Multifactorial analysis of the follicular environment is predictive of oocyte morphology in cattle.

Numerous attempts have been recently made in the search for a reliable, fast and noninvasive assay for selection of oocytes suitable for in vitro embryo production. Potential markers have been described in the follicle such as follicular fluid (FF) or cumulus cells (CCs). However, the reported findings are contradictory, which may reflect the complexity of metabolism of the ovarian follicle. In the present experiment, a data set from individual follicles of known diameter was obtained: cumulus-oocyte complex (COC) morphology, fatty acid composition and glucose concentration in FF as well as apoptotic index in CCs. The obtained data was statistically analyzed either separately (univariate analysis) or simultaneously (multivariate analysis) to examine its predictive value in morphology assessment of bovine COCs. Although the univariate analysis yielded a complex relation system of the selected parameters, no clear outcome could be established. In multivariate analysis, the concentration of the four fatty acids (C16:0, C16:1, C18:1cis9, C22:5n3) and Δ(9)-desaturase (16) as well as elongase activities were selected as covariates. This allowed prediction of the morphology of a COC with an accuracy of 72%, which is the most interesting finding of the experiment. The present study indicates that the multifactorial model comprising of selected parameters related to the follicle appeared more effective in predicting the morphology of a bovine COC, which may improve the effectiveness of in vitro production systems.

In vitro embryo production dates back to the 1950s/1960s, which is when the first convincing reports on the rabbit, mouse, hamster, rat and humans were published [1]. Since then, in vitro production (IVP) effectiveness has been significantly improved; however, still only 30–40% of in vitro fertilized bovine oocytes reach the blastocyst stage. It is claimed that one of the problems is a deficient strategy for oocyte selection, which, despite significant progress, is still mainly based on morphology evaluation of the cumulus-oocyte complex (COC). An alternative way seems to be prediction of oocyte quality basing on evaluation of other follicular components – mainly follicular fluid (FF) or cumulus cells (CC), which have an important influence on oocyte developmental competence [2]. The first experiments relating cumulus cells analysis to human oocyte quality showed that COCs containing less than 50% of luteinized cumulus cells and no vacuoles yielded the highest fertilization outcome [3]. Nowadays, mainly follicular fluid is described as a valuable source of data on oocyte quality. FF is a product of both transfer of blood plasma constituents and of secretory activity of granulosa and thecal cells [4]. It protects the oocyte against proteolysis and extrusion and acts as a buffer against negative influences of blood [5]. Since it has an essential impact on the oocyte, follicular fluid has emerged as one of the most promising sources of data predicting oocyte quality, which in turn directly affects the blastocyst yield [6]. Some experiments have also indicated cumulus cells as elements of the follicular environment, which may indirectly reflect oocyte quality.

In discussions of follicular fluid, the emphasis is placed on some metabolites, e.g., fatty acids or amino acids. According to recent human and bovine studies, the concentrations of particular fatty acids (FAs, e.g., stearic, palmitic, oleic, linoleic acids) in follicular fluid or in IVM medium were related to COC morphology, meiotic stage as well as cleavage and blastocysts yield [7,8,9,10,11,12,13]. On the other hand, Sinclair et al. [8] observed correlation between the concentrations of palmitic and stearic fatty acids (FAs) and morphology of the bovine COC; however, no relation with blastocyst rate has been found. A recent manuscript of Matoba et al. [14] described relations among FF composition (e.g., fatty acids, amino acids, steroids), mRNA relative abundance in cumulus, granulose and theca cells and blastocyst rate. It emphasized the metabolomic profile of follicular tissues as a source of information on oocyte quality. Other approaches focused on cumulus cell analysis revealed associations between CC apoptosis and COC morphology [15], meiotic stage [16] or cleavage rate [17]. On the other hand, Yuan et al. [18] showed no relationship between COC morphology and the apoptotic index (AI) in bovine cumulus cells. We have recently shown a lack of correlation between incidence of apoptosis in the CC and two oocyte-related parameters (COC morphology, meiotic stage) [19]. The studies referred to above present inconsistent data, which may reflect the complexity of the follicular metabolism, since interactions between the oocyte and somatic cells have an undoubted impact on oocyte growth [20]. Therefore, investigation based on individual parameters appears insufficient to predict oocyte quality, which requires a more complex approach.

