Metformin use linked to decreased risk of pelvic organ prolapse: insights from NHANES data.

This cross-sectional study used prospectively collected data from the NHANES. The participants were interviewed in their homes and underwent standardized physical examinations, interviews, and biological sample collection at a mobile examination center (MEC). The POP survey was included only in the 2005–2012 cycles, during which information on DM and antihyperglycemic agents was also available. Non-pregnant women aged ≥20 years were included. Participants who did not respond to the question regarding vaginal bulging were excluded. No other specific exclusions were applied to maximize the sample size. A comparison of baseline characteristics between the included and excluded participants was conducted to assess potential selection bias. The sample size was determined by available data within the survey cycles, consistent with large-scale epidemiological studies. The National Centre for Healthcare Statistics Research Ethics Review Board approved the NHANES surveys, and this study was exempt from the Institutional Review Board approval as a secondary analysis of existing public data. POP was defined as an affirmative response to “Experience bulging in the vaginal area” [ 31 ], i.e., symptomatic POP, which was derived from the Pelvic Floor Distress Inventory [ 32 ], and was commonly used in NHANES studies [ 3 , 33 ]. DM was defined as follows: (1) self-reported diagnosis by a physician or health care professional (other than during pregnancy), (2) HbA1c ≥ 6.5%, or (3) fasting plasma glucose (FPG) ≥ 126 mg/dL [ 34 ]. Controlled DM was defined as HbA1c < 7.0% [ 35 ]. Prediabetes was defined as follows: (1) self-reported diagnosis (prediabetes/impaired fasting glucose/impaired glucose tolerance/borderline diabetes) by a physician or health care professional [ 36 ], (2) HbA1c 5.7–6.4%, or (3) FPG 100–125 mg/dl [ 37 ]. Medication use was confirmed by the interviewer’s inspection of medication containers. Data on the use of the following medicines were extracted: metformin, thiazolidinediones (TZDs, including pioglitazone, rosiglitazone, and troglitazone), sulfonylureas (SUs, including glimepiride, glipizide, glyburide, chlorpropamide, and tolazamide), glinides (including repaglinide and nateglinide), alpha-glucosidase inhibitors (AGIs, including acarbose and miglitol), and insulin. Inflammatory biomarkers included C-reactive protein (CRP, 2005–2010 cycles only), the systemic immune-inflammation index (SII, computed as platelet count × neutrophil count/lymphocyte count [ 38 ]), ferritin (2005–2010 cycles only), and alkaline phosphatase (ALP). Oxidative stress biomarkers included serum bilirubin, albumin, iron [ 39 ], γ-glutamyl transferase (GGT), and uric acid (UA) [ 40 ]. The covariates included age, race/ethnicity, education level, body mass index (BMI), smoking history, vaginal deliveries, cesarean deliveries, hysterectomy history, and postmenopausal status, which were selected based on previous POP-related studies [ 31 , 41 ]. Age was assessed in years. Self-reported race/ethnicity was categorized as non-Hispanic white, non-Hispanic black, other Hispanic, Mexican American, or other races. Education level was categorized as less than high school, high school diploma including general education development, or more than high school; BMI was calculated as kg/m 2 and categorized as less than 24.9 (underweight/normal weight), 25.0–29.9 (overweight), or 30.0 or more (obesity). Smoking history was defined as an affirmative response to “Smoked at least 100 cigarettes in your lifetime?”