Review

Metabolic Profile in Dairy Cows With Cystic Ovarian Follicles: A Systematic and Meta-Analysis of Available Evidence from USA, Iran, and Indonesia

Viski Fitri Hendrawan1,2, Epy Muhammad Luqman3*, Rimayanti Rimayanti4, Widjiati Widjiati3, Iwan Sahrial Hamid5, Moh. Anam Al-Arif6, Hani Plumeriastuti7, Tri Wahyu Suprayogi4

¹Doctoral Program of Veterinary Science, Faculty of Veterinary Medicine, Universitas Airlangga, Surabaya, 60115, East Java, Indonesia; ²Division of Veterinary Reproduction, Faculty of Veterinary Medicine, Universitas Brawijaya, Malang, 65161, East Java, Indonesia; ³Division of Veterinary Science, Faculty of Veterinary Medicine, Universitas Airlangga, Surabaya, 60115, East Java, Indonesia; ⁴Division of Veterinary Reproduction, Faculty of Veterinary Medicine, Universitas Airlangga, Surabaya, 60115, East Java, Indonesia; ⁵Division of Veterinary Pharmacology, Faculty of Veterinary Medicine, Universitas Airlangga, Surabaya, 60115, East Java, Indonesia; ⁶Division of Veterinary Science (Animal Nutrition), Faculty of Veterinary Medicine, Universitas Airlangga, Surabaya, 60115, East Java, Indonesia; ⁷Division of Veterinary Pathology, Faculty of Veterinary Medicine, Universitas Airlangga, Surabaya, 60115, East Java, Indonesia.

Received | October 10, 2025; Accepted | November 25, 2025; Published | March 04, 2026

*Correspondence | Epy Muhammad Luqman, Division of Veterinary Science, Faculty of Veterinary Medicine, Universitas Airlangga, Surabaya, 60115, East Java, Indonesia; Email: [email protected]

Citation | Hendrawan VF, Luqman EM, Rimayanti R, Widjiati W, Hamid IS, Al-Arif MA, Plumeriastuti H, Suprayogi TW (2026). Metabolic Profile in Dairy Cows With Cystic Ovarian Follicles: A Systematic and Meta-Analysis of Available Evidence from USA, Iran, and Indonesia. J. Anim. Health Prod. 14(2): 399-406.

DOI | https://dx.doi.org/10.17582/journal.jahp/2026/14.2.399.406

ISSN (Online) | 2308-2801

Copyright: 2026 by the authors. Licensee ResearchersLinks Ltd, England, UK.

This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

INTRODUCTION

Cystic ovarian follicles (COF) are common in dairy herds and are associated with disrupted estrous cyclicity, reduced conception, prolonged calving intervals, and increased culling risk, together driving substantial economic losses (Bello, 2021; Gümen and Wiltbank, 2021). Pathogenesis is multifactorial, involving dysregulation of the hypothalamic–pituitary–ovarian axis and transition-period metabolic stressors notably negative energy balance (NEB), insulin resistance, and endocrine perturbations (Jafari-Dehkordi et al., 2015; Keskin et al., 2016; Lee and Kim, 2019). Recent clustering analyses based on β-hydroxybutyrate trajectories in early-lactating dairy cows further emphasize the dynamic nature of energy metabolism in relation to postpartum reproductive disorders (Zachut et al., 2024). Metabolic differences in follicular fluid composition between ovulatory and cystic follicles also reflect systemic metabolic disturbances and endocrine dysregulation (Wang et al., 2025). Clinically, COF are diagnosed by transrectal ultrasonography and/or rectal palpation and are broadly categorized as follicular or luteal cysts (Lima et al., 2019). Given the sparse literature directly addressing COF in dairy cows, we adopted an inclusive evidence strategy while prespecifying dairy or COF diagnosed populations as the target of inference and conducting sensitivity analyses to test robustness when broader ovarian dysfunction populations were considered.

