
TE D
Szent István University, Faculty of Veterinary Science, Department and Clinic for Production
*corresponding author: Zoltán Szelényi, Hungary-2225 Üllő-Dóra major, +36302967012,
18
[email protected]
19
AC C
17
20
Abstract
21
Progesterone (P4) and bovine pregnancy-associated glycoprotein-1 (bPAG-1) concentrations
22
during gestation are dependent on the number of corpora lutea and fetuses, respectively. The
23
objective of this present study was to measure and evaluate the usefulness of measuring the P4 1
ACCEPTED MANUSCRIPT
and bPAG-1 concentrations in cases of single vs. twin pregnancies and one vs. two corpora lutea
25
at the first four months of gestation. We hypothesized that both the number of the corpus luteum
26
and the number of fetus might have an effect on progesterone and pregnancy protein
27
concentrations and we can set up clinically useful threshold levels in order to predict twin
28
gestations. Eighty-four Holstein-Friesian dairy cows were enrolled in this prospective
29
observational clinical trial. Blood was collected at time-point 1: between Days 29 to 42, time-
30
point 2: between Days 57 to 70, time-point 3: between Days 85 to 98, and time-point 4: between
31
Days 113 to 126 of gestation, respectively and bPAG-1 and P4 concentrations were measured.
32
Binary logistic regression analyzing serum P4 concentrations differed at time-point 2 compared
33
to baseline level, but the area under the curve (AUC) had low sensitivity. The bPAG-1
34
concentrations were statistically different at each time-point of gestation. The AUC cut-off values
35
of serum bPAG-1 concentrations were sufficiently sensitive to differentiate between twin
36
gestations from singleton ones. At time-points 3 (cut-off value of 3.4 ng/mL) and 4 (cut-off value
37
of 56.5 ng/mL) statistically significant differences with low sensitivity, high specificity and a
38
high AUC were found. Based on these results the diagnosis of twin pregnancy using pregnancy
39
protein measurements is clinically insufficient prior Day 85 of gestation, however the ability to
40
confirm the early twin pregnancy diagnosis with bPAG-1 measurements appears to be promising.
41
To achieve high sensitivity, further studies are required.
43 44
SC
M AN U
TE D
EP
AC C
42
RI PT
24
Keywords: cattle, twin, progesterone, pregnancy-associated glycoprotein, RIA
45
1. Introduction
46
Several pregnancy diagnosing methods are intensively used worldwide to diagnose whether a 2
ACCEPTED MANUSCRIPT
pregnant cattle is carrying multiple fetuses or not. The different rectal palpation techniques
48
(palpation of the uterine fluctuation, amniotic vesicle, or fetal membrane slip) [1] are useful and
49
widespread however they are really dependent on the skill of the examiner. An experienced
50
veterinarian/technician even from Day 35 with the fetal membrane slip technique is able to set up
51
pregnancy diagnoses [2]. However, the manual techniques are limited in diagnosing twins:
52
usually only the bilateral twins can be diagnosed accurately. The prediction accuracy of unilateral
53
twinning is low [3].
SC
RI PT
47
54
Ultrasonography in the early period (around Day 30) of gestation is also widely used with high
56
accuracy to diagnose early pregnancy. At this time usually twin pregnancies are also visible.
57
When dealing with twin pregnancy at a herd level, we always have to take into consideration the
58
phenomenon of late embryonic/early fetal mortality [4]. This can influence the accuracy of our
59
diagnostic tests.
TE D
60
M AN U
55
Another possibility for diagnosing early pregnancy in the field is the measurement of pregnancy
62
specific proteins, like bovine pregnancy-associated glycoprotein-1 (bPAG-1), bovine pregnancy-
63
specific protein B (bPSPB) or pregnancy serum protein (PSP-60), which are secreted by the
64
binucleate cells originating from the bovine trophoblast. In contrast to progesterone (P4)
65
measurements, they are probably good indicators of the presence of an alive conceptus [5,6,7].
66
Several assays have been developed, and nowadays not only blood but milk tests can be used for
67
an accurate pregnancy diagnosis [8]. Numerous factors are influencing the levels of these
68
proteins throughout gestation, twin pregnancy is one of these factors [9,10]. There are several
69
studies [11,12] focusing on the measurements of pregnancy proteins in case of bovine twin
AC C
EP
61
3
ACCEPTED MANUSCRIPT
70
pregnancies. Regarding these studies the question is still open: whether we can determine a cut-
71
off value for the pregnancy proteins at different time periods of gestation by which twin pregnant
72
animals can be distinguished from singleton carrying ones.
