Introduction Non-alcoholic fatty liver disease (NAFLD) is a disease of the liver that primarily characterized by excessive fat deposition in hepatocytes due to non-alcohol and other well-defined factors damaging the liver. The prevalence of NAFLD has increased steadily in recent years, with a global prevalence of approximately 25% in the general population(1). Even more alarmingly, a meta-analysis in 2022 showed that the global prevalence of NAFLD was higher than previously estimated and growing at an alarming rate(2). NAFLD is one of the most important causes of chronic liver disease, which can progress from nonalcoholic simple fatty liver to nonalcoholic steatohepatitis, hepatic fibrosis, and finally to cirrhosis and hepatocellular carcinoma (3).

A new hypothesis, the “hepatoovarian axis”, has recently been proposed as a result of the growing evidence of a strong association between the liver and the female ovaries(4). Complex lipids are strongly associated with aging, for example, bis(monoacylglycero) phosphate (BMP), a characteristic lipid of aging, increases significantly with the aging process, and a short-term healthy lifestyle can reduce BMP levels in postmenopausal women (5). Genetic studies based on multi-omics data have found a decrease in healthy cardiolipin (CL) and an increase in unhealthy CL in NAFLD. Perhaps, changes in CL levels may be one of the underlying causes of hepatic steatosis and may even be a biomarker of NAFLD disease (6).

Animal experiments have demonstrated that ovarian hormones in rats play a role in regulating lipid accumulation in the liver(7). Compared with sham oophorectomized female rats, true-oophorectomized rats not only had increased systemic fat weight and peripheral insulin resistance, but also developed severe hepatic triglyceride accumulation and were more susceptible to inducing hepatic steatosis(8). Ovarian estrogen levels in the liver are not only involved in lipid metabolism, but also have anti-inflammatory, anti-fibrotic, and anti-steatosis (9, 10). Lipid regulation disorders can directly affect ovarian hormone levels, accelerating female ovarian aging(11). First menarche is a sign of the onset of puberty and indicates the establishment of positive feedback from estrogen to pituitary gonadotropins(12).

Natural menopause, suggesting the decline of ovarian function, marks the end of female reproductivefunction.Duringthefemalereproductivecourse, hormone levels fluctuate and change with the menstrual cycle, and reproductive events. Female reproductive factors (including age at menarche, age at natural menopause, age at first birth, menopause, etc.) are closely related to women’s health status. Female reproductive factors have a significant impact on cardiovascular disease, diabetes, osteoporosis and other diseases(13). Earlier age at menarche has been associated in previous Coronary Artery Risk Development in Young Adults (CARDIA) study not only with an increased risk of NAFLD in adults, but also with visceral and subcutaneous abdominal ectopic fat depots in mid-adulthood(14). Menopause is associated with progression and severity of NAFLD.

In a cross sectional study involving 488 menopausal women with NAFLD(15), both premature menopause and time since menopause were associated with an increased risk of NAFLD fibrosis. Premature menopause, and time after menopause were both correlated with an increased risk of more severe liver fibrosis. Evidence from the previous NHANES study suggests that women with AFB <26 years are at greater risk of developing NAFLD compared to women with a first birth at 30-32 years old(16). These findings show a strong association between female reproductive factors with NAFLD. However, there are no clinical studies investigating the causal relationship between female reproductive factors and NAFLD. The liver is an essential metabolic and immune organ of the human body(17).

Immune cells account for 10-20% of total liver cells(18). T-lymphocyte subsets have an important role in immunomodulation. Peripheral CD8 cells, natural killer cells, and CD4 T cells are biased toward activation in NAFLD patients compared to healthy populations(19). Macrophages are involved in steatosis, inflammation and fibrosis in NAFLD(20). The number of immune cells in the female genital tract varies with the menstrual cycle, pregnancy and menopause(21). Immune cells are involved in the continuous remodeling of the endometrium and the regulation of the uterine microenvironment during the more than 400 months betweentheonsetofmenstruationandmenopause(22).From the above studies, we have revealed the specific causality of the hepato-ovarian axis through two diseases and have elicited a “hepato-ovarian axis” related to aging.

Immune cells have a major role in the liver and ovary. Therefore, immune cells may be the “mediators” of communication between the liver and ovarian organs. Mendelian randomization (MR) is a method of causal inference in epidemiology based on genetic variation, whereby GWAS (genome-wide association studies) are analyzed to find SNPs with genetic variants associated with biological factors, and then the genetic variants are used as instrumental variables (IVs) to derive causal associations between the exposures and the outcomes. Therefore, we applied MR to explore the potential causal relationship between AAM, ANM, AFB and NAFLD to determine their underlying mechanisms. In addition, we propose the hypothesis that female reproductive factors may lead to alterations in certain immune cells, which in turn mediate female reproductive factor induced NAFLD.

In this study, we utilized data from 731 immunophenotypes in the GWAS database(23), using mediation analysis to identify potential immune cells. The results of the study may shed light on the causal relationship between female reproductive factors and NAFLD and the association in terms of potential immune cells, and contribute to the development of clinical screening and prevention strategies for NAFLD.

Methods Study design The two-sample MR study examined the causal relationship between AAM, AMP, AFB and NAFLD in two sample MR designs based on different GWAS data, including three assumptions(24): (1) robust correlations between instrumental variables (IVs) and exposures (associations); (2) IVs are not associated with confounders (independence); (3) the effects of IVs on outcomes can only be realized through exposures (exclusion restriction criteria). See Fig 1. Figure 1 : MR analysis flow chart Note: The total effect of AAM on NAFLD, β1, was derived by univariate MR (i.e., genetically predicted AAM as exposure and NAFLD as outcome). A two-step MR design was used (where β2 is the effect of AAM on immune cells and β23 is the effect of immune cells on NAFLD (adjusted for exposure)).

