Association Between Serum ANGPTL8 Levels and PCOS and Its Clinical Utility in Diagnosis

Objective To analyze the correlation between angiopoietin‐like protein 8 (ANGPTL8) and polycystic ovary syndrome (PCOS) and explore the clinical diagnostic value of ANGPTL8 in PCOS.

In this retrospective case–control study, we included 94 PCOS patients (study group) from our hospital's Reproductive Center from June 2023 to December 2024 and 116 age‐matched women with normal ovarian function (control group) from the Gynecology Department. Serum levels of anti‐Müllerian hormone (AMH), sex hormones, and ANGPTL8 were measured in all participants.

The study group exhibited higher AMH, LH, LH/FSH ratio, and E2 levels compared to the control group (p < 0.001). Serum ANGPTL8 levels were lower in PCOS patients (p < 0.001). Logistic multivariate regression analysis identified BMI, AMH, LH, and ANGPTL8 as independent influencing factors for PCOS. Elevated ANGPTL8 levels were associated with reduced PCOS risk (p < 0.001), while increased BMI, AMH, and LH levels correlated with higher PCOS risk (p < 0.05). At a cutoff value of 12.871 ng/mL, ANGPTL8 demonstrated a sensitivity of 90.4%, specificity of 77.6%, and an area under the curve (AUC) of 0.881 (95% CI: 0.836–0.926, p < 0.001), indicating moderate‐to‐high diagnostic efficacy. The combined use of ANGPTL8 with AMH or sex hormones improved diagnostic performance compared to individual markers. Notably, the combination of ANGPTL8 and sex hormones yielded a higher AUC (0.980) than ANGPTL8 with AMH (AUC = 0.928).

Serum ANGPTL8 levels are significantly lower in PCOS patients and inversely associated with PCOS risk. ANGPTL8 shows promise as a novel diagnostic biomarker for PCOS, especially when integrated into multi‐marker panels with existing hormonal parameters.

angiopoietin‐like protein 8, anti‐Müllerian hormone, diagnostic biomarker, polycystic ovary syndrome, sex hormones 1. Introduction Polycystic ovary syndrome (PCOS) is a heterogeneous polygenic and multifactorial disorder characterized by ovulatory dysfunction, hyperandrogenism, and polycystic ovarian morphology. It represents the most common gynecological reproductive endocrine and metabolic disorder among women of reproductive age [1]. In recent years, due to lifestyle changes, rising obesity rates, and delayed childbearing, the global age‐standardized incidence of PCOS has reached 82.44 per 100 000, reflecting a 1.45% increase since 2007 [2], with a trend toward younger onset and rising prevalence [3]. Currently, diagnostic criteria for PCOS remain inconsistent, leading to underdiagnosis in clinical practice. The 2003 Rotterdam criteria are widely adopted internationally [4], however, scholars argue that these criteria may be overly broad for Chinese populations, potentially resulting in overdiagnosis and overtreatment. Common auxiliary diagnostic indicators include elevated anti‐Müllerian hormone (AMH) levels [5], insulin resistance (IR) [6], an increased luteinizing hormone (LH)/follicle‐stimulating hormone (FSH) ratio [7], and elevated body weight or body mass index (BMI) [8]. As PCOS progresses, it may exacerbate menstrual irregularities, hyperandrogenic symptoms, IR, Type 2 diabetes risk, and severe infertility [9]. Therefore, identifying potential biomarkers for early PCOS diagnosis is critical for timely intervention, fertility preservation, and disease progression management. The angiopoietin‐like protein (ANGPTL) family comprises secretory glycoproteins structurally similar to angiopoietins, with eight identified members (ANGPTL1‐ANGPTL8). Although ANGPTLs share structural homology with angiopoietins, they lack binding affinity for classical angiopoietin receptors, leading to distinct biological roles [10]. ANGPTLs are involved in lipid metabolism, glucose homeostasis, inflammation, and tumorigenesis. Recent studies highlight their involvement in PCOS pathogenesis through IR and glucose dysregulation. Among the ANGPTL family members, ANGPTL3, ANGPTL4, and ANGPTL6 have been reported to be dysregulated in PCOS, often associated with IR and metabolic disturbances [11, 12, 13, 14]. However, research on ANGPTL8, a unique member lacking the fibrinogen‐like domain, remains limited and contradictory. Structurally, ANGPTL1‐ANGPTL7 contain an N‐terminal coiled‐coil domain, a linker region, and a C‐terminal fibrinogen‐like domain. In contrast, ANGPTL8 lacks the fibrinogen‐like domain but shares homology with the N‐terminal region of ANGPTL3, making it an exception within the family [10]. Current evidence regarding ANGPTL8 in PCOS is inconsistent. A systematic review and meta‐analysis by Varikasuvu et al. [15] suggested that PCOS independently increases circulating ANGPTL8 levels, irrespective of BMI. Conversely, other studies reported lower ANGPTL8 levels in PCOS patients compared to healthy controls [16]. These discrepancies may stem from methodological variations, including differences in ELISA kits used for measurement as well as population heterogeneity and PCOS phenotypic diversity. Given these controversies and the unexplored diagnostic potential of ANGPTL8 for PCOS, this study aimed to: (1) compare serum ANGPTL8 levels between well‐phenotyped PCOS patients and matched controls; (2) evaluate the independent association between ANGPTL8 and PCOS risk through multivariate analysis; (3) assess the diagnostic performance of ANGPTL8 alone and in combination with established biomarkers (AMH and sex hormones); and (4) explore the potential clinical utility of ANGPTL8 as part of a multi‐marker diagnostic strategy. 