RI PT
73
Progesterone (P4) assays are also widely used to measure the blood P4 concentration even at a
75
herd level. The disadvantage of these tests is that the P4 itself is non-specific for pregnancy,
76
therefore performing pregnancy diagnoses by means of P4 measurement is not resulting in an
77
accurate sensitivity and specificity [4,5,6].
M AN U
78
SC
74
We have hypothesized that serum P4 concentrations will be different between the single and twin
80
pregnancies and this will be a sensitive and specific clinical indicator for the prediction of the
81
number of corpora lutea. Further we have also hypothesized that serum bPAG-1 concentrations
82
will be different between single and twin pregnant groups at each of the time-points and thus will
83
serve as a sensitive and specific clinical indicator for the prediction of the number of fetuses.
TE D
79
84
The objectives of this study were to compare serum P4 (1) and bPAG-1 (2) concentrations at
86
different time-points of gestation in cattle with the accurate determination of one or two corpora
87
lutea; to assess whether or not serum P4 concentration is a useful clinical indicator for the
88
determination of the number of corpus lutea at any of the previously mentioned time-points (3);
89
and to assess whether or not serum bPAG-1 concentration is a useful clinical indicator for the
90
determination of the number of fetuses at any of the above mentioned time-points (4).
AC C
EP
85
91 92
2. Materials and methods 4
ACCEPTED MANUSCRIPT
93
2.1 Animals
95
Altogether 84 Holstein-Friesian dairy cattle were included in our study with a confirmed
96
pregnancy by means of ultrasonography at the late embryonic period between Days 29 to 42 of
97
gestation (time-point 1). Animals were divided into 3 groups: the first group (TWIN Group,
98
n=29) had two viable embryos in the uterine horns and two mature corpora lutea on the ovaries at
99
the time of pregnancy diagnosis. Animals with one viable embryo in the pregnant uterine horn
100
and two corpora lutea on the ovaries formed the second group (DCL Group, n=35), while animals
101
with a singleton pregnancy (one viable embryo and one corpus luteum) served as the control
102
group (CON Group, n=20). All animals were kept in free barns with cubicles and were fed
103
according to the NRC recommendations [13].
M AN U
SC
RI PT
94
104
2.2 Ultrasonographic examination and blood sampling
106
Animals were sampled four times during the study. First, at the time of early pregnancy diagnosis
107
(time-point 1) which was performed by a real-time B-mode diagnostic ultrasound scanner
108
(EasiScan, BCF technologies, Bellshill, UK) equipped with a 5 to 9 MHz linear-array rectal
109
transducer, then monthly for three additional occasions (namely: between Days 57 to 70 (time-
110
point 2), Days 85 to 98 (time-point 3), and Days 113 to 126 (time-point 4) of gestation,
111
respectively). For the diagnosis of early pregnancy and ovarian structures 7.5 MHz transducer
112
was used. All ultrasonographic examinations and blood samplings were performed by the same
113
operator. Individual animals were examined in stalls in the milking house with dim ambient
114
lighting. After removal the feces, the transducer was inserted into the rectum and was slowly
115
moved from the cervix to the right, and then to the left uterine horn until the entire uterus was
AC C
EP
TE D
105
5
ACCEPTED MANUSCRIPT
imaged. No manipulation of the uterus took place in any of the investigations. Scanning of the
117
uterus and interpretation the echoscopic images have been described previously [14]. At each
118
examination, the operator was required to record the number and the presence of embryos in the
119
uterine horn and the number of the corpus luteum. At every time-point, when it was possible, the
120
viability of the fetus was confirmed by the detection of the heartbeat. In those cases, when
121
scanning of the entire uterus especially at time-points 3 and 4 was not feasible, the presence and
122
the movement of the fetus was used to confirm the viable pregnancy. Following pregnancy
123
diagnosis animals were handled according to the reproductive protocol of the particular farm. At
124
the time of drying-off pregnancy was confirmed by rectal palpation. Finally our ultrasonographic
125
diagnoses were confirmed by calving records. Animals were excluded from the study if repeated
126
examinations yielded inconsistent diagnoses.
