A causal examination of the correlation between hormonal and reproductive factors and low back pain

Background The relationship between hormonal fluctuations in the reproductive system and the occurrence of low back pain (LBP) has been widely observed. However, the causal impact of specific variables that may be indicative of hormonal and reproductive factors, such as age at menopause (ANM), age at menarche (AAM), length of menstrual cycle (LMC), age at first birth (AFB), age at last live birth (ALB) and age first had sexual intercourse (AFS) on low back pain remains unclear. Methods This study employed Bidirectional Mendelian randomization (MR) using publicly available summary statistics from Genome Wide Association Studies (GWAS) and FinnGen Consortium to investigate the causal links between hormonal and reproductive factors on LBP. Various MR methodologies, including inverse-variance weighted (IVW), MR-Egger regression, and weighted median, were utilized. Sensitivity analysis was conducted to ensure the robustness and validity of the findings. Subsequently, Multivariate Mendelian randomization (MVMR) was employed to assess the direct causal impact of reproductive and hormone factors on the risk of LBP. Results After implementing the Bonferroni correction and conducting rigorous quality control, the results from MR indicated a noteworthy association between a decreased risk of LBP and AAM (OR=0.784, 95% CI: 0.689-0.891; p=3.53E-04), AFB (OR=0.558, 95% CI: 0.436-0.715; p=8.97E-06), ALB (OR=0.396, 95% CI: 0.226-0.692; p=0.002), and AFS (OR=0.602, 95% CI: 0.518-0.700; p=3.47E-10). Moreover, in the reverse MR analysis, we observed no significant causal effects of LBP on ANM, AAM, LMC and AFS. MVMR analysis demonstrated the continued significance of the causal effect of AFB on LBP after adjusting for BMI. Conclusion Our study explored the causal relationship between ANM, AAM, LMC, AFB, AFS, ALB and the prevalence of LBP. We found that early menarche, early age at first birth, early age at last live birth and early age first had sexual intercourse may decrease the risk of LBP. These insights enhance our understanding of LBP risk factors, offering valuable guidance for screening, prevention, and treatment strategies for at-risk women.


Introduction
Low back pain (LBP) is a prevalent public health issue, affecting approximately 60-80% of individuals at various stages of their lives (1,2).Intervertebral disc degeneration is a major contributing factor to LBP and a noticeable trend towards its occurrence at younger ages has been observed (3,4).The prevalence of LBP is generally higher among women than men, which can be attributed to factors such as increased pain sensitivity, variations in the menstrual cycle, physiological responses to pregnancy and childbirth, and abdominal weight gain during the perimenopausal phase (5)(6)(7)(8)(9)(10)(11).Some studies have found a higher propensity for LBP among postmenopausal women compared to men of equivalent age (12).There was also evidence of an increased likelihood of LBP in individuals undergoing postmenopausal hormone therapy (13,14).However, conflicting perspectives exist, with some suggesting potential positive outcomes associated with hormone therapy (15)(16)(17).These divergent views highlighted the potential significance of hormonal and reproductive factors in the pathogenesis and progression of LBP.
A strong connection has been established between hormonal factors, such as age at menopause (ANM), age at menarche (AAM), length of menstrual cycle (LMC), and age at first birth (AFB), and the occurrence of LBP.Various studies have identified associations between these factors and the risk of developing LBP, although the causal relationship between these remains unclear (18)(19)(20)(21).
Observational studies on this subject were prone to bias due to confounding factors and reverse causality.To overcome these limitations, researchers have proposed the use of Mendelian randomization (MR) analysis.MR is a genetic epidemiological approach that uses single nucleotide polymorphisms (SNPs) as instrumental variables (IVs) for risk factors, allowing for the assessment of potential causal effects of exposure on outcomes (22).This method is based on Mendel's second law (23), asserting that alleles are randomly allocated during meiosis and are typically unaffected by environmental influences (22,24).
Prior to this investigation, MR analyses had not been used to explore the causal relationship between hormonal and reproductive factors and LBP.Therefore, we conducted the MR analysis focusing on four female hormonal and reproductive factors (ANM, AAM, LMC, AFB, AFS and ALB), examining their associations with LBP.Subsequently, Multivariate Mendelian randomization (MVMR) was employed to assess the direct causal impact of reproductive and hormone factors on the risk of LBP.These findings may enhance our understanding of the hormonal and reproductive mechanisms underlying LBP, guiding future research towards developing potential therapeutic or preventative strategies.