2. Materials and Methods 2.1. Data Source This retrospective case–control study was conducted at The First Hospital of Hebei Medical University from June 2023 to December 2024. A total of 94 patients with PCOS admitted to the Reproductive Center of our hospital from June 2023 to December 2024 were enrolled as the study group. Their ages ranged from 22 to 42 years (mean: 32.28 ± 4.28 years), with a mean height of 160.98 ± 4.86 cm, weight of 64.63 ± 9.33 kg, and BMI of 24.97 ± 3.64 kg/m2. For comparison, 116 age‐matched women with normal ovarian function treated in the Gynecology Department during the same period were selected as the control group. The control group had an age range of 21–44 years (mean: 31.59 ± 4.20 years), mean height of 160.87 ± 4.59 cm, weight of 59.05 ± 8.86 kg, and BMI of 22.85 ± 3.52 kg/m2. This study was approved by the Institutional Ethics Committee prior to its initiation. Diagnostic Criteria: PCOS patients met the revised Rotterdam diagnostic criteria established by the American Society for Reproductive Medicine (ASRM) [4]. (1) Oligo‐ovulation or anovulation, (2) clinical and/or biochemical hyperandrogenism, (3) polycystic ovarian morphology: > 12 follicles measuring 2–9 mm in diameter in one or both ovaries and/or ovarian volume ≥ 12 mL, and (4) meeting at least two of the three criteria above, after excluding other causes of hyperandrogenism (e.g., adrenal tumors, hyperprolactinemia). Inclusion criteria for the control group: (1) normal androgen levels, (2) regular menstrual cycles (21–35 days), and (3) no evidence of polycystic ovarian morphology on gynecological ultrasound. Exclusion criteria for all participants: (1) reproductive disorders affecting fertility, including ovarian tumors, uterine fibroids, endometriosis, or congenital uterine anomalies, (2) comorbid endocrine diseases (e.g., thyroid dysfunction, Cushing's syndrome), (3) pregnancy, ovulation induction therapy, or ovarian surgery within the past 3 months, (4) hormonal medication use within the past 3 months, and (5) conditions associated with abnormal ANGPTL8 expression (e.g., pathological cardiac hypertrophy, chronic inflammatory diseases). This study was approved by the Institutional Ethics Committee, and written informed consent was obtained from all participants. Sample size justification: As this was a retrospective study, a formal a priori power analysis was not conducted. However, we evaluated the statistical power of our cohort post hoc using G*Power 3.1. Based on the observed difference in ANGPTL8 levels between groups (control: 17.81 ± 7.79 ng/mL, PCOS: 7.37 ± 4.51 ng/mL; and effect size d = 1.60), post hoc calculation (two‐sample t‐test, α = 0.05, two‐tailed) indicated that our study had 100% power to detect this difference. Even for a more conservative medium effect size (d = 0.65), our sample size (94 PCOS, 116 controls) provided > 99.99% power. This confirms the adequacy of our sample size for the primary comparison. 2.2. Specimen Collection Fasting peripheral venous blood samples (3–4 mL) were collected from all participants on Days 2–4 of the menstrual cycle during the early morning. Blood samples were placed in anticoagulant tubes and centrifuged at 3000 rpm for 10 min using a Hettich EBA270 horizontal centrifuge (Hettich Scientific Instruments, Germany) to isolate serum. The serum was aliquoted and stored at −20°C until further analysis. 2.3. Measurement of Related Indicators After obtaining the serum supernatant, AMH levels were quantified using the Access2 chemiluminescence immunoassay analyzer (Beckman Coulter Commerce [China] Co. Ltd.) with the manufacturer's AMH detection kit (Cat. No. B13127, Beckman Coulter, USA). LH, FSH, E2 levels were measured using compatible reagent kits. Serum ANGPTL8 levels were determined via enzyme‐linked immunosorbent assay (ELISA) using a VersaMax automated microplate reader (Model 80 416‐KBHL, Molecular Devices, USA) and a human angiopoietin‐like protein 8 ELISA kit (Cat. No. GSB‐EL028, Wuhan Huamei Biological Engineering Co. Ltd., China). This sandwich ELISA employs polyclonal antibodies against full‐length human ANGPTL8. All samples and standards were measured in duplicate. The intra‐assay coefficient of variation (CV) was < 8%, and the inter‐assay CV was < 12%, based on quality control samples measured across different plates. The assay detection range was 0.5–50 ng/mL. No sample dilution was required as all values fell within the standard curve range. 