Decoding the potential association of liver steatosis with female infertility: an analysis from NHANES 2013-2018.

NHANES (National Health and Nutrition Examination Survey) is a national survey led by the National Center for Health Statistics (NCHS) under the Centers for Disease Control and Prevention (CDC) [ 16 ]. Its core objective is to assess the health and nutritional status of the U.S. population and to provide a scientific basis for public health policymaking and disease prevention. NHANES is an annual continuous survey that includes approximately 5,000 participants each year, covering all age, racial, and gender groups to ensure that the data are representative. The subjective reports, including demographic characteristics, health history and lifestyle behaviors, and objective measures, including physiological measurements and laboratory tests, of the participants are collected through home interviews and Mobile Examination Center (MEC) physical examinations. Ethical approval for the survey was obtained from the Ethics Review Committee of the National Center for Health Statistics. Written informed consent was provided by all participants in the study. Given that the NHANES data used in this study are publicly available and accessible at no cost, additional approval from an Institutional Review Board is not required for this research. The dataset can be freely obtained from the following website: https://wwwn.cdc.gov/nchs/nhanes/Default.aspx . For this research, we extracted data from the NHANES database covering four survey cycles from 2013 to 2018, which included a total of 29,400 participants. On the basis of the exclusion criteria in Fig.  1 , unqualified participants were excluded from the study: [ 1 ] females under 20 years of age or over 44 years of age and male ( n  = 25,693), [ 2 ] participants with incomplete infertility data ( n  = 586), [ 3 ] participants with incomplete data on FLI, HSI and ZJU index ( n  = 1,817), [ 4 ] participants with missing data on important covariates ( n  = 103), [ 5 ] participants with incomplete weight data or a weight equal to 0 ( n  = 34). Consequently, the final study population included 1,167 women aged 20 to 44 years who had complete data available. Fig. 1 The flow chart of inclusion and exclusion criteria in the study The flow chart of inclusion and exclusion criteria in the study The Reproductive Health Questionnaire (RHQ section) in NHANES is a number of female-only questions about reproductive health including menstrual history, pregnancy history, and other relevant reproductive conditions. Female infertility is assessed from two questions about pregnancy history (questionnaire RHQ-074 and questionnaire RHQ-076) [ 17 ]. The questionnaire RHQ-074 asks: “Have you tried to conceive in the last 12 months but were unsuccessful?” The questionnaire RHQ-076 asks: “Have you visited a doctor or sought help from another healthcare professional because of your failure to conceive?” The participants responded to both questionnaires in four ways: “yes”, “no”, “refused to answer” and “don’t know”. The women who answered “refused” and “don’t know” were excluded from the study, as were women who did not participate in the questionnaires. The women who answered “yes” were considered to be suffering from infertility, while the women who answered “no” were considered to be non-infertile. We used FLI, HSI and ZJU index to assess liver steatosis. The larger values of these indices typically indicate a higher degree of fat deposition in the liver and a greater degree of liver steatosis [ 18 ]. These indices provide a non-invasive method for clinical assessment and monitoring of the progression of fatty liver. They are composite indicators that combine physical measurements and laboratory tests, and are calculated as follows. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm{FLI}=\frac{\mathrm e^{0.953\ast In\mathit(TG\mathit)\mathit+\mathit0\mathit.\mathit{139}\mathit\ast BMI\mathit+\mathit0\mathit.\mathit{718}\mathit\ast In\mathit(GGT\mathit)\mathit+\mathit0\mathit.\mathit{053}\mathit\ast WC\mathit-\mathit{15}\mathit.\mathit{745}}}{1+\mathrm e^{0.953\ast\mathrm{In}\mathit(\mathrm{TG}\mathit)\mathit+\mathit0\mathit.\mathit{139}\mathit\ast\mathrm{BMI}\mathit+\mathit0\mathit.\mathit{718}\mathit\ast\mathrm{In}\mathit(\mathrm{GGT}\mathit)\mathit+\mathit0\mathit.\mathit{053}\mathit\ast\mathrm{WC}\mathit-\mathit{15}\mathit.\mathit{745}}}\ast100^{\left(\mathrm a\right)}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm{HSI}=8\ast\frac{\mathrm{ALT}}{\mathrm{AST}}+\mathrm{BMI}+\mathrm\alpha^{\mathrm{Diabetes}}+\mathrm\beta^{\mathrm{Gender}}\left(\mathrm b\right)$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm{ZJU}\;\mathrm{index}=\mathrm{BMI}+\mathrm{FPG}+3\ast\frac{\mathrm{ALT}}{\mathrm{AST}}+\mathrm{TG}+\mathrm\beta^{\mathrm{Gender}}\left(\mathrm c\right)$$\end{document} (a) TG: triglyceride (mg/dL); BMI: body mass index (kg/m 2 ); GGT: Gamma Glutamyl Transferase (IU/L); WC: waist circumference (cm). (b) ALT: Alanine Aminotransferase (U/L); AST: Aspartate Aminotransferase (U/L); BMI: body mass index (kg/m 2 ). α Diabetes is a coefficient related to the state of diabetes. If an individual has diabetes, α Diabetes equals 2; if not, α Diabetes equals 0. β Gender is a gender-related coefficient. If the individual is female, β Gender equals 2; if the individual is male, β Gender equals 0. (c) BMI: body mass index (kg/m 2 ); FPG: fasting plasma glucose (mmol/L); ALT: Alanine Aminotransferase (U/L); AST: Aspartate Aminotransferase (U/L); TG: triglyceride (mmol/L). β Gender is a gender-related coefficient. If the individual is female, β Gender equals 2; if the individual is male, β Gender equals 0. Additional, these participants were categorized into fatty and non-fatty livers based on the fatty liver cut-off values for each index. According to Bedogni et al.’s research, FLI ≥ 60 is regarded as an effective cut-off value for indicating a higher possibility of having fatty liver [ 19 ]. Lee