. Women self-reported the number of vaginal deliveries, cesarean deliveries, and hysterectomy and oophorectomy history. Vaginal and cesarean deliveries were used as continuous variables. Postmenopausal status was defined as follows: (1) aged ≥ 40 years, with either both of the ovaries removed and/or no period in the past year due to hysterectomy and/or menopause [ 41 ]; or (2) aged >55 years and no period in the past year (denied or absent). HTN was defined as a self-reported diagnosis by a physician or health care professional (other than during pregnancy). The NHANES applied a complex, stratified, multi-stage, probability cluster design [ 41 ]. Sampling probability weights were used to balance the representativeness of the sample, allowing extrapolation to the US population and correcting possible biases introduced in population sampling. Data analysis was carried out via R and R Studio (R Foundation for Statistical Computing, Version 4.4.2) with the Survey package. Weighted prevalence odds ratios (ORs) and 95% confidence intervals (CIs) were calculated by incorporating the design effect, appropriate sample weights (WTMEC2YR of MEC examination as instructed), stratification, and clustering of the complex NHANES sample design. The sample weights were adjusted for unequal probabilities of selection and nonresponse. Missing data were handled by complete-case analysis in view of the cross-sectional design, the sampling weights, and the reliability of estimates in subgroup analyses. Continuous variables were presented as means and standard deviations, and categorical variables were presented as counts and weighted percentages. The Rao‒Scott adjusted χ 2 test and rank sum test were performed to calculate differences between the two groups (POP data available vs. POP data missing, and symptomatic POP vs. no POP symptoms). Two‐sided P  < 0.05 was considered significant. (1) Risk analysis of POP in specific populations was performed via four logistic regression models. Model 1 was unadjusted. Model 2 adjusted for demographic factors (age, race, and education level). Model 3 added comorbidity factors (DM/ prediabetes, HTN, smoking history, and BMI); Model 4 added obstetric and gynecologic factors (vaginal and cesarean deliveries, hysterectomy history, and postmenopausal status). (2) The correlation between POP and the usage of hypoglycemic agents in subgroups was examined via multivariable regression models, with adjustments for HbA1c, FPG (both as continuous variables), and covariates mentioned above. (3) Correlations between POP and inflammatory and oxidative stress biomarkers were analyzed as continuous variables via three logistic regression models. Model A was unadjusted. Model B was adjusted for covariates. If the result of Model B was significant, further analysis in the metformin subgroup was performed in Model C.

From the NHANES 2005–2012 cycles, 11131 non-pregnant women aged 20 years and older were included, and 1787 participants without POP information were excluded. Therefore, our final analysis included a comprehensive dataset of 9344 eligible participants. The enrollment flowchart is shown in Supplemental Fig. 1. Comparison between the included and the excluded: As presented in Supplemental Table 1, compared with excluded 1787 women without POP data, the included 9344 women with POP data were more educated ( P  < 0.001), more often non-Hispanic white ( P  < 0.001), had a higher BMI ( P  = 0.034), were more