In this manuscript, we proposed a multivariate statistical model based on data describing follicular fluid components (FF, CC) of individual bovine follicles. The experiment aimed at describing a set of parameters characterizing the individual follicle: fatty acid composition and glucose content in FF as well as incidence of apoptosis in CCs. The obtained set of data was statistically analyzed either separately (univariate analysis) or simultaneously (multivariate analysis) to examine its predictive value in morphology assessment of bovine COCs. Although the univariate analysis yielded a complex relation system between analyzed parameters, no clear outcome could be established. More essential information was provided by the multivariate analysis. The concentrations of four fatty acids (C16:0, C16:1, C18:1cis9, C22:5n3) and Δ9-desaturase (16) activity were selected as covariates for the prediction of COC morphology, whereas the remaining parameters were excluded. This allowed prediction of the morphology of a COC with a probability of 0.72, which is the most interesting finding of this experiment. The present study indicates that the multifactorial model comprising selected parameters related to follicular components appeared more effective in predicting the morphology of the bovine COC.

Material and Methods

Sample collection

Cumulus oocyte complexes were recovered from bovine ovaries of slaughterhouse origin. The diameter of each follicle was measured, and three size categories were defined: small (<6 mm, 70 follicles), medium (6–8 mm, 60 follicles) and large (>8 mm, 56 follicles). Each follicle was aspirated separately with 1 ml syringe. Individually aspirated FF was transferred to a Petri dish. The morphological grade of each COC was assessed according to a modified criteria described by Stojkovic et al. [21]: grade 1–homogenous ooplasm, complete, compact and multilayered cumulus cell mass; grade 2–homogeneous ooplasm with some irregular pigmentation, >5 layers of compact cumulus cells; grade 3–heterogeneous, partially vacuolated ooplasm, 3–5 layers of cumulus cells; grade 4–heterogeneous, pigmented ooplasm, expanded cumulus cell mass. Each sample analyzed in this experiment comprised of 3 follicular components – CCs, oocytes and FF – and represented an individual follicle. Analyses of apoptosis in CCs, glucose concentration in FF and fatty acid concentration were performed.

Fatty acids composition in follicular fluid

The fatty acid composition of the follicular fluid was analyzed using gas chromatography according to the procedure described by Cieslak et al. [22] and modified by Warzych et al. [23]. The activities of Δ9-desaturase (16), Δ9-desaturase (18) and elongase were calculated according to Bender et al. [24].

Cumulus cells apoptosis

A terminal TUNEL assay kit was used to detect apoptosis in cumulus cells (DeadEndTM Fluorometric TUNEL System, Promega Biosciences, Madison, WI, USA) with minor modifications described by Warzych et al. [19]. The apoptotic index was calculated as a percentage of FITC-positive cells within all DAPI-positive cells.

Glucose concentration

Glucose concentration was measured in 2 µl of aspirated FF with an Accu-Chek glucometer (Roche) according to the manufacturer’s recommendations. Statistical analysis included the physiological concentration of glucose in bovine FF, which ranges from 1.4 to 5 mM [25].

Statistical analysis

Both univariate analysis, which evaluates the effect of one factor at a time, and multivariate analysis, in which several factors are investigated simultaneously, were used to analyze the data. The effect of follicle size and COC morphology grade on the concentration of fatty acids was tested by Kruskal-Wallis rank sum test, followed by nonparametric Tukey-type multiple comparisons using the Miller (1981) test. Spearman’s rank correlation coefficients were used to relate apoptotic index and fatty acids concentration in follicular fluid. Logistic regression in SAS PROC LOGISTIC with backward variable selection was used to find a set of variables predictive of oocyte morphology. All calculations were done using the R package (R Development Core Team, 2009) and SAS (SAS Institute, 2002–2003).