2.4. Statistical Analysis Data analysis was performed using SPSS 27.0 (IBM Corp., USA). Continuous variables were first tested for normality using the Shapiro–Wilk test. Normally distributed data are presented as ± SD and analyzed with independent samples t‐tests for intergroup comparisons. Pearson correlation analysis was applied for normally distributed variables. Non‐normally distributed data are expressed as M (Q1, Q3) and analyzed using the Mann–Whitney U test for intergroup comparisons, with Spearman correlation analysis for associations. Multivariate logistic regression analysis (backward stepwise regression method) was employed to identify independent influencing factors. Receiver operating characteristic (ROC) curve analysis was conducted to evaluate diagnostic performance, calculating sensitivity, specificity, area under the curve (AUC), Youden's index (YI), and cutoff values. Diagnostic efficacy was interpreted according to Swets' criteria [17]: AUC = 0.5–0.7 indicates low diagnostic value, AUC = 0.7–0.9 indicates moderate value, AUC > 0.9 indicates high diagnostic value. Statistical significance was set at α = 0.05, with p < 0.05 considered significant. The deidentified data used for this analysis are provided as Data S1. No artificial intelligence‐generated content (AIGC) tools, such as ChatGPT or other large language models (LLMs), were used in the development of any portion of this manuscript. 3. Results 3.1. Baseline Characteristics of Study Participants The demographic and anthropometric characteristics of both groups are presented in Table 1. The PCOS and control groups were well‐matched for age and height (p > 0.05). As expected, PCOS patients had significantly higher weight and BMI compared to controls (p < 0.001, Table 1). TABLE 1. | Variables | Control group (n = 116) | Research group (n = 94) | t | p | |---|---|---|---|---| | Age (year) | 31.59 ± 4.20 | 32.28 ± 4.28 | −1.16 | 0.247 | | Height (cm) | 160.87 ± 4.59 | 160.98 ± 4.86 | −0.17 | 0.869 | | Weight (kg) | 59.05 ± 8.86 | 64.63 ± 9.33 | −4.44 | < 0.001 | | BMI (kg/m2) | 22.85 ± 3.52 | 24.97 ± 3.64 | −4.29 | 0.05). The study group exhibited significantly higher levels of AMH, LH, LH/FSH ratio, and E2 compared to the control group (p < 0.001). Conversely, ANGPTL8 levels were significantly lower in the study group (p < 0.001). Detailed results are presented in Table 2. In post hoc analyses, we categorized PCOS patients according to Rotterdam phenotypes: Phenotype A (classic, n = 52, 55.32%), Phenotype B (ovulatory, n = 22, 23.40%), Phenotype C (non‐hyperandrogenic, n = 14, 14.89%), and Phenotype D (mild, n = 6, 6.38%). ANGPTL8 levels did not significantly differ among phenotypes (p > 0.05), though Phenotype A tended to have the lowest levels (Figure S1 and Table S1). TABLE 2. | Variables | Control group (n = 116) | Research group (n = 94) | t/Z | p | |---|---|---|---|---| | AMH (ng/mL) | 4.44 ± 2.36 | 8.23 ± 4.70 | −7.12 | < 0.001 | | ANGPTL8 (ng/mL) | 17.81 ± 7.79 | 7.37 ± 4.51 | 12.14 | < 0.001 | | LH (mIU/mL) | 4.70 (3.37, 5.85) | 14.14 (10.26, 18.30) | −10.98 | < 0.001 | | FSH (mIU/mL) | 7.19 ± 1.47 | 7.03 ± 1.50 | 0.78 | 0.438 | | LH/FSH | 0.61 (0.48, 0.87) | 2.08 (1.38, 2.72) | −10.71 | < 0.001 | | E2 (pg/mL) | 37.21 ± 17.78 | 46.38 ± 16.64 | −3.83 | < 0.001 | Abbreviations: t, t‐test; Z, Mann–Whitney U test. 3.3. Identification of Independent Influencing Factors for PCOS Using PCOS diagnosis (1 = PCOS, 0 = control) as the dependent variable, variables with significant differences in univariate analysis (Tables 1 and 2) including BMI, AMH, LH, LH/FSH ratio, E2, and ANGPTL8—were incorporated into a multivariate logistic regression model. The Omnibus test (χ 2 = 227.221, p < 0.001) confirmed the overall statistical significance of the model. The Nagelkerke R 2 (0.855) indicated that the model explained 85.5% of the variance in the dependent variable. The Hosmer–Lemeshow test (χ 2 = 7.794, p = 0.999) demonstrated good calibration between predicted and observed outcomes, confirming robust explanatory power and model fit. The results identified BMI, AMH, ANGPTL8, and LH as independent influencing factors for PCOS (p < 0.05). Serum ANGPTL8 emerged as an independent protective factor, with each unit increase associated with a 0.71‐fold reduction in PCOS risk (p < 0.001). Conversely, BMI, AMH, and LH were independent risk factors, with each unit increase corresponding to 1.33‐, 1.42‐, and 1.80‐fold increases in PCOS risk, respectively (p < 0.05, Table 3). TABLE 3. | Variables | β | SE | Z | p | OR (95% CI) | |---|---|---|---|---|---| | BMI | 0.28 | 0.11 | 2.54 | 0.011 | 1.33 (1.07–1.65) | | AMH | 0.35 | 0.11 | 3.17 | 0.002 | 1.42 (1.14–1.77) | | ANGPTL8 | −0.35 | 0.08 | −4.27 | < 0.001 | 0.71 (0.60–0.83) | | LH | 0.59 | 0.13 | 4.40 | < 0.001 | 1.80 (1.39–2.34) | | E2 | 0.03 | 0.02 | 1.63 | 0.103 | 1.04 (0.99–1.08) | 3.4. Diagnostic Efficacy of ANGPTL8 and AMH Alone or in Combination for PCOS Using logistic regression‐derived β coefficients, a combined diagnostic equation was established: 1.36 + 0.334 × AMH − 0.304 × ANGPTL8. ROC curves were plotted with sensitivity as the Y‐axis and 1 − specificity as the X‐axis, using AMH, ANGPTL8, and their combined diagnostic score as test variables, and PCOS diagnosis as the state variable (Figure 1). ANGPTL8 alone: At a YI of 0.680, the optimal cutoff was 12.871 ng/mL, yielding a sensitivity of 90.4%, specificity of 77.6%, and AUC of 0.881 (95% CI: 0.836–0.926, p < 0.001). AMH alone: At a YI of 0.467, the optimal cutoff was 7.425 ng/mL, with a sensitivity of 55.3%, specificity of 91.4%, and AUC of 0.752 (95% CI: 0.682–0.822, p < 0.001). Combined model: At a YI of 0.736, the optimal probability cutoff was 0.316, achieving a sensitivity of 92.6%, specificity of 81.0%, and AUC of 0.928 (95% CI: 0.895–0.961, p < 0.001). These results demonstrate that serum ANGPTL8 exhibits moderate‐to‐high diagnostic value for PCOS, outperforming AMH alone. The combined use of AMH and ANGPTL8 significantly enhanced diagnostic accuracy compared to individual markers (Table 4). TABLE 4. | Variables | AUC | SE | 95% CI | p | Sensitivity | Specificity | YI | Cutoff value | |---|---|---|---|---|---|---|---|---| | ANGPTL8 | 0.881 | 0.023 | 0.836–0.926 | < 0.001 | 90.4% | 77.6% | 0.680 | 12.871 | | AMH | 0.752 | 0.036 | 0.682–0.822 | < 0.001 | 55.3% | 91.4% | 0.467 | 7.425 | | Combined Diagnosis | 0.928 | 0.017 | 0.895–0.961 | < 0.001 | 92.6% | 81.0% | 0.736 | 0.316 | 3.5. Diagnostic Efficacy of ANGPTL8 and Sex Hormones Alone or in Combination for PCOS A combined diagnostic equation was derived using logistic regression β coefficients: −1.922 − 0.049 × FSH + 0.562 × LH + 0.026 × E2 − 0.301 × ANGPTL8. ROC curves were generated with sex hormones (LH, FSH, E2), ANGPTL8, and their combined diagnostic score as test variables, and PCOS diagnosis as the state variable. Sensitivity was plotted on the Y‐axis and 1 − specificity on the X‐axis (Figure 2). LH alone: At a YI of 0.817, the optimal cutoff was 8.560 mIU/mL, yielding a sensitivity of 85.1%, specificity of 96.6%, and AUC of 0.941 (95% CI: 0.906–0.976, p 0.05). E2 alone: At a YI of 0.305, the optimal cutoff was 28.645 pg/mL, achieving a sensitivity of 92.6%, specificity of 37.9%, and AUC of 0.667 (95% CI: 0.594–0.740, p < 0.001). Combined model: At a YI of 0.852, the optimal probability cutoff was 0.304, demonstrating a sensitivity of 94.7%, specificity of 90.5%, and AUC of 0.980 (95% CI: 0.965–0.994, p < 0.001). In an additional analysis, the combination of ANGPTL8 with LH alone (excluding other sex hormones) yielded an AUC of 0.979 (95% CI: 0.964–0.994, p < 0.001), with 95.7% sensitivity and 90.5% specificity at the optimal cutoff (Figure S2 and Table S2). Serum ANGPTL8 exhibited higher diagnostic value for PCOS compared to FSH and E2 alone but was inferior to LH alone. The combined use of sex hormones and ANGPTL8 significantly improved diagnostic accuracy over individual markers (Table 5). TABLE 5. | Variables | AUC | SE | 95% CI | p | Sensitivity | Specificity | YI | Cutoff value | |---|---|---|---|---|---|---|---|---| | ANGPTL8 | 0.881 | 0.023 | 0.836–0.926 | < 0.001 | 90.4% | 77.6% | 0.680 | 12.871 | | LH | 0.941 | 0.018 | 0.906–0.976 | < 0.001 | 85.1% | 96.6% | 0.817 | 8.560 | | FSH | 0.462 | 0.040 | 0.383–0.541 | 0.342 | 24.5% | 80.2% | 0.047 | 8.415 | | E2 | 0.667 | 0.037 | 0.594–0.740 | < 0.001 | 92.6% | 37.9% | 0.305 | 28.645 | | Combined Diagnosis | 0.980 | 0.007 | 0.965–0.994 | < 0.001 | 94.7% | 90.5% | 0.852 | 0.304 | 3.6. Comparison of Diagnostic Efficacy Between ANGPTL8 Combined With AMH and ANGPTL8 Combined With Sex Hormones for PCOS The results demonstrated that the combination of serum ANGPTL8 with sex hormones exhibited significantly higher diagnostic efficacy for PCOS compared to its combination with AMH. Detailed comparisons are illustrated in Figure 3 and summarized in Tables 4 and 5. 