likely to be smokers ( P  = 0.009), had fewer vaginal deliveries ( P  < 0.001), were more likely to be postmenopausal ( P  < 0.001), less likely to have a history of hysterectomy ( P  = 0.002), and more likely to have HTN (P < 0.001). There were no differences in age, number of cesarean deliveries, or DM distributions between the two groups. Comparison between participants with and without symptomatic POP: Among 9344 women, 330 (weighted prevalence 2.9%) reported symptomatic POP. As shown in Table  1 , women with POP were significantly older, less educated, had a higher BMI, had more vaginal deliveries, were more likely to be postmenopausal, and were more likely to have a history of hysterectomy. Ethnicity was significantly different between the two groups. There were no differences in POP symptoms according to smoking history or cesarean delivery. POP was more prevalent in those with HTN, prediabetes, and DM, especially in people with uncontrolled DM. These variables were all adjusted in the subsequent analysis of risk factors. Table 1 Participant characteristics among non-pregnant women aged ≥20 years by reported POP, NHANES 2005–2012 Characteristics Overall, N  = 9344 Symptomatic POP P value No, N  = 9014 Yes, N  = 330 Age (years) 48.26 ± 17.02 48.01 ± 17.03 57.89 ± 14.41  <0.001 Race 0.009  Non-Hispanic white 4300 (71%) 4164 (71%) 136 (69%)  Non-Hispanic black 2055 (12%) 2006 (12%) 49 (9.0%)  Mexican American 1452 (6.9%) 1368 (6.7%) 84 (11%)  Other Hispanic 902 (4.9%) 852 (4.8%) 50 (7.4%)  Other races 635 (6.0%) 624 (6.1%) 11 (3.5%) Education  <0.001  High school 4798 (60%) 4672 (61%) 126 (47%) BMI 0.001   <25.0 2886 (36%) 2819 (36%) 67 (22%)  25.0–29.9 2659 (28%) 2546 (28%) 113 (37%)   ≥30.0 3697 (36%) 3551 (36%) 146 (40%) Smoking history 3523 (40%) 3409 (41%) 114 (38%) 0.400 Vaginal delivery 2.08 ± 1.84 2.36 ± 2.30 3.11 ± 2.30  <0.001 Cesarean delivery 0.65 ± 0.95 0.66 ± 0.95 0.52 ± 0.93 0.051 Postmenopausal 4756 (46%) 4526 (45%) 230 (72%)  <0.001 Hysterectomy 2203 (25%) 2080 (24%) 123 (44%)  <0.001 Hypertension 3469 (32%) 3297 (32%) 172 (49%)  <0.001 Diabetes 1506 (11%) 1434 (11%) 72 (16%) 0.031 Uncontrolled diabetes 623 (4.5%) 588 (4.4%) 35 (8.0%) 0.032 Prediabetes 2931 (30%) 2793 (29%) 138 (45%)  0.900 SUs 455 (3.2%) 431 (3.2%) 24 (3.5%) 0.700 Glinides 12 (0.1%) 12 (0.1%) 0 (0%) 0.600 AGIs 8 (<0.1%) 8 (<0.1%) 0 (0%) 0.700 Bold values indicate statistical significance ( P  < 0.05) Data are presented in mean ± SD for continuous variables and n (%) for categorical variables. POP, pelvic organ prolapse; TZDs, thiazolidinediones; SUs, sulfonylureas; AGIs, alpha-glucosidase inhibitors Participant characteristics among non-pregnant women aged ≥20 years by reported POP, NHANES 2005–2012 Bold values indicate statistical significance ( P  < 0.05) Data are presented in mean ± SD for continuous variables and n (%) for categorical variables. POP, pelvic organ prolapse; TZDs, thiazolidinediones; SUs, sulfonylureas; AGIs, alpha-glucosidase inhibitors Among 668 metformin users (630 with diabetes, 36 with prediabetes, and 2 were unknown status), 37 had POP. Among 455 SUs users (445 with diabetes, 7 with prediabetes, and 3 were unknown status), 24 had POP. Among 183 TZDs users (178 with diabetes, 4 with prediabetes, and 1 were unknown status), 6 had POP. A few participants used glinides or AGIs, so they were not included in further analyses. Given the low prevalence of POP, we collectively referred to diabetes and prediabetes as (pre)diabetes to make the most of the drug data. There was no difference