et al. demonstrated that HSI > 36 has a good screening performance for fatty liver, which is suitable for the assessment of fatty liver in the general population [ 20 ]. Wang et al.’s research verified the effectiveness of the ZJU index in evaluating non-alcoholic fatty liver disease. ZJU > 38 was determined as the appropriate critical value for differentiating the presence of fatty liver [ 21 ]. The potential covariates considered in this study included demographic characteristics, body measurements, lifestyle, and medical history, which included age, race, education level, marital status, poverty-to-income ratio (PIR), body mass index (BMI, kg/m 2 ), waist circumference (WC, cm), smoking status, alcohol consumption, physical activity, hypertension, diabetes mellitus, age at first menstruation, menstrual regularity, and pelvic infection disease. Age was categorized as 20–34 and 35–44 using 35 years as the cut-off point, and poverty was defined as a PIR of less than 1. Obesity was categorized as normal or low weight (BMI < 25 kg/m 2 ), overweight (25 kg/m 2 ≤ BMI < 30 kg/m 2 ), and obese (BMI ≥ 30 kg/m 2 ). Smoking status was categorized as never, current and quit using a cut-off of 100 cigarettes to date. Alcohol consumption was considered to be greater than or equal to 12 times drinks in the past 12 months. Participants who had engaged in moderate or vigorous exercise, fitness, or recreational activities within 1 week were considered physically active. The diagnosis of hypertension depended on meeting any of the following: self-reported hypertension, currently taking antihypertensive medication, average of systolic blood pressure greater than or equal to 140 mmHg and average of diastolic blood pressure greater than or equal to 90 mmHg after three or four consecutive measurements. Diabetes mellitus was defined by meeting any of the following: self-reported history of diabetes mellitus, currently taking glucose-lowering medication or insulin injections, glycosylated hemoglobin greater than or equal to 6.5%, and FPG ≥ 126 mg/dL [ 22 ]. Age at menarche was categorized as < 15 years and ≥ 15 years according to the questionnaire RHQ-010: “How old were you when you had your first period?” A participant was considered to be menstruating regularly if she had at least one period in the past 12 months, excluding bleeding from other causes (an affirmative answer to questionnaire RHQ-031). Pelvic infection disease meant that she had been treated for a tubal, uterine, or ovarian infection in the past (RHQ-078). This study strictly followed the analysis guidelines for NHANES data issued by the NCHS of the United States. To correct for complex sampling designs and non-response bias, all analyses used the sampling weights provided by NCHS. When choosing weights, the principle of “least common denominator” should be followed, and the fasting weights wtsaf2 year for specific sub-samples should be adopted. The baseline characteristics of the study participants were delineated on the basis of infertility status. The continuous variables were examined via weighted Mann-Whitney U tests and are expressed as medians and quartiles. The categorical variables were compared via weighted Chi-square test and are described as number (n) and percentage (%). The weighted multivariate logistic regression analysis was employed to examine the association of FLI, HSI and ZJU with female infertility across three distinct models, each with a different set of covariates for adjustments. Model 1 adjusted for no variables. Model 2 adjusted for age (categorization), race, education level, poverty and marital status. Model 3 adjusted for age (categorization), race, education level, poverty and marital status, smoking status, alcohol consumption, physical activity, menstrual regularity, pelvic infection disease and age at menarche. The analysis results are presented in terms of the odds ratios (OR) along with its 95% confidence interval (95%CI). The dose-response relationship between these liver steatosis and female infertility was depicted via a weighted restricted cubic spline (RCS) curve in the fully adjusted Model 3. The weighted subgroup analysis was conducted to assess potential factors that could affect the association between liver steatosis and female infertility in Model 3. A cross-product term was included in the weighted logistic regression models to explore the interactions tests with potential modification effects. Furthermore, we identified the fatty liver population among these female participants based on the fatty liver cutoff values for each of these three indices. Multivariate logistic regression analysis was applied to explore the association between fatty liver and female infertility in the same three model. Eventually, we plotted the receiver operating characteristic (ROC) curves to examine the predictability of liver steatosis for female infertility. The area under the curve (AUC) and the delong test were used to compare the differences between them. To determine whether the association between hepatic steatosis indicators and infertility is independent of overall obesity, we conducted a sensitivity analysis. Based on the fully adjusted Model 3, we separately included BMI and WC as covariates to re-evaluate the association between FLI, HSI and ZJU index and female infertility. By comparing the results before and after adjustment, we can better understand the role of obesity in this association. In this study, we utilized the R software environment to carry out the statistical analyses ( https://www.r-project.org/ ; version 4.3.3). The “survey” package is used for complex sampling design, the “svyglm” function for analyzing weighted multivariate logistic regression and subgroup analysis, the “ggplot2” package for plotting ROC plots, the “forestplotter” and “ggplot2” package for plot forest plots, and the “rms” package for plotting RCS curves. A P -value of less than 0.05 from a two-tailed test was considered to indicate statistical significance.