4. Discussion This study is the first to report a negative association between serum ANGPTL8 levels and PCOS in a Chinese population, and systematically evaluates the diagnostic value of ANGPTL8 alone and in combination with AMH and sex hormones for PCOS. The key findings include: (1) significantly lower serum ANGPTL8 levels in PCOS patients compared to healthy controls; (2) confirmation that ANGPTL8 is an independent protective factor for PCOS; (3) demonstration of moderate‐to‐high diagnostic performance of ANGPTL8 for PCOS (AUC = 0.881); and (4) innovative proposal and validation of diagnostic models combining ANGPTL8 with AMH or sex hormones, which significantly improve diagnostic accuracy (AUC reaching 0.928 and 0.980, respectively). These findings provide new insights into the pathophysiological mechanisms of PCOS and suggest the potential of ANGPTL8 as a novel clinical diagnostic biomarker. The following discussion will delve into its mechanistic relevance, clinical value, and similarities and differences with previous research. The ANGPTL family plays a key role in metabolic regulation, but existing research shows significant disagreement regarding the specific function of ANGPTL8 in PCOS. Our results indicate that serum ANGPTL8 levels are significantly decreased in PCOS patients and that it is an independent protective factor for PCOS, which aligns with the findings of Eksi Haydardedeoglu et al. [16] who also reported lower serum betatrophin (ANGPTL8) in PCOS patients. However, this finding contradicts the conclusions summarized in the systematic review by Navaeian et al. and the meta‐analysis by Varikasuvu et al., which indicate elevated circulating ANGPTL8 levels in PCOS patients, positively correlated with IR and hyperandrogenemia [15, 18]. Similarly, Wu et al.'s genetic polymorphism study found that PCOS patients carrying the T risk allele of ANGPTL8 rs2278426 had higher serum ANGPTL8 levels, associated with more severe metabolic disorder phenotypes [19]. These discrepancies may stem from several factors. First, PCOS is a highly heterogeneous disease encompassing various metabolic and endocrine phenotypes. The PCOS patients included in our study had an average BMI of 24.97 kg/m2, falling into the overweight category, whereas Wu et al.'s study highlighted the significance of elevated ANGPTL8 levels and genotype effects in the obese subgroup [19]. Navaeian's review also notes that ANGPTL8 expression is strongly regulated by nutritional status, obesity, and IR [7]. Therefore, differences in BMI distribution, IR severity, and the composition of PCOS phenotypes (A–D) across study populations may lead to divergent trends in ANGPTL8 levels. Second, ELISA kits from different manufacturers may vary in their recognition of different isoforms or fragments of the ANGPTL8 protein, leading to inconsistent measurements. Gómez‐Ambrosi et al. [20] pointed out poor correlation between results from different commercial ELISA kits for ANGPTL8, possibly due to recognition of different immunoreactive forms or fragments. This methodological heterogeneity complicates cross‐study comparisons and underscores the need for standardization using validated reference methods or harmonized protocols. Finally, from a biological mechanism perspective, the observed decrease in ANGPTL8 levels in this study may interact with the metabolic disturbances characteristic of PCOS. It is known that ANGPTL8 regulates lipid metabolism by inhibiting lipoprotein lipase (LPL) activity [21]. Reduced ANGPTL8 levels may enhance LPL activity, potentially exacerbating the dyslipidemia common in PCOS [22]. Furthermore, ANGPTL8 influences insulin sensitivity via the PI3K/Akt signaling pathway and immunomodulation of adipose tissue [10, 23]. Therefore, decreased ANGPTL8 may reflect or contribute to the IR characteristic of PCOS. However, these mechanistic links still require experimental validation in appropriate models. Additionally, basic research suggests that ANGPTL8 may play a “double‐edged sword” role in glucose and lipid metabolism. Short‐term or physiological increases may improve insulin sensitivity by activating the PI3K/Akt pathway [10, 23, 24, 25, 26]; however, under chronic metabolic stress (e.g., sustained hyperglycemia, obesity), excessive ANGPTL8 expression might worsen IR by exacerbating hepatic gluconeogenesis and lipid accumulation [27, 28, 29, 30, 31]. The decreased ANGPTL8 levels observed in this study may reflect a compensatory downregulation in a specific metabolic context or be related to dysregulated feedback in the liver‐adipose axis. Despite the controversy regarding the direction of change, our study, along with the research by Navaeian and Wu, affirms the core association between ANGPTL8 and metabolic disturbances in PCOS. Wu et al. found that the risk genotype (CT + TT) of rs2278426 was associated with higher BMI, TG, HOMA‐IR, and ANGPTL8 levels [19], supporting from a genetic standpoint the importance of ANGPTL8 in PCOS metabolic complications. Integrating our results, we speculate that ANGPTL8 dysregulation (whether elevated or decreased) may be a critical node in the metabolic dysregulation network of PCOS, with the specific direction of change potentially depending on the disease stage, predominant phenotype, and individual genetic background. The current clinical diagnosis of PCOS still relies on the Rotterdam criteria and lacks a single, highly efficient