Because energy balance is tightly linked to fertility, numerous studies have examined serum metabolic biomarkers in COF non-esterified fatty acids (NEFA), β-hydroxybutyrate (BHBA), blood urea nitrogen (BUN), insulin, cortisol, and IGF-1 but findings remain inconsistent across production systems, climates, and methodologies (Mann, 2022; Rico, 2021), especially as emerging metabolomics approaches begin to distinguish adaptive metabolic signatures across management systems (Kumar et al., 2025). The diagnostic accuracy of ultrasonographic classification of follicular and luteal cysts has recently been validated in field conditions (Silva et al., 2023). Geographical context likely contributes substantially to this heterogeneity: Intensive temperate dairies may buffer metabolic disturbances, whereas tropical and smallholder settings face nutritional constraints and environmental stressors that can alter biomarker patterns (Gümen and Wiltbank, 2021). However, with only three eligible studies, quantitative generalization is inherently limited; thus, this review emphasizes transparent study-level reporting and narrative synthesis, with meta-analysis treated as exploratory.

This systematic review and meta-analysis synthesizes quantitative evidence on serum biomarkers in cows with COF versus healthy controls and explicitly explores between-country variability using studies from the United States, Iran, and Indonesia (Lima et al., 2019; Jafari Dehkordi et al., 2015; Rosadi et al., 2018). Our aims were to estimate pooled standardized mean differences (SMD), quantify heterogeneity (Q, τ², I²), and evaluate country-level subgroup effects to determine whether biomarker–COF associations are robust across contexts or predominantly context dependent. We additionally prespecified sensitivity analyses that (i) exclude non-dairy and non-COF primary diagnoses and (ii) focus on NEFA-only effects to align with the target population and pathophysiology. However, only three eligible studies were identified across all databases, which severely limits the statistical generalizability of the findings. Therefore, the quantitative synthesis in this paper is intended to be exploratory and hypothesis generating rather than confirmatory. Although the search strategy was global and unrestricted by country, only three studies from the United States, Iran, and Indonesia met the eligibility criteria based on diagnostic clarity, target population (dairy cows with COF), and availability of quantitative biomarker data. These countries were not pre-selected but rather represent the existing evidence base available for synthesis. In addition, this study contributes to Sustainable Development Goals (SDGs) 2 (Zero Hunger) and 3 (Good Health and Well-being) by addressing animal reproductive health and nutritional productivity in dairy systems.

MATERIALS AND METHODS

Search strategy

A systematic literature search was performed in PubMed, Scopus, and Web of Science up to December 2023. Search terms included combinations of “cystic ovarian follicles,” “cystic ovarian disease,” “dairy cows,” “metabolic profile,” and “reproduction.” Reference lists of relevant articles were also screened to identify additional eligible publications. The search and selection processes followed the PRISMA 2020 guidelines. We preregistered a concise search algorithm and screening checklist prior to data extraction to ensure reproducibility. Screening forms explicitly captured species production type (dairy vs beef) and primary diagnosis (COF vs other ovarian dysfunction) to enable sensitivity restrictions to the target population. Although the evidence base was limited (k = 3), a meta-analysis was conducted to provide an initial quantitative overview of between-country differences, while acknowledging that results would be interpreted cautiously under extreme heterogeneity.

Inclusion and exclusion criteria

Studies were included if they met the following criteria: (i) conducted in dairy cows diagnosed with COF using clear diagnostic methods (e.g., rectal palpation or transrectal ultrasonography); (ii) included a comparative group of healthy or cycling cows; and (iii) reported quantitative blood metabolite data, including non-esterified fatty acids (NEFA), β-hydroxybutyrate (BHBA), blood urea nitrogen (BUN), insulin, or cortisol, and/or COF prevalence data amenable to meta-analysis. Articles published in English between 2010 and 2023 were considered. Exclusion criteria comprised narrative or systematic reviews without primary data, single case reports, studies on species other than dairy cows, and publications lacking numerical data convertible

 

Table 1: Input data from eligible studies for meta-analysis using OpenMEE

Study	Ne	Xe (Experimental mean)	SDe	Nc	Xc (Control mean)	SDc	Country	Treatment	Cattle Breed
Lima et al., 2019	6	0.4	0.2	11	0.3	0.17	AS	Plasma NEFA (COF mid–late lactation vs. control)	Holstein (University of Illinois herd)
Jafari Dehkordi et al., 2015	20	0.71	0.04	20	0.43	0.07	Iran	Metabolic profile (cystic vs. cycling cows)	High-yielding Holstein
(Rosadi et al., 2018)	56	0.3768	0.0396	150	0.6232	0.0396	Indonesia (Jambi)	Postpartum anestrus longer than 3 months	Local beef cattle

 

Description: This table summarizes the characteristics of the studies (sample size, mean values, and standard deviations of the experimental and control groups) extracted for the analysis of metabolic profiles and cystic ovarian follicles in dairy cows.