Mediated effect (β2*β3). Mediated ratio is indirect effect divided by total effect (β2 × β3/β1). Abbreviations: AAM = age at menarche; ANM = age at natural menopause; AFB = age at first birth, NAFLD = non-alcoholic fatty liver disease; IVW = inverse variance weighted; SNP = single nucleotide polymorphism; GWAS = genome-wide association study. Datasets A detailed overview of all data sources is presented in Table1. We obtained publicly available summary statistics from GWAS for MR analysis. Data on AAM were extracted from a GWAS study that included 182,416 European women from 57 studies, incorporating 2,441,816 single nucleotide polymorphisms (SNPs)(12). The ANM summary data included 69,360 European women and 2,418,696 associated SNPs(25).

A recent GWAS examined 542,901 female cases of AFB containing 9,702,772 associated SNPs(26). The NAFLD data were derived from a large GWAS-aggregated dataset with a sample size of 778,614 cases (8,434 cases and 770,180 controls)(27). Genetic data on immune cells were derived from a population of European ancestry and contained 731 immunophenotypes (GCST0001391 GCST0002121)(23). These 731 immunophenotypes consisted of four types: morphological parameters (MP), absolute cell count (AC) relative cell count (RC), median fluorescence intensity (MFI) of surface antigens, and, the numbers of which ranged from small to large were 32, 118, 192, and 389, respectively. Table 1 : The GWAS data source details in our study Phenotype Consortium Participants Population PMID Type of variables AAM ReproGen 182,416 European 25231870 Exposure ANM ReproGen 69,360 European 26414677 Exposure AFB NA 542,901 European 34211149 Exposure NAFLD NA 778,614 European 34841290 Outcome Triglycerides UK Biobank 441,016 European 32203549 Confounder Hypertension UK Biobank 461,880 European NA Confounder BMI UK Biobank 454,884 European NA Confounder Selection of IVs The number of single nucleotide polymorphisms (SNPs) were screened to fulfill the following conditions : 1)SNPS: P < 5×10−8 , 2) removed linkage disequilibrium(LD) (r2 < 0.001, distance threshold = 10000kb), 3) The Phenosanner database was queried to remove confounders, 4) Removed palindromic SNPs with intermediate allele frequencies, 5) The F-statistic data of the selected SNPs exceeds 10 (F = beta2/se2)(28).

F > 10 suggests that the relationship between IV and exposure is robust and that the results of MR analysis are not influenced by weak instrumental bias. Finally, the data from the database were extracted, organized, and merged. The effect alleles were subsequently compared to correspond to the same effect allele for both the exposure and outcome effect values. Based on prior studies(29), the IV significance threshold with each immune trait was set at 1 × 10-5 and then r2 < 0.001, with a distance threshold = 10,000 kb, to ensure that no linkage disequilibrium (LD) correlations existed. Mendelian randomization analysis MR analyses were conducted with Inverse Variance Weighting (IVW),MREgger,WeightedMedian,SimpleModeandWeighted Mode.

We used IVW estimators with multiplicative random effects for our main analyses (if they were not heterogeneous, fixed effects IVW were applied). IVW was utilized as the primary outcome to assess the causality between exposure and results, with statistical significance set at a p-value <0.05. Since the result was binary, we translated it into a ratio-ratio (OR). In the absence of pleiotropy, IVW accurately appraises each valid IV and provides reliable causal estimates by incorporating the Wald ratio. The weighted median method calculates the causal estimate as the median estimate of the ratio of each genetic variant, weighted by the inverse of its variance. The weighted model assigns causal estimates to each genetic variant by the inverse of its variance.

The simple model considers each genetic variant individually to estimate causal effects. We conducted sensitivity analyses with several methods robust to horizontal pleiotropy, which include MR Egger, weighted mode, weighted median, and MR pleiotropy residual sum and outlier test (MR-PRESSO). Different assumptions were made about the validity of IV using each method. If half of the IVs are invalid, perform the estimation of the median weights. We employed sensitivity analysis to verify the plausibility and stability of the MR results. The slope of the MR Egger regression provides an estimate of the causal relationship between exposure and outcome. In addition, the MR-Egger intercept represents the average pleiotropic effect of genetic variation, with an intercept P < 0.05 indicating the presence of pleiotropy.

Leave-one-out analysis was performed by eliminating each SNP one by one to test whether the results were affected by any individual SNP. We utilized the Benjamini-Hochberg method for calibration and calculated corrected p values for false discovery rate (FDR). A strongly causal relationship was considered when the raw P < 0.05 and PFDR < 0.05. If raw P-value < 0.05 and PFDR > 0.05, it indicates that there may be a potential causal relationship between the exposure and the outcome. We utilized a two-step MR design for mediation analyses to investigate whether immune cells mediate causality from female reproductive factors to NAFLD. The direct effect of exposure on outcome (to obtain the coefficient β1) and the reverse causality of outcome on exposure were determined in step 1.

In Step 2, the effect of exposure on immune cells (to obtain the coefficient β2). Step 3, the effect of immune cells on the outcome (to obtain the coefficient β3). Finally, the mediating effect (β2*β3) was calculated. We calculated the percentage mediated by the mediating effect by dividing the indirect effect by the total effect. Finally, 95% confidence intervals were calculated using the delta method. Finally, the results of the MR analysis were visualized utilizing scatter plots and forest plots. All MR analyses were conducted using R software packages “Two-sample MR” and “MRPRESSO”. All data were analyzed by R (version 3.1.5).

