Introduction

The association between childbearing and pelvic organ prolapse (POP) giving rise to symptoms of pelvic floor dysfunction (PFD) is commonly recognised [1]. However, POP is present not only in women giving birth but also in premenopausal, nulliparous women, in whom the prevalence of prolapse is similar to women post-Caesarean section (CS), with the latter delivery mode considered to be partially protective compared with vaginal delivery [2]. Initially it was shown that POP is associated with reduction in total collagen content, with higher levels of weaker immature collagen and no change in collagen type I/III ratios [3]. However, a later study demonstrated that levels of collagen type I(providing strength and being mainly present in bone structures and, to a lesser extent, in soft tissue) is minimally changed in women with POP; collagen type III (providing elasticity) is increased, leading to decreased collagen I/III ratio [4]. POP is also associated with joint hypermobility and various medical conditions linked with abnormal collagen [5, 6]. All these factors are indicative of the systemic manifestation of collagen abnormalities and suggest a role for congenital predisposition in POP aetiology [2, 7, 8]. Additionally, prolapse is associated with levator ani muscle (LAM) trauma [9].

We hypothesised that postnatal POP in premenopausal primiparous women is not a condition caused by pregnancy alone but could be a clinical manifestation of a pre-existing, undiagnosed condition or a congenital predisposition triggered by pregnancy and delivery. The 4P-study reported here (Prevalence and Predictors of Pelvic Floor Dysfunction in Primips) aimed to elucidate the natural history of POP by investigating the role of a congenital component in postnatal POP in premenopausal primiparous women while at the same time considering such major confounding factor as LAM trauma .

Materials and methods

The 4P is a prospective cohort study nested within the parent Screening for Pregnancy Endpoints (SCOPE Ireland study; www.scopestudy.net), previously described in detail [10]. The study was approved by the Clinical Research Ethics Committee of the Cork Teaching Hospitals (CREC) Ireland and performed in a tertiary maternity hospital attending approximately 9,000 deliveries a year. All participants, 872 nulliparous women, completed the Australian Pelvic Floor Questionnaire [11] in early pregnancy and 1 year postnatally. The recruitment phase occurred between February 2008 and March 2011. All participants were invited for clinical follow-up between March and December 2012. We had a response rate of 60.8 % (530), and the proposal for follow-up was accepted by 46.8 % (408). Participants who had more than one delivery or were repeatedly pregnant at the time of follow-up [23.6 % (206)] were excluded from the study (Online Appendix A1). Clinical follow-up of 202 (23.2 % of all invited) participants who met the inclusion criteria consisted of POP-Q assessment measured on rest and on maximal Valsalva, joint hypermobility assessment using the Beighton score, transperineal 2D/4D ultrasound (US) scan for prolapse quantification. blood serum collection for procollagen quantification and a collection of personal and/or family history of collagen-related or other diseases previously shown to be associated with PFD and POP (Online Appendix A2). The Beighton joint mobility score is a system proposed to quantify joint hypermobility and is mainly used in rheumatological and orthopaedic practice. A generalised joint laxity is considered to be present with a score of four or more [12].

According to Dietz and Lekskulchai, POP can be quantified on 2D transperineal US scan (Tp-USS) [13]. Two images are acquired—one at rest and one at maximal Valsalva maintained for not less than 6 s in order to achieve the greatest pelvic organ descent [14]. The inferior margin of the pubic bone is considered as a reference line. The distance from the most distant bladder point or anorectal junction is measured to the reference line. Prolapse is then classified into significant and nonsignificant categories. For significant cystocele, a we proposed a cutoff level of 10 mm below the symphysis pubis (outside the pelvis) and 15 mm for rectocele; there was no cutoff level for uterine prolapse. LAM trauma, especially its puborectalis aspect, has a recognised association with POP. US-diagnosed avulsion is considered to be present if the urethro-LAM insertion distance is >25 mm on tomographic US investigation (TUI) [15]. We measured procollagen type III N-terminal propeptide (PIIINP) level in 96 participants, using commercial enzyme-linked immunosorbent assay (ELISA) tests by Uscn Life Science Inc. Wuhan. Participants with highest and lowest values for point C (leading edge of cervix on POP-Q assessment) were selected from the cohort for procollagen quantification. The main outcome measures were cystocele, rectocele and uterine prolapse.

