- Email: [email protected]
Factors affecting plasma progesterone concentration and the retrospective determination of time of ovulation in cyclic mares
Theriogenology 61 (2004) 203–214
Factors affecting plasma progesterone concentration and the retrospective determination of time of ovulation in cyclic mares P. Nagya,1, Gy. Huszeniczaa, J. Reiczigelb, J. Juha´szc,1, M. Kulcsa´ra, K. Abava´rya, D. Guillaumed,* a
Department of Obstetrics and Reproduction, Szent Istvan University, Faculty of Veterinary Science, Budapest, Hungary b Department of Biomathematics, Szent Istvan University, Faculty of Veterinary Science, Budapest, Hungary c Research Institute, Szent Istvan University, Faculty of Veterinary Science, Budapest, Hungary d I.N.R.A., Reproductive Physiology and Behaviour, Nouzilly, France Received 9 July 2001; accepted 2 October 2001
Abstract Factors influencing plasma progesterone concentration were investigated in seven mares. Two-phase logistic curves were fitted (r ¼ 0:98) to plasma progesterone concentrations of blood samples collected once daily. In addition to the effect of time (P < 0:001), there were differences (P < 0:01) among mares in the peak height of the progesterone plateau and in the (area under the curve) AUC. Plasma progesterone concentrations were higher (P < 0:001) after a multiple versus single ovulation. There was an effect of season (P < 0:001), but no significant effect of luteal morphology. The retrospective determination of time of ovulation was carried out using a linear model on the seven mares and 25 additional mares. Linear regression on the measured values or on the ratio to the average concentration from D5 to D10, was calculated with the day of cycle between D0 and D4. The ovulation date was then calculated using both of these equations, whether blood sampling was performed twice or thrice weekly on 25 postpartum mares. The accuracy to predict day of ovulation (1 day) ranged from 88 to 97%. In conclusion, the retrospective estimation of time of ovulation in mares was possible, although the technique had some limitations. # 2003 Elsevier Inc. All rights reserved. Keywords: Mare; Progesterone; Logistic regression; Ovulation; Corpus luteum
* Corresponding author. Tel.: þ33-2-47-42-77-00; fax: þ33-2-47-42-77-43. E-mail address: [email protected] (D. Guillaume). 1 Present address: Central Veterinary Research Laboratory, Dubai, U.A.E.
0093-691X/$ – see front matter # 2003 Elsevier Inc. All rights reserved. doi:10.1016/S0093-691X(03)00211-5
204
P. Nagy et al. / Theriogenology 61 (2004) 203–214
1. Introduction In mares, intrafollicular concentration of progesterone increased in the dominant follicle 2 days prior to ovulation [1] but plasma progesterone concentrations were low during the follicular phase. With frequent blood sampling, the first significant rise in plasma progesterone concentrations occurred 10–12 h after ovulation [2–5], but there was substantial variability (6–60 h) among mares. With once-daily blood sampling, this plasma progesterone increase was detected 24–48 h after ovulation [4,6–8,9–12]. Therefore, variation among mares in the postovulation rise in peripheral progesterone concentrations made it difficult to accurately estimate time of ovulation (in a specific mare) based on plasma progesterone concentration [2]. In equine embryo transfer, it is important to synchronize (1 day) the time of ovulation between the donor and recipient mare. Although a method has been proposed to retrospectively determinate the time of ovulation based on progesterone concentration [13], the accuracy (approximately 70%) may be inadequate for many applications. Plasma progesterone curves can be summarised by multi-phasic logistic curves. A twophase logistic model was used for jennies [14] and a combined logistic and exponential model was used in cows [15]. However, the latter combined model had a disruptive point (between the two parts of the curve) that was difficult to estimate and therefore this model was not retained. During the luteal phase, maximum peripheral progesterone concentration is highly variable among mares [11]. Some ultradian but not diurnal changes were also observed during this period [16,17]. One study [18] found some seasonal differences in both luteal and follicular phase progesterone values. The morphology of the corpus luteum (as determined by ultrasonography) did not significantly affect peripheral progesterone concentration [5]. However, in some reports, mares with a double ovulation had a higher plasma progesterone concentration than those with only a single ovulation [8,19–22]. The objectives of the present study were to (a) model plasma progesterone profiles of mares with an appropriate logistic curve and to use these curves to determine factors influencing plasma progesterone concentrations; and (b) retrospectively estimate the date of ovulation from plasma progesterone concentrations when samples were taken twice or thrice weekly.
2. Materials and methods 2.1. Animals In the first part of the study, seven trotter mares from the Research Institute of the University, in good body condition (mean S:E:M:, 9:1 2:04 years of age, 581:4 50:47 kg of body weight) were monitored daily from June to October. They were kept in an open barn during the night and on pasture during the day and fed hay and oats twice a day. Examinations were started on the same day for all mares, but only data following the first detected ovulation were included in the final evaluation. Rectal palpation and ultrasonography2 were done each morning. The number and size of follicles and the 2
Aloka SS-210 DX with 5 MHz transducer, Aloka Inc., Japan.
P. Nagy et al. / Theriogenology 61 (2004) 203–214
205
size and quality of the corpus luteum were recorded. The day of ovulation was defined as Day 0 (D0). Double ovulations were considered synchronous if both ovulations were detected within 24 h. In case of asynchronous multiple ovulations, D0 was the day of the first ovulation. Two types of luteal glands were distinguished based on their ultrasonic appearance: uniformly echogenic and centrally non-echogenic [5]. Blood samples were collected each morning and immediately centrifuged and plasma was frozen until assayed for progesterone. In the second part of the study, for the retrospective determination of the time of ovulation, data from 26 mixed-breed postpartum mares (4–19 year, foaled from February to June) from another study were added to the data from the seven trotter mares previously described. Ovarian activity was monitored by estrus detection, rectal palpation and ultrasonography.3 Blood samples (for progesterone analysis) were collected thrice weekly, from 5 day after foaling to confirmation of a new pregnancy. Blood was processed as described above. 2.2. Progesterone assay Plasma progesterone concentrations were determined with a radioimmunoassay validated for equine plasma, without extraction, with 3 H progesterone as tracer. The antiserum was highly specific for progesterone, with a particularly low cross reactivity with steroids (0.0001–0.6%) other than 11a-OH-progesterone (96%) [23]. The sensitivity of the assay was 0.88 nmol/l progesterone (20 ml plasma per tube). The intra- and inter-assay coefficients of variations for plasma low (3–4 nmol/l), medium (6–7 nmol/l) and high (15– 19 nmol/l) ranges were 14.3, 8.2, 7.1 and 15.2, 9.1, 8.2%, respectively. This assay was compared to a radioimmunoassay (with extraction) using the same antibody [24]; there was a close relationship between the two assays (r ¼ 0:94). 2.3. Statistical analysis The Eq. (1), difference between two logistic functions (Eq. (2) and Eq. (3)) was fitted to daily plasma progesterone concentrations after ovulation. PðtÞ ¼ gðtÞ hðtÞ
(1)
gðtÞ ¼
A1 1 þ expfA2 ðt þ A3 Þg
(2)
hðtÞ ¼
A4 1 þ expfA5 ðt þ A6 Þg
(3)
where P(t) is progesterone plasma concentration (nmol/l) at time t (day). A1–A6 are curve parameters estimated by the non-linear regression procedure of SAS software using the D.U.D. algorithm [25]. A1 and A4 are positive in nmol/l; A2 and A5 are negative in inverse of time (1 day1); A3 and A6 are negative in days. The simple logistic function, g(t), increases continuously from 0 to A1; at time 0, its value is A1/[1 þ expðA2 A3 Þ] and the slope at the 3
Pie Medical 450 with 5 MHz transducer, Pie Medical Inc., The Netherlands.