 

to effect sizes. From each included study, we extracted sample size (n), means, and standard deviations (SD) for case and control groups to compute effect sizes. The input data used in OpenMEE are summarized in Table 1 (study identifiers, group sizes, means, SDs, country, outcome/measure, and breed details).

To minimize double counting, where multiple reports used overlapping cohorts, we retained the most comprehensive dataset. Because the literature is limited, one included study (Rosadi et al., 2018) involved local beef cattle with prolonged postpartum anestrus. We justified provisional inclusion on pathophysiological grounds (overlap of ovarian dysfunction with metabolic perturbation) but did not treat it as representative of dairy COF; therefore, we performed a pre-specified sensitivity analysis excluding this study and interpreted pooled estimates with caution.

Eligibility criteria

Eligible studies were observational investigations conducted on dairy cows diagnosed with cystic ovarian follicles (COF) that reported quantitative outcomes as blood metabolites namely NEFA, BHBA, BUN, insulin, and cortisol or incidence data of COF, included a healthy control group for comparison, and provided sufficient summary statistics (sample size, mean, and standard deviation) or convertible data (e.g., SEM to SD). Studies were excluded if they were review papers or case reports, involved experimental work on non-bovine species, or lacked an appropriate control group. Diagnostic definitions for COF were accepted as reported by primary studies and extracted verbatim (e.g., follicular diameter cut-offs, persistence criteria); and considered in the heterogeneity assessment. For transparency, the target population of inference is dairy cows with COF diagnosed by palpation/ultrasound; analyses restricted to this target set are reported as sensitivity findings.

Data extraction

Two reviewers independently extracted data using a standardized form, capturing first author, year, country, study design, sample size, breed, COF diagnostic method, and lactation stage. Outcome data included n, mean, and SD for case and control groups. When a study reported mean ± SEM, SD was derived using: SD = SEM × √n. For incidence data, event rates were treated as proportions and SDs were computed using SD = √(p × (1 − p)/n), where p is the event rate and n is the total sample size. When results were available only as graphs, numeric values were obtained from the appendix figure or appendix tables. Discrepancies were resolved through discussion until consensus was achieved. For studies reporting mean ± SEM, SD was calculated using the formula:

Explanation: SD= standard deviation (variability of the data within a group), SEM= standard error of the mean (often reported instead of SD in some articles), n = sample size of the group.

For incidence data, event rates were calculated as proportions, and standard deviations of proportions were estimated using:

Explanation: SD= standard deviation of the proportion, p= proportion of events (event rate), calculated as number of cases ÷ total sample size, n= total number of animals in the group.

For studies reporting multiple metabolic biomarkers, we calculated a study-level standardized mean difference (SMD) for each biomarker separately and then derived a within-study composite SMD by inverse-variance weighting of the biomarker-specific SMDs to obtain one effect size per study used in the primary meta-analysis. Because biomarker measurements within a study may be correlated, this approach is conservative with respect to variance; we therefore performed sensitivity analyses using (i) NEFA-only SMDs (pre-specified as the pathophysiologically most relevant and most consistently reported biomarker) and (ii) leave-one biomarker out recomputation of the composite. The primary pooled estimates presented in Figures 2–4 are based on the composite SMDs, while NEFA-only results are provided as sensitivity analyses. We also flagged population mismatch (dairy vs beef; COF vs other) at extraction to operationalize sensitivity constraints.

Study quality assessment

Risk of bias for observational studies was evaluated using the Newcastle–Ottawa Scale (NOS), which assesses three domains: selection, comparability, and outcome validity. The maximum score is nine, with higher scores indicating lower risk of bias. Two assessors scored each study independently; disagreements were resolved by consensus. We explored whether NOS score ( ≥7 vs <7 ) explained between-study heterogeneity in subgroup analyses. Population alignment (dairy/COF vs other) was treated as a methodological modifier in qualitative appraisal.