biomarker. Traditional indicators such as AMH and the LH/FSH ratio are widely used but have limited diagnostic sensitivity or specificity [7]. This study is the first to systematically evaluate the diagnostic performance of ANGPTL8 for PCOS, finding an AUC of 0.881 for its standalone use, significantly superior to AMH alone (AUC = 0.752), demonstrating its potential as a novel biomarker. The combined diagnostic models proposed in this study show significant advantages. The combination of ANGPTL8 with sex hormones yielded the highest diagnostic value (AUC = 0.980). This aligns with the high specificity of LH for PCOS diagnosis in clinical practice. Although LH alone showed excellent diagnostic performance (AUC = 0.941), the addition of ANGPTL8 further improved the model's comprehensive sensitivity (from 85.1% to 94.7%) and achieved a better balance without significantly compromising specificity (Youden index increased from 0.817 to 0.852). Notably, the combination of ANGPTL8 with LH alone also performed excellently (AUC = 0.979), confirming the efficacy of LH as a key diagnostic marker. In contrast, the diagnostic performance of the ANGPTL8‐AMH combination (AUC = 0.928), while slightly lower than the combination with sex hormones, remained significantly superior to any single indicator. Our strategic decision to highlight the ANGPTL8‐AMH model alongside others was based on the aim to explore combinations spanning different pathophysiological domains (metabolism and ovarian reserve) rather than combining markers within the same hormonal axis. Furthermore, the stability of AMH throughout the menstrual cycle, compared to the cycle‐dependent variability of LH, offers practical advantages for clinical testing schedules. Our results support the view that “multi‐marker panels can improve PCOS diagnostic performance” [32], providing direction for developing more precise laboratory diagnostic tools. This study has several limitations. First, as a single‐center study with a relatively small sample size (n = 210) and a population from a single region in Northern China, the generalizability of the conclusions requires caution. Second, although standardized ELISA methods were used to measure ANGPTL8, differences between kits may affect result comparability and cutoff value determination, necessitating validation through multicenter studies using standardized reagents. Third, the cross‐sectional design precludes establishing a causal relationship between ANGPTL8 level changes and PCOS onset. Finally, the lack of mechanistic studies (e.g., animal or cellular models) limits the ability to elucidate the specific pathways through which ANGPTL8 exerts protective or detrimental effects in PCOS. Future research should: (1) conduct large‐sample, multicenter prospective cohort studies to validate the diagnostic value of ANGPTL8 and establish population‐specific reference ranges; (2) utilize gene‐edited animal models or cell experiments to explore the direct impact of ANGPTL8 on PCOS features (e.g., abnormal folliculogenesis, IR, hyperandrogenism) in both ovarian and systemic metabolism; and (3) track dynamic changes in ANGPTL8 before and after interventional therapies to assess its potential as a predictor or monitoring indicator for treatment efficacy. In summary, this study found that serum ANGPTL8 levels are significantly negatively associated with PCOS and that it is an independent protective factor for PCOS. Although discrepancies exist with some previous studies, likely due to differences in populations, methodologies, and disease heterogeneity, this further underscores the important role of ANGPTL8 in the complex metabolic network of PCOS. ANGPTL8 shows good potential for PCOS diagnosis, and its combined application with existing biomarkers can significantly enhance diagnostic accuracy. These results provide new clues and foundations for exploring the pathological mechanisms of PCOS and optimizing clinical diagnostic strategies. Author Contributions Yanli Wang: writing – original draft, writing – review and editing. Chongbo Du: writing – original draft, methodology. Xuejia Guo: validation. Xiaowen Ma: formal analysis. Yanping Feng: data curation. Yuxiao Zhang: conceptualization, writing – review and editing. Yandong Deng: data curation. Funding This work was supported by the Medical Science Research Project of Hebei Province (Grant No. 20231034). Ethics Statement This study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of The First Hospital of Hebei Medical University (Approval No. 20220709). Consent Written informed consent was obtained from all participants involved in the study. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Acknowledgments The authors have nothing to report. Data Availability Statement The data that supports the findings of this study are available in the Supporting Information of this article.