in the use of these oral antihyperglycemic agents or insulin between the two groups. Before exploring metformin in specific populations who are already known to benefit from metformin, as mentioned above, we need to determine whether these comorbidities or specific conditions are risk factors for POP, including (pre)diabetes, smoking, HTN, and overweight/obesity. We performed incremental logistic regressions. Table 2 shows that (pre)diabetes was strongly associated with POP (in Model 1, OR = 2.33, 95% CI: 1.75–3.10, P  < 0.001) when unadjusted, and remained significant when adjusted for demographic and comorbidity factors (in Models 2, OR = 1.63, 95% CI: 1.21–2.20, P  < 0.001, and in Models 3, OR = 1.52, 95% CI: 1.11–2.08, P  = 0.007, respectively). Overweight/obesity also exhibited strong associations when unadjusted (in Model 1, OR = 1.95, 95% CI: 1.39–2.73, P  < 0.001) and adjusted for demographic and comorbidity factors (in Model 2, OR = 1.70, 95% CI: 1.20–2.41, P  = 0.002). However, the factors diminished to non-significance when further adjusted for obstetric and gynecologic factors. HTN was significantly associated with POP in the unadjusted model (in Model 1, OR = 2.08, 95% CI: 1.57–2.76, P  < 0.001), but the association was attenuated and became non-significant after adjusting for covariates. Smoking history was not significantly associated in any models. Table 2 Analysis of the POP risk in populations who are already known to benefit from metformin Characteristics Model 1 Model 2 Model 3 Model 4 OR (95% CI) P OR (95% CI) P OR (95% CI) P OR (95% CI) P (Pre)diabetes 2.33 (1.75, 3.10)  <0.001 1.63 (1.21, 2.20)  <0.001 1.52 (1.11, 2.08) 0.007 1.05 (0.67, 1.65) 0.800 Smoking history 0.88 (0.65, 1.19) 0.400 0.89 (0.64, 1.22) 0.500 0.88 (0.63, 1.22) 0.400 0.72 (0.43, 1.19) 0.200 Overweight/obesity 1.95 (1.39, 2.73)  <0.001 1.70 (1.20, 2.41) 0.002 1.52 (1.07, 2.16) 0.016 1.26 (0.70, 2.27) 0.400 HTN 2.08 (1.57, 2.76)  <0.001 1.40 (0.98, 2.00) 0.059 1.33 (0.92, 1.93) 0.12 1.23 (0.72, 2.08) 0.400 Model 1: unadjusted. Model 2: adjusted for demographic factors. Model 3: adjusted for demographic and comorbidity factors. Model 4: fully adjusted for demographic, comorbidity, and obstetric and gynecologic factors Bold values indicate statistical significance ( P  < 0.05) DM diabetes, HTN hypertension Analysis of the POP risk in populations who are already known to benefit from metformin Model 1: unadjusted. Model 2: adjusted for demographic factors. Model 3: adjusted for demographic and comorbidity factors. Model 4: fully adjusted for demographic, comorbidity, and obstetric and gynecologic factors Bold values indicate statistical significance ( P  < 0.05) DM diabetes, HTN hypertension We then examined whether metformin had a beneficial association with POP beyond its antihyperglycemic effect. To account for potential confounding factors, HbA1c and FPG were adjusted. Other antihyperglycemic agents, including insulin, TZDs, and SUs, were compared. Subgroup analyses were performed on (pre)diabetes participants based on smoking history, HTN, and BMI. As demonstrated in Table  3 , metformin use was associated with lower POP prevalence in (pre)diabetes subgroups with HTN (OR = 0.32, OR = 0.32, 95% CI: 0.10–1.00, P  = 0.045) or smoking history (OR = 0.09, 95% CI: 0.01–0.75, P  = 0.022), but not in non-hypertensive or non-smoking subgroups, neither in subgroups with or without overweight/obesity. Other agents (TZDs, SUs, insulin) showed no significant associations. Table 3 Multivariable-adjusted