Outcomes

Primary outcomes were blood metabolite parameters reflecting energy metabolism and endocrine function (NEFA, BHBA, BUN, insulin, cortisol). Secondary outcomes were the occurrence or prevalence of COF in dairy cow populations. Consistent with this specification, continuous biomarkers informed the primary pooled SMDs, and COF prevalence was synthesized separately as pooled proportions. The primary target analysis pertains to dairy COF studies; inclusion of broader ovarian-dysfunction populations is reported separately as exploratory.

Statistical analysis

Analyses were performed using OpenMEE. For continuous outcomes, effect sizes were computed as standardized mean differences (SMD; Hedges’ g) with 95% confidence intervals under a random-effects model (REML) to account for between-study variability. For prevalence data, meta-analysis used logit-transformed proportions to stabilize variance. Heterogeneity was evaluated using Cochran’s Q, between-study variance (τ²), and the percentage of variability due to true heterogeneity (I²). I² values >75% were considered indicative of high heterogeneity. Subgroup analyses were conducted by country (USA, Iran, Indonesia) to explore sources of heterogeneity. Sensitivity analysis (leave-one-out) assessed robustness by excluding one study at a time and comparing effect estimates. A p-value <0.05 was considered statistically significant.

For the primary model, we pooled one composite SMD per study, obtained by inverse-variance weighting of biomarker-specific SMDs. Sensitivity analyses included NEFA-only pooling, leave-one-out outlier exclusion, and fixed-effect re-estimation. Prevalence was synthesized as logit-transformed proportions under random-effects; we report back-transformed estimates with 95% CIs and heterogeneity (Q, τ², I²= 99.31% where applicable). With k= 3, the pooled estimate is hypothesis-generating; therefore, we prioritized narrative synthesis, ran a dairy COF-only sensitivity (excluding Rosadi et al., 2018), and interpreted all pooled effects under extreme heterogeneity. Small-study effects were inspected visually (funnel plot); Egger’s test was not performed (k < 10). Computations were duplicated, and numeric inputs (n, means, SDs) appear in Table 1. Country subgroups are descriptive only (one study each).

Registration and ethics

The review protocol was not registered in PROSPERO. As this study used secondary data from published literature, no additional ethical approval was required. No individual animal data were used, and all analyses were based on aggregated published statistics.

 

RESULTS

Study selection

The PRISMA flow diagram (Figure 1) summarizes study identification and selection. The database search retrieved 356 records, with an additional 12 identified through manual screening (total= 368). After removing 85 duplicates, 283 records were screened by title and abstract; 260 were excluded as irrelevant. Twenty-three full-text articles were assessed for eligibility; 20 were excluded due to lack of quantitative data (n= 8), incomplete outcome reporting (n= 5), single case reports (n= 4), or inappropriate study design (n= 3). Three studies met the inclusion criteria and were included in the systematic review and meta-analysis (9–11). Key study characteristics (country, herd type, breeds, diagnostics, outcomes, and summary statistics) are presented in Table 1. For each included study, we computed biomarker-specific SMDs and then obtained a composite within-study SMD via inverse-variance weighting to represent the overall metabolic profile, which served as the study’s effect size in the primary meta-analysis. COF prevalence was extractable for two studies; results were pooled and are reported in appendix table or figure.

Random-effects meta-analysis (Model: continuous outcomes)

Figure 2 shows study weights and the pooled effect under a random-effects model. Across the three studies, there was no overall significant difference in metabolic parameters between cows with COF and healthy controls (pooled SMD= −0.302; 95% CI −6.867 to 6.264; p= 0.928). Heterogeneity was very high (Q= 287.794, df= 2, p < 0.001; τ² = 33.404; I² = 99.31%), indicating that most variability reflected true between-study differences rather than sampling error. Approximate study weights were balanced Lima et al. (2019) (33.3%), Jafari-Dehkordi et al. (2015) (33.2%), and Rosadi et al. (2018) (33.5%) suggesting that no single study dominated the pooled estimate despite divergent directions and magnitudes of effect (see Table 1 for study-level inputs). NEFA-only sensitivity pooling yielded effect estimates with the same direction of association, indicating that the primary conclusions were not driven by the choice of composite vs single-biomarker specification.