Results

Scope of the study

The apoptotic index was evaluated in 174 samples of CCs. A CC sample was excluded from the analysis when fewer than 50 cells could be evaluated. The average AI for the analyzed CC samples was 6.71 and varied from 0 to 97.62.

FA composition was investigated in 132 samples of FF. Samples originating from small follicles could not be analyzed due to insufficient FF volume (<100 µl). The average concentration of total FA was 460.29 µg/ml. The most abundant FAs were C18:2n6 (121.97 µg/ml), C16:0 (91.73 µg/ml), C18:1cis9 (82.45 µg/ml) and C18:0 (64.92 µg/ml).

Glucose concentration was analyzed in 101 FF samples, and the mean value was 3.51 mM (min. <0.6 mM, max. 6.55 mM). Samples originating from small follicles could not be analyzed. In the case of three FF samples, the glucose concentration was lower than the detection threshold of the glucometer (0.6 mM).

Although the collected set of data was not complete for all samples (due to, e.g., insufficient FF volume), the multivariate model was employed only for samples with a complete set of parameters.

Univariate analysis

Follicle size in relation to other parameters: The following follicular components of the ovarian follicles were analyzed with regard to follicle size: follicular fluid (fatty acid composition and glucose content) and cumulus cells (apoptotic index). Significant differences were observed only for fatty acid composition. FF originating from medium follicles contained more of the selected FAs (C18.1cis9, ΣC18.1cis, C18.2n6 or C22.5n3). This, in consequence, affected the level of certain groups of FAs such as UFA, PUFA, n6 or n3. The highest apoptotic index was noted for cumulus cells originating from small follicles, however, it was not statistically significant due to a big variation within each group.

The glucose concentration was similar in medium and large follicles (Table 1). The rate of FF samples with a physiological concentration of glucose (from 1.4 to 5 mM) was similar in medium (30/46, 65%) and large follicles (36/50, 72%).

Table 1.

Follicle size in relation to apoptotic index in cumulus cells, glucose concentration and fatty acid composition of follicular fluid

	Small follicles	Medium follicles	Large follicles	P
Apoptotic index in CC
	8 ± 1.84	6.47 ± 1.8	5.27 ± 1.49	NS
Glucose concentration (mM) in FF
	n.a.	3.15 ± 0.22	3.84 ± 0.21	NS
Fatty acid composition (µg/ml) of FF
TOTAL FA	n.a.	480.87 ± 18.14	439.71 ± 18.93	NS
C14:0	n.a.	15.3 ± 1.07	17.4 ± 1.22	NS
C15:0	n.a.	3.57 ± 0.18	3.48 ± 0.16	NS
C16:0	n.a.	90.47 ± 4.22	92.99 ± 4.73	NS
C16:1	n.a.	7.58 ± 0.43	7.19 ± 0.45	NS
C18:0	n.a.	61.62 ± 2.53	68.23 ± 3.06	NS
C18:1cis9	n.a.	86.46 ± 4.08	78.43 ± 4.2	<0.05
ΣC18:1trans	n.a.	9.83 ± 0.76	8.54 ± 1.18	NS
ΣC18:1cis	n.a.	94.61 ± 4.3	84.87 ± 4.44	<0.01
C18:2n6	n.a.	137.95 ± 6.5	105.99 ± 4.51	<0.01
C18:3n3	n.a.	19.67 ± 1.09	15.31 ± 0.74	NS
C20:4n6	n.a.	2.54 ± 0.25	2.19 ± 0.15	NS
C22:5n3	n.a.	1.9 ± 0.21	1.7 ± 0.3	<0.05
SFA	n.a.	193.79 ± 8.14	193.79 ± 0.69	NS
UFA	n.a.	287.08 ± 11.53	237.81 ± 10.06	<0.01
MUFA	n.a.	117.44 ± 4.89	108 ± 5.27	NS
PUFA	n.a.	169.63 ± 7.63	129.81 ± 5.46	<0.01
n6	n.a.	141.69 ± 6.61	109.18 ± 4.6	<0.01
n3	n.a.	22.76 ± 1.19	17.83 ± 0.97	<0.01
Δ9-des. (16) activity	n.a.	7.9 ± 0.33	7.4 ± 0.33	NS
Δ9-des. (18) activity	n.a.	57.74 ± 0.89	53.53 ± 0.90	<0.01
Elongase activity	n.a.	60.42 ± 0.59	59.68 ± 0.59	NS