A total of 1,167 women aged 20 to 44 were included in this study, representing 43,669,098 women of the same age in the United States. Among them, 161 (13.80%) women suffer from infertility, representing 6,155,882 infertile women in the United States (Supplementary Table S1 ). The weighted prevalence and quartile range of infertility were 14.10% (11.47%, 16.73%). Table  1 summarizes the weighted baseline characteristics of the study population. The median age of all women is 32 years old, and the majority are concentrated in the range of 20–34 years old. Among these women, white people account for the majority, they often receive college education and above. Most of them are married or cohabiting, and a few are economically poor. Most of them are obese, most of whom have never smoked or no-drinking. More than half of them participate in physical activities. A few people suffer from hypertension or diabetes. Most of them have menarche before the age of 15 years. Nearly 90% of them have regular menstruation, and very few have a history of pelvic infection. Compared with women without infertility, infertile women tend to be older, married or cohabiting, have a higher BMI and WC, do not drink alcohol, have diabetes and hypertension, as well as higher FLI, HSI and ZJU index. Table 1 The weighted baseline characteristics of female participants by infertility status Variables Total Infertility p-value (N =  1167 ) No (N =  1006 ) Yes (N =  161 ) Age (years) 32.0 (25.0, 38.0) 31.0 (25.0, 37.0) 34.0 (29.0, 40.0) < 0.001 Age, n(%) a 0.035  20-34 years 691 (62.3%) 614 (64.1%) 77 (51.6%)  35-44 years 476 (37.7%) 392 (35.9%) 84 (48.4%) Race, n(%) a 0.5  Mexican American 192 (11.3%) 165 (11.4%) 27 (10.8%)  Non-Hispanic White 411 (57.5%) 348 (56.6%) 63 (62.5%)  Non-Hispanic Black 231 (12.7%) 200 (12.9%) 31 (11.6%)  Other Race 333 (18.5%) 293 (19.0%) 40 (15.1%) Education levels, n(%) a 0.6  Below high school 178 (11.3%) 152 (10.9%) 26 (13.8%)  High school graduation 221 (20.0%) 191 (20.2%) 30 (18.9%)  College or above 768 (68.7%) 663 (68.9%) 105 (67.3%) PIR 2.2 (1.1, 4.0) 2.2 (1.1, 4.0) 2.4 (1.2, 4.7) 0.3 Poverty, n(%) a 0.2  Yes 324 (22.4%) 288 (23.0%) 36 (18.4%)  No 843 (77.6%) 718 (77.0%) 125 (81.6%) Marital status, n(%) a < 0.001  Married/living with partner 672 (59.9%) 554 (56.7%) 118 (79.2%)  Widowed/divorced/separated 134 (10.6%) 115 (10.6%) 19 (10.4%)  Never married 361 (29.6%) 337 (32.7%) 24 (10.4%) BMI (kg/m²) 27.9 (23.3, 34.5) 27.3 (23.1, 33.6) 32.5 (25.0, 38.7) < 0.001 BMI, n(%) a < 0.001  < 25 kg/m 2 405 (34.7%) 361 (36.3%) 44 (24.9%)  25–30 kg/m 2 280 (23.6%) 258 (25.1%) 22 (14.4%)  ≥ 30 kg/m 2 482 (41.7%) 387 (38.6%) 95 (60.7%) WC (cm) a 93.9 (82.0, 107.2) 92.5 (81.4, 106.2) 101.1 (89.7, 117.3) < 0.001 Smoking status, n(%) a 0.7  Never 814 (65.7%) 706 (65.6%) 108 (66.0%)  Current 204 (19.3%) 177 (19.7%) 27 (16.8%)  Quit 149 (15.1%) 123 (14.7%) 26 (17.1%) Alcohol consumption, n(%) a 0.048  No 544 (44.4%) 455 (42.9%) 89 (53.4%)  Yes 442 (43.8%) 387 (44.5%) 55 (39.7%)  Unknown 181 (11.8%) 164 (12.6%) 17 (7.0%) Physical Activity, n(%) a 0.3  Inactive 509 (41.0%) 431 (40.0%) 78 (46.6%)  Active 658 (59.0%) 575 (60.0%) 83 (53.4%) Diabetes, n(%) a 0.004  No 1088 (94.1%) 944 (95.0%) 144 (88.2%)  Yes 79 (5.9%) 62 (5.0%) 17 (11.8%) Hypertension, n(%) a < 0.001  No 984 (84.8%) 869 (87.4%) 115 (69.2%)  Yes 183 (15.2%) 137 (12.6%) 46 (30.8%) Age at menarche, n(%) a 0.3  < 15 years 1020 (87.7%) 875 (87.3%) 145 (90.3%)  ≥ 15 years 147 (12.3%) 131 (12.7%) 16 (9.7%) Menstrual regularity, n(%) a 0.031  No 110 (10.4%) 91 (9.1%) 19 (18.5%)  Yes 1057 (89.6%) 915 (90.9%) 142 (81.5%) Pelvic infection disease, n(%) a 0.031  No 1117 (95.9%) 968 (96.4%) 149 (92.7%)  Yes 50 (4.1%) 38 (3.6%) 12 (7.3%) FLI 32.6 (8.7, 74.8) 29.5 (8.3, 71.7) 62.7 (14.3, 88.9) < 0.001 HSI 39.2 (33.7, 46.2) 38.4 (33.6, 45.3) 44.4 (35.9, 50.0) < 0.001 ZJU index 38.9 (33.5, 46.6) 38.1 (33.3, 45.4) 44.4 (35.3, 50.0) < 0.001 The continuous variables were expressed as median and quartiles, and the categorical variables were expressed as number and percentage PIR Poverty income ratio, BMI Body mass index, WC Waist circumference, FLI Fatty liver index, HSI Hepatic steatosis index, ZJU Zhejiang university index a Unweighted frequency counts and weighted percentages are shown The weighted baseline characteristics of female participants by infertility status The continuous variables were expressed as median and quartiles, and the categorical variables were expressed as number and percentage PIR Poverty income ratio, BMI Body mass index, WC Waist circumference, FLI Fatty liver index, HSI Hepatic steatosis index, ZJU Zhejiang university index a Unweighted frequency counts and weighted percentages are shown The results of weighted multivariate logistic regression analysis demonstrate the association between liver steatosis and infertility in women aged 20–44 in Table  2 . We will analyze FLI, HSI and ZJU index as continuous and categorical variables, respectively. In the continuous analysis, we found that FLI (OR = 1.01, 95%CI: 1.01–1.02; P  < 0.001), HSI (OR = 1.04, 95%CI: 1.02–1.07; P  < 0.001) and ZJU index (OR = 1.04, 95%CI: 1.02–1.06; P  < 0.001) were significantly and positively associated with the prevalence of female infertility in the unadjusted Model 1. The association of these indices with female infertility remained robust in the adjusted Model 2 and Model 3. We transformed these indices from continuous variables to quartile categorical variables for trend test analysis. Compared to the lowest quartile group (Q1), these women in the highest quartile group (Q4) of FLI (OR = 2.35, 95%CI: 1.10–4.99; P  = 0.028), HSI (OR = 2.35, 95%CI: 1.20–4.61; P  = 0.015), and ZJU index (OR = 2.73, 95%CI: 1.31–5.67; P  = 0.009) may have a higher risk of infertility prevalence in Model 3. The prevalence of female infertility increases with the increase of quartiles of these indices (all P for trend < 0.05). Table 2 The weighted multivariate logistic regression analysis between liver steatosis and infertility Variables Count, n Model 1 Model 2 Model 3 OR(95%CI) P -value OR(95%CI) P -value OR(95%CI) P -value FLI 1167 1.01(1.01,1.02) < 0.001 1.01(1.01,1.02) < 0.001 1.01(1.00,1.02) 0.003  Q1(0.69,8.37) 292 Ref Ref Ref  Q2(8.37,33.11) 292 0.70(0.34,1.46) 0.332 0.58(0.27,1.25) 0.161 0.57(0.26,1.24) 0.147  Q3(33.11,75.01) 291 1.74(0.95,3.19) 0.074 1.44(0.73,2.82) 0.280 1.41(0.69,2.90) 0.337  Q4(75.01,100) 292 2.67(1.45,4.92) 0.002 2.40(1.20,4.77) 0.014 2.35(1.10,4.99) 0.028  P for trend < 0.001 0.002 0.005 HSI 1167 1.04(1.02,1.07) < 0.001 1.04(1.02,1.07) 0.001 1.04(1.01,1.07) 0.005  Q1(23.20,33.70) 294 Ref Ref Ref  Q2(33.70,39.44) 290 0.63(0.32,1.23) 0.170 0.50(0.25,1.01) 0.053 0.49(0.24,0.98) 0.045  Q3(39.44,46.09) 291 1.26(0.70,2.24) 0.432 1.04(0.55,1.97) 0.895 0.96(0.50,1.87) 0.904  Q4(46.09,82.10) 292 2.66(1.49,4.73) 0.001 2.36(1.27,4.40) 0.008 2.35(1.20,4.61) 0.015  P for trend < 0.001 0.001 0.004 ZJU 1167 1.04(1.02,1.06) < 0.001 1.04(1.02,1.07) 0.001 1.04(1.01,1.07) 0.004  Q1(24.77,33.53) 292 Ref Ref Ref  Q2(33.53,39.09) 292 0.90(0.44,1.85) 0.761 0.70(0.32,1.54) 0.369 0.66(0.30,1.45) 0.288  Q3(39.09,46.42) 291 1.36(0.71,2.62) 0.344 1.11(0.54,2.27) 0.762 1.07(0.50,2.32) 0.852  Q4(46.42,97.20) 292 3.10(1.69,5.71) < 0.001 2.72(1.38,5.37) 0.005 2.73(1.31,5.67) 0.009  P for trend < 0.001 0.001 0.004 Model 1 adjusted for no variables Model 2 adjusted for age (categorization), race, education level, poverty and marital status Model 3 adjusted for age (categorization), race, education level, poverty and marital status, smoking status, alcohol consumption, physical activity, menstrual regularity, pelvic infection disease and age at menarche The weighted multivariate logistic regression analysis between liver steatosis and infertility Model 1 adjusted for no variables Model 2 adjusted for age (categorization), race, education level, poverty and marital status Model 3 adjusted for age (categorization), race, education level, poverty and marital status, smoking status, alcohol consumption, physical activity, menstrual regularity, pelvic infection disease and age at menarche The weighted restricted cubic spline (RCS) plots after fully adjusted Model 3 were used to visualize the dose-response relationship between liver steatosis and female infertility in Fig.  2 . FLI ( P for nonlinear = 0.566), HSI ( P for nonlinear = 0.858) and ZJU index ( P for nonlinear = 0.860) all exhibited significant linear relationships with infertility. Fig. 2 The dose-response relationship of FLI ( A ), HSI ( B ) and ZJU index ( C ) with female infertility. The OR (red solid lines) and 95%CI (red shaded areas) in the RCS was adjusted for age (categorization), race, education level, poverty and marital status, smoking status, alcohol consumption, physical activity, menstrual regularity, pelvic infection disease and age at menarche The dose-response relationship of FLI ( A ), HSI ( B ) and ZJU index ( C ) with female infertility. The OR (red solid lines) and 95%CI (red shaded areas) in the RCS was adjusted for age (categorization), race, education level, poverty and marital status, smoking status, alcohol consumption, physical activity, menstrual regularity, pelvic infection disease and age at menarche We performed subgroup analyses and interactions for stratified variables such as age, race, education level, economic status, marital status, abdominal obesity (WC > 88 cm) [ 23 ]smoking status, alcohol consumption, physical activity, diabetes mellitus, hypertension, and age at menarche in order to observe differences in the association between liver steatosis and female infertility in populations with different characteristics in Fig.  3 . Subgroup analyses were adjusted for all covariates in Model 3, except for the stratification variable itself. Fig. 3 The subgroup analysis of the relationship of FLI ( A ), HSI ( B ) and ZJU index ( C ) with female infertility. Each subgroup analysis was adjusted for age (categorization), race, education level, poverty and marital status, smoking status, alcohol consumption, physical activity, menstrual regularity, pelvic infection disease and age at menarche, except for stratified variables The subgroup analysis of the relationship of FLI ( A ), HSI ( B ) and ZJU index ( C ) with female infertility. Each subgroup analysis was adjusted for age (categorization), race, education level, poverty and marital status, smoking status, alcohol consumption, physical activity, menstrual regularity, pelvic infection disease and age at menarche, except for stratified variables A positive correlation between hepatic steatosis (FLI, HSI and ZJU index) and female infertility was observed in almost all subgroups. A positive correlation between hepatic steatosis and female infertility was observed in almost all subgroups. It is worth noting that there are significant interactions in some subgroups: among them, the positive correlation between FLI and infertility is more obvious in the age ( P for interaction = 0.023), race ( P for interaction = 0.044) and physical activity ( P for interaction = 0.015) subgroups; The correlation of HSI is stronger in people with marital status ( P for interaction = 0.030) and physical activity ( P for interaction = 0.010). The correlation of the ZJU index was most significant in the physical activity ( P for interaction = 0.006) subgroup. Specifically, the association strength between FLI and infertility was more significant in the 20–34 age, Mexican Americans and non-Hispanic white, and active physical activity group. The association between HSI and infertility is more significant among married/cohabitating and widowed/divorced/separated women, as well as active physical activity people. The association between the ZJU index and infertility is more significant among active physical activity people. Multivariate logistic regression was performed to explore the association between fatty liver and female infertility in Table  3 . Different populations were defined as having fatty liver based on the respective fatty liver cutoff values for the FLI, HSI, and ZJU index. 