 

Forest plot

Figure 3 illustrates study-level effects. Lima et al. (USA) reported a small, non-significant SMD (0.526; 95% CI −0.485 to 1.536), consistent with broadly similar metabolic profiles between COF and control cows under research-herd management. Jafari-Dehkordi et al. (2015)(Iran) observed a significant positive effect (SMD= 4.814; 95% CI 3.590 to 6.037), suggesting notable metabolic dysregulation potentially linked to postpartum negative energy balance in high-producing cows. Conversely, Rosadi et al. (2018) (Indonesia) reported a significant negative effect (SMD= −6.199; 95% CI −6.872 to −5.527), indicating a different disturbance pattern under tropical, predominantly smallholder conditions. These contrasting effects emphasize the high heterogeneity and the need for context-specific interpretation (study design details are summarized in Table 1). Influence diagnostics identified the Indonesian study as an outlier in magnitude but not in direction relative to its subgroup context; exclusion did not reverse the overall conclusion.

 

 

Subgroup analysis by country

Figure 4 (subgroup forest plots) shows distinct patterns by country. The United States subgroup yielded a small, non-significant difference (SMD 0.526; 95% CI −0.485 to 1.536), aligning with intensive nutrition and health management. The Iranian subgroup showed a significant positive effect (SMD 4.814; 95% CI 3.590 to 6.037), consistent with a stronger link between NEB and COF in high-yielding Holsteins. The Indonesian subgroup demonstrated an opposite, significantly negative effect (SMD −6.199; 95% CI −6.872 to −5.527). By convention, SMDs were computed as COF minus control; therefore, a negative value indicates lower composite metabolite concentrations in COF cows relative to controls. Although the pooled estimate remained non-significant (SMD = −0.302; 95% CI −6.867 to 6.264), heterogeneity was consistently high (I² = 99.31%). Country-level subgrouping explained part of the dispersion, but residual heterogeneity remained high, suggesting additional modifiers (e.g., days in milk at sampling, parity, assay methods) not fully captured by available data. Interpretation of the Indonesian subgroup: the markedly negative estimate in predominantly smallholder tropical herds may reflect a distinct metabolic pattern compared with intensive systems, potentially driven by (i) forage-based diets and chronic heat load that blunt lipid mobilization (lower NEFA/BHBA), (ii) inclusion of cows with prolonged postpartum anestrus (altered endocrine milieu not dominated by classic postpartum NEB), and (iii) sampling timing later in lactation. These hypotheses are developed further in the Discussion.

DISCUSSION

Summary of main findings

Given the very limited number of studies available, the present meta-analysis should be viewed as a preliminary synthesis designed to map knowledge gaps rather than to draw definitive conclusions. The extreme heterogeneity observed (I²= 99.31%) reflects both biological and methodological differences among studies. The meta-analysis did not reveal a consistent overall difference in serum metabolic biomarkers between cows with COF and healthy controls; however, country specific effects were apparent. Balanced study weights argue against dominance by any single dataset (Lima et al., 2019; Jafari Dehkordi et al., 2015; Rosadi et al., 2018), while the study-level directions and subgroup estimates underscore contextual influences. Primary pooling used a within-study composite SMD; NEFA-only sensitivity supported the same qualitative conclusion. By convention, SMDs were computed as COF minus control; thus, negative values indicate lower metabolite concentrations in COF cows. Between-study heterogeneity was extreme (I²= 99.31%), so pooled estimates warrant cautious, context-specific interpretation. We confirmed robustness with NEFA-only, leave-one-out, and fixed-effect checks. Apparent differences across countries likely reflect nutrition/heat load, endocrine milieu (including prolonged postpartum anestrus), assay protocols, and sampling timing. A random-effects prevalence synthesis in the appendix supports contextual heterogeneity at the population level. Given k= 3 and I²= 99.31%, the pooled estimate is hypothesis-generating; conclusions rely on narrative synthesis and sensitivity checks (NEFA-only; dairy COF-only).