206
P. Nagy et al. / Theriogenology 61 (2004) 203–214
inflection point (A3; A1/2) is (A1; A2/4). In Eq. (1), the first logistic function, g(t), characterises the production rate of the corpus luteum; the slope at the inflection point represents the maximum rate of progesterone production. On the other hand, h(t) characterises the disappearance of progesterone from the blood compartment. To test the effect of different factors on plasma progesterone, ANOVA with an unbalanced model was used. The parameters of the logistic function and the area under the curve (AUC) [26] were used as dependent variables to test the following main effects: (a) mare, (b) number of ovulations, (c) synchronous and asynchronous multiple ovulations, (d) ultrasonic morphology of the corpus luteum; and (e) season of ovulation. The effect of these factors as well as day of the cycle on progesterone concentrations were also analysed using ANOVA for repeated measures. In the case of significant effects (P < 0:05), paired comparisons were also carried out according to the day of cycle. This model with two logistic functions cannot be used to retrospectively determine the time of ovulation. Although the postovulation increase in plasma progesterone was approximately linear (this straight line was used to predict this ovulation), its slope depended on blood sampling frequencies. Linear regression of the progesterone concentration by day of cycle (from D0 to D4) was calculated with both absolute and relative values. The relative value was defined as the ratio of the measured value to the mean of D5– D10 values. The resulting regression equations were applied to the trotter mares (using only twice-weekly blood samples) and to the postpartum mares (using thrice-weekly sampling). Three methods were used to calculate the day of cycle on the basis of progesterone concentrations. For the three methods, the first increase was defined as a rise of at least 0.5 nmol/l after low values. Method 1 and 2 used relative values. In Method 1, the maximal concentration was defined as the average of the second and third high values (sampling twice a week) or the average of the third and fourth (sampling three times a week) high values. In Method 2, the second or third (twice versus thrice weekly sampling, respectively) high value were regarded as the maximum value of the cycle. In Method 3, gross measured values were used. The estimated day of cycle was compared to the true day of cycle. The results of Methods 1, 2 and 3 were compared using the Chi-square test.
3. Results The two-phase logistic curves were fitted to a total of 37 cycles of the seven trotter mares. The calculated curves exhibited excellent correlation with the measured profiles (r ¼ 0:98; Fig. 1). For one cycle, a rise in plasma progesterone at the end of diestrus affected curve fitting; therefore, data for this cycle were excluded from the statistical analysis. 3.1. Effect of day of cycle There was an effect (P < 0:001) of day of cycle on plasma progesterone concentrations. Mean plasma concentration on the day of ovulation was below assay sensitivity (0.18 nmol/l); it increased above the detection limit by D1 (1.32 nmol/l) and continued to rise until D5. Mean plasma progesterone concentrations remained high until D13, then decline, and reached values below assay sensitivity by D16.
P. Nagy et al. / Theriogenology 61 (2004) 203–214
Progesterone (nmol/L)
25
SO
DO
SO
DO
SO
207
ADO
20 15 10 5 0 21/6
6/7
21/7
5/8
20/8
4/9
19/9
4/10
19/10
Date
Fig. 1. Plasma progesterone-profile of a mare throughout the study (calculated curve); (^) indicates measured values and the arrow shows the time of diestrus ovulation; SO: single ovulation; DO: synchronous double ovulation; ADO: asynchronous double ovulation.
3.2. Effect of mare There were differences among mares, in both the upper asymptotic values (A1 and A4) of the calculated curves (P < 0:01) and the AUC (P < 0:01), irrespective of the number of ovulations. None of the other curve parameters were significantly different. Two-way ANOVA of daily progesterone values with the day of cycle and the mare as main effects gave the same results (P < 0:001), without a significant day of cycle by mare interaction. 3.3. Effect of the number of ovulations Twenty six (60.5%) cycles had a single ovulation, 15 (35%) double and 2 (5%) triple ovulations. Regarding the 17 multiple ovulations, 12 (28%) were synchronous; the remaining 5 (12%) that were asynchronous were detected on D2, D4 (two cases), D6 and D7, respectively. Multiple ovulations occurred in each month of the study but there was a trend towards a higher incidence in July and August. Means and standard deviations of estimated parameters of the progesterone profiles according to single and multiple synchronous or asynchronous ovulations are shown in Table 1. The number of ovulations (single versus multiple) had an effect on the upper asymptotic values (A1, A4; P < 0:001), on the locations of the inflection points (A3, A6; P < 0:05) of the calculated curves, and on the AUC (P < 0:01). The ratio of AUC of multiple ovulation to AUC of single ovulation was 1.36. Analysis of variance of daily progesterone concentrations to D12 revealed not only the effect of the number of ovulations (P < 0:001), but interactions of day of cycle by number of ovulations and mare by number of ovulations (P < 0:001 for each). In the latter part of the interovulatory interval (after D12), plasma progesterone concentrations were similar for cycles with single versus multiple ovulations. In the case of synchronous multiple ovulations, there was a trend toward higher mean plasma progesterone concentrations as early as 2 day after ovulation, but the difference became significant only after D4 (and remained high until D10). Mean plasma progesterone profiles and logistic curves for
208
Ovulation Single (n ¼ 22) Synchronous double (n ¼ 10) Asynchronous double (n ¼ 5)
A1
A2 a
12.21 3.67 15.92 5.85b 17.37 4.44b
A3 c
1.53 0.49 1.44 0.33c 0.77 0.38d
A4 a
2.63 0.54 2.62 0.27a 4.74 1.60b
A5 a
12.34 3.83 15.98 5.73b 17.24 4.31b
A1–A6 are estimated parameters defined in Eq. (2) and Eq. (3). Within a column, values with different superscripts (a–d) are different (a, b: P < 0:05; c, d: P < 0:01).
3.26 3.81 2.37 1.40 3.05 0.99
A6
AUC c
12.79 1.01 13.03 1.05c 14.68 0.51d
121.5 36.9a 162.1 58.5b 171.0 46.7b
P. Nagy et al. / Theriogenology 61 (2004) 203–214
Table 1 Mean S:D: curve parameters and areas under the curve (AUC) in cycles with single, multiple synchronous, or multiple asynchronous ovulations
Progesterone (nmol/L)
P. Nagy et al. / Theriogenology 61 (2004) 203–214
209
20 15 10 5 0 0
5
10
15
20
Day of cycle (0 = day of ovulation) Fig. 2. Mean plasma progesterone profiles of single and synchronous multiple ovulations (calculates curves); (&) and ( ) indicated mean values for single and multiple ovulation cycles, respectively; the arrow shows the period of significant difference.
Progesterone (nmol/L)
single and synchronous multiple ovulation cycles are shown in Fig. 2. The interovulatory interval was not significantly different for single versus multiple ovulations (17.82 day versus 19.13 day). In the case of asynchronous multiple ovulations, the shape of the estimated profiles was different (A2, P < 0:01; A3, P < 0:05; A6, P < 0:01). In these cycles, progesterone increased slowly after the first ovulation, but the increase was more pronounced after the second ovulation and progesterone reached the same plateau as in the case of synchronous multiple ovulations (no difference in A1 and AUC; Fig. 3). Daily plasma progesterone concentration was significantly higher in the increasing phase (D1–D4, P < 0:01) for synchronous multiple ovulations, but this difference disappeared after D5. In comparison to single ovulations, asynchronous multiple ovulations were different for both curve parameters (A2, P < 0:01; A3, P < 0:05) and plasma progesterone increase during the first 5 day after ovulation (P < 0:05). The slope of the curve was lower after asynchronous multiple ovulations than after a single ovulation.