Values are presented as means ± SEM. n.a. – not analyzed. NS – not significant. SFA=saturated fatty acid (C14:0, C15:0, C16:0, C18:0, C20:0, C21:0, C22:0). UFA=unsaturated fatty acid (C16:1, C17:1, C18:1 trans-9, C18:1 trans-10, C18:1 trans-11, C18:1 trans-12, C18:1 trans-13–14, C18:1 cis-9, C18:1 cis-11, C18:1 cis-12, C20:1 cis-9, C18:2n-6, C18:3n-3, C20:3n-3, C20:4n-6, C20:5n-3). MUFA=monounsaturated fatty acid (C16:1, C17:1, C18:1 trans-9, C18:1 trans-10, C18:1 trans-11, C18:1 trans-12, C18:1 trans-13–14, C18:1 cis-9, C18:1 cis-11, C18:1 cis-12, C20:1 cis-9). PUFA= polyunsaturated fatty acid (C18:2n-6, C18:3n-3, C20:3n-3, C20:4n-6, C20:5n-3). C18:1trans=C18:1 trans-9, C18:1 trans-10, C18:1 trans-11, C18:1 trans-12, C18:1 trans-13–14. C18:1cis=C18:1 cis-9, C18:1 cis-11, C18:1 cis-12, C20:1 cis-9.

COC morphological grade in relation to other parameters: Although the highest AI was observed in the grade 4 COCs, the difference was not significant. FF accompanying the grade 3 COCs had a significantly higher glucose concentration when compared with follicles having grade 4 COCs (Table 2). The rate of FF samples with a physiological concentration of glucose (from 1.4 to 5 mM) was not affected by the COC morphology [grade 1 COCs (9/14, 64%), grade 2 COCs (20/25, 80%), grade 3 COCs (24/40, 60%) and grade 4 COCs (14/18, 78%)]. Also, the FA composition in FF was not related to COC morphology (Table 2).

Table 2.

COC morphology in relation to apoptotic index in cumulus cells, glucose concentration and fatty acid concentrations in follicular fluid