Sources of heterogeneity

Heterogeneity likely stems from differences in management and nutrition across production systems, diagnostic criteria (ultrasonography vs. palpation), stage of lactation, laboratory methods, and genetic backgrounds (Keskin et al., 2016; Lee and Kim, 2019; Walsh et al., 2011). Risk of bias appraisal using the Newcastle–Ottawa Scale is summarized in Table 2, highlighting variable methodological quality that may contribute to between-study variability. Additional sources may include timing of blood sampling relative to parturition and follicular dynamics, heat load, and feed quality (including potential mycotoxin exposure), which can differentially influence NEFA/BHBA and nitrogen metabolites across ecologies. Population mismatch (beef vs. dairy; COF vs. anestrus) likely contributed materially to between-study variability and was addressed through restriction sensitivity analyses.

Interpretation of the forest plot

Neutral to small effects in U.S. research herds (Lima et al., 2019), strongly positive effects in Iranian high-yielding Holsteins (Jafari-Dehkordi et al., 2015), and strongly negative effects in Indonesian smallholder settings (Rosadi et al., 2018) collectively indicate that biomarker COF associations are context dependent. Differences in COF definitions and assay protocols likely added variability (Bello, 2021; Rico, 2021). Study design and cohort details supporting this interpretation are consolidated in Table 1. These findings align with those of Santos (2020), Wang et al. (2020), and Walsh et al. (2011), who emphasized ecological influences on ovarian physiology.

Interpretation for the Indonesian subgroup. The markedly negative SMD suggests that, in predominantly smallholder tropical systems, cows classified with COF may exhibit lower concentrations in the composite metabolic profile relative to controls. This could reflect (i) a distinct metabolic adaptation wherein negative energy balance is less dominant than in intensive systems, with greater influence of chronic heat stress and forage-based diets on insulin and nitrogen metabolism; (ii) phenotypic heterogeneity linked to inclusion of cows with prolonged postpartum anestrus,

 

Table 2: Quality assessment of included studies using the Newcastle–Ottawa Scale (NOS)

Study	Selection (Max 4)	Comparability (Max 2)	Outcome (Max 3)	Total score (Max 9)	Quality rating
Lima et al., 2019 (USA)	4	2	2	8	Low risk of bias
Jafari Dehkordi et al., 2015 (Iran)	3	1	2	6	Moderate risk of bias
Rosadi et al., 2018 (Indonesia)	2	1	2	5	Moderate risk of bias

 

Description: Quality appraisal of the included studies using the Newcastle–Ottawa Scale (NOS), evaluating selection, comparability, and outcome domains. Total scores (0–9) indicate overall risk of bias: low (7–9), moderate (5–6), high (<5).

 

which may present a different endocrine milieu than classic postpartum NEB; and/or (iii) timing of sampling later in lactation when lipid mobilization (NEFA) has subsided but other stress axes (e.g., cortisol) differ. These hypotheses require validation in larger cohorts with harmonized definitions and synchronized sampling windows.We emphasize that country-specific effects cannot be inferred from single-study subgroups, the apparent differences reflect three individual, divergent studies rather than reproducible country effects.

Appropriateness of meta-analysis

With only three heterogeneous studies, the decision to meta-analyze warrants caution. We therefore present the quantitative pooling as exploratory, prioritize narrative synthesis, and explicitly report sensitivity analyses (NEFA-only; dairy COF-only). This approach aligns with guidance that extreme heterogeneity can render pooled effects statistically and clinically uninformative (Egger et al., 1997; Ashwell et al., 2022). From a methodological perspective, performing a meta-analysis with only three heterogeneous studies is statistically fragile and clinically uninformative. The pooled estimate (SMD= –0.302) therefore cannot be interpreted as an effect size but rather as a demonstration of inconsistency among contexts. This exercise primarily served to quantify the magnitude of heterogeneity rather than to establish a true overall effect.

Potential publication bias

Formal publication-bias tests (e.g., funnel plot, Egger’s test) were not conducted due to the small number of studies (<10). Nonetheless, selective publication remains a possibility and should be considered when interpreting pooled estimates. Visual inspection of funnel plots suggested asymmetry; however, we refrain from formal inference given k = 3 (Egger et al., 1997).