Statistical analysis

All statistical analyses were performed using IBM SPSS 19 and Stata Software 10.0. All statistical tests were two sided, and a p value <0.05 was considered statistically significant. Log-linear binomial regression was used to estimate the relative risk (RR) of developing various types of POP in relation to mode of delivery (MoD). All regression models were adjusted for maternal age, body mass index (BMI), education, smoking and marital status. When convergence was not achieved, a recognised problem with this model, log-linear Poisson regression with “robust” estimation variance was used [16]. In separate models, we classified MoD into spontaneous vaginal, instrumental vaginal and CS deliveries to increase statistical power of the analysis. The association between various POP types and collagen level was assessed using Student’s t test. The association between POP and various medical conditions was analysed using chi-square test. Correlation between Beighton hypermobility score and POP was tested with Kruskal–Wallis test and Spearman rank correlation, as appropriate. We investigated the correlation between hypermobility score and presence of various types of prolapse in isolation as well as in combination. An ordered logistic regression was used to assess the impact of LAM trauma and MoD, among other risk factors (RFs), on main outcome measures. RFs investigated were: LAM hiatus >25 cm2, LAM avulsion, subpubic arch angle, collagen type III level, family and personal history of collagen diseases, foetal birthweight and head circumference, duration of first and second stage of labour, MoD, type of maternity care, induction of labour, acceleration of labour with oxytocin, epidural analgesia, perineal tear, episiotomy and perineal suturing. As a reference group, we used patients with no prolapse, who had a vaginal delivery, had an intact LAM, a levator hiatal area <25 cm2, absence of any medical history of collagen disease, no induction of labour, no epidural analgesia, no episiotomy etc. and who had a subpubic angle >90° and <100°.

Results

Clinical follow-up was attended by 202 participants (33.3 % of all eligible for follow-up). The study population consisted of 99.5 % Caucasian women, with mean age 31.2 years, mean BMI 25.1 kg/m2 and mean weight 68.0 kg. All demographic characteristics of women who participated in the 4P clinical follow-up study were similar to those in participants of the Screening for Obstetric and Pregnancy Endpoints (SCOPE) Ireland (Table 1).

Table 1 Demographic characteristics of the population in the Prevalence and Predictors of Pelvic Floor Dysfunction in Primips (4p) study reported here and the Screening for Obstetric and Pregnancy Endpoints (SCOPE) Ireland study
Full size table

We found a high prevalence of various types of POP. Most prevalent was cystocele, present in 90 % of participants, followed by uterine prolapse (89 %) and rectocele (70 %). None of the participants had prolapse grade 3 or higher (Table 2). In the majority of cases (65 %), there was a coexistence of the anterior-, central- and posterior-compartment prolapse. Two-compartment involvement was the second most common (25 %), with cystocele and uterine prolapse being the most frequently associated conditions. Solitary POP was the rarest finding in this study group (Table 3).

Table 2 Prevalence of various types of pelvic organ prolapse (POP) (n = 202)
Full size table
Table 3 Number of compartments involved (n = 197)
Full size table

Despite the high prevalence of POP on POP-Q examination, only 20 % of participants with prolapse were symptomatic. The prevalence of symptoms is presented in Table 4 as binary variables. There was a better association between prolapse diagnosis and presence of symptoms on Tp-USS, which outlines patients with significant prolapse only. On Tp-USS, 33 % of participants with POP were symptomatic (Table 4). We found a statistically significant correlation between procollagen type III (PIIINP) levels and uterine prolapse. Mean PIIINP level (SD) was higher in participants with uterine vs. nonuterine prolapse: 101(39) μg/ml vs. 69 (25) μg/ml. Mean difference was 32 μg/ml [95 % confidence interval (CI) 3–60, p = 0.01)] However, there was no significant association between collagen type III levels and presence of cystocele or rectocele.

Table 4 Prevalence of prolapse symptoms in association with various types of prolapse (n = 202)
Full size table

On examining risk factors in medical history, we found a statistically significant association between various types of POP and family history of uterine prolapse, cystocele, varicose veins and personal history of vertebral hernia, varicose veins and asthma (p < 0.05) (Online Appendix A2). A statistically significant association was found between Beighton score and rectocele (p = 0.036) but not for cystocele and uterine prolapse. However, when we analysed the correlation with a combined score for all prolapses, a statistically significant association was detected (p = 0.022). We analysed the correlation between various POPs and different risk factors (Table 5). A particular POP was associated with presence of other types of prolapse; however, the most persistent and significant risk factor was LAM trauma. During analysis of the impact of MoD on prolapse, in order to increase statistical power, we combined all instrumental deliveries and all CS (Table 6). We found that CS reduced the risk of cystocele and rectocele but not of uterine prolapse. Instrumental deliveries slightly but not significantly reduced the risk of rectocele.