407 women were defined as having fatty liver by the cutoff value of FLI; 728 women were defined as having fatty liver by the cutoff value of HSI; and 632 women were defined as having fatty liver by the ZJU index. In Model 1, women with fatty liver defined by the FLI (OR = 2.31, 95%CI: 1.51–3.54; P  < 0.001), HSI (OR = 2.01, 95%CI: 1.26–3.21; P  = 0.005) and ZJU index (OR = 2.46, 95%CI: 1.57–3.86; P  < 0.001) all have a significantly higher risk of infertility than women without fatty liver. The association between fatty liver defined by HSI and female infertility did not reach statistical significance after adjustment for Model 3 (OR = 1.74, 95%CI: 0.98–3.08; P  = 0.057). Fatty liver as defined by the FLI and ZJU index still showed a significant correlation with female infertility after adjustment for Model 2 and 3. These results suggest that FLI and ZJU index may be more reliable biomarkers for assessing the association of fatty liver with female infertility. Table 3 Weighted multivariate logistic regression analysis between fatty liver by different biomarkers and infertility Fatty liver Count,n Model 1 Model 2 Model 3 OR(95%CI) P -value OR(95%CI) P -value OR(95%CI) P -value FLI < 60 760 Ref Ref Ref FLI ≥ 60 407 2.31(1.51,3.54) < 0.001 2.40(1.53,3.77)  36 728 2.01(1.26,3.21) 0.005 1.83(1.09,3.09) 0.024 1.74(0.98,3.08) 0.057 ZJU ≤ 38 535 Ref Ref Ref ZJU > 38 632 2.46(1.57,3.86) < 0.001 2.37(1.46,3.86) 0.001 2.33(1.36,3.98) 0.003 Model 1 adjusted for no variables Model 2 adjusted for age (categorization), race, education level, poverty and marital status Model 3 adjusted for age (categorization), race, education level, poverty and marital status, smoking status, alcohol consumption, physical activity, menstrual regularity, pelvic infection disease and age at menarche FLI ≥ 60: FLI ≥ 60 is regarded as an effective cut-off value for indicating a higher possibility of having fatty liver HSI > 36: HSI > 36 has a good screening performance for fatty liver, which is suitable for the assessment of fatty liver in the general population ZJU > 38: ZJU > 38 was determined as the appropriate critical value for differentiating the presence of fatty liver Weighted multivariate logistic regression analysis between fatty liver by different biomarkers and infertility Model 1 adjusted for no variables Model 2 adjusted for age (categorization), race, education level, poverty and marital status Model 3 adjusted for age (categorization), race, education level, poverty and marital status, smoking status, alcohol consumption, physical activity, menstrual regularity, pelvic infection disease and age at menarche FLI ≥ 60: FLI ≥ 60 is regarded as an effective cut-off value for indicating a higher possibility of having fatty liver HSI > 36: HSI > 36 has a good screening performance for fatty liver, which is suitable for the assessment of fatty liver in the general population ZJU > 38: ZJU > 38 was determined as the appropriate critical value for differentiating the presence of fatty liver The ROC curves were used to show the predictive ability of liver steatosis for female infertility in Fig.  4 . The AUCs of FLI, HSI and ZJU were 0.607 (95%CI: 0.559–0.655) vs. 0.611 (95%CI: 0.563–0.659) vs. 0.609 (95%CI: 0.561–0.658), respectively. The optimal cutoff values for each of them to predict infertility were 38.71 (FLI), 41.81 (HSI) and 40.07 (ZJU). The results of the delong test showed that the P -value between the AUCs of FLI and HSI is 0.591, the P -value between the AUCs of FLI and ZJU is 0.698, and the P -value between the AUCs of HSI and ZJU is 0.710 (Supplementary Table S2). This means that there is no significant difference in the predictive ability of the three indices for female infertility. Fig. 4 ROC curves for liver steatosis to predict female infertility ROC curves for liver steatosis to predict female infertility Based on the fully adjusted Model 3, we separately included BMI and WC as covariates. The results indicated that when BMI or WC was included, the associations between FLI, HSI and ZJU index and female infertility were no longer significant (Supplementary Table S3). These results indicate that the association between hepatic steatosis indicators and female infertility is significantly weakened after controlling obesity.

Our research has revealed a significant association between hepatic steatosis and infertility in women aged 20 to 44, particularly in terms of FLI and ZJU indices. These relationships may vary due to differences in age, race, marital status and physical activity. However, the relatively low AUC value limits their clinical application as independent screening tools for female infertility. These findings suggest that metabolic health may be associated with reproductive outcomes and suggest the potential value of fatty liver screening for infertile women. However, due to the limitations of the cross-sectional design of this study, the causal relationship could not be determined. The future longitudinal studies or intervention studies are crucial for determining whether reducing hepatic steatosis can improve fertility outcomes.