Strengths and limitations

Strengths include a PRISMA-guided process, quantitative synthesis, and explicit examination of geographic heterogeneity. Limitations include the small number of eligible studies, variable NOS scores (Table 2), extreme heterogeneity (I² > 99.31%), and incomplete alignment of populations and outcomes (e.g., inclusion of a cohort with prolonged postpartum anestrus). Furthermore, the within-study composite approach assumes independence among biomarkers and may underestimate variance if biomarkers are positively correlated; our NEFA-only and leave-one-biomarker-out sensitivity analyses mitigate this concern but cannot eliminate it. Major limitations are the very small evidence base (k = 3), a population mismatch in one study (beef or anestrus), and extreme heterogeneity, all of which preclude firm quantitative conclusions. The pooled estimate (SMD= –0.302) lacks statistical and clinical significance, reinforcing the interpretive limits of this dataset (Sirmans and Pate, 2014; Fujii et al., 2022).

Implications and future directions

Future work should prioritize multicenter designs with standardized COF definitions, harmonized biomarker panels and laboratory methods, and adequate statistical power across production ecologies. Integrating molecular or genomic markers with metabolic profiling and testing targeted nutritional or reproductive interventions would further clarify mechanisms and improve prevention strategies (Wang et al., 2020; Ashwell et al., 2022). Prospective studies should prespecify primary biomarkers (e.g., NEFA) with synchronized sampling relative to parturition and follicular dynamics, and report both continuous biomarker distributions and COF prevalence to enable joint modeling of risk and pathophysiology.

CONCLUSION

This meta-analysis of three studies (USA, Iran, Indonesia) found no overall significant differences in metabolic parameters between cows with COF and healthy controls; however, heterogeneity was extremely high (I² > 99%). Subgroup analysis revealed differing patterns: neutral in the USA, significantly positive in Iran, and significantly negative in Indonesia. These findings emphasize that the relationship between COF and metabolic profiles is influenced by local factors. Multicenter studies with standardized diagnostic and measurement protocols are required to strengthen the evidence base. Given the very small and heterogeneous evidence base, the pooled estimate should be regarded as exploratory. Results are best interpreted as a systematic review with narrative synthesis, indicating that current data are insufficient for definitive quantitative inference about country effects or overall biomarker differences. Larger, harmonized, dairy COF specific studies are needed. These results highlight the scarcity of harmonized metabolic data in cows with COF and demonstrate that current evidence is insufficient for clinical interpretation. Future systematic reviews should expand the database inclusion period and integrate unpublished or gray-literature datasets to improve power and representativeness.

ACKNOWLEDGEMENTS

The authors express their gratitude to the Faculty of Veterinary Medicine, Universitas Brawijaya, for providing support and research resources that facilitated this review. Special thanks are also extended to colleagues in the Department of Animal Reproduction for their valuable ideas and input during manuscript preparation. The authors further acknowledge the contributions of researchers and practitioners whose work formed the foundation of this review.

Novelty Statement

The present study is the first systematic review and meta-analysis exploring serum metabolic biomarkers in cows with cystic ovarian follicles (COF) across multiple countries (USA, Iran, Indonesia). It highlights how geographical production systems shape biomarker patterns and COF interpretation.

AUTHOR’S CONTRIBUTION

VFH contributed to study conceptualization, literature review, data collection, data analysis, drafting, and revision of the manuscript. EML was responsible for methodology design, supervision, and critical manuscript review. R performed data extraction, statistical analysis, and contributed to writing. W provided validation of results, constructive feedback, and editorial support. ISH contributed to data interpretation and review of the discussion. MAA-A assisted with literature search and manuscript editing. HP provided supervision, conceptual guidance, and critical evaluation of the study. TWS served as senior supervisor, offering overall guidance and final manuscript approval.

Generative AI and AI-assisted technology statement

No AI tools were used in the writing, editing, or data analysis of this manuscript. All tasks were performed manually by the authors.

Conflict of interest

The authors have declared no conflict of interest.

REFERENCES

Ashwell MS, Mooney CS, Gonzalez-Pena D, Sattar A, Black BL, BonDurant RH, 2022). Meta-analytic strategies for evaluating reproductive traits in cattle. Anim. Reprod. Sci., 239: 107006.

Bello NM (2021). Hormonal treatment strategies for cystic ovarian disease in dairy cows. J. Dairy Res., 88(3): 345–56.