20 15 10 5 0 0
5 10 15 Day of cycle (0 = day of ovulation)
20
Fig. 3. Mean plasma progesterone profiles of synchronous and asynchronous multiple ovulations (calculated curves); (&) and ( ) indicated mean values for synchronous and asynchronous multiple ovulation cycles, respectively; the arrow shows the period of significant difference.
210
P. Nagy et al. / Theriogenology 61 (2004) 203–214
3.4. Effect of corpus luteum morphology Twenty six (69.2%) single ovulations resulted in an uniformly echogenic corpus luteum, whereas in eight cases (30.8%), there was a non-echogenic central area that remained within the corpus luteum throughout most of its lifespan. The frequency distribution of the two types of luteal morphology was similar among mares. There was no effect of the morphology on any parameters of the calculated profiles or on the AUC. Analysis of variance of plasma progesterone in the first 10 days of the cycle failed to detect any significant effect of morphology or day of cycle by morphology interaction. However, there was a difference in mean daily progesterone on D12–D14 (12.08 nmol/l versus 7.30 nmol/l, 9.03 nmol/l versus 2.41 nmol/l and 5.07 nmol/l versus 0.33 nmol/l for uniform and fluidfilled corpora lutea, respectively; P < 0:05). Although this suggested that luteolysis of fluid-filled corpora lutea occurred earlier, the length of the interovulatory interval was not significantly different (18.14 and 17.25 day for uniform and fluid-filled corpora lutea, respectively). 3.5. Effect of season There were 26 single ovulations; 16 in the summer and 10 in the autumn. No effect of season was detected, either on the parameters of the calculated profiles or on AUC. However, when day of cycle, mare and season were included together as main effects in the model, there was a negative effect of season on daily progesterone (P < 0:001), with no day of cycle by season or mare by season interactions. 3.6. Description of progesterone increase after ovulation
Progesterone (relative values) (P4/P4 (D5-D10))
Both measured and relative values were used to describe the progesterone increase after ovulation. Progesterone increase was linear between D1 and D4 (Fig. 4). Linear regression
1.2 1 0.8 0.6 0.4 0.2 0 -10
-5
0
5
10
15
Day of cycle (0 = day of ovulation) Fig. 4. Increase in plasma progesterone after ovulation; daily concentrations are expressed as relative values (ratio of the maximum concentration); (&) indicates mean relative value; regression line: Y ¼ 0:22, X 0:06.
P. Nagy et al. / Theriogenology 61 (2004) 203–214
211
Table 2 Accuracy (%) of Methods 1, 2 and 3 for determining the true day of cycle with different sampling protocols Method
1 2 3
Blood sampling twice weekly
Blood sampling thrice weekly
Exact match
1 day
Exact match
1 day
47.4 53.3 46.6
97.7 96.7 91.4
61.8 58.8 52.9
88.2 88.2 88.2
between the relative value and the day of cycle from D0 to D4 resulted in the following equation: Y ¼ 0:22t 0:06
ðr ¼ 0:89Þ
(4)
where t ¼ the day of cycle and Y ¼ the ratio of daily plasma progesterone concentrations to maximum concentrations. Linear regression between the daily measured values and the day of cycle from D0 through D4 resulted in the following equation: Y ¼ 2:7t 0:67
ðr ¼ 0:83Þ
(5)
where t ¼ the day of cycle and Y ¼ the daily plasma progesterone concentrations. 3.6.1. Comparison of methods for calculating the time of ovulation Using these equations and assuming blood sampling twice weekly, the day of cycle was recalculated from daily progesterone values between D0 and D4 using the three different methods. The true day of cycle was compared to the calculated day. Calculated values were around the true day except for two cases with Method 1, (in one case the 20th, in the other case the 12th day of the cycle were calculated, these were the consequence of low progesterone value in the 3rd high sample after ovulation). Chi-square test revealed no significant difference in the distribution of calculated days according to the different methods. 3.6.2. Accuracy of the determination of time of ovulation The calculated days were the true days of the cycle in 46.6–58.8% of the cases. Accuracy was higher when blood samples were taken more frequently (thrice weekly). However, the 1 day accuracy was >88% in all cases, irrespective of the sampling protocol. Accuracy data for Methods 1, 2 and 3 are shown in Table 2.
4. Discussion The fitting of a sum of logistic curves (Eq. (1)) well describes the individual progesterone curve, as previously reported for jennies [14]. The effect of the day of cycle was entirely consistent with numerous preliminary reports [5–12]. Progesterone concentrations increased from D1 to D5 and thereafter reached a plateau until luteolysis. With frequent sampling protocols, the time of the first increase occurred as early as 8–10 [4], 12 [3], 12–24 [5] and 12–18 h [2] after ovulation. The high
212
P. Nagy et al. / Theriogenology 61 (2004) 203–214
variability of the peak (plateau) concentrations among mares with a single ovulation mares was confirmed [11,27]; this may be due to the secretory capacity of the corpus luteum or the rate of progesterone catabolism; the latter could explain higher plasma progesterone concentrations in the first 5 day after ovulation in ponies compared to mares [5]. The incidence (27.9%) of synchronous multiple ovulations did not differ from that reported earlier for this breed [28]. Two corpora lutea secreted more progesterone than one; differences were significant both in the concentrations during the plateau and in the AUC (more than 35%). In the case of synchronous multiple ovulations, progesterone reached a higher plateau at the same time as for a single ovulation. In this case, the two corpora lutea stopped secreting progesterone 2 days before a single ovulation. These results confirm those obtained by some authors [8,19,21,22] but not by others [12,20]. With asynchronous double ovulations, the initial rise in progesterone was followed by an additional increase after the second ovulation. The shape of the curve was different from synchronous multiple ovulations, but the maximum concentration was the same. Since all diestrus ovulations occurred before D8, no prolonged luteal activity was detected. After this second ovulation, plasma progesterone seemed to increase slower than after a single ovulation, but this is probably an artifact of the model. The logistic model or the straight line model do not take into account this type of asynchronous ovulation. A previous report [5] on corpus luteum scanning morphology and progesterone concentrations did not describe any difference in progesterone level between uniform and fluid-filled corpora lutea; the present results confirm this report. Corpora lutea with central non-echogenic areas started to decrease their production rate earlier than corpora lutea with homogenous echotexture. However, there was no significant difference in interovulatory intervals. There are reports [5,18,29] of an influence of season on circulating progesterone concentrations. These authors agree on a progressive seasonal decline in diestrus plasma progesterone to the anovulatory period. Our results seemed to support this hypothesis. In our study, all mares stopped cycling during the following winter. This plasma progesterone decline was perhaps the consequence of the decline in the LH surge. However, recent data indicated that amplitude of the LH surge was not implicated in the maintenance of corpus luteum [30]. Therefore, other explanations are needed to explain this seasonal effect. The plasma progesterone increase during the first 4 days after ovulation was approximately linear (Fig. 4), so it was simple and convenient to use this straight line to predict the time of ovulation. Methods 1 and 2 could not be immediately applied, because they required additional samples to determine maximal concentration. If the time of ovulation was calculated using these methods, the results would often be available too late for many clinical applications. When the luteal phase is short, Method 1 is inappropriate because the second sample used for estimating the maximum concentration may be too low. In Method 3, measured values are used without further transformation. Owing to the greater variation, the correlation coefficient of Eq. (5), was lower (0.83 versus 0.89) than in Eq. (4), which used relative values. However, there was no difference between methods in calculating the day of cycle (compared to the true day of the cycle); all models provided 1 day accuracy >88% of the time. When errors occurred, the actual day of cycle was usually underestimated; furthermore, the same postovulatory periods tended to cause problems in all methods. One typical problem was related to asynchronous multiple (diestrus) ovulations.