In this study, we investigated the association between liver steatosis indices, including the FLI, HSI and ZJU index, and female infertility using NHANES data with 1,167 participants. Our results demonstrated that higher FLI, HSI and ZJU index values were significantly associated with an increased prevalence of female infertility. These associations remained robust even after adjusting for multiple potential confounders, including demographic, lifestyle factors and past personal history. Furthermore, the dose-response relationship analysis suggested a linear correlation between these indices and infertility risk. Additionally, when defining fatty liver using standard cutoff values, we found that FLI-defined and ZJU index-defined fatty liver remained significantly associated with infertility, while HSI-defined fatty liver lost statistical significance after adjusting for confounders. Notably, subgroup analyses indicated significant interactions between these indices and factors such as age, race, marital status and physical activity, suggesting that the relationship between liver steatosis and infertility may vary across different population subgroups. A low AUC value indicates that the predictive ability of these indices in clinical applications is limited and they are not suitable to be used as independent predictive tools for diagnosing or ruling out infertility. On the contrary, these indices may be more suitable to be used as risk markers to indicate the possible risk of infertility in women. The results of the sensitivity analysis indicated that when BMI or WC was additionally included in the fully adjusted Model 3, the associations between FLI, HSI and ZJU index and female infertility were no longer significant. Our findings align with and expand upon previous studies exploring the relationship between fatty liver and reproductive dysfunction. Recently, some studies have explored the association between hepatic steatosis and female infertility. Wu et al. found that HSI was positively correlated with the incidence of female infertility (OR = 1.02, 95%CI:1.01–1.04; P  = 0.005)[ 24 ]. Tian et al. found that the ZJU index was significantly correlated with the prevalence of infertility (OR = 1.04, 95%CI: 1.00-1.07, P  < 0.05)[ 25 ]. These findings are very similar to the results of this study, enhancing the credibility of the analysis. In contrast, our study went further by evaluating multiple fatty liver indices and conducting more in-depth subgroup analyses and ROC analyses. This further indicates that the association between hepatic steatosis and infertility is not an accidental result. It also helps to illustrate the contribution of this study to the existing field: not only did it verify that the FLI and ZJU index (except for HSI) also showed similar associations, but its predictive ability was also tested. However, most current studies tend to focus more on the interrelationship between fatty liver, PCOS and reduced reproductive capacity [ 26 , 27 ]. PCOS is a common hormonal disorder that often leads to difficulty in conception and usually coexists with non-alcoholic fatty liver disease (NAFLD) and metabolic syndrome [ 28 ]. PCOS may be a common cause of fatty liver and infertility. It is important to note that although we took irregular menstruation into account in our analysis, not all PCOS patients will show abnormal menstruation. Therefore, some of the women included in the study may have undiagnosed PCOS, which may affect the observed association between hepatic steatosis and infertility. Recent studies have shown that fatty liver not only increases the risk of PCOS but may also affect the fertility of women with PCOS through multiple mechanisms, such as IR, changes in sex hormone levels, oxidative stress and chronic inflammation [ 29 ]. A genetic piece of evidence confirms that NAFLD increases the risk of PCOS. IR and hyperandrogenemia may play a mediating role in this causal pathway, and there may be a “liver-ovarian axis” between the two. However, the impact of PCOS on the risk of NAFLD is not significant [ 30 ]. Therefore, the close relationship between fatty liver and PCOS is similar to our research conclusion, which is that for women of childbearing age with higher degree of hepatic steatosis, the risk of infertility is extremely high. Unlike previous studies that primarily focused on PCOS populations, our research included a broader cohort of reproductive-age women, allowing us to assess the generalizability of liver steatosis indices in predicting infertility beyond PCOS-related mechanisms. The mechanisms linking hepatic steatosis to female infertility are multifaceted and may involve several interrelated metabolic and endocrine pathways. Insulin resistance is an important risk factor for female infertility, and hepatic steatosis is associated with insulin resistance [ 31 , 32 ]. Some studies have shown that in the state of hepatic steatosis, lipid accumulation within liver cells may activate inflammatory pathways, thereby affecting key molecules in the insulin signaling pathway and leading to insulin resistance [ 33 ]. The compensatory insulin hypersecretion by pancreatic β-cells and reduced hepatic insulin clearance further contribute to hyperinsulinemia [ 34 ]. On the one hand, hyperinsulinemia activates the insulin receptors in ovarian theca cells, increasing the activity of cytochrome P450c17 (17α-hydroxylase), a key enzyme in androgen synthesis [ 35 ]. The significant increase in androgen synthesis leads to hyperandrogenemia. On the other hand, hyperinsulinemia stimulates the hypothalamus to secrete gonadotropin-releasing hormone (GnRH), disrupting the normal function of the Hypothalamic-Pituitary-Ovarian (HPO) axis and leading to excessive secretion of luteinizing hormone (LH) by the pituitary gland [ 36 ]. Elevated LH levels further promote androgen synthesis in ovarian theca cells, while the relative decrease in follicle-stimulating hormone (FSH) secretion results in an increased LH/FSH ratio [ 37 ]. Additionally, hyperinsulinemia inhibits the synthesis of sex hormone-binding globulin (SHBG) in the liver, a plasma protein responsible for binding and transporting sex hormones [ 38 ]. When SHBG levels decline, free androgen levels rise, exacerbating hyperandrogenemia. Hyperandrogenemia and an elevated LH/FSH ratio disrupt normal folliculogenesis and ovulation, leading to irregular menstruation and infertility [ 39 ]. In addition, hyperandrogenism can alter the endometrial environment and reduce endometrial receptivity, which is not conducive to embryo implantation and increases the risk of infertility [ 40 ]. This mechanism may be able to explain the potential link between hepatic steatosis and female infertility, but more research is still needed to confirm it. Chronic inflammation is associated with female infertility [ 41 ]. Hepatic steatosis is often accompanied by a chronic inflammatory state, manifested as elevated levels of pro-inflammatory cytokines such as interleukin-6 (IL-6) and tumor necrosis factor-α (TNF-α) [ 42 ]. These pro-inflammatory factors may inhibit the activity of aromatase in ovarian granulosa cells, an enzyme crucial for converting androgens to estrogens [ 43 ]. Reduced aromatase activity leads to decreased estrogen synthesis and increased androgen accumulation, which in turn affects normal follicular development. These pro-inflammatory cytokines can also damage mitochondrial function in oocytes, impairing their energy metabolism and quality, thereby reducing oocyte fertilization capacity and embryonic developmental potential [ 44 ]. TNF-α inhibits the secretion of GnRH from the hypothalamus, subsequently reducing the secretion of LH and FSH [ 45 ]. Additionally, IL-6 affects LH receptor expression, reducing follicular responsiveness to gonadotropins [ 46 ]. TNF-α can also activate proteolytic enzymes in the follicular wall, leading to abnormal follicular rupture and affecting the release of mature oocytes [ 47 ]. Furthermore, inflammatory cytokines interfere with endometrial decidualization, reducing endometrial receptivity and