Egger M, Davey Smith G, Schneider M, Minder C (1997). Bias in meta-analysis detected by a simple, graphical test. Br. Med. J., 315(7109): 629–634. https://doi.org/10.1136/bmj.315.7109.629

Fujii S, Hirota Y, Kawamura K, Matsuo M, Matsumoto T (2022). Involvement of steroid hormones and prostaglandin signaling in endometrial decidualization. Reprod. Med. Biol., 21: e12453.

Gümen A, Wiltbank MC (2021). Cystic ovarian disease in dairy cattle: A review. Theriogenology, 165: 93–100.

Jafari Dehkordi A, Mirshokraei P, Dehghani A (2015). Metabolic profiles of high-yielding dairy cows with ovarian cysts formation. Iran J. Vet. Med., 9(4): 241–248.

Kawashima C, Hamano S, Matsui M, Shimizu T, Matsunaga N, Ishii M, Kida, K., Miyamoto, A. (2006). Relationships between metabolic status and reproductive hormones in postpartum dairy cows. J. Reprod. Dev., 58(1): 88–95.

Keskin A, Yilmazbas-Mecitoglu G, Gumen A, Karakaya E, Darici R, Okut H (2016). Relationships among serum IGF-1 concentrations, metabolic status, and reproductive performance in dairy cows. Theriogenology, 86(1): 138–1345.

Kumar A, Jadhav A, Chakravarty PK (2025). Applications of metabolomics in cow health assessment: Emerging biomarkers and analytical perspectives. Metabolomics, 21(1): e1130. https://doi.org/10.1007/s11306-025-02364-7

Lee HJ, Kim SY (2019). Apoptosis and autophagy regulation in ovarian follicle development and cystic ovarian disease. Reprod. Biol. Endocrinol., 17(1): 95.

Lima FS, Acosta DAV, Egan TR, Skenandore C, Sulzberger S, French DD, Cardoso FC (2019). Steroidogenic, metabolic, and immunological markers in dairy cows diagnosed with cystic ovarian follicles at early and mid-late lactation. Front. Vet. Sci., 6: 324. https://doi.org/10.3389/fvets.2019.00324

Mann GE (2022). Economic implications of reproductive disorders in dairy herds. Anim. Reprod. Sci. 233: 106114.

Rico C (2021). Diagnosis and treatment of ovarian cysts in dairy cows. Vet. J., 275: 105705.

Rosadi R, Isnaini N, Gunawan A, Purwantara B (2018). Identification of ovarian disorders in beef cows with prolonged postpartum anestrus. J. Vet., 19(3): 422–429.

Santos JEP (2020). Reproductive management of dairy cows with cystic ovarian disease. Anim. Reprod. Sci., 220: 106519.

Silva R, Giordano JO, Wiltbank MC (2023). Cystic ovarian disease in dairy cattle: Diagnostic accuracy when using ultrasonography. J. Dairy Sci., 106(8): 7345–7356.

Sirmans SM, Pate KA (2014). Epidemiology, diagnosis, and management of polycystic ovary syndrome. Clin. Epidemiol., 6: 1–13. https://doi.org/10.2147/CLEP.S37559

Walsh SW, Williams EJ, Evans ACO (2011). A review of the causes of poor fertility in high milk producing dairy cows. Anim. Reprod. Sci., 123(3–4): 127–138. https://doi.org/10.1016/j.anireprosci.2010.12.001

Wang J, Chen L, Zhao Y (2025). Metabolic characteristics of follicular fluid in cows: A review. Theriogenology, 228: 113459.

Wang RS, Yeh CY, Chuang LY, Guh JY (2020). Comparative gene expression profiling of the involvement of steroidogenic acute regulatory protein (STAR) in inflammation and PCOS. J. Ovarian Res., 13(1):5.

Wathes DC, Fenwick M, Cheng Z, Bourne N, Llewellyn S, Morris DG, Steeneveld W, Wicaksono AS, Edwardes MD (2007). Influence of negative energy balance on cyclicity and fertility in the high producing dairy cow. Theriogenology, 68: S232–41. https://doi.org/10.1016/j.theriogenology.2007.04.006

Zachut M, Shabat G, Dekel R, Moallem U (2024). Metabolic clusters of early-lactating dairy cows based on blood β-hydroxybutyrate trajectories. J. Dairy Sci., 107(5): e10342.