logistic regression analysis of the associations between antihyperglycemic agents and POP Subset Metformin ( n  = 666) TZDs ( n  = 182) SUs ( n  = 452) Insulin ( n  = 299) OR (95% CI) P OR (95% CI) P OR (95% CI) P OR (95% CI) P (Pre)diabetes 0.42 (0.14, 1.22) 0.100 0.57 (0.12, 2.58) 0.400 0.44 (0.18, 1.09) 0.069 0.73 (0.20, 2.61) 0.600 (Pre)diabetes + HTN 0.32 (0.10, 1.00) 0.045 0.86 (0.15, 4.99) 0.900 0.58 (0.27, 1.26) 0.200 1.22 (0.42, 3.55) 0.700 (Pre)diabetes + without HTN 1.31 (0.20, 8.73) 0.800 0.38 (0.02, 7.22) 0.500 0.24 (0.02, 2.83) 0.200 0.04 (0.00, 3.43) 0.140 (Pre)diabetes + smoker 0.09 (0.01, 0.75) 0.022 0.49 (0.06, 4.10) 0.500 0.59 (0.15, 2.36) 0.400 0.38 (0.05, 2.97) 0.300 (Pre)diabetes + non-smoker 0.68 (0.22, 1.98) 0.500 0.52 (0.09, 3.16) 0.500 0.52 (0.19, 1.41) 0.200 0.69 (0.14, 3.38) 0.600 (Pre)diabetes + BMI ≥ 25 0.36 (0.12, 1.11) 0.068 0.49 (0.09, 2.73) 0.400 0.42 (0.16, 1.10) 0.069 0.76 (0.21, 2.78) 0.700 (Pre)diabetes + BMI < 25 1.34 (0.19, 9.61) 0.900 5.53 (0.79, 38.7) 0.077 1.43 (0.18, 11.1) 0.700 NA NA HbA1c, fasting blood glucose, demographic (age, race, and education level), comorbidity (HTN, smoking history, and BMI), and obstetric and gynecologic factors (vaginal deliveries, hysterectomy history, and postmenopausal status) were adjusted Bold values indicate statistical significance ( P  < 0.05) TZD thiazolidinediones, SU sulfonylureas, NA the subgroups were too small for analysis Multivariable-adjusted logistic regression analysis of the associations between antihyperglycemic agents and POP HbA1c, fasting blood glucose, demographic (age, race, and education level), comorbidity (HTN, smoking history, and BMI), and obstetric and gynecologic factors (vaginal deliveries, hysterectomy history, and postmenopausal status) were adjusted Bold values indicate statistical significance ( P  < 0.05) TZD thiazolidinediones, SU sulfonylureas, NA the subgroups were too small for analysis To investigate the potential mechanisms underlying the beneficial effect of metformin, we analyzed the associations between inflammatory and oxidative stress biomarkers and POP with three incremental logistic regression models. As shown in Table  4 , for inflammatory biomarkers, CRP was significantly elevated in individuals with POP (0.56 ± 0.98 vs 0.44 ± 0.73 mg/dL). CRP was associated with POP when unadjusted (OR = 1.15, 95% CI: 1.00–1.32, P  = 0.048), and strengthened when fully adjusted (OR = 1.30, 95% CI: 1.03–1.64, P  = 0.020). The association between CRP and POP remained significant in the (pre)diabetes subgroup (OR = 1.34, 95% CI: 1.01–1.78, P  = 0.033) and the (pre)diabetes subgroup with a smoking history (OR = 1.50, 95% CI: 1.15–1.96, P  = 0.002). Importantly, for smokers, CRP was no longer significant for POP in the metformin subgroup. In the (pre)diabetes subgroup with HTN, CRP was not significantly associated with POP (OR = 0.45, 95% CI: 0.16–1.25; P  = 0.110). Additionally, ALP exhibited a weak but significant association with POP in the unadjusted model (OR = 1.01, 95% CI: 1.00–1.01, P  < 0.001), which attenuated after adjustments. Neither the SII nor ferritin demonstrated significant associations in any model. Table 4 Logistic regression analysis of inflammatory and oxidative stress biomarkers for POP Biomarker Model All (Pre)diabetes (Pre)diabetes + HTN (Pre)diabetes + smoker Inflammatory SII A 1.00 (1.00, 1.00) 1.00 (1.00, 1.00) 1.00 (1.00, 1.00) 1.00 (1.00, 1.00) B 1.00 (1.00, 1.00) 1.00 (1.00, 1.00) 1.00 (1.00, 1.00) 1.00 (1.00, 1.00) CRP (mg/dL) A 1.15 (1.00, 1.32) * 1.03 (0.78, 1.37) 0.77 (0.51, 1.15) 1.05 (0.69, 1.59) B 1.30 (1.03, 1.64) * 1.34 (1.01, 1.78) * 0.45 (0.16, 1.25) 1.50 (1.15, 1.96) * C NA 1.32 (1.04, 1.67) * NA 1.29 (0.97, 1.73) Ferritin (ng/mL) A 1.00 (0.99, 1.00) 1.00 (0.99, 1.00) 0.97 (0.94, 1.00) 0.99 (0.98, 1.01) B 1.00 (0.99, 1.01) 1.00 (0.99, 1.01) 0.97 (0.92, 1.02) 0.99 (0.95, 1.04) ALP (U/L) A 1.01 (1.00, 1.01) * 1.00 (0.99, 1.01) 1.00 (0.98, 1.01) 1.00 (0.98, 1.01) B 1.00 (0.99, 1.01) 1.00 (0.99, 1.01) 1.00 (0.99, 1.02) 1.00 (0.99, 1.01) Oxidative stress Total bilirubin (mg/dL) A 0.39 (0.17, 0.88) * 0.62 (0.23, 1.68) 1.01 (0.36, 2.80) 0.61 (0.17, 2.10) B 0.26 (0.07, 0.95) * 0.28 (0.07, 1.09) 0.69 (0.16, 2.91) 0.53 (0.05, 5.22) Albumin (g/L) A 0.99 (0.95, 1.04) 1.06 (0.99, 1.14) 1.03 (0.96, 1.11) 1.06 (0.96, 1.18) B 0.99 (0.91, 1.06) 1.03 (0.90, 1.19) 1.10 (0.95, 1.27) 0.93 (0.75, 1.15) Iron (μmol/L) A 0.97 (0.96, 0.99) * 0.99 (0.96, 1.01) 0.99 (0.95, 1.04) 0.96 (0.92, 1.01) B 0.97 (0.92, 1.02) 0.98 (0.92, 1.05) 0.99 (0.91, 1.07) 0.96 (0.88, 1.05) GGT (IU/L) A 1.00 (1.00, 1.00) 0.99 (0.99, 1.00) * 0.99 (0.98, 1.00) 1.00 (0.99, 1.00) B 1.00 (1.00, 1.00) 1.00 (0.99, 1.00) 1.00 (0.99, 1.01) 1.00 (0.99, 1.01) UA (mg/dL) A 1.02 (0.89, 1.18) 0.94 (0.78, 1.13) 0.96 (0.79, 1.17) 0.83 (0.62, 1.12) B 0.99 (0.80, 1.22) 0.99 (0.76, 1.27) 0.98 (0.72, 1.33) 1.13 (0.75, 1.72) Model A: unadjusted. Model B: adjusted for demographic, comorbidity, and obstetric and gynecologic factors. Model C: adjusted in the metformin subgroup based on Model B. NA, Model C was not applicable for the total population, as metformin was used only in the (pre)diabetes subgroup; CRP was not significant in the (pre)diabetes subgroup with HTN in Model B, so no further analysis was performed. * P  < 0.05 SII systemic immune-inflammation index, CRP C-reactive protein, ALP alkaline phosphatase, GGT γ-glutamyl transferase, UA uric acid Logistic regression analysis of inflammatory and oxidative stress biomarkers for POP Model A: unadjusted. Model B: adjusted for demographic, comorbidity, and obstetric and gynecologic factors. Model C: adjusted in the metformin subgroup based on Model B. NA, Model C was not applicable for the total population, as metformin was used only in the (pre)diabetes subgroup; CRP was not significant in the (pre)diabetes subgroup with HTN in Model B, so no further analysis was performed. * P  < 0.05 SII systemic immune-inflammation index, CRP C-reactive protein, ALP alkaline phosphatase, GGT γ-glutamyl transferase, UA uric acid For oxidative stress biomarkers, bilirubin was lower in individuals with POP in Model A (OR = 0.39, 95% CI: 0.17–0.88, P  = 0.020) and Model B (OR = 0.26, 95% CI: 0.07–0.95, P  = 0.033), but the CIs widened in subgroups. Iron (OR = 0.97, 95% CI: 0.96–0.99, P  = 0.002) and GGT (OR = 0.99, 95% CI: 0.99–1.00, P  = 0.040) were significant in the unadjusted model, but the ORs and narrow CIs were very close to 1 and attenuated after adjustments. Albumin and UA levels were not significantly associated.

Our study provides valuable insights into the potentially protective role of metformin in POP despite of antihyperglycemic effect in (pre)diabetes subgroups with HTN or smoking history, probably via the suppression of inflammation. Given the limitations and potential biases inherent to the cross-sectional study design, this association does not establish causation. This epidemiological finding may serve as a basis for future research to elucidate the underlying mechanisms and identify effective prevention strategies for diverse populations.