M AN U
SC
RI PT
116
127
Following transrectal ultrasonographic examinations blood samples were collected into
129
vacutainer tubes (Monovette 9ml, Sarstedt, Nümbrecht, Germany) at every time-point. The tubes
130
were transported within 2 h into the laboratory in a cool-box containing icebags, centrifuged at
131
3000 x g/ 10 min) and the serum samples were kept at -20 C until measurements of P4 and
132
bPAG-1 concentrations.
EP
AC C
133
TE D
128
134
2.3 P4 and bPAG-1 assay procedures
135
Plasma concentrations of P4 were estimated by validated solid-phase
136
(RIA) method (Coat-A-Count TKPG; Diagnostic Products Corporation) as described by [15].
137
The sensitivity was 0.15 nmol/L, and the intra- and inter-assay coefficients of variation were 8.9
138
and 11%, respectively.
125
I radioimmunoassay
6
ACCEPTED MANUSCRIPT
139
Plasma bPAG-1 measurements were performed according to the method of Zoli et al. [16] and
141
subsequently modified by Perényi et al. [17]. A polyclonal antiserum was used [16] which was
142
produced against PAG I67kDa. A PAG I67kDa preparation, purified according to the protocol of
143
Zoli et al. [18] was used as standard and tracer. The sensitivity of the bPAG-1 RIA test was 0.06
144
ng/mL, and the intra- and inter-assay coefficients of variation were 5.0 and 5.5%, respectively.
RI PT
140
SC
145
2.4 Statistical analysis
147
The data were analyzed using a commercially available software (Minitab 16, State College, PA,
148
USA). Normal distribution of the data was tested with the Shapiro-Wilk test. Serum P4 and
149
bPAG-1 concentrations were compared within groups (TWIN, DCL, CON) and different time-
150
points (Time-point 1: between Days 28 to 42, Time-point 2: between Days 57 to 70, Time-point
151
3: between Days 85 to 98, and Time-point 4: between Days 113 to 126, respectively) using a
152
generalized linear model (GLM), due to the unbalanced design. Post-hoc evaluation was
153
performed with the Bonferroni correction.
TE D
EP
154
M AN U
146
Regarding to P4 concentrations, after evaluating the animals in a generalized linear model, the
156
TWIN and the DCL group (cows having 2 CL and one embryo) were condensed together and
157
compared to the CON group (1 CL and one embryo) at the four different time points. Binary
158
logistic regression was used to reveal any association between serum P4 concentration and the -
159
number of corpus luteum at each of the 4 time-points. At each time-points, when the regression
160
model was significant, a receiver operating curve (ROC) was created to find the most specific
161
and sensitive cut-off value for discrimination of cows with 2 CLs vs. 1 CL. Area under the curve
AC C
155
7
ACCEPTED MANUSCRIPT
162
(AUC) was also calculated.
163
To find the accuracy of pregnancy protein levels, the CON group and the DCL (1 embryo/fetus
165
with 2 CL) group were also condensed and compared to the TWIN group (2 embryos/fetuses) at
166
each of the four time-points regarding to pregnancy protein concentrations. Binary logistic
167
regression analysis was used to reveal any association between serum bPAG-1 concentration and
168
number of embryos/fetuses at each time-points. A receiver operating curve was created to reveal
169
the most specific and sensitive cut-off value for the determination of the number of
170
embryos/fetuses. Area under the curve (AUC) was also calculated.