	COCs morphological grade
	1st	2nd	3rd	4th	P
Apoptotic index in CC
	6 ± 1.58	4.35 ± 1.31	5.44 ± 1.82	13.24 ± 4.47	NS
Glucose concentration (mM) in FF
	3.44 ± 0.42	3.2 ± 0.24	3.99a ± 0.26	2.9b ± 0.29	<0.05
Fatty acid composition (µg/ml) of FF
TOTAL FA	486.01 ± 30.86	462.68 ± 36.17	451.85 ± 23.67	469.5 ± 29.03	NS
C14:0	17.22 ± 2.03	19.83 ± 2.13	15.77 ± 1.46	16.54 ± 1.74	NS
C15:0	3.67 ± 0.27	3.85 ± 0.31	3.51 ± 0.23	3.46 ± 0.26	NS
C16:0	94.71 ± 6.79	93.11 ± 7.0	91.04 ± 6.21	96.04 ± 6.83	NS
C16:1	8.61 ± 0.83	7.89 ± 1.15	7.24 ± 0.51	6.66 ± 0.48	NS
C18:0	71.3 ± 4.8	64.43 ± 4.71	63.93 ± 3.53	66.66 ± 4.45	NS
C18:1cis9	86.61 ± 6.34	84.46 ± 7.53	81.39 ± 4.88	77.58 ± 5.63	NS
ΣC18:1trans	7.51 ± 0.91	7.7 ± 1.19	9.61 ± 1.12	11.61 ± 2.19	NS
ΣC18:1cis	93.98 ± 6.74	91.18 ± 8.14	88.77 ± 5.2	85.03 ± 6.07	NS
C18:2n6	125.78 ± 9.67	120.71 ± 13.09	118.52 ± 6.96	126.79 ± 9.11	NS
C18:3n3	18.51 ± 1.58	17.87 ± 1.9	17.7 ± 1.27	17.62 ± 1.47	NS
C20:4n6	2.95 ± 0.37	2.63 ± 0.55	2.31 ± 0.25	1.93 ± 0.24	NS
C22:5n3	2.44 ± 0.65	1.56 ± 0.48	1.42 ± 0.21	1.94 ± 0.33	NS
SFA	211.88 ± 14.7	203.06 ± 14.83	193.42 ± 11.42	205.7 ± 14.02	NS
UFA	274.13 ± 18.39	259.62 ± 23.71	258.43 ± 13.66	263.8 ± 16.93	NS
MUFA	117.1 ± 7.99	111.06 ± 9.68	112.37 ± 6.62	110.4 ± 6.92	NS
PUFA	157.03 ± 12.27	148.56 ± 16.08	146.05 ± 8.05	153.4 ± 10.69	NS
n6	129.98 ± 10.01	124.24 ± 13.25	122.03 ± 7.06	129.83 ± 9.27	NS
n3	22.45 ± 2.05	20.5 ± 2.39	20.04 ± 1.38	20.3 ± 1.63	NS
Δ9-des. (16) activity	8.35 ± 0.57	7.5 ± 0.7	7.76 ± 0.47	6.77 ± 0.35	NS
Δ9-des. (18) activity	54.58 ± 1.26	56.15 ± 2.02	55.62 ± 1.06	53.15 ± 1.97	NS
Elongase activity	60.6 ± 1.01	59.73 ± 0.8	60.23 ± 0.62	58.73 ± 0.96	NS

Values are presented as means ± SEM. NS – not significant. Values with different superscripts differ significantly (P<0.05). SFA=saturated fatty acid (C14:0, C15:0, C16:0, C18:0, C20:0, C21:0, C22:0). UFA=unsaturated fatty acid (C16:1, C17:1, C18:1 trans-9, C18:1 trans-10, C18:1 trans-11, C18:1 trans-12, C18:1 trans-13–14, C18:1 cis-9, C18:1 cis-11, C18:1 cis-12, C20:1 cis-9, C18:2n-6, C18:3n-3, C20:3n-3, C20:4n-6, C20:5n-3). MUFA=monounsaturated fatty acid (C16:1, C17:1, C18:1 trans-9, C18:1 trans-10, C18:1 trans-11, C18:1 trans-12, C18:1 trans-13–14, C18:1 cis-9, C18:1 cis-11, C18:1 cis-12, C20:1 cis-9). PUFA=polyunsaturated fatty acid (C18:2n-6, C18:3n-3, C20:3n-3, C20:4n-6, C20:5n-3). C18:1trans=C18:1 trans-9, C18:1 trans-10, C18:1 trans-11, C18:1 trans-12, C18:1 trans-13–14. C18:1cis=C18:1 cis-9, C18:1 cis-11, C18:1 cis-12, C20:1 cis-9.

Fatty acid composition of FF in relation to apoptotic index in CC: Statistical analysis showed negative correlation between the apoptotic index in CCs and the concentrations of some FAs in FF (Fig. 1). It concerned the following fatty acids or FA groups: C15:0, C16:0, C16:1, C18:0, ΣC18.1cis, C18:2n6, C18:3n3 or total FA, SFA, UFA, MUFA, PUFA and n6.

Fig. 1.

Spearman’s rank correlations between apoptotic index in the CC and fatty acids concentrations in follicular fluid. Correlations marked with black bars are significantly different from 0 (P<0.05).

Fatty acid composition of FF in relation to glucose: The glucose concentration was negatively correlated with the concentrations of the following FAs: C16:0, C16:1, C18:3n3 and n3 (Fig. 2).

Fig. 2.

Correlations between glucose (mM) and fatty acid (µg/ml) concentrations in follicular fluid (striped bars – P<0.05, black bars – P<0.01).