Results Two-sample MR analysis We utilized IVs that were significantly associated with MR exposure in both samples. Furthermore, we excluded SNPs with strong correlations with BMI and type 2 diabetes, and then calculated F-statistics for the remaining SNPs (See Supplementary Tables S1-3). The F-statistics for each instrument exceeded 10. The Univariate MR analyses showed that AAM, ANM, AFB and NAFLD had 56, 37, and 43 SNPs. Genetically determined AAM and AFB showed a possible causal effect on NAFLD, as validated by the Benjamini- Hochberg method. AAM is strongly associated with NAFLD (OR = 0.82, 95% CI: 0.74-0.91, P < 0.0001, PFDR = 0.0002). AFB has a protective effect on NAFLD (OR = 0.91, 95% CI: 0.83- 0.99, P = 0.0286, PFDR = 0.0465).

However, there is no causal relationship between ANM and NAFLD (OR = 1.01, 95% CI: 0.98- 1.04, P = 0.539, PFDR = 0.564). The inverse MR of NAFLD with AAM demonstrated no causal association between both (OR = 0.99, 95% CI: 0.92-1.16, P = 0.75). See Fig 2. With the exception of ANM, for which there was horizontal pleiotropy (MR-Egger P<0.05), the P-values of the intercepts in the MR-Egger were greater than 0.05, indicating no horizontal pleiotropic effect. MR PRESSO did not detect any outliers for SNP effects, and the Cochran Q-values indicated no heterogeneity (see Table 2). The “Leave-one-out plot” showed that none of the SNPs had a dominant effect on the estimated causal relationship between AAM, AFB and NAFLD.

See Supplementary Fig 2. Table 2: Heterogeneity and pleiotropy of AAM, ANM, AFB and NAFLD Exposure Heterogeneity tests Pleiotropic test MR egger IVW Egger intercept se P Q P Q P AAM 51.46434 0.572782 52.11228 0.585664 -0.00746 0.009272 0.42438 ANM 36.22883 0.411088 42.68432 0.205764 0.018092 0.007245 0.017367 AFB 37.00325 0.648868 37.20587 0.681141 -0.00541 0.012008 0.654983 Additionalfile1:TableS1:CharacteristicsoftheInstrumentalSNPsassociatedwithSNPbeta.exposurese.exposurepval. exposure F statistic SNP effect_allele. exposure other_allele. exposure beta.exposure se.exposurepval. exposure F statistic rs2344508 A G 0.034 0.005747 3.30E-09 34.99828 rs466639 C T 0.075 0.008706 7.00E-18 74.21662 rs618678 T C -0.034 0.006148 3.20E-08 30.58231 rs633715 C T -0.051 0.00725 2.00E-12 49.48395 rs6694738 A C -0.044 0.007885 2.40E-08 31.14062 rs6747380 A G 0.065 0.007658 2.10E-17 72.04821 rs6758290 C T -0.04 0.006322 2.50E-10 40.0307 rs1518080 G C -0.051 0.006004 2.00E-17 72.14463 rs4369815 G T -0.08 0.011888 1.70E-11 45.28888 rs895526 C T 0.044 0.007553 5.70E-09 33.93447 rs2947411 G A -0.052 0.007655 1.10E-11 46.1417 rs2687729 G A 0.044 0.006575 2.20E-11 44.78416 rs6770162 A G 0.036 0.005733 3.40E-10 39.4302 rs3870341 G A -0.043 0.006449 2.60E-11 44.45699 rs11715566 T C 0.052 0.005631 2.60E-20 85.27228 rs3914188 C G 0.044 0.006702 5.20E-11 43.10049 rs7642134 G A 0.038 0.005912 1.30E-10 41.30851 rs17351680 G C 0.044 0.007557 5.80E-09 33.90062 rs1482853 A C -0.038 0.006043 3.20E-10 39.54859 rs10938397 G A -0.038 0.005923 1.40E-10 41.16371 rs3733632 G A 0.049 0.007862 4.60E-10 38.84004 rs9997604 C A 0.039 0.008427 4.80E-08 29.79595 rs740077 C A -0.046 0.007045 6.60E-11 42.63401 rs13179411 T G 0.06 0.007813 1.60E-14 58.97123 rs3115627 G A 0.038 0.006453 3.90E-09 34.673 rs13215865 T C -0.042 0.002077 1.00E-200 1373.869 Research Article rs4840086 G A -0.036 0.00562 1.50E-10 41.02883 rs11756454 A T 0.034 0.005703 2.50E-09 35.53902 rs2179786 T G -0.039 0.00572 9.20E-12 46.49183 rs9373571 A T 0.034 0.005926 9.60E-09 32.92064 rs2184968 C T -0.036 0.005672 2.20E-10 40.28045 rs6933660 A C -0.036 0.008444 1.50E-17 72.71226 rs11767400 A C 0.035 0.006295 2.70E-08 30.91197 rs7821178 A C -0.045 0.006196 3.80E-13 52.74396 rs888345 A G -0.044 0.007315 1.80E-09 36.179 rs2617056 T A -0.036 0.005933 1.30E-09 36.81333 rs4242496 A T -0.033 0.005696 6.90E-09 33.56274 rs1516883 A G -0.091 0.002455 1.00E-200 1373.877 rs913588 A G -0.034 0.005683 2.20E-09 35.78795 rs12003641 T C 0.082 0.010555 7.90E-15 60.36044 rs7853970 C T -0.037 0.00626 3.40E-09 34.9401 rs1874984 C G 0.037 0.006005 7.20E-10 37.96574 rs7103411 T C -0.043 0.007056 1.10E-09 37.13905 rs7944630 A G 0.047 0.005728 2.30E-16 67.32707 rs10840031 A G 0.038 0.006406 3.00E-09 35.18383 rs12291726 G A 0.057 0.008333 7.90E-12 46.79035 rs11022756 C A -0.048 0.006481 1.30E-13 54.8514 rs16938437 T C -0.067 0.010348 9.50E-11 41.9215 rs7119712 A G -0.041 0.006445 2.00E-10 40.46668 rs9565073 C T 0.034 0.005881 7.40E-09 33.42673 rs9555810 G C 0.047 0.006492 4.50E-13 52.41183 rs10144321 G A -0.042 0.006618 2.20E-10 40.28039 rs10483727 C T -0.037 0.005785 1.60E-10 40.90267 rs12148769 A G -0.055 0.009752 1.70E-08 31.81021 rs3743266 C T -0.045 0.006233 5.20E-13 52.12785 rs12915845 T C -0.035 0.005729 1.00E-09 37.32486 rs1659127 A G 0.044 0.006372 5.00E-12 47.68702 rs12598642 G A 0.044 0.005681 9.50E-15 59.99695 rs9939609 A T -0.042 0.00572 2.10E-13 53.90896 rs9635759 A G 0.058 0.006362 7.70E-20 83.12575 rs1398217 C G 0.046 0.005767 1.50E-15 63.63162 rs2303100 T C 0.038 0.005684 2.30E-11 44.69706 rs4801589 G C 0.032 0.005628 1.30E-08 32.33138 rs852069 G A 0.036 0.005907 1.10E-09 37.13897 rs2836950 G C -0.035 0.006206 1.70E-08 31.8101 Research Article Figure 2 : Univariate MR analysis Supplementary materials Supplementary Figure 1 : Scatter plots showing the effect of AAM, ANM, AFB on the risk of NAFLD A: AAM; B: ANM; C: AFB.