P. Nagy et al. / Theriogenology 61 (2004) 203–214
213
In these cycles, there was a slower increase in the first 5 days and the maximum concentrations were higher than in single-ovulation cycles. The other common source of mistakes was the difference in progesterone plateau among mares. If a mare typically had a low maximum concentration, then the day of cycle calculated using Method 3 would be much less than the actual day. Any of the methods mentioned above could be used for the retrospective determination of the time of ovulation, (with some limitations) when blood samples are collected twice or thrice weekly. However, for cyclicity studies, Method 3 seems to be easier. Additional studies on the management of recipient mares would be required to test the effectiveness of these methods under field conditions. In conclusion, the factors affecting plasma progesterone concentrations in cyclic mares were the day of cycle, the number of ovulations, the time of the second ovulation and season. Luteal morphology had no significant effect. The retrospective estimation of time of ovulation based on a postovulatory progesterone increase was possible; the majority of errors were attributed to the great variability in plasma progesterone concentrations during the first few days after ovulation.
Acknowledgements The authors wish to thank the Somogysa´ rd Police Stud, Dr. Frigyes Janza and Dr. Ja´ nos Seregi for all their efforts in providing the mares and to Ms. Erzse´ bet Kiss, Ms. Ibolya Simon and Mr. Be´ la Budai for the progesterone assays, Mr E. Palmer for his encouragements, Ms. A.M Wall (INRA) for the English correction. This study was partly funded by Grants of the National Research Found (OTKA, project numbers: 16473, F 013077).
References [1] Belin F, Goudet G, Duchamp G, Ge´ rard N. Intrafollicular concentrations of steroids and steroidogenic enzymes in relation to follicular development in the mare. Biol Reprod 2000;62:1335–43. [2] Koskinen E, Lindeberg H, Kuntsi H, Katila T. Milk and serum progesterone levels in mares after ovulation. Acta Vet Scand 1990;31:441–4. [3] Pipkin JL, Forrest DW, Kraemer DC, Wilson JM. Plasma steroid concentrations after conception in mares. J Reprod Fertil 1987;35:711–4. [4] Plotka ED, Foley CW, Witherspoon DM, Schmoller GC, Goetsch DD. Periovulatory changes in peripheral plasma progesterone and oestrogen concentrations in the mare. Am J Vet Res 1975;36:1359–62. [5] Townson DH, Pierson RA, Ginther OJ. Characterization of plasma progesterone concentrations for two distinct luteal morphologies in mares. Theriogenology 1989;32:197–204. [6] Allen WE, Hadley JC. Blood progesterone concentrations in pregnant and non-pregnant mares. Equine Vet J 1974;6:87–93. [7] Noden PA, Oxender WD, Hafs HD. The cycle of oestrus, ovulation and plasma levels of hormones in the mare. J Reprod Fertil 1975;23:189–92. [8] Palmer E, Jousset B. Urinary oestrogen and plasma progesterone levels in non-pregnant mares. J Reprod Fertil 1975;23:213–21. [9] Plotka ED, Witherspoon DM, Foley CW. Luteal function in the mare as reflected by progesterone concentrations in peripheral blood plasma. Am J Vet Res 1972;33:917–20. [10] Sharp DC, Black DL. Changes in peripheral plasma progesterone throughout the oestrus cycle of the pony mare. J Reprod Fertil 1973;33:535–8.
214
P. Nagy et al. / Theriogenology 61 (2004) 203–214
[11] Smith ID, Bassett JM, Williams T. Progesterone concentrations in the peripheral plasma of the mare during the oestrus cycle. J Endocrinol 1970;47:523–4. [12] Stabenfeldt GH, Hughes JP, Evans JW. Ovarian activity during the oestrus cycle of the mare. Endocrinology 1972;90:1379–84. [13] Hinrichs K, Sertich PL, Solorzano NM, Caldwell LA. Use of an immediate, qualitative progesterone assay for determination of day of ovulation in an equine embryo transfer program. Theriogenology 1988;29: 1123–30. [14] Meira C, Ferreira JCP, Papa FO, Tornero MTT, Bicudo SD. Study of the oestrus cycle in donkeys (Equus asinus) using scan and plasma progesterone concentrations. Biol Reprod 1995;(Mono 1):403–10. [15] Yenikoye A, Mariana JC, Ley JP, Jolivet E, Terqui M, Lemon-Resplandy M. Mode`le mathe´ matique de l’e´ volution de progeste´ rone chez la vache: application et mise en e´vidence de diffe´ rence entre races. Reprod Nutr Dev 1981;21:561–75. [16] Cockrill T, Allen WE. Diurnal variation of plasma progestagen concentrations in pony mares. Vet Rec 1978;102:503. [17] Evans JW. Biorhythms in plasma progesterone concentrations and absence of any correlation to LH biorhythms during different stages of the equine oestrus cycle. J Reprod Fertil 1991;(Suppl 44):684–5. [18] Oko´ lski A, Bielan˜ ski AB. Seasonal fluctuations of the plasma progesterone levels in mares with similar ovarian morphology. Int Cong Anim Reprod Artif Insem Madrid 1980;4:13–5. [19] Henry M, Coryn M, Vandeplassche M. Multiple ovulation in the mare. Zbl Vet Med A 1982;29:170–84. [20] Roser JF, Evans JW. Luteal luteinizing hormone receptors during the postovulatory period in the mare. Biol Reprod 1983;29:499–510. [21] Squires EL, McKinnon AO, Carnevale EM, Morris R, Nett TM. Reproductive characteristics of spontaneous single and double ovulating mares and superovulated mares. J Reprod Fertil 1987;35: 399–403. [22] Urwin VE, Allen WR. Follicle stimulating hormone, luteinising hormone and progesterone concentrations in the blood of Thoroughbred mares exibiting single and twin ovulations. Equine Vet J 1983;15:325–9. [23] Csernus V. Antibodies of high affinity and specificity for radioimmunological determination of progesterone, testosterone, estradiol-17b and cortisol. In: Go¨ ro¨ g S, editor. Advances in steroid analysis I. Budapest: Academy Press; 1982. p. 171–7. [24] Huszenicza Gy, Ba´ ba A, Nagy P, Juha´ sz J, Kulcsa´ r M, Abava´ ry MK, et al. Characteristics of the ovarian function in non-pregnant breeding mares at the beginning of the official breeding season. Hung Vet J 1993;48:587–92. [25] SAS User’s Guide, Statistics, Version 6 edition. Cary, NC: SAS Institute Inc.; 1986. [26] Matthews JNS, Altman DG, Campbell MJ, Royston P. Analysis of serial measurements in medical research. Br Med J 1990;300:230–5. [27] Squires EL, Wentworth BC, Ginther OJ. Progesterone concentration in blood of mares during the oestrus cycle, pregnancy and after hysterectomy. J Anim Sci 1974;39:759–67. [28] Ginther OJ. Reproductive biology of the mare. 2nd ed. Cross Plains, WI: Equiservices; 1992. p. 218–40. [29] King SS, Nequin LG, Drake S, Hebner TS, Roser JF, Evans JW. Progesterone levels correlate with impending anoestrus in the mare. J Equine Vet Sci 1988;8:109–11. [30] Briant C, Ottogali M, Morel M, Guillaume D. Use of GnRH antagonist, antarelix, associated or not with hCG, to control ovulation in cyclic pony mares. Domest Anim Endocrinol 2003;53:1–18.
