The Role of Gestational Diabetes Mellitus and Pelvic Floor 3D-Ultrasound Markers at the Second and Third Trimester of Pregnancy Predicting 6-18 Months Postpartum Urinary Incontinence: Preliminary Results From a Prospective Cohort

Carlos Izaias Sartorão Filho São Paulo State University Fabiane Affonso Pinheiro São Paulo State University Luiz Takano São Paulo State University Raghavendra Hallur Lakshmana Shetty São Paulo State University Sthefanie K. Nunes São Paulo State University Adriely Bittencourt Morgenstern Magyori São Paulo State University Claudia G. Magalhães São Paulo State University Baerbel Junginger Charité University Medicine Berlin Zsuzsanna Ilona Katalin de Jármy Di Bella Federal University of São Paulo Iracema M P Calderon São Paulo State University Angélica M P Barbosa São Paulo State University Marilza Vieira Cunha Rudge (  marilzarudge@gmail.com ) São Paulo State University

1. Introduction pharmacotherapy on the results. Moreover, we considered only those submitted to C-section before the second period of labor to prevent the recognized interference of parturition in the outcome UI. We calculated the sample size using the prevalence of long-term UI in non-GDM of 18%, [1] the relative risk of UI in the GDM group of 2.73 [2], a type 2 error of 20%, and a 95% con dence interval, resulting in 33 participants per group. Statistical analyses were performed using the software SPSS 23.0 (IBM, NY).
Eligible women were included between 24-28 weeks of gestation, completed a detailed questionnaire on sociodemographic information, past medical, regular physical activity (150 minutes/week), initial pregnancy weight in Kg, height in cm, Body Mass Index, and a physical examination. Ethnicity was de ned as Caucasian or non-Caucasian. The same researcher performed a PFM transperineal 3DUS at 24-28 and 36-40 weeks of gestation (CISF). The 3DUS data imaging was stored for posterior blinded and o ine analysis. Long-term follow-up continued until April 2020. The same clinical and anthropometric evaluation was done at 36-40 weeks of gestation.
The fasting plasma glucose in the rst trimester in mg/dL, 75 grams-Oral Glucose Tolerance Test (OGTT) in the second trimester of pregnancy, gestational age (in weeks) at the two-time points of 3DUS, the ISI, and the ICIQ-SF questionnaire scores for PS-UI at 36-40 weeks of pregnancy were recorded. In addition, at 6-18 months after C-section, breastfeeding duration, regular physical activity, postpartum smoking, newborn weight, and gestational age at delivery, to classify newborn as appropriate, large or small for gestational age, and the occurrence of Urinary Incontinence plus ISI and ICIQ-SF severity scores were collected.
The primary outcome was the occurrence of UI between 6-18 months after the C-section. In addition, we investigated the association between UI and the risk factors, including maternal and perinatal characteristics, the rst-trimester fasting glucose, and the OGTT results as positive or negative for GDM diagnosis, the occurrence and severity PS-UI, and the PFM 3DUS biometry.
Women included in this study with hyperglycemia in the second and third trimester of pregnancy met the Fasting Blood Glucose (FBG) criteria in the rst trimester (1-13 gestational weeks) was lower than 5.1 mmol/L. [15] All pregnant women underwent 75-g OGTT between 24-28 weeks. The cutoffs were 92, 180, and 153 mg/dL, respectively, fasting one and two hours after glucose overloading. The GDM was de ned by one or more values equal to or more than the respective cutoffs. [16,17] For mild GDM, a glucose pro le was performed during a 1-day hospital stay with the woman on 2840 kcal-diet fractionated in ve meals. Plasma glucose was measured every two hours, from 8 AM to 6 PM. The thresholds used were 90 mg/dL for fasting (8h) and 130 mg/dL for any postprandial level. To maintain normoglycemia, all pregnant women received the same nutritional and lifestyle interventions as the cornerstone of treatment during the last 10-15 weeks of pregnancy. After these instructions, when glycemia remains elevated after 1-2 weeks of lifestyle interventions, daily glucose testing was continued, pharmacological treatment was initiated, and the patient was excluded from this study. The exposition and a follow-up period were at 24-28 weeks of gestation (T1), 36-40 weeks (T2), birth, and 6-18 months after a planned C-section. PS-UI was considered as any urinary leakage that started during pregnancy. [18] The PF 3DUS examination was done according to the protocol proposed by Dietz et al. [11] at 24-28 weeks (T1) and 36-40 weeks (T2). The functional maneuvers were previously explained. The variables of 3DUS in T1 and T2, at rest, contraction, and distension, were collected after completing the study, o ine, by the same observer, blinded (CISF). The measurements: Hiatal area in square cm; Antero-Posterior (AP) diameter in cm; Transversal diameter in cm; variation of AP from T1-T2 in cm; variation of Hiatal Area T1-T2 in square cm; variation of Transversal diameter from T1-T2 in cm; the relationship between rest and contraction in T1, in T2 and the T1-T2 variation from all biometry; the relationship between rest and distension in T1, T2, and the T1-T2 variation; the relationship between contraction and distension in T1, T2, and the T1-T2 variation; the percentual variation from each biometry analysis were calculated. [6,19] We used the GE Voluson system with a volumetric curved array three-dimensional transducer (GE Healthcare, Zipf, Austria).