Supplementary Fig 2 : Leave-one-out plots which detect outliers SNPs in AAM, ANM, AFB on the risk of NAFLD A: AAM; B: ANM; C: AFB. Supplementary Fig.3 Leave-one-out plots which detect outliers SNPs in male BT and CVD A: coronary artery disease; B: hypertension; C: myocardial infarction; D: heart failure; E: atrial fibrillation; F: peripheral atherosclerosis Supplementary Figure 4 Supplementary Figure.4 Leave-one-out plots which detect outliers SNPs in male SHBG and CVD A: coronary artery disease; B: hypertension; C: myocardial infarction; D: heart failure; E: atrial fibrillation; F: peripheral atherosclerosis Supplementary Figure .5 Research Article Supplementary Fig.5 Leave-one-out plots which detect outliers SNPs in female TT and CVD A: coronary artery disease; B: hypertension; C: myocardial infarction; D: heart failure; E: atrial fibrillation; F: peripheral atherosclerosis Supplementary Figure 6 : Supplementary Fig.6 Leave-one-out plots which detect outliers SNPs in female BT and CVD A: coronary artery disease; B: hypertension; C: myocardial infarction; D: heart failure; E: atrial fibrillation; F: peripheral atherosclerosis Supplementary Fig.7 Supplementary Figure 7 Leave-one-out plots which detect outliers SNPs in female SHBG and CVD A: coronary artery disease; B: hypertension; C: myocardial infarction; D: heart failure; E: atrial fibrillation; F: peripheral atherosclerosis Mediation analysis We performed a two-step MR analysis to estimate the mediating role of immune cells between AAM and NAFLD.

AAM was causally associated with CD25 on switched memory B cell (OR = 0.83, 95% CI: 0.71-0.97, P = 0.016) and CD127 on CD28+ CD4- CD8- T cell (OR = 0.82, 95% CI: 0.68-0.98, P = 0.028. When immune cells were the exposure and NAFLD was the outcome, we found that CD25 on switched memory B cell was protective against NAFLD (OR = 0.94, 95% CI: 0.89-0.99, P = 0.012). However, CD127 on CD28+ CD4-CD8- T cell increased the risk of NAFLD (OR = 1.06, 95% CI: 1.00-1.13, P = 0.045). See Fig 2. The results revealed that CD25 on switched memory B cell had a mediating effect on AAM and NAFLD (β = 0.01231), with a mediating effect percentage of 6.23%.

CD127 on CD28+ CD4 CD8- T cell might mediate the transition from AAM to NAFLD (β = -0.0125), with a mediating effect percentage of 6.68%.

Discussion We investigated potential causal relationships between female reproductive factors and NAFLD risk using Two-sample MR analysis. Univariate MR findings revealed a significant negative causal association between AAM, AFB and NAFLD, suggesting that early AAM, AFB may increase the risk of NAFLD. However, we did not find a clear association between ANM and NAFLD. These results provide valuable insights into the underlying biological mechanisms of NAFLD and suggest that AAM, AFB may be a potential risk factor for NAFLD. Furthermore, immune cells may mediate the effects of AAM on NAFLD. Our findings on the causal relationship between AAM, AFB, and NAFLD confirm previous epidemiologic studies. A paired cohort study including a multiethnic population showed that age at menarche was associated with the risk of NAFLD; women with age at menarche >11 years had a lower risk of NAFLD compared with women with age at menarche <11 years(30).

A retrospective cohort study in the United States found that the odds of hepatic steatosis were significantly lower when women had an AFB >35 years, which may be related to waist circumference(31). Dividing women by age revealed that women with AFB <18, 18-20, 21-23, and 24-26 years of age had a 54%, 40%, 60%, and 33% increased risk of NAFLD, respectively, compared with participants with AFB aged 30-32 years(16)o Relevant previous studies have shown that the prevalence of NAFLD in women becomes larger with age, with 3.9% in the 21-39 year old group; 7.6% in the 40-49 year old group; and a gradual increase in prevalence after age 50 ( about 14%), peaking at age 60-69 years ( about 18.9%), and a gradual increase in prevalence after age 60-69 years ( about 18.9%) and decreases after the age of 70 years(32).