- 1. Azziz R., “Polycystic Ovary Syndrome,” Obstetrics & Gynecology 132, no. 2 (2018): 321–336. [DOI] [PubMed] [Google Scholar] - 2. Liu J., Wu Q., Hao Y., et al., “Measuring the Global Disease Burden of Polycystic Ovary Syndrome in 194 Countries: Global Burden of Disease Study 2017,” Human Reproduction 38, no. 6 (2023): 1125–1135. [DOI] [PMC free article] [PubMed] [Google Scholar] - 3. Liu L. J., Shao T., Fang Y., Zhu H. H., Song X. Q., and Wei J. C., “Analysis of the Incidence Trend of Polycystic Ovary Syndrome in China From 1990 to 2019,” Chinese Journal of Preventive Medicine 24, no. 3 (2023): 259–263. [Google Scholar] - 4. Rotterdam ESHRE/ASRM‐Sponsored PCOS Consensus Workshop Group , “Revised 2003 Consensus on Diagnostic Criteria and Long‐Term Health Risks Related to Polycystic Ovary Syndrome (PCOS),” Human Reproduction 19, no. 1 (2004): 41–47. [DOI] [PubMed] [Google Scholar] - 5. Homburg R., Ray A., Bhide P., Gudi A., Shah A., and Timms P., “The Relationship of Serum Anti‐Müllerian Hormone With Polycystic Ovarian Morphology and Polycystic Ovary Syndrome: A Prospective Cohort Study,” Human Reproduction 28, no. 4 (2013): 1077–1083. [DOI] [PubMed] [Google Scholar] - 6. Diamanti‐Kandarakis E. and Dunaif A., “Insulin Resistance and the Polycystic Ovary Syndrome Revisited: An Update on Mechanisms and Implications,” Endocrine Reviews 33, no. 6 (2012): 981–1030. [DOI] [PMC free article] [PubMed] [Google Scholar] - 7. Legro R. S., Arslanian S. A., Ehrmann D. A., et al., “Diagnosis and Treatment of Polycystic Ovary Syndrome: An Endocrine Society Clinical Practice Guideline,” Journal of Clinical Endocrinology & Metabolism 98, no. 12 (2013): 4565–4592. [DOI] [PMC free article] [PubMed] [Google Scholar] - 8. Teede H. J., Misso M. L., Costello M. F., et al., “Recommendations From the International Evidence‐Based Guideline for the Assessment and Management of Polycystic Ovary Syndrome,” Fertility and Sterility 110, no. 3 (2018): 364–379. [DOI] [PMC free article] [PubMed] [Google Scholar] - 9. Chai Y. H., Yang P. P., and Zhang X. H., “Mechanism of Gut Microbiota in Insulin Resistance of Polycystic Ovary Syndrome,” Journal of International Reproductive Health/Family Planning 40, no. 4 (2021): 334–338. [Google Scholar] - 10. Zhang Z., Wu H., Dai L., Li J., and Chen Y., “ANGPTL8 Enhances Insulin Sensitivity by Directly Activating Insulin‐Mediated AKT Phosphorylation,” Gene 749 (2020): 144707. [DOI] [PubMed] [Google Scholar] - 11. Vatannejad A., Salimi F., Moradi N., Fadaei R., and Alvandi E., “Evaluation of Angiopoietin‐Like Protein 3 (ANGPTL3) Levels in Polycystic Ovary Syndrome,” Life Sciences 263 (2020): 118595. [DOI] [PubMed] [Google Scholar] - 12. Günes M. and Bukan N., “Examination of Angiopoietin‐Like Protein 4, Neuropeptide Y, Omentin‐1 Levels of Obese and Non‐Obese Patients With Polycystic Ovary Syndrome,” Gynecological Endocrinology 31, no. 11 (2015): 903–906. [DOI] [PubMed] [Google Scholar] - 13. Ying X. J., Yu L., Wang Y., Li Y., and Zhang L., “Serum Angptl2 and Angptl4 Levels in Patients With Polycystic Ovary Syndrome: Associations With Obesity and Glucose‐Lipid Metabolism,” Chinese Journal of Family Planning 27, no. 8 (2019): 1069–1072. [Google Scholar] - 14. Elci E., Kaya C., Cim N., Yildizhan R., and Sahin H., “Evaluation of Cardiac Risk Marker Levels in Obese and Non‐Obese Patients With Polycystic Ovaries,” Gynecological Endocrinology 33, no. 1 (2017): 43–47. [DOI] [PubMed] [Google Scholar] - 15. Varikasuvu S. R., Panga J. R., Satyanarayana M. V., and Varshney S., “Circulating Angiopoietin‐Like 8 Protein (ANGPTL8/Betatrophin) in Patients With Polycystic Ovary Syndrome: A Systematic Review and Meta‐Analysis,” Gynecological Endocrinology 35, no. 3 (2019): 190–197. [DOI] [PubMed] [Google Scholar] - 16. Eksi Haydardedeoglu F., Bagir G. S., Haydardedeoglu B., Kilic N., and Kilic S., “Serum Betatrophin Levels Are Reduced in Patients With Full‐Blown Polycystic Ovary Syndrome,” Gynecological Endocrinology 35, no. 3 (2019): 224–227. [DOI] [PubMed] [Google Scholar] - 17. Swets J. A., “Measuring the Accuracy of Diagnostic Systems,” Science 240, no. 4857 (1988): 1285–1293. [DOI] [PubMed] [Google Scholar] - 18. Navaeian M., Asadian S., Ahmadpour Yazdi H., and Gheibi N., “ANGPTL8 Roles in Proliferation, Metabolic Diseases, Hypothyroidism, Polycystic Ovary Syndrome, and Signaling Pathways,” Molecular Biology Reports 48, no. 4 (2021): 3719–3731, 10.1007/s11033-021-06270-8. [DOI] [PubMed] [Google Scholar] - 19. Wu H., Wang H., Sun L., et al., “Association Between rs2278426 Polymorphism of the ANGPTL8 Gene and Polycystic Ovary Syndrome,” Diabetes, Metabolic Syndrome and Obesity: Targets and Therapy 17 (2024): 1749–1760, 10.2147/DMSO.S455274. [DOI] [PMC free article] [PubMed] [Google Scholar] - 20. Gómez‐Ambrosi J., Pascual E., Catalán V., et al., “Circulating Betatrophin Concentrations Are Decreased in Human Obesity and Type 2 Diabetes,” Journal of Clinical Endocrinology and Metabolism 99, no. 10 (2014): E2004–E2009, 10.1210/jc.2014-1568. [DOI] [PubMed] [Google Scholar] - 21. Zhang R., Abou‐Samra A. B., and Lin J. D., “ANGPTL8 in Lipid Metabolism and Metabolic Diseases,” Journal of Lipid Research 61, no. 8 (2020): 1209–1217. [Google Scholar] - 22. Carmina E., Oberfield S. E., and Lobo R. A., “Obesity and PCOS: Implications for Diagnosis and Treatment,” Endocrine Practice 25, no. 6 (2019): 611–619. [Google Scholar] - 23. Luo D., Chen X., Yang W., Li Y., Zhang Y., and Wang L., “Angiopoietin‐Like 8 Improves Insulin Resistance and Attenuates Adipose Tissue Inflammation in Diet‐Induced Obese Mice,” Experimental and Clinical Endocrinology & Diabetes 128, no. 5 (2020): 290–296. [DOI] [PubMed] [Google Scholar] - 24. Hao Q., Zheng A., Zhang H., Li J., Wang Y., and Liu Y., “Down‐Regulation of Betatrophin Enhances Insulin Sensitivity in Type 2 Diabetes Mellitus Through Activation of the GSK‐3β/PGC‐1α Signaling Pathway,” Journal of Endocrinological Investigation 44, no. 9 (2021): 1857–1868. [DOI] [PubMed] [Google Scholar] - 25. Zhao Z., Deng X., Jia J., Li S., Zhang Q., and Wang H., “Angiopoietin‐Like Protein 8 (Betatrophin) Inhibits Hepatic Gluconeogenesis Through PI3K/Akt Signaling Pathway in Diabetic Mice,” Metabolism 126 (2022): 154921. [DOI] [PubMed] [Google Scholar] - 26. Chen Y. Q., Pottanat T. G., Siegel R. W., Jansson L., and Lagerstedt J. O., “Angiopoietin‐Like Protein 4 (E40K) and ANGPTL4/8 Complex Have Reduced, Temperature‐Dependent LPL‐Inhibitory Activity Compared to ANGPTL4,” Biochemical and Biophysical Research Communications 534 (2021): 498–503. [DOI] [PubMed] [Google Scholar] - 27. Foroozanfard F., Rafiei H., Samimi M., Ghadimi R., Hosseinpanah F., and Azizi F., “The Effects of DASH Diet on Weight Loss, Anti‐Müllerian Hormone and Metabolic Profiles in Women With Polycystic Ovary Syndrome: A Randomized Clinical Trial,” Clinical Endocrinology 87, no. 1 (2017): 51–58. [DOI] [PubMed] [Google Scholar] - 28. Wang J., Zhang Y., Wang W. D., et al., “The Effect of Anti‐Müllerian Hormone Concentration on Pregnancy Outcomes in Infertile Patients,” Chinese Journal of Maternal and Child Health Care 32, no. 23 (2017): 5959–5961. [Google Scholar] - 29. Greenwood E. A., “Anti‐Müllerian Hormone as a Biomarker of Metabolic Parameters in Polycystic Ovary Syndrome,” Fertility and Sterility 109, no. 3 (2018): e7–e8. [Google Scholar] - 30. Tata B., Mimouni N. E. H., Barbotin A. L., et al., “Elevated Prenatal Anti‐Müllerian Hormone Reprograms the Fetus and Induces Polycystic Ovary Syndrome in Adulthood,” Nature Medicine 24, no. 6 (2018): 1298–1302. [DOI] [PMC free article] [PubMed] [Google Scholar] - 31. Zoltán K. I., Ildikó K., and Melinda K., “Current Aspects of Polycystic Ovary Syndrome I: Definition, Pathophysiology, Clinical Manifestations, Diagnosis and Complications,” Bulletin of Medical Sciences 91, no. 1 (2018): 5–18. [Google Scholar] - 32. Joham A. E., Norman R. J., Stener‐Victorin E., et al., “Polycystic Ovary Syndrome and the Risk of Type 2 Diabetes,” Nature Reviews Endocrinology 11, no. 10 (2015): 552–553. [Google Scholar] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Data Availability Statement The data that supports the findings of this study are available in the Supporting Information of this article.