Study design and data sources
We conducted a comprehensive analysis using publicly accessible Genome-Wide Association Studies (GWAS) database to explore the potential causal relationship between ANM, AAM, LMC, AFB, AFS and ALB, and the occurrence of LBP.A comprehensive overview of the proposed hypotheses was presented in (Figure 1).The current study adhered to the three fundamental assumptions essential for MR analyses (25): assumption 1, all chosen IVs exhibit a strong correlation with the exposure; assumption 2 the selected instrumental variables are independent of both exposure and outcome confounders; assumption 3, the selected instrumental variables impact the outcome solely through exposure.Previous MR studies have established BMI as a risk factor for LBP (26).And a reverse MR analysis was conducted to evaluate potential reverse causality.Consequently, we conducted MVMR to address this potential confounding factor.The exposure data were obtained from the GWAS database (https://gwas.mrcieu.ac.uk/).Data on LBP was sourced from the FinnGen Consortium (https://finngen.fi).The summary data for the GWAS of LBP from the FinnGen Consortium comprises 177,860 participants of European ancestry (13,178 cases and 164,682 controls).Summary information for all datasets were presented in (Table 1) .All participants were of European origin, and informed consent was obtained from each.Since our data were derived from publicly accessible GWAS summary statistics, no ethical approval was necessary.

Selection of instrumental variables
Firstly, we carefully selected SNPs that demonstrated a strong association with exposure (P < 5 × 10 -8 ) and excluded SNPs with Fvalues < 10, ensuring significance and mitigating weak instrumental variable bias (27).Secondly, we utilized specific parameters (r 2 < 0.001, kb = 10,000 kb) to eliminate strong linkage disequilibrium, thus guaranteeing instrumental variable independence (28).Thirdly, we excluded SNPs associated with confounders and results using Phenoscanner V2.Additionally, palindromic SNPs with moderate allele frequencies were subsequently removed.Ultimately, we assessed the instrument strength through the F parameter, calculated using the formula F = R² × (n -2)/(1 -R²), where R² represents the proportion of variance in instruments.The formula for R² is given by R² = 2 × effect allele frequency × (1 -effect allele frequency) × (Beta/SD², with SD equaling 1), and n denotes the sample size.An F statistic exceeding 10 indicated a diminished likelihood of weak instrument bias.

Statistical analysis
In MR and MVMR analyses, the primary method employed was inverse variance weighting (IVW), complemented by MR-Egger, weighted median, simple mode, and weighted mode (29).In the absence of weak IVs, the primary outcome was determined using the IVW method, with the alternative methods considered as secondary outcomes.We employed MVMR as a statistical approach to incorporate SNP-phenotype associations into the analysis, facilitating the estimation of each phenotype's direct impact on the outcome.As indicated by previous studies (26), in MVMR, we adjusted for body mass index (BMI) to clarify the causal impact of hormonal and reproductive factors on LBP.

Heterogeneity and sensitivity test
Cochrane's Q-test was utilized to detect heterogeneity, while funnel plots indicated heterogeneity through symmetry (30).The MR-Egger intercept test and the MR polytomous residuals and outliers (MR-PRESSO) global test were employed to assess pleiotropy (31).If significant pleiotropy was identified through the MR-PRESSO method, we will mitigate this concern by addressing outlier variability and subsequently reiterating the MR analysis.Lastly, the leave-one-out test was conducted to evaluate the sensitivity of the results.We utilized the TwoSample MR, MVMR, and MR-PRESSO packages in R software (version 4.3.1).Statistically significant associations were defined by results with a p-value < 0.05.

Instrumental variables selection
After conducting a comprehensive quality assessment, we incorporated SNPs as reliable IVs for ANM, AAM, LMC, AFB, AFS, ALB and BMI.Detailed information regarding these IVs were provided in Supplementary Tables S1-S9.Notably, all the selected SNPs utilized as IVs possess F values exceeding 10, indicating their effectiveness as IVs.
In MVMR analysis adjusting for BMI, AFB (OR=0.522,95% CI: 0.313-0.869;p=0.012) exhibited a significant association with LBP.The MR-Lasso test results remained unaffected by the removal of heterogeneous SNPs.Nevertheless, associations between AAM, ALB, and AFS with LBP did not persist after further adjustment for BMI.Detailed MVMR results are presented in Figure 3.