Factors affecting plasma progesterone concentration and the retrospective determination of time of ovulation in cyclic mares P. Nagya,1, Gy. Huszeniczaa, J. Reiczigelb, J. Juha´szc,1, M. Kulcsa´ra, K. Abava´rya, D. Guillaumed,* a
Department of Obstetrics and Reproduction, Szent Istvan University, Faculty of Veterinary Science, Budapest, Hungary b Department of Biomathematics, Szent Istvan University, Faculty of Veterinary Science, Budapest, Hungary c Research Institute, Szent Istvan University, Faculty of Veterinary Science, Budapest, Hungary d I.N.R.A., Reproductive Physiology and Behaviour, Nouzilly, France Received 9 July 2001; accepted 2 October 2001
Abstract Factors influencing plasma progesterone concentration were investigated in seven mares. Two-phase logistic curves were fitted (r ¼ 0:98) to plasma progesterone concentrations of blood samples collected once daily. In addition to the effect of time (P < 0:001), there were differences (P < 0:01) among mares in the peak height of the progesterone plateau and in the (area under the curve) AUC. Plasma progesterone concentrations were higher (P < 0:001) after a multiple versus single ovulation. There was an effect of season (P < 0:001), but no significant effect of luteal morphology. The retrospective determination of time of ovulation was carried out using a linear model on the seven mares and 25 additional mares. Linear regression on the measured values or on the ratio to the average concentration from D5 to D10, was calculated with the day of cycle between D0 and D4. The ovulation date was then calculated using both of these equations, whether blood sampling was performed twice or thrice weekly on 25 postpartum mares. The accuracy to predict day of ovulation (1 day) ranged from 88 to 97%. In conclusion, the retrospective estimation of time of ovulation in mares was possible, although the technique had some limitations. # 2003 Elsevier Inc. All rights reserved. Keywords: Mare; Progesterone; Logistic regression; Ovulation; Corpus luteum
* Corresponding author. Tel.: þ33-2-47-42-77-00; fax: þ33-2-47-42-77-43. E-mail address: [email protected] (D. Guillaume). 1 Present address: Central Veterinary Research Laboratory, Dubai, U.A.E.
0093-691X/$ – see front matter # 2003 Elsevier Inc. All rights reserved. doi:10.1016/S0093-691X(03)00211-5
204
P. Nagy et al. / Theriogenology 61 (2004) 203–214
1. Introduction In mares, intrafollicular concentration of progesterone increased in the dominant follicle 2 days prior to ovulation [1] but plasma progesterone concentrations were low during the follicular phase. With frequent blood sampling, the first significant rise in plasma progesterone concentrations occurred 10–12 h after ovulation [2–5], but there was substantial variability (6–60 h) among mares. With once-daily blood sampling, this plasma progesterone increase was detected 24–48 h after ovulation [4,6–8,9–12]. Therefore, variation among mares in the postovulation rise in peripheral progesterone concentrations made it difficult to accurately estimate time of ovulation (in a specific mare) based on plasma progesterone concentration [2]. In equine embryo transfer, it is important to synchronize (1 day) the time of ovulation between the donor and recipient mare. Although a method has been proposed to retrospectively determinate the time of ovulation based on progesterone concentration [13], the accuracy (approximately 70%) may be inadequate for many applications. Plasma progesterone curves can be summarised by multi-phasic logistic curves. A twophase logistic model was used for jennies [14] and a combined logistic and exponential model was used in cows [15]. However, the latter combined model had a disruptive point (between the two parts of the curve) that was difficult to estimate and therefore this model was not retained. During the luteal phase, maximum peripheral progesterone concentration is highly variable among mares [11]. Some ultradian but not diurnal changes were also observed during this period [16,17]. One study [18] found some seasonal differences in both luteal and follicular phase progesterone values. The morphology of the corpus luteum (as determined by ultrasonography) did not significantly affect peripheral progesterone concentration [5]. However, in some reports, mares with a double ovulation had a higher plasma progesterone concentration than those with only a single ovulation [8,19–22]. The objectives of the present study were to (a) model plasma progesterone profiles of mares with an appropriate logistic curve and to use these curves to determine factors influencing plasma progesterone concentrations; and (b) retrospectively estimate the date of ovulation from plasma progesterone concentrations when samples were taken twice or thrice weekly.
2. Materials and methods 2.1. Animals In the first part of the study, seven trotter mares from the Research Institute of the University, in good body condition (mean S:E:M:, 9:1 2:04 years of age, 581:4 50:47 kg of body weight) were monitored daily from June to October. They were kept in an open barn during the night and on pasture during the day and fed hay and oats twice a day. Examinations were started on the same day for all mares, but only data following the first detected ovulation were included in the final evaluation. Rectal palpation and ultrasonography2 were done each morning. The number and size of follicles and the 2
Aloka SS-210 DX with 5 MHz transducer, Aloka Inc., Japan.
P. Nagy et al. / Theriogenology 61 (2004) 203–214
205
size and quality of the corpus luteum were recorded. The day of ovulation was defined as Day 0 (D0). Double ovulations were considered synchronous if both ovulations were detected within 24 h. In case of asynchronous multiple ovulations, D0 was the day of the first ovulation. Two types of luteal glands were distinguished based on their ultrasonic appearance: uniformly echogenic and centrally non-echogenic [5]. Blood samples were collected each morning and immediately centrifuged and plasma was frozen until assayed for progesterone. In the second part of the study, for the retrospective determination of the time of ovulation, data from 26 mixed-breed postpartum mares (4–19 year, foaled from February to June) from another study were added to the data from the seven trotter mares previously described. Ovarian activity was monitored by estrus detection, rectal palpation and ultrasonography.3 Blood samples (for progesterone analysis) were collected thrice weekly, from 5 day after foaling to confirmation of a new pregnancy. Blood was processed as described above. 2.2. Progesterone assay Plasma progesterone concentrations were determined with a radioimmunoassay validated for equine plasma, without extraction, with 3 H progesterone as tracer. The antiserum was highly specific for progesterone, with a particularly low cross reactivity with steroids (0.0001–0.6%) other than 11a-OH-progesterone (96%) [23]. The sensitivity of the assay was 0.88 nmol/l progesterone (20 ml plasma per tube). The intra- and inter-assay coefficients of variations for plasma low (3–4 nmol/l), medium (6–7 nmol/l) and high (15– 19 nmol/l) ranges were 14.3, 8.2, 7.1 and 15.2, 9.1, 8.2%, respectively. This assay was compared to a radioimmunoassay (with extraction) using the same antibody [24]; there was a close relationship between the two assays (r ¼ 0:94). 2.3. Statistical analysis The Eq. (1), difference between two logistic functions (Eq. (2) and Eq. (3)) was fitted to daily plasma progesterone concentrations after ovulation. PðtÞ ¼ gðtÞ hðtÞ
(1)
gðtÞ ¼
A1 1 þ expfA2 ðt þ A3 Þg
(2)
hðtÞ ¼
A4 1 þ expfA5 ðt þ A6 Þg
(3)
where P(t) is progesterone plasma concentration (nmol/l) at time t (day). A1–A6 are curve parameters estimated by the non-linear regression procedure of SAS software using the D.U.D. algorithm [25]. A1 and A4 are positive in nmol/l; A2 and A5 are negative in inverse of time (1 day1); A3 and A6 are negative in days. The simple logistic function, g(t), increases continuously from 0 to A1; at time 0, its value is A1/[1 þ expðA2 A3 Þ] and the slope at the 3
Pie Medical 450 with 5 MHz transducer, Pie Medical Inc., The Netherlands.