Unlike past studies, our MR study did not find a correlation between age at menopause and NAFLD. Early menarche can have negative effects on women’s health, such as psychiatric disorders, metabolic syndrome, and cardiovascular disease(33, 34). The age at menarche has a significant impact on the body’s metabolic function. Early menarche is a risk factor for metabolic syndrome. When a woman’s age at menarche decreases by one year, the risk of metabolic syndrome increases by 8%(35). NAFLD involves multi-system disease and is not only a predictor of metabolism but is also strongly associated with metabolic syndrome(36). It hasalsobeensuggestedthatthenegativeassociationbetween AAM and NAFLD is mediated by obesity(37). A meta-analysis of genome-wide association studies found that several novel loci for age at menarche overlap with genetic loci related to weight regulation, BMI.

In addition, studies of the majority of alleles associated with higher BMI and waist-to-hip ratio also showed an association with early menarche(38). Obesity leads to the release of large amounts of inflammatory adipokines and excess free fatty acids in the body, which promote hepatic steatosis and lipotoxicity(39). In female rats with obesity due to overfeeding, not only do they show precocious puberty, but also lead to increased levels of synthesized ceramides in the hypothalamus(40). Ceramides have an important role in the onset and progression of NAFLD. Ceramides are a class of bioactive sphingolipids composed of fatty acid chains that act as signaling molecules that are not only involved in cell proliferation, apoptosis, and senescence, and modulate stress responses, but also cause lipids to accumulate in the liver(41).

The continuous accumulation of ceramides in the liver induces insulin resistance, mediates apoptosis, increases oxidative stress and inflammation, and ultimately leads to lipid deposition, inflammation and fibrosis(42). Early AFB causes metabolic disturbances in the body, which increase the risk of obesity, hypertension, and metabolic diseases(43). These risks are closely associated with NAFLD. The immature physical development of adolescents, changes in hormone levels during pregnancy, and increased insulin resistance induced by hyperglycemia may be detrimental to normal organ development(44). Evidence from previous cross-sectional studies suggests that overweight adolescents are associated with an increased risk of NAFLD in multivariate regression models, with the prevalence of NAFLD in overweight adolescents as high as 34.2%.

In addition, from a socioeconomic perspective, women with early AFB are more likely to have low levels of education and mental immaturity, which is frequently accompanied by health-hazardous behaviors such as smoking, violence, and alcohol abuse(45). Estrogen inhibits glucagon secretion, regulates lipogenesis and fatty acid oxidation, and intervenes in reverse cholesterol output(46–48). Therefore, estrogen is considered a protective factor in NAFLD. After menopause, estrogen levels decline severely in women. Numerous studies have shown that the risk of NAFLD is significantly increased in menopausal women (15, 32). Our study did not find an association between ANM and NAFLD, which may be related to hormone replacement therapy (HRT). The results of a randomized controlled trial on HRT showed that low doses of HRT were effective in reducing serum liver function enzyme concentrations.

Results of a randomized controlled trial of HRT showed that low doses of HRT were effective in reducing serum liver function enzyme concentrations(49). Another reported that waist circumference, HOMA-IR index, ferritin levels, and glutamyltransferase were all lower in postmenopausal women on HRT compared to postmenopausal women who did not use HRT(50). Liver histology using abdominal ultrasound found that The incidence of NAFLD was lower in women taking hormone replacement therapy than in women not taking hormone replacement therapy(51). We observed an association between CD25 on switched memory B cell and a reduced risk of NAFLD. CD127 on CD28+ CD4-CD8-TcellincreasestheriskofNAFLD.Thesetwoimmune cells may mediate the effects of AAM on NAFLD. Memory B cells are a specialized subtype of B cells that differentiate to counteract their original recognition function after the first infection, and play an important role in generating an accelerated and more potent antibody-mediated immune response (also known as the secondary immune response) in the event of re-infection(52).

T cells are lymphocytes that mature in the thymus. Conventional T cells can be broadly categorized into CD8 and CD4 T cells Regulation of CD8 T cell abundance, activity, or function has been associated with the evolution of nonalcoholic steatohepatitis (NASH)(53). CD4 T cells are recruited to the liver through integrin-ligand interactions, exacerbating hepatic inflammation and fibrosis and ultimately accelerating the development of NAFLD(54). For women with AAM, valuing immune cells can help better research and treatment of NAFDL. The rate of NAFLD has been increasing yearly, causing enormous harm to public health and socio economics. Therefore, the identification and screening of people with high risk of NAFLD has become particularly important.

In this MR study, we found a significant association between AAM, AFB and NAFLD. More attention from multidisciplinary physicians and educators is necessary for women with early AAM, AFB. Understand the dangers associated with early menarche and premature childbearing with special attention to active monitoring and management of NAFLD in this group. Finally, further research is required on the link between female reproductive factors and NAFLD in order to provide solutions for policy makers to attack NAFLD. Strengths and Limitations Our findings show a causal relationship between AAM, AFB and NAFLD and that women’s first menstrual period as well as young age at first child birth may increase the risk of NAFLD.

MR has the advantage of controlling for confounding factors compared to traditional observational studies. Using recently published large-scale GWAS data, we carried out several sensitivity analyses and F-corrections to adjust for confounders to assess the robustness of our results. There are several limitations of this study to note. First, we selected a population of European ancestry, which reduces the influence of population stratification bias on the results but also limits the applicability of our results to other ethnic populations. Second, we used NAFLD as the only outcome indicator without delving into the associated mechanisms of action and the impact of AAM, ANM, and AFB on other liver diseases. Finally, our study design was not stratified by age, and thus we were unable to obtain the effect of reproductive factors on the risk of NAFLD in women at different ages.

Despite revealing a causal relationship between AAM, AFB and NAFLD in women, the underlying mechanisms are unknown.

Conclusion In conclusion, our findings indicated that reproductive factors (AAM, AFB) play an important role in NAFLD risk. Clinical trials and observational studies are still needed to discover more information about the mechanisms involved. Acknowledgments This research was funded by National Natural Science Foundation of China (82174517). Disclosures The authors have no financial conflicts of interest.