To our knowledge, this is the first clinical study exploring the association between metformin use and POP risk. We have demonstrated that metformin use is associated with reduced POP prevalence in specific high-inflammatory subgroups (hypertensive or smoking (pre)diabetes patients), independent of glycemic control. In addition, we found that CRP was significantly elevated in the POP population, especially in the (pre)diabetes (with or without smoking history) subgroup. In the smoking subgroup, the risk could be alleviated by metformin, indicating a potential anti-inflammatory mechanism. The pathophysiology of POP has not been fully elucidated [ 42 ]. Research has revealed that POP patients exhibited increased levels of inflammatory cytokines and oxidative stress biomarkers in the pelvic floor [ 42 , 43 ]. While for (pre)diabetes, hyperglycemia can upregulate biomarkers of chronic inflammation and contribute to increased reactive oxygen species (ROS) generation [ 44 ]. Besides, inflammation and oxidative stress contribute to HTN [ 45 ], and much evidence supports a key role for smoking-induced ROS and the resulting oxidative stress in inflammation [ 46 ]. In this study, other antihyperglycemic and insulin-sensitizing agents, including TZDs, SUs, and insulin, did not show associations with POP. Similarly, a significantly lower risk of lower urinary tract symptoms was observed in males with type 2 DM who used metformin, while not in users of rosiglitazone and pioglitazone, probably due to a lack of strong anti-inflammatory effects [ 47 ]. Interestingly, metformin only seemed to have a potentially protective effect on the (pre)diabetes subgroups with HTN or smoking history, but not on the (pre)diabetes subgroup alone. As mentioned before, the association between DM and POP is controversial, as well as the role of smoking in POP [ 10 , 19 ]. Besides, a systematic review revealed a positive association between HTN and POP [ 13 ]. In our study, (pre)diabetes was a risk factor for POP, but only when obstetric and gynecologic factors were not adjusted. HTN was a non-independent risk factor, and smoking was not a risk factor. However, it should be taken into account that the sample sizes of each subgroup are relatively small that when too many confounding factors are corrected, the results may be affected. We wondered whether the combination of subgroups exacerbated the levels of inflammation in the respective diseases that promote POP, and this pathway could be significantly inhibited by metformin, but this was not evident in separate subgroups. Low-grade systemic inflammation is characterized by a two-to-threefold increase in systemic plasma concentrations of cytokines, such as CRP [ 48 ]. The association of persistent modest elevations in plasma CRP levels with chronic diseases, such as HTN, DM, and smoking [ 48 , 49 ], has attracted considerable clinical interest and often contradictory interpretations. CRP was found to be elevated in late gestation sows at high risk for POP compared with low-risk ones [ 50 ], but there has been no similar research in humans. A CRP level greater than 0.2 mg/dL was considered elevated [ 48 ]. Our study showed significantly higher CRP (0.56 ± 0.98 mg/dL) in the POP population, which indicated that, in addition to local inflammation in the pelvic floor, POP also features low-grade systemic inflammation. Metformin seemed to exert its anti-inflammatory effect by reducing pro-inflammatory cytokine secretion, especially CRP [ 51 , 52 ]. A randomized-controlled study revealed that metformin significantly decreased CRP and improved quality of life in rheumatoid arthritis patients [ 53 ]. Studies also found metformin durably lowered CRP in participants at high risk for DM [ 54 ]. Our findings are consistent with this anti-inflammatory hypothesis, as we observed that the positive association between CRP and POP was attenuated in metformin users, particularly among smokers. This mechanism was reasonable in smoking individuals with (pre)diabetes; nevertheless, although we also found POP was less prevalent in hypertensive patients with (pre)diabetes who used metformin, elevated CRP was not associated with POP risk in this population. It remains unclear whether this discrepancy arises from distinct underlying mechanisms or is merely a consequence of limited sample size. The biggest limitation of this analysis is the inherent constraints of the cross-sectional study design, which cannot establish causality. Although the duration of diabetes and metformin use was available in the dataset, the course of POP was lacking, so the absence of chronology made causality difficult to establish. We also cannot determine how long one needs to take metformin before it begins to have a beneficial effect on POP. Besides, women who had been previously treated for POP were not identified. Second, NHANES only offered symptomatic POP data without a POP quantification examination. Although symptoms are of vital importance for the assessment of POP that previous NHANES studies for POP based on this definition were widely recognized [ 3 , 33 ]. However, we would not know whether metformin reduces the occurrence of POP or alleviates the severity of POP in the absence of quantitative indicators. Third, in the dataset from 2005 to 2012, almost all participants who used metformin had (pre)diabetes. While metformin has been increasingly widely applied in clinical practice nowadays. Fourth, differences between the included and excluded participants mean that the results may not be generalizable to all population groups. Finally, it should be noted that the number of POP patients included was not large, especially in each subgroup analysis when multiple covariates were considered, leading to wide confidence intervals, which should be interpreted with caution in clinical contexts. Therefore, further large-scale prospective cohorts or randomized-controlled trials with diabetic and nondiabetic populations that use metformin should confirm the effect in long-term outcomes, in which full-scale POP information should be collected. Mechanisms should also be explored in tissue and animal experiments. If validated, metformin could be considered for POP prevention or alleviation in high-risk women with (pre)diabetes and inflammation-related conditions. We also wonder whether metformin might have benefits in the post-operative POP population, such as reducing recurrence and complications, especially mesh exposure.