MR analysis of each feature related to hormonal and reproductive factors on LBP (validation analysis)
After implementing the Bonferroni correction, the results from MR indicated a noteworthy association between a decreased risk of LBP and AAM, AFB, ALB and AFS.Nevertheless, no significant association was observed between ANM and LMC with LBP (Figure 5).Scatter plots, funnel plots and leave-one-out plots illustrating the association between reproductive and hormonal factors and LBP were presented in Supplementary Figures S7-S9.Heterogeneity and pleiotropy are depicted in Table 4.

Discussion
Our study utilized a two-sample MR analysis to evaluate the potential causal effects of six hormonal and reproductive factors on the development of LBP.We uncovered novel insights regarding the influence of AAM, AFB, ALB and AFS on LBP.Through Bonferroni correction, we identified a negative causal relationship between these factors and the aforementioned spinal conditions.Specifically, early menarche, early age at first birth, early age at last live birth and early age first had sexual intercourse may elevate the risk of LBP.The verification results were consistent with the initial findings.After controlling for BMI, the association between AFB and LBP persisted, while the correlation between AAM, ALB, AFS and LBP did not endure.These insights underscore the importance of investigating hormonal and reproductive factors in spinal health, providing valuable directions for future research and clinical applications.We also recommend enhanced monitoring of women with these characteristics to proactively manage LBP.
Numerous observational studies have substantiated the connection between hormonal factors, reproductive factors and LBP.Nevertheless, there remains uncertainty regarding the potential influence of ANM, AAM, LMC, AFB, AFS and ALB on the development of LBP.The outcomes of the longitudinal cohort investigation aligned with our findings, affirming that an earlier AAM onset was associated with an increased likelihood of experiencing LBP (19).Other studies have also noted a positive association, with a cross-sectional study of more than 298,000 women discovering a positive link between early menarche and LBP (p<0.001) (32).Onset of menarche at age less than 11 years has been linked to a higher risk of experiencing LBP, as indicated by findings from both cross-sectional and cohort studies (18).However, it has also been shown that no association was found between ANM or AAM and risk of LBP (33).The existence of these contradictions could be attributed to potential bias in traditional epidemiological methods caused by confounding variables.Thus, employing MR methods could elucidate causality at the genetic level.
Many studies have shown that the prevalence of LBP in women was not significantly correlated with age, and the prevalence of LBP in the postmenopausal period was significantly different from that in the premenopausal period (34,35).However, the Mexican study  The causal relationship of genetically predicted LBP and Hormonal and Reproductive Factors.revealed that women with back pain were more likely to be older (36).Adera et al. conducted a population-based cross-sectional study that elucidates a noteworthy correlation between premature menopause and an escalated susceptibility to LBP (37).Our findings at the genetic level provide evidence that ANM was not causally associated with LBP, corroborating prior studies.Instead, the occurrence of LBP and IVDD in menopausal women might be related to a rapid decrease in androgen levels.Scholarly investigations have predominantly utilized menarche as a parameter in delineating pubertal onset.However, pubertal development was a complex process that entails a spectrum of changes across various bodily systems (38).Furthermore, researchers concur that the commencement of menarche may not be the optimal indicator, as a substantial portion of growth and the emergence of secondary sexual characteristics precede its occurrence (39, 40).Prolonged and heightened exposure to estrogen over an extended period was postulated as an additional contributory factor to the increased susceptibility to LBP among women displaying early onset of menarche (18,41).
A cross-sectional study showed that younger maternal age at the time of first birth (especially <20 years) was associated with chronic LBP, which was similar to our results (21).Meanwhile, in a prospective study, a statistically significant distinction was noted in the prevalence of LBP during pregnancy between younger and older women (42).Meanwhile, Heuch et al. have reported an association The causal relationship of genetically predicted Hormonal and Reproductive Factors and LBP (replication analysis).between the incidence of lumbar discomfort and advancing age, as well as the cumulative instances of pregnancies (43).Our investigation revealed an observation wherein a heightened susceptibility to dorsal discomfort was discerned among youthful females.The MR method employed mitigates biases arising from various factors, including confounding, through genetic allelic assignment principles.This method corroborates, at the genetic level, the notion that an early AFB constitutes a risk factor for LBP.This phenomenon could stem from elevated hormone levels that impact the soft tissues supporting the spine, potentially leading to enduring laxity in joints and ligaments (43)(44)(45)(46).This correlation aligns with an elevated risk of LBP observed in women undergoing hormone replacement therapy or using oral contraceptives (17, 20).Additionally, younger women demonstrate heightened sensitivity to hormonal variations in estrogen and relaxin, leading to more pronounced collagen relaxation (47,48).This sensitivity may elucidate the increased risk of LBP among women giving birth at a younger age.Moreover, compression of the uterus on the developing spine during the first childbirth in younger girls may contribute to the onset of low back pain (43).The prospective study by Brynhildsen et al. found that hormonal fluctuations during the menstrual cycle do not influence LBP (21).In contrast, Wijnhoven et al. identified a link between chronic LBP and irregular or prolonged menstrual cycles (20).Our study aligns with Brynhildsen et al.'s conclusion that shorter menstrual cycles are not associated with an increased risk of lower back pain LBP.This suggests that the menstrual cycle length is not a risk factor for LBP.
Several studies have documented the increasing severity of IVDD in women as they age (49-51), with a notably more rapid degeneration observed in females after the age of 60 compared to males (52).Epidemiological evidence supports the notion that disc degeneration correlates with age (53).This phenomenon was similarly observed by De Schepper et al. (54).The role of estrogen in IVD metabolism and its expression in annulus fibrosus and nucleus pulposus cells may explain these observations (55).IVD is the primary cause of LBP, with hormone levels playing a crucial role.Further investigation is needed to understand the specific mechanism of action.
Our study possesses several strengths.It marks the inaugural application of MR to investigate the causal relationship between hormonal and reproductive factors and LBP.Encompassing six distinct reproductive characteristics, our study offers a comprehensive understanding of the reproductive period.Utilizing data from a diverse range of cohorts enhances the reliability of our findings and minimizes overlap.Employing the principle of random allele assignment, we conducted a Bidirectional MR study to validate the robustness of these results.Furthermore, we corroborated the reliability of our conclusions through MVMR, with adjustments made for BMI.
However, the exclusive reliance on European GWAS data may limit the generalizability of our findings to other ethnic or geographic populations.Besides, the inclusion of both genders in the outcome data might also weaken the observed associations.The inclusion of both genders in the dataset introduced gender heterogeneity and potential bias.Ideally, the association between SNPs and outcome estimates should display gender heterogeneity.However, in the LBP GWAS database we used, with women comprising over 60%, it represents a predominantly female-led GWAS, thereby minimizing the likelihood of bias.Future MR studies should consider validating these results within female-only samples by appropriate stratification.