M AN U
SC
RI PT
164
171
The AUC was regarded as follows: AUC = 0.5 (no discrimination); 0.7 ≤ AUC < 0.8 (acceptable
173
discrimination); 0.8 ≤ AUC < 0.9 (excellent discrimination); and AUC > 0.9 (outstanding
174
discrimination). The goodness of the fit of the logistic regression models were tested with the
175
Hosmer-Lemeshow method. Level of significance was set at P
S0093-691X(15)00087-4
DOI:
10.1016/j.theriogenology.2015.02.014
Reference:
THE 13101
To appear in:
Theriogenology
Received Date: 2 November 2014 Revised Date:
8 February 2015
Accepted Date: 11 February 2015
Please cite this article as: Szelényi Z, Répási A, Melo de Sousa N, Beckers JF, Szenci O, Accuracy of diagnosing double corpora lutea and twin pregnancy by measuring serum progesterone and bovine pregnancy-associated glycoprotein-1 in the first trimester of gestation in dairy cows, Theriogenology (2015), doi: 10.1016/j.theriogenology.2015.02.014. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Revised Highlighted
1 2
Accuracy of diagnosing double corpora lutea and twin pregnancy by measuring serum
4
progesterone and bovine pregnancy-associated glycoprotein-1 in the first trimester of
5
gestation in dairy cows
6
RI PT
3
Zoltán Szelényi1,2*, Attila Répási2, Noelita Melo de Sousa3, Jean Francois Beckers3, Otto
8
Szenci1,2
SC
7
M AN U
9 10
1
11
Animals, H-2225 Üllő-Dóra major, Hungary
12
2
MTA-SZIE Large Animal Clinical Research Group, H-2225 Üllő-Dóra major, Hungary
13
3
University of Liege, Faculty of Veterinary Medicine, Fundamental and Applied Research for
14
Animals & Health (FARAH), Laboratory of Animal Endocrinology and Reproduction, Liege,
15
Belgium
EP
16
TE D
Szent István University, Faculty of Veterinary Science, Department and Clinic for Production
*corresponding author: Zoltán Szelényi, Hungary-2225 Üllő-Dóra major, +36302967012,
18
[email protected]
19
AC C
17
20
Abstract
21
Progesterone (P4) and bovine pregnancy-associated glycoprotein-1 (bPAG-1) concentrations
22
during gestation are dependent on the number of corpora lutea and fetuses, respectively. The
23
objective of this present study was to measure and evaluate the usefulness of measuring the P4 1
ACCEPTED MANUSCRIPT
and bPAG-1 concentrations in cases of single vs. twin pregnancies and one vs. two corpora lutea
25
at the first four months of gestation. We hypothesized that both the number of the corpus luteum
26
and the number of fetus might have an effect on progesterone and pregnancy protein
27
concentrations and we can set up clinically useful threshold levels in order to predict twin
28
gestations. Eighty-four Holstein-Friesian dairy cows were enrolled in this prospective
29
observational clinical trial. Blood was collected at time-point 1: between Days 29 to 42, time-
30
point 2: between Days 57 to 70, time-point 3: between Days 85 to 98, and time-point 4: between
31
Days 113 to 126 of gestation, respectively and bPAG-1 and P4 concentrations were measured.
32
Binary logistic regression analyzing serum P4 concentrations differed at time-point 2 compared
33
to baseline level, but the area under the curve (AUC) had low sensitivity. The bPAG-1
34
concentrations were statistically different at each time-point of gestation. The AUC cut-off values
35
of serum bPAG-1 concentrations were sufficiently sensitive to differentiate between twin
36
gestations from singleton ones. At time-points 3 (cut-off value of 3.4 ng/mL) and 4 (cut-off value
37
of 56.5 ng/mL) statistically significant differences with low sensitivity, high specificity and a
38
high AUC were found. Based on these results the diagnosis of twin pregnancy using pregnancy
39
protein measurements is clinically insufficient prior Day 85 of gestation, however the ability to
40
confirm the early twin pregnancy diagnosis with bPAG-1 measurements appears to be promising.
41
To achieve high sensitivity, further studies are required.
43 44
SC
M AN U
TE D
EP
AC C
42
RI PT
24
Keywords: cattle, twin, progesterone, pregnancy-associated glycoprotein, RIA
45
1. Introduction
46
Several pregnancy diagnosing methods are intensively used worldwide to diagnose whether a 2
ACCEPTED MANUSCRIPT
pregnant cattle is carrying multiple fetuses or not. The different rectal palpation techniques
48
(palpation of the uterine fluctuation, amniotic vesicle, or fetal membrane slip) [1] are useful and
49
widespread however they are really dependent on the skill of the examiner. An experienced
50
veterinarian/technician even from Day 35 with the fetal membrane slip technique is able to set up
51
pregnancy diagnoses [2]. However, the manual techniques are limited in diagnosing twins:
52
usually only the bilateral twins can be diagnosed accurately. The prediction accuracy of unilateral
53
twinning is low [3].
SC
RI PT
47
54
Ultrasonography in the early period (around Day 30) of gestation is also widely used with high
56
accuracy to diagnose early pregnancy. At this time usually twin pregnancies are also visible.
57
When dealing with twin pregnancy at a herd level, we always have to take into consideration the
58
phenomenon of late embryonic/early fetal mortality [4]. This can influence the accuracy of our
59
diagnostic tests.