Additional analysis showed that the concentration of C18:0 FA was significantly lower in FF from follicles with a physiological glucose concentration of 1.4 to 5 mM when compared with FF samples having a glucose concentration outside that range (63.7 vs. 66.0 µg/ml respectively, P<0.05).

Prediction of COC morphology based on the multivariate model

In the present experiment, we decided to apply a statistical multivariate model, which was based on parameters characterizing FF as well as CCs from the same follicle.

Using logistic regression with backward variable selection, the model with variables included in Table 3 was chosen to describe the probability of oocytes falling into one of two selected categories: 1) low-quality oocytes (morphology grade 3 or 4) and 2) high-quality oocytes (morphology grade 1 or 2).

Table 3.

Parameter estimates from the multivariate logistic regression model with backwards variable selection

Parameter	Estimate	SE	Chi-square	Odds ratio	95% CI Lower	95% CI Upper	P
Intercept	–33.8553	11.6608	8.4294				0.0037
∆9 –des. (16) activity	0.7378	0.3105	5.646	2.091	1.138	3.843	0.0175
Elongase activity	0.3966	0.1524	6.773	1.487	1.103	2.004	0.0093
C16:0	0.1278	0.0352	13.1843	1.136	1.061	1.218	0.0003
C16:1	–0.7883	0.2959	7.0965	0.455	0.255	0.812	0.0077
C18:1cis9	–0.0628	0.0241	6.78	0.939	0.896	0.985	0.0092
C22:5n3	–0.3155	0.1397	5.097	0.729	0.555	0.959	0.024

Estimate – the log odds change per unit of change in the explanatory variable, its standard error (SE) and Chi-square statistics were applied for significance testing. Odds ratio – the ratio of probabilities of falling into the low-quality oocytes category (vs. high quality) after increasing a given enzyme or fatty acid level by one unit. CI – confidence interval. P – Pr>ChiSq.

The P value for the overall model test of all regression coefficients equal to 0 was 0.006 with all individual regression coefficients significant at P=0.05. Seventy-two percent of the observations were concordant with the predicted probabilities. Thus, based on FA composition, it was possible to determine in 72% of the analyzed samples, whether the obtained COC was of proper morphology. Low-quality oocytes were characterized by a combination of high activity of the two enzymes (∆9 -desaturase (16) and elongase), high C16:0 content and lower concentrations of C16:1, C18:1cis9 and C22:5n3. According to supplementary analysis, non of the individual FAs was a good predictor of COC morphology, and it is impossible to set an individual threshold for each FA, since it is a combination of them that creates a predictable outcome (data not shown). Besides, 72% of predictions being correct seems fairly good for a complex trait.

Discussion

A lot of attempts have been made in the search for a reliable, fast and noninvasive assay for selection of oocytes suitable for in vitro embryo production. Potential markers have been found in the follicular components such as follicular fluid or cumulus cells. The reported findings are, however, contradictory, which may reflect the complexity of metabolism of the ovarian follicle. In the present experiment, two statistical models were applied in order to search for the most accurate approach for predicting the morphological grade of bovine oocytes: 1) the univariate model showed relations between particular parameters of the follicular environment and oocyte morphology or follicular diameter, which did not give evident conclusions, and 2) the multivariate model including FF parameters like concentrations of the four FAs (C16:0, C16:1, C18:1cis9, C22:5n3) and the activities of two enzymes (Δ9-desaturase (16) and elongase), enabled prediction of the morphology of an oocyte with an accuarcy of 72%. Therefore, the multivariate model may be a promising tool for selecting good quality COCs and thus improving the efficiency of the IVP protocol in cattle.