REFERENCES

1. Lazarus, J. V., H. E. Mark, Q. M. Anstee, J. P. Arab. 2022. Advancing the global public health agenda for NAFLD: a consensus statement. Nat. Rev. Gastroenterol. Hepatol. 19: 60–78.

2. Riazi, K., H. Azhari, J. H. Charette, F. E. Underwood, J. A. King, E. E. Afshar, M. G. Swain, S. E. Congly, G. G. Kaplan, and A.-A. Shaheen. 2022. The prevalence and incidence of NAFLD worldwide: a systematic review and metaanalysis. Lancet Gastroenterol. Hepatol. 7: 851–861.

3. Perumpail, B. J., M. A. Khan, E. R. Yoo, G. Cholankeril, D. Kim, and A. Ahmed. 2017. Clinical epidemiology and disease burden of nonalcoholic fatty liver disease. World J. Gastroenterol. 23: 8263–8276.

4. Targher, G., M. Rossini, and A. Lonardo. 2016. Evidence that non-alcoholic fatty liver disease and polycystic ovary syndrome are associated by necessity rather than chance: a novel hepato-ovarian axis? Endocrine 51: 211–221.

5. Janssens, G. E., M. Molenaars, K. Herzog, L. Grevendonk, C. M. E. Remie, M. A. T. Vervaart, H. L. Elfrink, E. J. M. Wever, B. V. Schomakers, S. W. Denis, H. R. Waterham, M. L. Pras-Raves, M. van Weeghel, A. H. C. van Kampen, A. Tammaro, L. M. Butter, S. van der Rijt, S. Florquin, A. Jongejan, P. D. Moerland, J. Hoeks, P. Schrauwen, F. M. Vaz, and R. H. Houtkooper. 2024. A conserved complex lipid signature marks human muscle aging and responds to short-term exercise. Nat. Aging .

6. Jha, P., M. T. McDevitt, R. Gupta, P. M. Quiros, E. G. Williams, K. Gariani, M. B. Sleiman, L. Diserens, A. Jochem, A. Ulbrich, J. J. Coon, J. Auwerx, and D. J. Pagliarini. 2018. Systems Analyses Reveal Physiological Roles and Genetic Regulators of Liver Lipid Species. Cell Syst. 6: 722-733.e6.

7. Barsalani, R., A. Pighon, R. Rabasa-Lhoret, S. Yasari, and J.-M. Lavoie. 2008. Liver of ovariectomized rats is resistant to resorption of lipids. Physiol. Behav. 95: 216– 221.

8. Malinská, H., M. Hüttl, D. Miklánková, J. Trnovská, I. Zapletalová, M. Poruba, and I. Marková. 2021. Ovariectomy-Induced Hepatic Lipid and Cytochrome P450 Dysmetabolism Precedes Serum Dyslipidemia. Int. J. Mol. Sci. 22: 4527.

9. Lee, C., J. Kim, and Y. Jung. 2019. Potential Therapeutic Application of Estrogen in Gender Disparity of Nonalcoholic Fatty Liver Disease/Nonalcoholic Steatohepatitis. Cells 8: 1259.

10. Anagnostis,P.,P.Galanis,V.Chatzistergiou,J.C.Stevenson, I. F. Godsland, I. Lambrinoudaki, M. Theodorou, and D. G. Goulis. 2017. The effect of hormone replacement therapy and tibolone on lipoprotein (a) concentrations in postmenopausal women: A systematic review and meta-analysis. Maturitas 99: 27–36.

11. Wang, Y., and Y. Wang. 2022. Accelerated Ovarian Aging Among Type 2 Diabetes Patients and Its Association With Adverse Lipid Profile. Front. Endocrinol. 13: 780979.

12. Perry, J. R., F. Day, C. E. Elks, P. Sulem, D. J. Thompson, T. 2014. Parent-of-origin-specific allelic associations among 106 genomic loci for age at menarche. Nature 514: 92–97.

13. Zhang, X., Z. Huangfu, and S. Wang. 2023. Review of mendelian randomization studies on age at natural menopause. Front. Endocrinol. 14: 1234324.

14. Mueller, N. T., M. A. Pereira, E. W. Demerath, J. G. Dreyfus, R. F. MacLehose, J. J. Carr, J. G. Terry, and D. R. Jacobs. 2015. Earlier menarche is associated with fatty liver and abdominal ectopic fat in midlife, independent of young adult BMI: The CARDIA study. Obes. Silver Spring Md 23: 468–474.

15. Klair, J. S., J. D. Yang, M. F. Abdelmalek, C. D. Guy, R. M. Gill, K. Yates, A. Unalp-Arida, J. E. Lavine, J. M. Clark, A. M. Diehl, A. Suzuki, and Nonalcoholic Steatohepatitis Clinical Research Network. 2016. A longer duration of estrogen deficiency increases fibrosis risk among postmenopausal women with nonalcoholic fatty liver disease. Hepatol. Baltim. Md 64: 85–91.

16. Yang, H.-H., G.-C. Chen, M.-G. Zhou, L.-F. Xie, Y.-Y. Jin, H.- T. Chen, Z.-K. Chen, Y.-H. Kong, C.- Z. Yuan, and Z.-H. Li. 2023. Association of age at first birth and risk of nonalcoholic fatty liver disease in women: evidence from the NHANES. Hepatol. Int. 17: 303–312.

17. Heymann,F.,andF.Tacke.2016.Immunologyintheliver- -from homeostasis to disease. Nat. Rev. Gastroenterol. Hepatol. 13: 88–110.

18. Racanelli, V., and B. Rehermann. 2006. The liver as an immunological organ. Hepatol. Baltim. Md 43: S54-62.

19. Lin, S.-Z., and J.-G. Fan. 2022. Peripheral immune cells in NAFLD patients: A spyhole to disease progression. EBioMedicine 75: 103768.