206
P. Nagy et al. / Theriogenology 61 (2004) 203–214
inflection point (A3; A1/2) is (A1; A2/4). In Eq. (1), the first logistic function, g(t), characterises the production rate of the corpus luteum; the slope at the inflection point represents the maximum rate of progesterone production. On the other hand, h(t) characterises the disappearance of progesterone from the blood compartment. To test the effect of different factors on plasma progesterone, ANOVA with an unbalanced model was used. The parameters of the logistic function and the area under the curve (AUC) [26] were used as dependent variables to test the following main effects: (a) mare, (b) number of ovulations, (c) synchronous and asynchronous multiple ovulations, (d) ultrasonic morphology of the corpus luteum; and (e) season of ovulation. The effect of these factors as well as day of the cycle on progesterone concentrations were also analysed using ANOVA for repeated measures. In the case of significant effects (P < 0:05), paired comparisons were also carried out according to the day of cycle. This model with two logistic functions cannot be used to retrospectively determine the time of ovulation. Although the postovulation increase in plasma progesterone was approximately linear (this straight line was used to predict this ovulation), its slope depended on blood sampling frequencies. Linear regression of the progesterone concentration by day of cycle (from D0 to D4) was calculated with both absolute and relative values. The relative value was defined as the ratio of the measured value to the mean of D5– D10 values. The resulting regression equations were applied to the trotter mares (using only twice-weekly blood samples) and to the postpartum mares (using thrice-weekly sampling). Three methods were used to calculate the day of cycle on the basis of progesterone concentrations. For the three methods, the first increase was defined as a rise of at least 0.5 nmol/l after low values. Method 1 and 2 used relative values. In Method 1, the maximal concentration was defined as the average of the second and third high values (sampling twice a week) or the average of the third and fourth (sampling three times a week) high values. In Method 2, the second or third (twice versus thrice weekly sampling, respectively) high value were regarded as the maximum value of the cycle. In Method 3, gross measured values were used. The estimated day of cycle was compared to the true day of cycle. The results of Methods 1, 2 and 3 were compared using the Chi-square test.
3. Results The two-phase logistic curves were fitted to a total of 37 cycles of the seven trotter mares. The calculated curves exhibited excellent correlation with the measured profiles (r ¼ 0:98; Fig. 1). For one cycle, a rise in plasma progesterone at the end of diestrus affected curve fitting; therefore, data for this cycle were excluded from the statistical analysis. 3.1. Effect of day of cycle There was an effect (P < 0:001) of day of cycle on plasma progesterone concentrations. Mean plasma concentration on the day of ovulation was below assay sensitivity (0.18 nmol/l); it increased above the detection limit by D1 (1.32 nmol/l) and continued to rise until D5. Mean plasma progesterone concentrations remained high until D13, then decline, and reached values below assay sensitivity by D16.
P. Nagy et al. / Theriogenology 61 (2004) 203–214
Progesterone (nmol/L)
25
SO
DO
SO
DO
SO
207
ADO
20 15 10 5 0 21/6
6/7
21/7
5/8
20/8
4/9
19/9
4/10
19/10
Date
Fig. 1. Plasma progesterone-profile of a mare throughout the study (calculated curve); (^) indicates measured values and the arrow shows the time of diestrus ovulation; SO: single ovulation; DO: synchronous double ovulation; ADO: asynchronous double ovulation.
3.2. Effect of mare There were differences among mares, in both the upper asymptotic values (A1 and A4) of the calculated curves (P < 0:01) and the AUC (P < 0:01), irrespective of the number of ovulations. None of the other curve parameters were significantly different. Two-way ANOVA of daily progesterone values with the day of cycle and the mare as main effects gave the same results (P < 0:001), without a significant day of cycle by mare interaction. 3.3. Effect of the number of ovulations Twenty six (60.5%) cycles had a single ovulation, 15 (35%) double and 2 (5%) triple ovulations. Regarding the 17 multiple ovulations, 12 (28%) were synchronous; the remaining 5 (12%) that were asynchronous were detected on D2, D4 (two cases), D6 and D7, respectively. Multiple ovulations occurred in each month of the study but there was a trend towards a higher incidence in July and August. Means and standard deviations of estimated parameters of the progesterone profiles according to single and multiple synchronous or asynchronous ovulations are shown in Table 1. The number of ovulations (single versus multiple) had an effect on the upper asymptotic values (A1, A4; P < 0:001), on the locations of the inflection points (A3, A6; P < 0:05) of the calculated curves, and on the AUC (P < 0:01). The ratio of AUC of multiple ovulation to AUC of single ovulation was 1.36. Analysis of variance of daily progesterone concentrations to D12 revealed not only the effect of the number of ovulations (P < 0:001), but interactions of day of cycle by number of ovulations and mare by number of ovulations (P < 0:001 for each). In the latter part of the interovulatory interval (after D12), plasma progesterone concentrations were similar for cycles with single versus multiple ovulations. In the case of synchronous multiple ovulations, there was a trend toward higher mean plasma progesterone concentrations as early as 2 day after ovulation, but the difference became significant only after D4 (and remained high until D10). Mean plasma progesterone profiles and logistic curves for
208
Ovulation Single (n ¼ 22) Synchronous double (n ¼ 10) Asynchronous double (n ¼ 5)
A1
A2 a
12.21 3.67 15.92 5.85b 17.37 4.44b
A3 c
1.53 0.49 1.44 0.33c 0.77 0.38d
A4 a
2.63 0.54 2.62 0.27a 4.74 1.60b
A5 a
12.34 3.83 15.98 5.73b 17.24 4.31b
A1–A6 are estimated parameters defined in Eq. (2) and Eq. (3). Within a column, values with different superscripts (a–d) are different (a, b: P < 0:05; c, d: P < 0:01).
3.26 3.81 2.37 1.40 3.05 0.99
A6
AUC c
12.79 1.01 13.03 1.05c 14.68 0.51d
121.5 36.9a 162.1 58.5b 171.0 46.7b
P. Nagy et al. / Theriogenology 61 (2004) 203–214
Table 1 Mean S:D: curve parameters and areas under the curve (AUC) in cycles with single, multiple synchronous, or multiple asynchronous ovulations
Progesterone (nmol/L)
P. Nagy et al. / Theriogenology 61 (2004) 203–214
209
20 15 10 5 0 0
5
10
15
20
Day of cycle (0 = day of ovulation) Fig. 2. Mean plasma progesterone profiles of single and synchronous multiple ovulations (calculates curves); (&) and ( ) indicated mean values for single and multiple ovulation cycles, respectively; the arrow shows the period of significant difference.
Progesterone (nmol/L)
single and synchronous multiple ovulation cycles are shown in Fig. 2. The interovulatory interval was not significantly different for single versus multiple ovulations (17.82 day versus 19.13 day). In the case of asynchronous multiple ovulations, the shape of the estimated profiles was different (A2, P < 0:01; A3, P < 0:05; A6, P < 0:01). In these cycles, progesterone increased slowly after the first ovulation, but the increase was more pronounced after the second ovulation and progesterone reached the same plateau as in the case of synchronous multiple ovulations (no difference in A1 and AUC; Fig. 3). Daily plasma progesterone concentration was significantly higher in the increasing phase (D1–D4, P < 0:01) for synchronous multiple ovulations, but this difference disappeared after D5. In comparison to single ovulations, asynchronous multiple ovulations were different for both curve parameters (A2, P < 0:01; A3, P < 0:05) and plasma progesterone increase during the first 5 day after ovulation (P < 0:05). The slope of the curve was lower after asynchronous multiple ovulations than after a single ovulation.