The univariate model referred the individual follicular components (FF, CC) to either COC morphology or follicular diameter, which are approved components of oocyte morphology. It revealed that the most promising marker of oocyte morphology is the fatty acids composition of the FF. The following traits were of particular interest: concentrations of C18:1cis9 (oleic acid), ∑C18:1cis, C18:2n6 (linoleic acid), C22:5n3 (EPA – eicosapentaenoic acid), UFA, PUFA, n6 and n3 and the ∆9-desaturase (18) activity. These parameters were present at higher concentrations in medium follicles (the source of better quality COCs) than in large follicles. However, the applied model did not enable the ability to distinguish a parameter that would be correlated with the COCs of the preferred morphology (grades 1 and 2). Glucose content in FF was the only parameter distinguishing follicles with COCs of reduced quality (grades 3 and 4). Therefore, we did not select a parameter with a distinct impact on COC quality. Besides, FF from medium follicles was enriched with respect to two FAs (oleic and linoleic acids) exerting opposite effects on oocyte maturation and embryo development [12], whereas linoleic acid negatively affected the rates of MII oocytes and blastocysts [10]. Hence, simultaneous high concentrations of the two FAs oppositely affected oocyte morphology in medium follicles, which was confusing.

Another aspect of the relations between COC morphology and FA composition was that the glucose concentration or CC apoptosis could be affected by gap junctions signalling. However, only limited data is available in the literature. In human white blood cells, PUFAs inhibit gap junctional communication [26], whereas oleic acid inhibits gap junctions permeability and increases glucose uptake in cultured rat astrocytes [27]. Glucose is transferred to the growing oocyte through gap junctions [28]. Also, it has been suggested that cell death signals are transmitted through gap junctions in many cell types [29]. Thus, it may be anticipated that the follicular factors analyzed in our experiment may influence COC morphology via gap junctional signalling pathways modifications.

Since investigating the follicular environment in relation to oocyte morphology may be deceptive when a single parameter is considered, we made an attempt to create a complex model, which included several parameters characterizing follicular complements. The principles of such approach were described by Aardema et al. [12], who observed interesting interactions among three fatty acids – oleic acid, palmitic acid and stearic acid. Oleic acid showed the ability to adverse the negative effects of palmitic and stearic acids in bovine FF. In similar analyses previously performed on human oocytes, multivariate analysis of follicular components was shown to be a promising predictor of pregnancy [30]. Such approach was also suggested by Arya et al. [31], who described the concentrations of selected biomolecules from follicular fluid in a multivariate model as a reliable predictor of oocyte quality.

In the present experiment, the multivariate logistic regression with backward variable selection allowed for the accurate (72% correct) prediction of the oocyte morphology. The initial analysis included 3 components: glucose and FA concentrations in FF as well as apoptotic index in cumulus cells. A significant contribution to the prediction model was observed for the following parameters: the activities of the two enzymes [∆9 -desaturase (16) and elongase] and the concentrations of four FAs (C16:0, C16:1, C18:1cis9, C22:5n3). The selected fatty acids have been previously described as FF components affecting oocytes and embryos. Palmitic acid (C16:0) exerted a negative effect on oocyte quality. Also, C16:0 supplemented into IVM medium negatively affected the postfertilization development [12, 13]. C16:0 was also described as a potential marker of reduced quality of immature bovine COCs [8, 31] and reduced blastocyst rate [14]. Another FA included into the multivariate model (oleic acid – C18:1cis9) exerted a positive effect on lipid accumulation, oocyte maturation and embryo development in cattle. Also, it showed an ability to adverse the negative effects of palmitic and stearic acids on oocyte maturation [12]. Although palmitoleic acid (C16:1) has not yet gained special attention as FF component, it was shown to inhibit the proapoptotic effect of saturated fatty acids in the rat pancreatic β-cells [32], thus, this particular FA seems to be worthy of attention. Eicosapentaenoic acid (EPA, C22:5n3) was the last FA included into the multivariate model. This FA has not yet been characterized with regard to follicle metabolism; however, as a precursor of prostanoids and inhibitor of the PGF2α secretion [33], it may play a significant role in reproduction.