TE D
60
M AN U
55
Another possibility for diagnosing early pregnancy in the field is the measurement of pregnancy
62
specific proteins, like bovine pregnancy-associated glycoprotein-1 (bPAG-1), bovine pregnancy-
63
specific protein B (bPSPB) or pregnancy serum protein (PSP-60), which are secreted by the
64
binucleate cells originating from the bovine trophoblast. In contrast to progesterone (P4)
65
measurements, they are probably good indicators of the presence of an alive conceptus [5,6,7].
66
Several assays have been developed, and nowadays not only blood but milk tests can be used for
67
an accurate pregnancy diagnosis [8]. Numerous factors are influencing the levels of these
68
proteins throughout gestation, twin pregnancy is one of these factors [9,10]. There are several
69
studies [11,12] focusing on the measurements of pregnancy proteins in case of bovine twin
AC C
EP
61
3
ACCEPTED MANUSCRIPT
70
pregnancies. Regarding these studies the question is still open: whether we can determine a cut-
71
off value for the pregnancy proteins at different time periods of gestation by which twin pregnant
72
animals can be distinguished from singleton carrying ones.
RI PT
73
Progesterone (P4) assays are also widely used to measure the blood P4 concentration even at a
75
herd level. The disadvantage of these tests is that the P4 itself is non-specific for pregnancy,
76
therefore performing pregnancy diagnoses by means of P4 measurement is not resulting in an
77
accurate sensitivity and specificity [4,5,6].
M AN U
78
SC
74
We have hypothesized that serum P4 concentrations will be different between the single and twin
80
pregnancies and this will be a sensitive and specific clinical indicator for the prediction of the
81
number of corpora lutea. Further we have also hypothesized that serum bPAG-1 concentrations
82
will be different between single and twin pregnant groups at each of the time-points and thus will
83
serve as a sensitive and specific clinical indicator for the prediction of the number of fetuses.
TE D
79
84
The objectives of this study were to compare serum P4 (1) and bPAG-1 (2) concentrations at
86
different time-points of gestation in cattle with the accurate determination of one or two corpora
87
lutea; to assess whether or not serum P4 concentration is a useful clinical indicator for the
88
determination of the number of corpus lutea at any of the previously mentioned time-points (3);
89
and to assess whether or not serum bPAG-1 concentration is a useful clinical indicator for the
90
determination of the number of fetuses at any of the above mentioned time-points (4).
AC C
EP
85
91 92
2. Materials and methods 4
ACCEPTED MANUSCRIPT
93
2.1 Animals
95
Altogether 84 Holstein-Friesian dairy cattle were included in our study with a confirmed
96
pregnancy by means of ultrasonography at the late embryonic period between Days 29 to 42 of
97
gestation (time-point 1). Animals were divided into 3 groups: the first group (TWIN Group,
98
n=29) had two viable embryos in the uterine horns and two mature corpora lutea on the ovaries at
99
the time of pregnancy diagnosis. Animals with one viable embryo in the pregnant uterine horn
100
and two corpora lutea on the ovaries formed the second group (DCL Group, n=35), while animals
101
with a singleton pregnancy (one viable embryo and one corpus luteum) served as the control
102
group (CON Group, n=20). All animals were kept in free barns with cubicles and were fed
103
according to the NRC recommendations [13].
M AN U
SC
RI PT
94
104
2.2 Ultrasonographic examination and blood sampling
106
Animals were sampled four times during the study. First, at the time of early pregnancy diagnosis
107
(time-point 1) which was performed by a real-time B-mode diagnostic ultrasound scanner
108
(EasiScan, BCF technologies, Bellshill, UK) equipped with a 5 to 9 MHz linear-array rectal
109
transducer, then monthly for three additional occasions (namely: between Days 57 to 70 (time-
110
point 2), Days 85 to 98 (time-point 3), and Days 113 to 126 (time-point 4) of gestation,
111
respectively). For the diagnosis of early pregnancy and ovarian structures 7.5 MHz transducer
112
was used. All ultrasonographic examinations and blood samplings were performed by the same
113
operator. Individual animals were examined in stalls in the milking house with dim ambient
114
lighting. After removal the feces, the transducer was inserted into the rectum and was slowly
115
moved from the cervix to the right, and then to the left uterine horn until the entire uterus was
AC C
EP
TE D
105
5
ACCEPTED MANUSCRIPT
imaged. No manipulation of the uterus took place in any of the investigations. Scanning of the
117
uterus and interpretation the echoscopic images have been described previously [14]. At each
118
examination, the operator was required to record the number and the presence of embryos in the
119
uterine horn and the number of the corpus luteum. At every time-point, when it was possible, the
120
viability of the fetus was confirmed by the detection of the heartbeat. In those cases, when
121
scanning of the entire uterus especially at time-points 3 and 4 was not feasible, the presence and
122
the movement of the fetus was used to confirm the viable pregnancy. Following pregnancy
123
diagnosis animals were handled according to the reproductive protocol of the particular farm. At
124
the time of drying-off pregnancy was confirmed by rectal palpation. Finally our ultrasonographic
125
diagnoses were confirmed by calving records. Animals were excluded from the study if repeated
126
examinations yielded inconsistent diagnoses.