Table 5 Correlation between different types of pelvic organ prolapse (POP) and various risk factors
Full size table
Table 6 Correlation between various types of pelvic organ prolapse (POP) and mode of delivery
Full size table

Discussion

In this study, we aimed to evaluate the role of childbirth in the development of postnatal POP by assessing primiparous women only at least 1 year postnatally to exclude transitional postnatal changes in the pelvic floor. Additionally, we intended to test the hypothesis stating that POP is a congenitally determined condition rather than a result of childbearing alone and especially of vaginal delivery. We hypothesised that if the condition were congenitally determined and caused by abnormal collagen quantity or quality, this change should have a systemic pattern, involving other parts of the body, as well having a family history of collagen-related disorders.

The prevalence of different types of POP appeared very high in our study. However, in the majority, prolapse grade was I–II and was asymptomatic. Our data are consistent with previous epidemiological studies, and the question of changing the prolapse classification to a more clinically relevant one has been raised in the past [18, 19]. There seems to be a better clinical correlation between presence of prolapse symptoms and US findings, where prolapse is classified into clinically significant or nonsignificant [13]. The actual POP-Q system comprehensively describes different types of POP. However, the system has a poor association with clinically meaningful prolapse, resulting in labelling the majority of patients as having prolapse and causing unnecessary anxiety and sometimes even surgical interventions.

Collagen plays an important role in the human body, offering biomechanical strength and elasticity to connective tissue. The process of collagen synthesis consists of five steps. The first two are intracellular, occurring in fibroblasts; the remaining three are extracellular [20]: When a procollagen molecule leaves the cell, it is transforming from procollagen into tropocollagen by proteolytic cleavage of carboxy and amino terminals. These terminals can be detected and quantified as markers of collagen synthesis in blood serum or frozen tissue homogenates [21]. Procollagen N-terminal propeptides (PNP) and C-terminal propeptides (PCP) are specific to collagen type and can be quantified by radioimmunoassay [22] or ELISA test [23]. There are 28 types of collagen; however only types I, II and III are major supporting collagens [24]. Type I is the most prevalent and widespread in the body, constituting 90 % of total body collagen, especially being abundantly present in bone tissue, where strength is required. Collagen type II is mostly found in cartilage and type III in soft tissue, e.g. skin, ligaments [24]. There is controversy regarding collagen content in patients with prolapse, the consensus being collagen type I is minimally changed or has a tendency to slightly lower levels in severely prolapsed patients; collagen type III seems to undergo the most important change, and its concentration appears to be significantly increased, which changes the type I/III ratio [20, 25]. Edwall et al. suggested this finding may be due to an increased collagen turnover in patients with prolapse following intensive breakdown, especially for collagen type III.

In our study, the majority of participants were relatively young (78 % <34 years), and POP was only grade I–II. We decided to investigate the synthesis of collagen type III only because, according to previous studies, this is where we would expect to find the most significant change [20, 25]. Our results were consistent with previous findings, showing a statistically significant increase in collagen type III synthesis in participants with uterine prolapse. However, we found no statistically significant association with cystocele or rectocele. This may be due to the fact that the dry weight of uterine attachment ligaments comprises 70–80 % collagen, where collagen type III has the highest content in the body; in fascial tissue, such as vesicovaginal or rectovaginal septum, collagen III content is lower [22, 24].

Collagen abnormality appears to be a systemic issue. Although genes coding collagen synthesis may have different expressions in various tissues, there is a body of evidences indicating that collagen-related problems seem to have a systemic manifestation. It has been demonstrated that POP is associated with joint hypermobility; abdominal, inguinal and vertebral hernias; varicose veins; mitral valve prolapse etc. [5, 6, 22, 26]. Our results are consistent with these findings, showing a statistically significant association between POP and various family and personal medical history risk factors. In clinical practice, introducing a relatively simple investigation for collagen levels, such as ELISA of blood serum, could help identify antenatally the patients at higher risk of POP. However, larger studies are needed to confirm the potential benefit of this investigation, which demonstrates that uterine prolapse, for instance, is associated with pregnancy per se, rather than with MoD.