Energy source is another important issue when discussing the follicular environment. Accumulation of fatty acids is an integral part of energy storage in oocytes, whereas glucose may be an alternative source of energy. Although the oocyte has a limited capacity to utilize glucose [34], the accompanying cumulus cells metabolize glucose and convert it into the pyruvate/lactate [25]. Although the glucose concentration was correlated with some of the analyzed traits (e.g. COC morphological grade, concentration of particular FA) in the present study, the results were contradictory. Both the highest and the lowest concentrations were observed in follicles containing COCs of reduced quality (grade 3 and 4). Also, negative correlation was observed with the concentrations of C16.0 FA (potential marker of reduced quality oocytes) and n3 FAs which have beneficial effects on COCs. Thus, we decided to distinguish two groups of FF samples – those containing a physiological concentration of glucose (1.4 to 5 mM) [25] and those that did not contain a physiological concentration of glucose. It appeared that neither COC morphology nor follicle diameter was related to physiological concentration of glucose. A similar observation was reported by Matoba et al. [14], who excluded glucose from the set of candidate markers of embryo quality in cattle.

Besides the follicular fluid, cumulus cells provide information that can be also used for oocyte assessment. Although a COC accompanied by a compact cumulus cell mass is considered healthy, early signs of atresia were correlated with higher quality for the oocyte [35]. An interesting result of the present study is the lack of correlations between AI in CC and COC morphology or follicle diameter, which shows the low utility of the data concerning apoptosis in oocyte quality prediction. This confirms the published evidence, which shows that the correlation between the extent of apoptosis in the CC and the quality of an oocyte is not consistent. Human studies revealed negative correlation between CC apoptosis and meiotic stage of the oocyte [16] as well as fertilization rate [17]. On the other hand, a positive correlation [36] or lack of correlations [37, 38] has also been observed. The data available for cattle is very limited and ambiguous. The results of our study indicate that the apoptotic index of cumulus cells was of limited value for predicting oocyte quality. Although the AI was not related to follicle diameter or COC morphology, a negative correlation with the fatty acid concentration in follicular fluid was observed. It is possible that FAs affect oocyte quality by inhibiting apoptosis in cumulus cells. However, it cannot be excluded that the high variation in apoptotic index significantly affected statistical analysis. A big variation in AI within CCs was previously reported [39]. The authors suggested a theoretical threshold for CC apoptosis that could trigger the process of oocyte degeneration; however, to our knowledge, it has not yet been defined. This may cause problems with interpretation of data on apoptosis in CCs.

In this study, we mainly focused on prediction of the morphology of bovine COCs by applying a multivariate model that included several follicular parameters. However, the experiment provided us with a set of data describing individual follicles; thus, we decided to point out some interesting interactions between FF components (FAs, glucose) and apoptosis in cumulus cells. An interesting, negative correlation between glucose and the majority of FA concentrations in follicular fluid was observed. There were significant differences with respect to the C16:0, C16:1, C18:3n3 and n3 FAs. Recently, Arya et al. [31] underlined energy pathways as possible essential factors affecting the fecundity. Glucose and fatty acids metabolic pathways were discussed in particular. They hypothesized that reduced glucose metabolism provokes other alternative energy pathways, which may utilize, e.g., fatty acids to cover the energy requirement. This suggests possible utilization of the C16:0, C16:1 and C18:3n3 FAs as an alternative energy source for the oocyte. This would explain the fact that C16:0 is the most abundant FA in cattle oocytes [40]. Moreover, interesting negative correlations were observed between the apoptotic index in the CC and the concentrations of the majority of FAs in FF. This was contrary to the experiments of Valholder et al. [41, 42], in which FAs supplementation (palmitic, stearic and oleic acid) during in vitro culture induced apoptosis in granulosa cells and theca cells.

In conclusion, the present experiment allowed us to obtain a complex set of data from individual bovine ovarian follicles. Statistical analysis showed that traditional univariate analysis, which is based on interpretation of a single follicular parameter, is not precise with regard to oocyte morphology estimation. Based on our results, the multivariate model in our study allowed prediction of the morphology of COCs with an accurancy of 72%. It should be noticed that the model included only selected data – the concentrations of the C16:0, C16:1, C18:1cis9 and C22:5n3 FAs and activities of two enzymes, ∆9 –desaturase (16) and elongase. The obtained data suggest that it is still possible to improve the selection of oocytes for in vitro embryo production; however, complex studies on follicular metabolism should be performed.