20 15 10 5 0 0
5 10 15 Day of cycle (0 = day of ovulation)
20
Fig. 3. Mean plasma progesterone profiles of synchronous and asynchronous multiple ovulations (calculated curves); (&) and ( ) indicated mean values for synchronous and asynchronous multiple ovulation cycles, respectively; the arrow shows the period of significant difference.
210
P. Nagy et al. / Theriogenology 61 (2004) 203–214
3.4. Effect of corpus luteum morphology Twenty six (69.2%) single ovulations resulted in an uniformly echogenic corpus luteum, whereas in eight cases (30.8%), there was a non-echogenic central area that remained within the corpus luteum throughout most of its lifespan. The frequency distribution of the two types of luteal morphology was similar among mares. There was no effect of the morphology on any parameters of the calculated profiles or on the AUC. Analysis of variance of plasma progesterone in the first 10 days of the cycle failed to detect any significant effect of morphology or day of cycle by morphology interaction. However, there was a difference in mean daily progesterone on D12–D14 (12.08 nmol/l versus 7.30 nmol/l, 9.03 nmol/l versus 2.41 nmol/l and 5.07 nmol/l versus 0.33 nmol/l for uniform and fluidfilled corpora lutea, respectively; P < 0:05). Although this suggested that luteolysis of fluid-filled corpora lutea occurred earlier, the length of the interovulatory interval was not significantly different (18.14 and 17.25 day for uniform and fluid-filled corpora lutea, respectively). 3.5. Effect of season There were 26 single ovulations; 16 in the summer and 10 in the autumn. No effect of season was detected, either on the parameters of the calculated profiles or on AUC. However, when day of cycle, mare and season were included together as main effects in the model, there was a negative effect of season on daily progesterone (P < 0:001), with no day of cycle by season or mare by season interactions. 3.6. Description of progesterone increase after ovulation
Progesterone (relative values) (P4/P4 (D5-D10))
Both measured and relative values were used to describe the progesterone increase after ovulation. Progesterone increase was linear between D1 and D4 (Fig. 4). Linear regression
1.2 1 0.8 0.6 0.4 0.2 0 -10
-5
0
5
10
15
Day of cycle (0 = day of ovulation) Fig. 4. Increase in plasma progesterone after ovulation; daily concentrations are expressed as relative values (ratio of the maximum concentration); (&) indicates mean relative value; regression line: Y ¼ 0:22, X 0:06.
P. Nagy et al. / Theriogenology 61 (2004) 203–214
211
Table 2 Accuracy (%) of Methods 1, 2 and 3 for determining the true day of cycle with different sampling protocols Method
1 2 3
Blood sampling twice weekly
Blood sampling thrice weekly
Exact match
1 day
Exact match
1 day
47.4 53.3 46.6
97.7 96.7 91.4
61.8 58.8 52.9
88.2 88.2 88.2
between the relative value and the day of cycle from D0 to D4 resulted in the following equation: Y ¼ 0:22t 0:06
ðr ¼ 0:89Þ
(4)
where t ¼ the day of cycle and Y ¼ the ratio of daily plasma progesterone concentrations to maximum concentrations. Linear regression between the daily measured values and the day of cycle from D0 through D4 resulted in the following equation: Y ¼ 2:7t 0:67
ðr ¼ 0:83Þ
(5)
where t ¼ the day of cycle and Y ¼ the daily plasma progesterone concentrations. 3.6.1. Comparison of methods for calculating the time of ovulation Using these equations and assuming blood sampling twice weekly, the day of cycle was recalculated from daily progesterone values between D0 and D4 using the three different methods. The true day of cycle was compared to the calculated day. Calculated values were around the true day except for two cases with Method 1, (in one case the 20th, in the other case the 12th day of the cycle were calculated, these were the consequence of low progesterone value in the 3rd high sample after ovulation). Chi-square test revealed no significant difference in the distribution of calculated days according to the different methods. 3.6.2. Accuracy of the determination of time of ovulation The calculated days were the true days of the cycle in 46.6–58.8% of the cases. Accuracy was higher when blood samples were taken more frequently (thrice weekly). However, the 1 day accuracy was >88% in all cases, irrespective of the sampling protocol. Accuracy data for Methods 1, 2 and 3 are shown in Table 2.
4. Discussion The fitting of a sum of logistic curves (Eq. (1)) well describes the individual progesterone curve, as previously reported for jennies [14]. The effect of the day of cycle was entirely consistent with numerous preliminary reports [5–12]. Progesterone concentrations increased from D1 to D5 and thereafter reached a plateau until luteolysis. With frequent sampling protocols, the time of the first increase occurred as early as 8–10 [4], 12 [3], 12–24 [5] and 12–18 h [2] after ovulation. The high
212
P. Nagy et al. / Theriogenology 61 (2004) 203–214
variability of the peak (plateau) concentrations among mares with a single ovulation mares was confirmed [11,27]; this may be due to the secretory capacity of the corpus luteum or the rate of progesterone catabolism; the latter could explain higher plasma progesterone concentrations in the first 5 day after ovulation in ponies compared to mares [5]. The incidence (27.9%) of synchronous multiple ovulations did not differ from that reported earlier for this breed [28]. Two corpora lutea secreted more progesterone than one; differences were significant both in the concentrations during the plateau and in the AUC (more than 35%). In the case of synchronous multiple ovulations, progesterone reached a higher plateau at the same time as for a single ovulation. In this case, the two corpora lutea stopped secreting progesterone 2 days before a single ovulation. These results confirm those obtained by some authors [8,19,21,22] but not by others [12,20]. With asynchronous double ovulations, the initial rise in progesterone was followed by an additional increase after the second ovulation. The shape of the curve was different from synchronous multiple ovulations, but the maximum concentration was the same. Since all diestrus ovulations occurred before D8, no prolonged luteal activity was detected. After this second ovulation, plasma progesterone seemed to increase slower than after a single ovulation, but this is probably an artifact of the model. The logistic model or the straight line model do not take into account this type of asynchronous ovulation. A previous report [5] on corpus luteum scanning morphology and progesterone concentrations did not describe any difference in progesterone level between uniform and fluid-filled corpora lutea; the present results confirm this report. Corpora lutea with central non-echogenic areas started to decrease their production rate earlier than corpora lutea with homogenous echotexture. However, there was no significant difference in interovulatory intervals. There are reports [5,18,29] of an influence of season on circulating progesterone concentrations. These authors agree on a progressive seasonal decline in diestrus plasma progesterone to the anovulatory period. Our results seemed to support this hypothesis. In our study, all mares stopped cycling during the following winter. This plasma progesterone decline was perhaps the consequence of the decline in the LH surge. However, recent data indicated that amplitude of the LH surge was not implicated in the maintenance of corpus luteum [30]. Therefore, other explanations are needed to explain this seasonal effect. The plasma progesterone increase during the first 4 days after ovulation was approximately linear (Fig. 4), so it was simple and convenient to use this straight line to predict the time of ovulation. Methods 1 and 2 could not be immediately applied, because they required additional samples to determine maximal concentration. If the time of ovulation was calculated using these methods, the results would often be available too late for many clinical applications. When the luteal phase is short, Method 1 is inappropriate because the second sample used for estimating the maximum concentration may be too low. In Method 3, measured values are used without further transformation. Owing to the greater variation, the correlation coefficient of Eq. (5), was lower (0.83 versus 0.89) than in Eq. (4), which used relative values. However, there was no difference between methods in calculating the day of cycle (compared to the true day of the cycle); all models provided 1 day accuracy >88% of the time. When errors occurred, the actual day of cycle was usually underestimated; furthermore, the same postovulatory periods tended to cause problems in all methods. One typical problem was related to asynchronous multiple (diestrus) ovulations.