20. Kazankov, K., S. M. D. Jørgensen, K. L. Thomsen, H. J. Møller, H. Vilstrup, J. George, D. Schuppan, and H. Grønbæk.2019.Theroleofmacrophagesinnonalcoholic fatty liver disease and nonalcoholic steatohepatitis. Nat. Rev. Gastroenterol. Hepatol. 16: 145–159.

21. Lee, S. K., C. J. Kim, D.-J. Kim, and J. Kang. 2015. Immune Cells in the Female Reproductive Tract. Immune Netw. 15: 16–26.

22. Wang, F., A. E. Qualls, L. Marques-Fernandez, and F. Colucci. 2021. Biology and pathology of the uterine microenvironment and its natural killer cells. Cell. Mol. Immunol. 18: 2101–2113.

23. Orrù, V., M. Steri, C. Sidore, M. Marongiu, V. Serra, S. Olla, G. Sole, S. Lai, M. Dei, A. Mulas, F. Virdis, M. G. Piras, M. Lobina, M. Marongiu, M. Pitzalis, F. Deidda, A. Loizedda, S. Onano, M. Zoledziewska, S. Sawcer, M. Devoto, M. Gorospe, G. R. Abecasis, M. Floris, M. Pala, D. Schlessinger, E. Fiorillo, and F. Cucca. 2020. Complex genetic signatures in immune cells underlie autoimmunity and inform therapy. Nat. Genet. 52: 1036–1045.

24. Davey Smith, G., and G. Hemani. 2014. Mendelian randomization: genetic anchors for causal inference in epidemiological studies. Hum. Mol. Genet. 23: R89-98.

25. Day, F. R., K. S. Ruth, D. J. Thompson, K. L. Lunetta, N. 2015. Large-scale genomic analyses link reproductive aging to hypothalamic signaling, breast cancer susceptibility and BRCA1-mediated DNA repair. Nat. Genet. 47: 1294–1303.

26. Mills, M. C., F. C. Tropf, D. M. Brazel, N. van Zuydam, A. Vaez, eQTLGen Consortium, BIOS Consortium, Human Reproductive Behaviour Consortium, T. H. Pers, H. Snieder, J. R. B. Perry, K. K. Ong, M. den Hoed, N. Barban, and F. R. Day. 2021. Identification of 371 genetic variants for age at first sex and birth linked to externalising behaviour. Nat. Hum. Behav. 5: 1717–1730.

27. Ghodsian, N., E. Abner, C. A. Emdin, É. Gobeil, N. Taba, M. E. Haas, N. Perrot, H. D. Manikpurage, É. Gagnon, J. Bourgault, A. St-Amand, C. Couture, P. L. Mitchell, Y. Bossé, P. Mathieu, M.-C. Vohl, A. Tchernof, S. Thériault, A. V. Khera, T. Esko, and B. J. Arsenault. 2021. Electronic health record-based genome-wide meta-analysis provides insights on the genetic architecture of nonalcoholic fatty liver disease. Cell Rep. Med. 2: 100437.

28. Li, J., L. Zhao, X. Ding, X. Cui, L. Qi, and Y. Chen. 2022. Obstructive sleep apnea and the risk of Alzheimer’s disease and Parkinson disease: A Mendelian randomization study OSA, Alzheimer’s disease and Parkinson disease. Sleep Med. 97: 55–63.

29. Wang, A., and J. Zhang. 2024. Causal role of immune cells in psoriasis: a Mendelian randomization analysis. Front. Immunol. 15: 1326717.

30. Wang, J., A. H. Wu, F. Z. Stanczyk, J. Porcel, M. Noureddin, N. A. Terrault, L. R. Wilkens, and V. W. Setiawan. 2021. Associations Between Reproductive and Hormone- Related Factors and Risk of Nonalcoholic Fatty Liver Disease in a Multiethnic Population. Clin. Gastroenterol. Hepatol. Off. Clin. Pract. J. Am. Gastroenterol. Assoc. 19: 1258-1266.e1.

31. Kim, C., S. D. Harlow, S. Kong, C. Karvonen-Gutierrez, K. Ylitalo, and B. Nan. 2015. Reproductive history and chronic hepatic steatosis in the Michigan Study of Women’s Health Across the Nation. J. Womens Health 2002 24: 42–48.

32. Younossi, Z. M., A. B. Koenig, D. Abdelatif, Y. Fazel, L. Henry, and M. Wymer. 2016. Global epidemiology of nonalcoholic fatty liver disease-Meta-analytic assessment of prevalence, incidence, and outcomes. Hepatol. Baltim. Md 64: 73–84.

33. Li, J., Y. Chongpison, J. Amornvit, S. Chaikittisilpa, S. Santibenchakul, and U. Jaisamrarn. 2023. Association of reproductive factors and exogenous hormone use with distal sensory polyneuropathy among postmenopausal women in the United States: results from 1999 to 2004 NHANES. Sci. Rep. 13: 9274.

34. Yu, Y., L. Hou, Y. Wu, Y. Yu, X. Liu, S. Wu, Y. He, Y. Ge, Y. Wei, F. Qian, Q. Luo, Y. Feng, X. Cheng, T. Yu, H. Li, and F. Xue. 2023. Causal associations between female reproductive behaviors and psychiatric disorders: a lifecourse Mendelian randomization study. BMC Psychiatry 23: 799.

35. Kim, Y., and Y. Je. 2019. Early Menarche and Risk of Metabolic Syndrome: A Systematic Review and Meta- Analysis. J. Womens Health 2002 28: 77–86.

36. Mikolasevic, I., S. Milic, T. Turk Wensveen, I. Grgic, I. Jakopcic, D. Stimac, F. Wensveen, and L. Orlic. 2016. Nonalcoholic fatty liver disease - A multisystem disease? World J. Gastroenterol. 22: 9488– 9505.