A multivariate analysis was performed to assess the correlation of significant risk factors described above and the impact of LAM trauma and MoD on POP, excluding most common confounders. The analysis showed that the associations between collagen level and family and personal history of collagen-related diseases were confounded. In multivariate analysis, levator avulsion was significantly associated with different types of POP, and the presence of rectocele was associated with uterine prolapse. The latter partially confirms the role of multicompartment involvement of POP in the majority of participants as a reflection of generalised pelvic floor weakness. Low yield of statistically significant results in multivariate analysis could probably be explained by the limited number of observations of these variables in our study. Multivariate analysis did not demonstrate that vaginal delivery increases the risk POP; however, surprisingly, we found that CS is significantly protective against the development of rectocele and cystocele only and does not affect the occurrence of uterine prolapse (Table 6). Attempting to explain this selective impact of MoD on different types of POP, we hypothesised that the significantly protective effect of CS against cystocele and rectocele, in contrast to uterine prolapse, could possibly be explained by the different attachment mechanism of the uterus on the one hand and vagina on the other. The main uterine support consists of cardinal and uterosacral ligaments. Their insertion point is concentrically focussed the around lower uterus and is perpendicular to the birth canal axis. Paravaginal support was demonstrated by DeLancy to have three levels of attachment [27]. Although level one, which is similar to the uterine attachment, plays an important supportive role, the area of level two is much larger. It consists of vesicovaginal fascia and rectovaginal septum, both running parallel to the birth canal axis. Level two resembles a sail framed by arcus tendineus and fascia of levator ani muscles. Thus, vesicovaginal fascia and rectovaginal septum are peripherally attached, whereas the midline part, projecting on the urogenital hiatus, is the most mobile and exposed area. In addition, there is a different distribution of pressure vectors during labour on supportive structures through the birth canal. During vaginal delivery, the uterine pressure vector is directed along the longitudinal uterine axis. Suspension ligaments, focussed around the exit of the fully dilated uterus, have opposing contradirectional resistance in a parallel plane. In the vagina, the pressure vector is perpendicular to vesicovaginal fascia and rectovaginal septum [28]. The main impact on uterine position appears to result from pregnancy, during which gravitation facilitates continuous overdistension of the uterine ligaments, which contain a higher level of abnormal collagen type III. In the case of rectocele and cystocele, damage is probably produced during labour, when the foetal head is progressing through the birth canal and is overstretching the vagina. This can lead to avulsion injury of the paravaginal supportive structures. Considering this, additional strategies apart from CS must be investigated to prevent POP and especially uterine prolapse; such strategies include education around avoiding risk factors, such as smoking, which has been shown to impair collagen quality [29], or performing life-long pelvic floor exercises to reinforce the pelvic floor.

Strengths and limitations

The main strength of our study is its comprehensive approach to the investigation of POP. It included detailed family and personal history of related medical conditions, use of a validated questionnaire, POP-Q assessment, transperineal scan and collagen quantification. Additionally, this study encompasses a large number of well-phenotyped nulliparous participants followed up to 1 year postnatally and is representative of the entire population. Although we had a clinical follow-up rate of 33 % of eligible participants, demographic characteristics of these women was similar to the SCOPE cohort (Table 1), which has been shown prepregnancy to be representative of the entire nulliparous population [10]. High homogeneity of the study population is another strength, allowing us to overcome naturally occurring confounders for POP, such as advanced age and interracial differences. The majority of participants was relatively young (78 % <34 years) and Caucasian (99 %), which enables a better understanding of the natural history of POP. Previous studies have shown that white women are considered more prone to develop prolapse and urinary incontinence (UI) [30]. All patients were delivered in the same hospital following similar protocols and obstetric approaches, which excludes various obstetric management confounders. The main limitation of the study is that our participants did not have a POP-Q and Tp-USS assessment in early pregnancy, i.e. at the time of recruitment. Also, we recognise that our study could be underpowered for making a definite conclusion on the impact of MoD on POP. The low clinical follow-up rate due to high exclusion numbers as a result of the next ongoing pregnancy is another limitation. This could also explain why the number of participants >34 years was slightly higher than in SCOPE data. However, we dealt with this issue by doing a multivariate logistic regression, including age as a confounder. The slight difference with SCOPE demographics can be also explained by the fact that all participants were at least 1 year older than at recruitment. In addition, we investigated the association between POP and 35 medical conditions. We acknowledge that having so many tests creates a possibility of chance findings. However, we investigated only conditions previously shown to be associated with POP.

Conclusion

POP has a very high prevalence among premenopausal primiparous women, with the majority having multicompartmental involvement. In the majority of our patients, prolapse was asymptomatic; however, it may well contribute to future pelvic floor morbidity in the postmenopausal period. The relationship between POP and family and personal history of collagen abnormalities, the association between uterine prolapse and collagen level and the lack of correlation between MoD and uterine prolapse grade is suggestive of an important congenital contribution to POP aetiology. Similarly, LAM trauma proved to be a paramount risk factor. CS seems to be protective against cystocele and rectocele, with no effect on uterine prolapse; this issue warrants further research. POP-Q is a very comprehensive classification system; however, we agree that perhaps it has a more valuable role in research than in clinical settings in which grade I–II prolapse is assessed. Considering the fact that POP seems to have a strong congenital predetermination, a thorough analysis of risk factors could help identify women at a higher risk of POP and implement some preventive measures.