Statistical Analysis
Descriptive analysis was applied for the demographic characteristics described as mean ± standard deviation (SD) and Minimum / Maximum values. We rst used the Cox univariate logistic regression analysis to test the association between each risk factor and the 6-18 months postpartum UI. Then, factors with a P-value under .20 were included in a multivariate logistic regression analysis. All signi cant maternal-pregnancy characteristics and the PFM 3DUS measurements from the univariate analysis were used as predictors. A multivariate logistic regression analysis after adjustments for potential confounding factors to identify the relationship between 6-18 months postpartum UI and clinical plus PFM 3DUS at the second and third trimesters of pregnancy was performed. Relative risk (RR) and 95% con dence intervals (95%CI) were reported. The RR and 95%CI were calculated for each clinical and 3DUS measure and adjusted for all other variables. These variables were excluded one by one (models 1 to 8) until reaching the nal best model, which was de ned by the impossibility to exclude any other variable without signi cant loss in adjustment. For all multivariate logistic tests, a P-value under .05 was considered.
Hiatal area in cm square. Table 3 demonstrates the percentual variation of 3DUS from T1 to T2. The primary feature is the T1-T2 percentual variation of HA from distension to contraction maneuvers, with a mean of -46.8 (SD 28.5).  Table 4 demonstrates Cox univariate and multivariate logistic regression analysis among clinical and HA measurements and 6-18 months postpartum UI. The UI risk was 5.4 times higher in women with positive 75g-OGTT; it decreased with increased weight gain; it decreased with an increased variation from T1-T2 in transversal diameter and HA at rest; it was 3.6 times higher in women who had PS-UI. All the variables with a P-value under .20 in the univariate were included in the multivariate regression analysis. We did not use the variable weight gain because its clinical interpretation was inconsistent; univariate demonstrates that a higher weight gain was related to a lower RR for 6-18 months postpartum UI, a condition not observed clinically. The multivariate analysis was performed after adjustments for GDM; PS-UI diagnosis; superior educational level; variation from T1 to T2 of transversal diameter and HA at rest; percentual variation of HA from contraction to rest and from distension to rest at T2; percentual variation of HA from contraction to rest and from distension to rest from T1 to T2; gestational ICIQ-SF and ISI scores; and newborn weight (model 1, the full model). In the backward model, some variables were eliminated, avoiding collinearity. Also, after adjustments for the newborn weight (model 2, P: .992), gestational ICIQ-SF (model 3; P: .778), the percentual of HA variation from contraction to rest at T2 (model 4; P: .576), PS-UI (model 5; P: .296), the variation from T1-T2 of transversal diameter at rest (model 6; P: .228), and for the percentual variation from T1 to T2 of HA from contraction to rest (model 7; P: .093). Adjusted analysis of potential confounders disclosed that 6-18 months postpartum UI development was signi cantly associated with GDM (model 8; Adjusted RR: 8.08; 95%CI 1.17-55.8; P: .034) and the percentual variation from T1 to T2 of the HA from distension to rest (model 8, the Best model; Adjusted RR: 0.96; 95%CI 0.93 -0.99; P: .023).