Pelvic organ prolapse (POP) is a common female disease, defined as the descent of the pelvic organ into the vagina, primarily due to pelvic floor dysfunction [ 1 ], and profoundly impairs quality of life [ 2 ]. According to the National Health and Nutrition Examination Survey (NHANES), approximately 3% of women in the United States report symptoms of vaginal bulging [ 3 ], and the prevalence is projected to increase by 46% from 2010 to 2050 [ 4 ]. Established risk factors for POP include parity, vaginal delivery and obstetrical trauma, age, obesity, connective tissue disorders, menopausal status, and chronic constipation [ 2 ]. Besides, smoking, hysterectomy history, certain races, family history, and genetic susceptibility have also been reported as risk factors [ 3 , 5 – 9 ]. Emerging evidence suggests that endocrine, metabolic, and inflammatory conditions may contribute to POP development [ 10 ]. The association between diabetes mellitus (DM) and POP remains controversial. While some studies have reported that DM is an independent risk factor for POP [ 11 – 13 ] and that DM may be linked to more advanced stages of POP [ 14 ], others have reported no significant associations [ 13 , 15 , 16 ]. Interestingly, hypertension (HTN) and DM combined (in metabolic syndrome) were reported to double the risk of POP, whereas HTN or DM alone was not independently associated with POP [ 17 – 19 ]. Metformin was first used to treat type 2 DM in the late 1950s and is currently the first-choice drug. The accumulation of positive preclinical and clinical data has stimulated interest in repurposing metformin in a variety of diseases, including obesity, HTN, rheumatoid arthritis, and senility [ 20 – 22 ]. Evidence suggests that metformin, via the suppression of pro-inflammatory pathways, protection of mitochondria and vascular function, and direct actions on neuronal stem cells, may protect against neurodegenerative diseases [ 23 ]; via oxidative stress inhibition and redox rebalancing, may improve the cardiometabolic status in populations with obesity, DM, smoking history, or psychiatric disorders [ 21 , 24 ]. Metformin is also increasingly used to prevent or treat a wide range of gynecological disorders, especially in polycystic ovarian syndrome (PCOS) [ 25 ]. PCOS has been recognized as a low-degree chronic inflammatory disease; its inflammatory biomarkers include C-reactive protein (CRP), interleukin-6, and so on [ 26 ]. Trial findings for endometriosis, premature ovarian failure, and uterine fibroids remain controversial and insufficient [ 25 ]. Only one preclinical study has reported the potential benefits of metformin for POP [ 27 ]. It focused on the molecular landscape of POP, in which transforming growth factor beta 1 (TGFB1) is one of the key regulatory factors on epithelial–mesenchymal transition, immune response, modulation of the extracellular matrix, and fibroblast function. Beneficially, metformin was found to downregulate the expression of TGFB1 target genes in vaginal fibroblasts from POP patients [ 27 ]. In POP, oxidative stress promotes inflammatory responses and extracellular matrix degradation by activating the MAPK/NF-κB pathway [ 28 ], which was related to the effect of TGFB1 [ 29 ], and metformin has also been shown to target the pathway [ 25 ]. Therefore, the beneficial effect of metformin on POP seems to be supported by reasonable mechanisms. Importantly, in this in vitro experiment, metformin [2 mM] was used to treat the cells [ 27 ], while circulating metformin concentrations are approximately 10–40 μM in both animals and humans [ 30 ]. The authors put forward that these findings may indicate the link between metformin use and decreased risk of POP, but by no means did it show clinical effectiveness. To date, there is no further evidence of the effectiveness of metformin in treating POP. A primary analysis in a large cross-sectional study may be the first step. The objective of this study was to use the NHANES dataset to explore the association between metformin use and POP. We hypothesized that metformin might reduce POP risk beyond glycemic control and assessed the potential roles of inflammatory and oxidative stress biomarkers.

Below is the link to the electronic supplementary material. Supplementary file1 (DOCX 98 KB) Supplementary file1 (DOCX 98 KB)