P. Nagy et al. / Theriogenology 61 (2004) 203–214
213
In these cycles, there was a slower increase in the first 5 days and the maximum concentrations were higher than in single-ovulation cycles. The other common source of mistakes was the difference in progesterone plateau among mares. If a mare typically had a low maximum concentration, then the day of cycle calculated using Method 3 would be much less than the actual day. Any of the methods mentioned above could be used for the retrospective determination of the time of ovulation, (with some limitations) when blood samples are collected twice or thrice weekly. However, for cyclicity studies, Method 3 seems to be easier. Additional studies on the management of recipient mares would be required to test the effectiveness of these methods under field conditions. In conclusion, the factors affecting plasma progesterone concentrations in cyclic mares were the day of cycle, the number of ovulations, the time of the second ovulation and season. Luteal morphology had no significant effect. The retrospective estimation of time of ovulation based on a postovulatory progesterone increase was possible; the majority of errors were attributed to the great variability in plasma progesterone concentrations during the first few days after ovulation.
Acknowledgements The authors wish to thank the Somogysa´ rd Police Stud, Dr. Frigyes Janza and Dr. Ja´ nos Seregi for all their efforts in providing the mares and to Ms. Erzse´ bet Kiss, Ms. Ibolya Simon and Mr. Be´ la Budai for the progesterone assays, Mr E. Palmer for his encouragements, Ms. A.M Wall (INRA) for the English correction. This study was partly funded by Grants of the National Research Found (OTKA, project numbers: 16473, F 013077).
References [1] Belin F, Goudet G, Duchamp G, Ge´ rard N. Intrafollicular concentrations of steroids and steroidogenic enzymes in relation to follicular development in the mare. Biol Reprod 2000;62:1335–43. [2] Koskinen E, Lindeberg H, Kuntsi H, Katila T. Milk and serum progesterone levels in mares after ovulation. Acta Vet Scand 1990;31:441–4. [3] Pipkin JL, Forrest DW, Kraemer DC, Wilson JM. Plasma steroid concentrations after conception in mares. J Reprod Fertil 1987;35:711–4. [4] Plotka ED, Foley CW, Witherspoon DM, Schmoller GC, Goetsch DD. Periovulatory changes in peripheral plasma progesterone and oestrogen concentrations in the mare. Am J Vet Res 1975;36:1359–62. [5] Townson DH, Pierson RA, Ginther OJ. Characterization of plasma progesterone concentrations for two distinct luteal morphologies in mares. Theriogenology 1989;32:197–204. [6] Allen WE, Hadley JC. Blood progesterone concentrations in pregnant and non-pregnant mares. Equine Vet J 1974;6:87–93. [7] Noden PA, Oxender WD, Hafs HD. The cycle of oestrus, ovulation and plasma levels of hormones in the mare. J Reprod Fertil 1975;23:189–92. [8] Palmer E, Jousset B. Urinary oestrogen and plasma progesterone levels in non-pregnant mares. J Reprod Fertil 1975;23:213–21. [9] Plotka ED, Witherspoon DM, Foley CW. Luteal function in the mare as reflected by progesterone concentrations in peripheral blood plasma. Am J Vet Res 1972;33:917–20. [10] Sharp DC, Black DL. Changes in peripheral plasma progesterone throughout the oestrus cycle of the pony mare. J Reprod Fertil 1973;33:535–8.
214
P. Nagy et al. / Theriogenology 61 (2004) 203–214
[11] Smith ID, Bassett JM, Williams T. Progesterone concentrations in the peripheral plasma of the mare during the oestrus cycle. J Endocrinol 1970;47:523–4. [12] Stabenfeldt GH, Hughes JP, Evans JW. Ovarian activity during the oestrus cycle of the mare. Endocrinology 1972;90:1379–84. [13] Hinrichs K, Sertich PL, Solorzano NM, Caldwell LA. Use of an immediate, qualitative progesterone assay for determination of day of ovulation in an equine embryo transfer program. Theriogenology 1988;29: 1123–30. [14] Meira C, Ferreira JCP, Papa FO, Tornero MTT, Bicudo SD. Study of the oestrus cycle in donkeys (Equus asinus) using scan and plasma progesterone concentrations. Biol Reprod 1995;(Mono 1):403–10. [15] Yenikoye A, Mariana JC, Ley JP, Jolivet E, Terqui M, Lemon-Resplandy M. Mode`le mathe´ matique de l’e´ volution de progeste´ rone chez la vache: application et mise en e´vidence de diffe´ rence entre races. Reprod Nutr Dev 1981;21:561–75. [16] Cockrill T, Allen WE. Diurnal variation of plasma progestagen concentrations in pony mares. Vet Rec 1978;102:503. [17] Evans JW. Biorhythms in plasma progesterone concentrations and absence of any correlation to LH biorhythms during different stages of the equine oestrus cycle. J Reprod Fertil 1991;(Suppl 44):684–5. [18] Oko´ lski A, Bielan˜ ski AB. Seasonal fluctuations of the plasma progesterone levels in mares with similar ovarian morphology. Int Cong Anim Reprod Artif Insem Madrid 1980;4:13–5. [19] Henry M, Coryn M, Vandeplassche M. Multiple ovulation in the mare. Zbl Vet Med A 1982;29:170–84. [20] Roser JF, Evans JW. Luteal luteinizing hormone receptors during the postovulatory period in the mare. Biol Reprod 1983;29:499–510. [21] Squires EL, McKinnon AO, Carnevale EM, Morris R, Nett TM. Reproductive characteristics of spontaneous single and double ovulating mares and superovulated mares. J Reprod Fertil 1987;35: 399–403. [22] Urwin VE, Allen WR. Follicle stimulating hormone, luteinising hormone and progesterone concentrations in the blood of Thoroughbred mares exibiting single and twin ovulations. Equine Vet J 1983;15:325–9. [23] Csernus V. Antibodies of high affinity and specificity for radioimmunological determination of progesterone, testosterone, estradiol-17b and cortisol. In: Go¨ ro¨ g S, editor. Advances in steroid analysis I. Budapest: Academy Press; 1982. p. 171–7. [24] Huszenicza Gy, Ba´ ba A, Nagy P, Juha´ sz J, Kulcsa´ r M, Abava´ ry MK, et al. Characteristics of the ovarian function in non-pregnant breeding mares at the beginning of the official breeding season. Hung Vet J 1993;48:587–92. [25] SAS User’s Guide, Statistics, Version 6 edition. Cary, NC: SAS Institute Inc.; 1986. [26] Matthews JNS, Altman DG, Campbell MJ, Royston P. Analysis of serial measurements in medical research. Br Med J 1990;300:230–5. [27] Squires EL, Wentworth BC, Ginther OJ. Progesterone concentration in blood of mares during the oestrus cycle, pregnancy and after hysterectomy. J Anim Sci 1974;39:759–67. [28] Ginther OJ. Reproductive biology of the mare. 2nd ed. Cross Plains, WI: Equiservices; 1992. p. 218–40. [29] King SS, Nequin LG, Drake S, Hebner TS, Roser JF, Evans JW. Progesterone levels correlate with impending anoestrus in the mare. J Equine Vet Sci 1988;8:109–11. [30] Briant C, Ottogali M, Morel M, Guillaume D. Use of GnRH antagonist, antarelix, associated or not with hCG, to control ovulation in cyclic pony mares. Domest Anim Endocrinol 2003;53:1–18.