the likelihood of implantation, which are important processes for pregnancy [ 48 ]. However, this underlying mechanism still requires further research to clarify its role in the relationship between hepatic steatosis and female infertility. Oxidative stress is associated with infertility in women with hepatic steatosis [ 49 ]. Studies have shown that lipid accumulation in the liver may be accompanied by increased lipid peroxidation and reactive oxygen species (ROS) generation [ 50 ]. Elevated ROS levels may affect the mitochondrial function of oocytes, thereby interfering with their energy metabolism and developmental capacity, which may be related to ovulatory dysfunction [ 51 ]. Oxidative stress affects the normal function of the HPO axis, causing abnormal secretion of gonadotropins and disrupting the balance between FSH and LH, resulting in irregular ovulation [ 52 ]. Oxidative stress not only affects oocyte quality but also negatively impacts embryonic development after fertilization, leading to delayed or arrested embryonic development, reduced fertilization rates, and decreased embryo implantation rates [ 53 ]. Oxidative stress is closely associated with several reproductive disorders, such as PCOS, endometriosis, and premature ovarian insufficiency (POI), all of which lead to infertility or reduced fertility [ 54 – 56 ]. Additionally, Chronic oxidative stress accelerates ovarian tissue damage, leading to diminished ovarian reserve, characterized by decreased oocyte quality and ovulatory dysfunction [ 57 ]. This study further reveals significant heterogeneity in the association strength between hepatic steatosis and infertility across different subgroups. We observed significant interactions with age, race, marital status, and physical activity, suggesting these factors may be linked to variations in the hepatic steatosis-infertility association. First, in the age-stratified analysis, the positive correlation between hepatic steatosis and infertility was more significant in young women (aged 20–34). This discovery may be due to the biological feature that the female reproductive system is potentially more sensitive to metabolic disorders during the young reproductive years. Hepatic steatosis, as a marker of early-onset metabolic disorders, has a more prominent association with infertility during the peak childbearing age group. In the group of older women, age-related changes in ovarian function and the mixture of other causes of infertility may relatively weaken the independent associated signals of hepatic steatosis. Second, a stronger hepatic steatosis-infertility association was observed among non-Hispanic White and Mexican American women compared to non-Hispanic Black and other race women. This variation may relate to differences in genetic susceptibility, patterns of hepatic fat accumulation, and lifestyle factors across racial groups [ 58 ]. Previous studies have shown significant differences in metabolic syndrome, insulin sensitivity, and fat distribution patterns among different races, which may further affect the impact of fatty liver on reproductive health [ 59 ]. Third, the differences in marital status reveal the profound influence of social behavioral factors on the definition of infertility. The stronger correlation among married/cohabitating and widowed/divorced/separated women essentially reflects the reliance of “infertility” diagnosis on the exposure of fertility intentions and medical seeking behaviors: the former is the core group of fertility attempts, while the latter often includes past fertility failure experiences or future fertility demands, making the reproductive damage of fatty liver explicit. However, due to the lack of regular and non-contraceptive sexual behavior, the potential reproductive dysfunction of unmarried women has not been included in the “infertility” statistics, resulting in an artificial underestimation. This bias warns us that epidemiological studies based on clinical diagnosis may underestimate the long-term reproductive health impact of metabolic risks on populations with unactivated fertility intentions. Finally, among physically active women, the positive correlation between hepatic steatosis and infertility is more significant. It is worth noting that the active group usually has a lower overall metabolic risk, and the presence of hepatic steatosis in this group suggests that there may be genetic susceptibility or intractable metabolic disorders beyond the protective effect of exercise. Under a relatively healthy metabolic background, the coexistence of hepatic steatosis and reproductive dysfunction is more prominent. Our findings suggest that FLI and ZJU index may be valuable tools for assessing infertility risk in clinical practice. These indices are easily calculable using routine clinical and laboratory data, making them accessible for use in primary care and reproductive health settings. Given the high prevalence of fatty liver and its association with infertility, incorporating FLI and ZJU index into routine reproductive health assessments could help identify women at higher risk of infertility, allowing for early interventions. The results of the sensitivity analysis indicated that although hepatic steatosis was associated with infertility, this association disappeared after obesity was controlled. This indicates that hepatic steatosis may be more of a sign of obesity rather than an independent causal factor. In other words, obesity may be the main driving factor for female infertility, while hepatic steatosis is merely a metabolic manifestation of obesity. Previous studies have shown that weight loss, dietary modifications, and increased physical activity can improve both fatty liver and reproductive outcomes [ 60 , 61 ]. Women with high FLI or ZJU index scores may benefit from targeted lifestyle interventions aimed at reducing liver fat accumulation and improving metabolic health. In assisted reproductive technology (ART) settings, identifying metabolic contributors to infertility can help optimize treatment protocols. Women with high FLI or ZJU index may require additional metabolic management to enhance their chances of successful conception. The use of NHANES data provides a diverse population sample, enhancing the generalizability of our findings. We accounted for numerous confounders, strengthening the robustness of our results. The use of RCS allowed for a nuanced understanding of the relationship between liver steatosis and infertility. Furthermore, it is necessary to note that when constructing the model, we did not choose to include obesity-related indicators (BMI and WC) as covariates in Model 3, as these indicators are fundamental components of hepatic steatosis related indicators (FLI, HSI and ZJU index). Similarly, the same is true for liver enzymes, blood lipids and blood sugar. Adjusting these indicators in the regression model would overly control the factors on the causal path, thereby possibly masking the potential effect between hepatic steatosis and female infertility. This method enables us to more accurately assess the association between hepatic steatosis as a comprehensive metabolic marker and female infertility. Despite its advantages, this study also has its limitations that should be acknowledged. Given the cross-sectional nature of our study, we cannot determine the temporal sequence of hepatic steatosis and infertility. It is possible that other factors, such as obesity, PCOS, or genetic predispositions, may contribute to both conditions. While our results suggest a significant association between hepatic steatosis and infertility, we cannot definitively conclude that hepatic steatosis directly causes infertility. Future longitudinal or intervention studies are needed to establish causality and to explore the potential mechanisms underlying this association. Another important consideration is the potential for reverse causation. Infertility itself, or its treatment, could lead to changes in weight or lifestyle that might exacerbate fatty liver. For example, the psychological stress of