37. Ryu, S., Y. Chang, Y. Choi, M.-J. Kwon, C.-W. Kim, K. E. Yun, H.-S. Jung, B.-K. Kim, Y. J. Kim, J. Ahn, Y. K. Cho, K.-H. Kim, E. C. Chung, H. Shin, and J. Cho. 2015. Age at menarche and non-alcoholic fatty liver disease. J. Hepatol. 62: 1164–1170.

38. Elks, C. E., J. R. B. Perry, P. Sulem, D. I. Chasman, N. 2010. Thirty new loci for age at menarche identified by a meta-analysis of genome-wide association studies. Nat. Genet. 42: 1077–1085.

39. Abranches, M. V., F. C. E. de Oliveira, L. L. da Conceição, and M. do C. G. Peluzio. 2015. Obesity and diabetes: the link between adipose tissue dysfunction and glucose homeostasis. Nutr. Res. Rev. 28: 121–132.

40. Heras, V., J. M. Castellano, D. Fernandois, I. Velasco, E. Rodríguez-Vazquez, J. Roa, M. J. Vazquez, F. Ruiz- Pino, M. Rubio, R. Pineda, E. Torres, M. S. Avendaño, A. Paredes, L. Pinilla, D. Belsham, C. Diéguez, F. Gaytán, N. Casals, M. López, and M. Tena-Sempere. 2020. Central Ceramide Signaling Mediates Obesity-Induced Precocious Puberty. Cell Metab. 32: 951-966.e8.

41. Mencarelli, C., and P. Martinez-Martinez. 2013. Ceramide function in the brain: when a slight tilt is enough. Cell. Mol. Life Sci. CMLS 70: 181–203.

42. Zhu, C., Q. Huai, X. Zhang, H. Dai, X. Li, and H. Wang. 2023. Insights into the roles and pathomechanisms of ceramide and sphigosine-1-phosphate in nonalcoholic fatty liver disease. Int. J. Biol. Sci. 19: 311–330.

43. Sim, J. H., D. Chung, J. S. Lim, M. Y. Lee, C. H. Chung, J. Y. Shin, and J. H. Huh. 2015. Maternal age at first delivery is associated with the risk of metabolic syndrome in postmenopausal women: from 2008-2010 Korean National Health and Nutrition Examination Survey. PloS One 10: e0127860.

44. Ryan, E. A. 2003. Hormones and insulin resistance during pregnancy. Lancet Lond. Engl. 362: 1777– 1778.

45. Nairz, J., A. Messner, S. J. Kiechl, B. Winder, C. Hochmayr, A. E. Egger, A. Griesmacher, R. Geiger, E. Griesmaier, R. Pechlaner, M. Knoflach, U. Kiechl-Kohlendorfer, and Early Vascular Ageing in the YOUth (EVA4YOU) Study Group. 2024. Determinants of non-alcoholic fatty liver disease in young people: Maternal, neonatal, and adolescent factors. PloS One 19: e0298800.

46. Zhu, L., J. Shi, T. N. Luu, J. C. Neuman, E. Trefts, S. Yu, B. T. Palmisano, D. H. Wasserman, M. F. Linton, and J. M. Stafford. 2018. Hepatocyte estrogen receptor alpha mediates estrogen action to promote reverse cholesterol transport during Western-type diet feeding. Mol. Metab. 8: 106–116.

47. Mandour, T., A. H. Kissebah, and V. Wynn. 1977. Mechanism of oestrogen and progesterone effects on lipid and carbohydrate metabolism: alteration in the insulin: glucagon molar ratio and hepatic enzyme activity. Eur. J. Clin. Invest. 7: 181–187.

48. Paquette, A., N. A. Chapados, R. Bergeron, and J.-M. Lavoie. 2009. Fatty acid oxidation is decreased in the liver of ovariectomized rats. Horm. Metab. Res. Horm. Stoffwechselforschung Horm. Metab. 41: 511–515.

49. McKenzie,J.,B.M.Fisher,A.J.Jaap,A.Stanley,K.Paterson, and N. Sattar. 2006. Effects of HRT on liver enzyme levels in women with type 2 diabetes: a randomized placebocontrolled trial. Clin.Endocrinol. (Oxf.) 65: 40–44.

50. Florentino, G. S. de A., H. P. Cotrim, C. P. Vilar, A. V. de A. Florentino, G. M. A. Guimarães, and V. S. T. Barreto. 2013. Nonalcoholic fatty liver disease in menopausal women. Arq. Gastroenterol. 50: 180–185.

51. Florentino, G., H. P. Cotrim, A. Florentino, C. Padilha, M. Medeiros-Neto, G. Bragagnol, and P. Schwingel. 2012. Hormone replacement therapy in menopausal women: risk factor or protection to nonalcoholic fatty liver disease? Ann. Hepatol. 11: 147–149.

52. Akkaya,M.,K.Kwak,andS.K.Pierce.2020.Bcellmemory: building two walls of protection against pathogens. Nat. Rev. Immunol. 20: 229–238.

53. Hirsova, P., A. O. Bamidele, H. Wang, D. Povero, and X. S. Revelo. 2021. Emerging Roles of T Cells in the Pathogenesis of Nonalcoholic Steatohepatitis and Hepatocellular Carcinoma. Front. Endocrinol. 12: 760860.

54. Rai, R. P., Y. Liu, S. S. Iyer, S. Liu, B. Gupta, C. Desai, P. Kumar, T. Smith, A. D. Singhi, A. Nusrat, C. A. Parkos, S. P. Monga, M. J. Czaja, F. A. Anania, and R. Raeman. 2020. Blocking integrin α4β7- mediated CD4 T cell recruitment to the intestine and liver protects mice from western diet-induced non- alcoholic steatohepatitis. J. Hepatol. 73: 1013–1022. Research Article