Discussion
First, we observed that GDM, even the mild presentation, treated only with nutritional and healthy lifestyle interventions, is a solid and independent risk factor for 6-18 months postpartum UI. Second, the higher PF HA distensibility in the third trimester of pregnancy is protective against 6-18 months postpartum UI.
Consistent with these ndings, our study reports a positive relation between GDM and increased 6-18 months postpartum UI and a negative relation between higher HA distension and decreased 6-18 months postpartum UI. Thus, both markers are opposed, as GDM is a causative agent, and the higher HA distensibility observed by 3DUS is a protective agent for 6-18 months postpartum UI development.
Pregnancy has been considered "an underutilized window of opportunity to improve long-term maternal and infant health." [20] GDM diagnosis nds a group of women and their offspring at higher risk of diabetes, obesity, and premature cardiovascular disease in the long term. [21] The role of GDM in the later development of UI is limited in the classical literature. [21,22] Multivariate regression analyses revealed that GDM, even when mild, was a solid and independent risk factor for 6-18 months postpartum UI In this study, the PF 3DUS assessment detected an increase in the PF HA distensibility at the end of the third trimester of pregnancy as a protective factor against 6-18 months postpartum UI. This unexpected functional observation identifying a PFM becoming more distensible at the end of the third trimester of pregnancy is not previously reported in the literature.
The strengths of our study include the longitudinal design with a 1-2 years follow-up rate, considering GDM as a transient condition with lifelong consequences. The same author did all PFM 3DUS, blinded.
We provided selection exclusion criteria as previous parity, labor induction, the second period of active labor, and vaginal delivery, the most considered postpartum UI risk factors. Finally, the statistical analyses considered all patients as a whole, without separation between GDM and non-GDM, and all GDM patients treated with hypoglycemic drugs were excluded. This analysis allows us to obtain the best model after eight backward models of multivariate analyses, con rming that even the mild GDM is a strong predictor of 6-18 months postpartum UI despite the short period of hyperglycemic status exposure. Limitations: First, the lost follow-up was randomly due to loss of contact information by cellphone number changes, a common situation in Brazil. Thus, most losses occurred at the outcome analysis on 6-18 months postpartum UI, and we considered our losses acceptable for a 1-2 year cohort. Second, our results may not be generalized to other populations, including those women with vaginal birth, multiparous, or GDM treated by insulin or other drugs. Third, our strict inclusion criteria of GDM women-only treated with nutritional and lifestyle interventions probably removed essential risk factors for 6-18 months postpartum UI. Fourth, we do not collect data on the prepregnancy and rst trimester PFM 3DUS, as the enrollment occurred at the second trimester of gestation. Moreover, although we observed changes in the PFM function during the pregnancy span, more studies are demanded to explore these ndings better.
Our preliminary results suggested that GDM diagnosis and PFM 3DUS functional analysis are good clinical indexes for postpartum UI prediction concerning GDM diagnosis and protective prediction of UI concerning higher distensibility of HA observed during pregnancy by 3DUS.

Conclusion
Our preliminary results provided objective data that GDM-pregnancies are positively related to increasing the risk for 6-18 months postpartum UI development, even in a population submitted to a planned Csection. Furthermore, mild GDM exposure is a solid and independent risk predictor for 6-18 months postpartum UI. While further studies are required to elucidate the mechanism, mild GDM may represent a modi able risk factor when detected earlier in pregnancy. Opposite results were found for higher PFM 3DUS distensibility from the second to third trimesters as positively related to decreasing risk of 6-18 months postpartum UI. Further studies may be designed to determine the role and consequences for the PFM after childbirth.

Declarations
Ethics approval, guidelines and consent to participate: Regulatory approval was obtained from the Institutional Review Board (number 1.716.895) by "Botucatu Medical School of São Paulo State University (Unesp)" Ethics Committee. Written informed consent was obtained, and the participants were handled adequately for their morbidities. The Strobe Statement guideline was used for reporting the manuscript. All methods were carried out in accordance with relevant guidelines and regulations.

Consent for publication: not applicable
Availability of data and materials: the raw database le is provided as a supplementary submission le.