infertility and the treatments involved could lead to weight gain or unhealthy lifestyle choices. Additionally, some treatments for infertility, such as ART, might involve hormonal interventions that could affect metabolic health. While our study design does not allow us to directly assess these potential reverse causal pathways, future research should explore these possibilities to provide a more comprehensive understanding of the relationship between hepatic steatosis and infertility. Infertility was defined based on self-reported data, specifically whether the participant had tried to conceive for 12 months without success and had sought medical help. While this definition is practical and aligns with clinical criteria, it has limitations. On the one hand, women who have never attempted to conceive would be automatically classified as ‘non-infertile,’ which could underestimate the true prevalence of infertility, particularly among younger or unmarried women who may have undiagnosed fertility issues. Conversely, some women classified as ‘infertile’ may eventually conceive, indicating that their infertility was temporary. Therefore, our results reflect self-reported difficulty conceiving within the past year rather than confirmed permanent infertility. This highlights the importance of interpreting our results with caution and recognizing the potential for misclassification. On the other hand, the NHANES dataset lacks etiological differentiation in defining infertility, which is a limitation of the data source. It is failed to distinguish between female factor infertility and infertility caused by male partner factors or a combination of both, while male factors account for a considerable proportion of infertility. This may lead to an overestimation of the association between female metabolic health and infertility. In addition, other factors such as blocked fallopian tubes may also lead to infertility and weaken this association. Unfortunately, this information is not available in the NHANES dataset and thus was not included in our analysis. If future research can incorporate relevant data from male partners and more detailed classifications of infertility causes, it will help to more accurately assess the direct relationship between female metabolic health and infertility. Although this study adjusted for numerous confounding factors, there are still some confounding factors that are difficult to include. PCOS is the main cause of anovulatory infertility and often coexists with fatty liver. It may be a common cause of hepatic steatosis and infertility. In addition, other factors such as thyroid diseases and fertility treatments may also have a certain impact on infertility. However, NHANES lacked data on PCOS, thyroid diseases and other related diseases, and thus could not be included in the analysis. Biomarkers measured at a single time point may not accurately reflect chronic diseases. Infertility data was collected through self-reported questionnaires, lacking detailed reproductive history or reproductive medical treatment information, which may introduce recall bias. Self-reported data may underestimate or overestimate the actual prevalence of infertility. Despite this, this data collection method is common in large-scale epidemiological investigations and can provide basic data for subsequent more in-depth research. In our research, we conducted subgroup analyses and interaction effect tests on multiple stratified variables. However, we did not further conduct multiple correction tests on the results, which might lead to the significance of some interactions being caused by accidental factors. In this study, FLI, HSI and ZJU index were used to evaluate hepatic steatosis, which are non-imaging alternative markers. Although these indicators have the advantages of low cost and easy accessibility, they cannot completely replace imaging diagnostic methods (such as ultrasound or transient elastography). Future research will verify the diagnostic efficacy of these alternative markers through imaging methods, which will further enhance the accuracy of the assessment.

Female infertility is defined as the inability to successfully conceive after 12 months of sexual activity without using any contraceptive measures [ 1 ]. According to the World Health Organization, in 2021, approximately 110 million women worldwide suffered from infertility, with a prevalence rate of 3.7%. From 1990 to 2021, the age-standardized prevalence rate of female infertility globally increased by an average of 0.68%, and this rate is projected to continue rising by 2040[ 2 ]. Infertility not only has a negative impact on physical health to women but also exerts significant psychological pressure. Multiple studies have consistently demonstrated that women with infertility face a substantially elevated risk of developing cardiovascular diseases, breast cancer, and endometrial cancer, compared to those who achieve normal pregnancies [ 3 – 5 ]. Moreover, infertility often places a tremendous psychological burden on affected women. Those with a history of infertility or relevant diagnoses, who face a substantially increased risk of developing mental disorders such as depression and anxiety [ 6 , 7 ]. From a societal perspective, the high incidence of infertility also has profound negative impacts on population structure and family stability [ 8 ]. Ovulatory disorders and tubal diseases are the main causes of female infertility [ 9 ]. In recent years, a growing body of research has highlighted that metabolic disorders are significant contributors to female reproductive dysfunction, including infertility, premature ovarian failure, and polycystic ovary syndrome (PCOS) [ 10 ]. Liver steatosis, characterized by excessive accumulation of fat within hepatocytes, is closely associated with metabolic disorders and often coexists with obesity, hyperglycemia, and dyslipidemia [ 11 , 12 ]. Notably, both obesity and insulin resistance have been recognized as significant independent contributors to female infertility, which implies a potential association between liver steatosis and female infertility [ 13 ]. However, the role of liver steatosis in female infertility remains unclear. Elucidating this relationship will not only help reveal the underlying mechanisms of infertility but also provide new perspectives for the prevention and treatment of female infertility. Imaging examinations, such as ultrasound and transient elastography, are important tools for diagnosing liver steatosis [ 14 ]. In recent years, with the advancement of research, non-invasive diagnostic indicators and models have gradually become essential for diagnosing hepatic steatosis. For example, the Fatty Liver Index (FLI), Hepatic Steatosis Index (HSI), and Zhejiang University Index (ZJU Index) integrate clinical indicators and blood test results to offer more convenient and cost-effective methods for screening and diagnosing liver steatosis [ 15 ]. From a clinical perspective, the diagnosis and management of infertility require more sensitive and specific risk prediction tools, especially for patients with obesity and metabolic disorders leading to infertility. Despite the potential link between liver steatosis and female infertility, no large-scale studies have yet investigated this relationship. In light of this, we selected women aged 20–44 from the National Health and Nutrition Examination Survey (NHANES) database as the study population, and evaluated liver steatosis using FLI, HSI and ZJU index to investigate the potential relationship between liver steatosis and female infertility. This study aims to provide foundational data and theoretical support in this underexplored area, thereby laying the groundwork for future research. Through risk assessment based on these indices, it is expected to provide clinicians with new auxiliary tools to help identify patients with metabolism-related infertility at an early stage and provide a scientific basis for the development of personalized intervention strategies.

Supplementary Material 1. Supplementary Material 1.