M AN U
SC
RI PT
116
127
Following transrectal ultrasonographic examinations blood samples were collected into
129
vacutainer tubes (Monovette 9ml, Sarstedt, Nümbrecht, Germany) at every time-point. The tubes
130
were transported within 2 h into the laboratory in a cool-box containing icebags, centrifuged at
131
3000 x g/ 10 min) and the serum samples were kept at -20 C until measurements of P4 and
132
bPAG-1 concentrations.
EP
AC C
133
TE D
128
134
2.3 P4 and bPAG-1 assay procedures
135
Plasma concentrations of P4 were estimated by validated solid-phase
136
(RIA) method (Coat-A-Count TKPG; Diagnostic Products Corporation) as described by [15].
137
The sensitivity was 0.15 nmol/L, and the intra- and inter-assay coefficients of variation were 8.9
138
and 11%, respectively.
125
I radioimmunoassay
6
ACCEPTED MANUSCRIPT
139
Plasma bPAG-1 measurements were performed according to the method of Zoli et al. [16] and
141
subsequently modified by Perényi et al. [17]. A polyclonal antiserum was used [16] which was
142
produced against PAG I67kDa. A PAG I67kDa preparation, purified according to the protocol of
143
Zoli et al. [18] was used as standard and tracer. The sensitivity of the bPAG-1 RIA test was 0.06
144
ng/mL, and the intra- and inter-assay coefficients of variation were 5.0 and 5.5%, respectively.
RI PT
140
SC
145
2.4 Statistical analysis
147
The data were analyzed using a commercially available software (Minitab 16, State College, PA,
148
USA). Normal distribution of the data was tested with the Shapiro-Wilk test. Serum P4 and
149
bPAG-1 concentrations were compared within groups (TWIN, DCL, CON) and different time-
150
points (Time-point 1: between Days 28 to 42, Time-point 2: between Days 57 to 70, Time-point
151
3: between Days 85 to 98, and Time-point 4: between Days 113 to 126, respectively) using a
152
generalized linear model (GLM), due to the unbalanced design. Post-hoc evaluation was
153
performed with the Bonferroni correction.
TE D
EP
154
M AN U
146
Regarding to P4 concentrations, after evaluating the animals in a generalized linear model, the
156
TWIN and the DCL group (cows having 2 CL and one embryo) were condensed together and
157
compared to the CON group (1 CL and one embryo) at the four different time points. Binary
158
logistic regression was used to reveal any association between serum P4 concentration and the -
159
number of corpus luteum at each of the 4 time-points. At each time-points, when the regression
160
model was significant, a receiver operating curve (ROC) was created to find the most specific
161
and sensitive cut-off value for discrimination of cows with 2 CLs vs. 1 CL. Area under the curve
AC C
155
7
ACCEPTED MANUSCRIPT
162
(AUC) was also calculated.
163
To find the accuracy of pregnancy protein levels, the CON group and the DCL (1 embryo/fetus
165
with 2 CL) group were also condensed and compared to the TWIN group (2 embryos/fetuses) at
166
each of the four time-points regarding to pregnancy protein concentrations. Binary logistic
167
regression analysis was used to reveal any association between serum bPAG-1 concentration and
168
number of embryos/fetuses at each time-points. A receiver operating curve was created to reveal
169
the most specific and sensitive cut-off value for the determination of the number of
170
embryos/fetuses. Area under the curve (AUC) was also calculated.
M AN U
SC
RI PT
164
171
The AUC was regarded as follows: AUC = 0.5 (no discrimination); 0.7 ≤ AUC < 0.8 (acceptable
173
discrimination); 0.8 ≤ AUC < 0.9 (excellent discrimination); and AUC > 0.9 (outstanding
174
discrimination). The goodness of the fit of the logistic regression models were tested with the
175
Hosmer-Lemeshow method. Level of significance was set at P