Conclusions
In conclusion, our study explored the causal relationship between ANM, AAM, LMC, AFB, AFS, ALB and the prevalence of LBP.We found that early menarche, early age at first birth, early age at last live birth and early age first had sexual intercourse may decrease the risk of LBP.These insights enhance our understanding of LBP risk factors, offering valuable guidance for screening, prevention, and treatment strategies for at-risk women.

FIGURE 1
FIGURE 1Scheme diagram of Mendelian randomization design.

FIGURE 3
FIGURE 3Causal estimates of Hormonal and Reproductive Factors on LBP in MVMR.

TABLE 1
Summary of GWAS data for instrumental variables.
ANM, age at menopause; AAM, age at menarche; LMC, length of menstrual cycle; AFB, age at first birth; ALB, Age at last live birth; AFS, Age first had sexual intercourse; BMI, Body mass index; LBP, low back pain.

TABLE 2
Sensitivity analysis of hormonal and reproductive factors causally linked to LBP.
ANM, age at menopause; AAM, age at menarche; LMC, length of menstrual cycle; AFB, age at first birth; ALB, Age at last live birth; AFS, Age first had sexual intercourse; BMI, Body mass index; LBP, low back pain.

TABLE 3
Sensitivity analysis of LBP causally linked to hormonal and reproductive factors.
ANM, age at menopause; AAM, age at menarche; LMC, length of menstrual cycle; AFB, age at first birth; ALB, Age at last live birth; AFS, Age first had sexual intercourse; BMI, Body mass index; LBP, low back pain.

TABLE 4
Sensitivity analysis of hormonal and reproductive factors causally linked to LBP (validation analysis)., age at menopause; AAM, age at menarche; LMC, length of menstrual cycle; AFB, age at first birth; ALB, Age at last live birth; AFS, Age first had sexual intercourse; BMI, Body mass index; LBP, low back pain. ANM