Urogynaecology: Collagen metabolism in the uterosacral ligaments and vaginal skin of women with uterine prolapse

Authors

  • Christian H Phillips,

    Corresponding author
    1. Faculty of Medicine Maternal, Fetal & Neonatal Physiology Group, Department of Gynecology, Princess Anne Hospital, Southampton, Hampshire, UK
    Search for more papers by this author
  • Fred Anthony,

    1. Faculty of Medicine Maternal, Fetal & Neonatal Physiology Group, Department of Gynecology, Princess Anne Hospital, Southampton, Hampshire, UK
    Search for more papers by this author
  • Chris Benyon,

    1. Faculty of Medicine Maternal, Fetal & Neonatal Physiology Group, Department of Gynecology, Princess Anne Hospital, Southampton, Hampshire, UK
    Search for more papers by this author
  • Ash K Monga

    1. Faculty of Medicine Maternal, Fetal & Neonatal Physiology Group, Department of Gynecology, Princess Anne Hospital, Southampton, Hampshire, UK
    Search for more papers by this author

CH Phillips, Faculty of Medicine Maternal, Fetal & Neonatal Physiology Group, Department of Gynecology, Princess Anne Hospital, Coxford Road, Southampton, Hampshire SO16 5YA, UK.

Abstract

Objective  To compare tissue markers of collagen metabolism in the uterosacral ligaments with those in vaginal tissue in women with uterine prolapse.

Design  Prospective observational experimental study.

Setting  A tertiary urogynaecology unit.

Population  Women referred for hysterectomy for prolapse or benign gynaecological disease.

Methods  Matrix metalloproteinase (MMP)-2 and -9 expression, tissue inhibitors of metalloproteinase (TIMP)-2 expression and hydroxyproline content were measured in the uterosacral ligaments and vaginal tissue from 14 women with prolapse compared with 14 controls.

Main outcome measures  Levels of MMP, TIMP and hydroxyproline in the uterosacral ligaments and vaginal tissue of women with prolapse and controls.

Results  Fourteen women with prolapse and 14 women without prolapse (controls) were included. A significant increase in pro MMP-2 expression was seen in vaginal tissue from women with prolapse (P < 0.05) but not activated MMP-2, MMP-9 and TIMP-2. For uterosacral ligaments, the differences were not statistically significant. No significant difference in hydroxyproline content was found between control and prolapse in either tissue. Significant correlations exist in expression of pro-MMP-2, activated MMP-2, MMP-9 and TIMP-2 in vaginal tissue with that in uterosacral ligaments.

Conclusions  Correlations existed between markers of collagen metabolism in the vaginal and uterosacral tissues. This suggests vaginal tissue reflects the endopelvic fascia. The changes which are more pronounced in vaginal tissue may be as a result of prolapse rather than cause.

Introduction

Genitourinary prolapse is a common complaint, with approximately 11% of all women requiring at least one corrective surgical procedure.1,2 The pelvic organs are supported by both the muscles of the pelvic floor and their intact attachments to the endopelvic fascia. Neuromuscular damage that occurs to the pelvic floor during childbirth is seen in some women who later develop prolapse, although this damage is not seen in all cases.3–5 Thus, it is thought that deficiencies in the strength of the endopelvic fascia may be a cause of genitourinary prolapse in some cases. The most important constituent of the endopelvic fascia is the glycoprotein collagen, which is a fibrous protein that forms cross-linkages between adjacent collagen fibres. The amount of collagen, type of collagen and degree of cross-linking all contribute to the properties and strength of the tissue.

Collagen is degraded by a family of enzymes called the matrix metalloproteinases (MMPs), which are regulated by locally produced tissue inhibitors of metalloproteinases (TIMPs).

Abnormalities in collagen metabolism have been identified in association with pelvic organ prolapse and stress incontinence.6,7 Reduced collagen content, altered ratios of different collagen types, increased tissue expression of MMPs and altered collagen cross-linkages have been found in samples of vaginal tissue taken from women with prolapse compared with controls.8 It has therefore been suggested that increased degradation of collagen may lead to a decrease in mechanical strength and predispose the individual to prolapse. Chen et al.9 found increased MMP-1 and reduced TIMP-1 expression in vaginal tissue of women with stress urinary incontinence. However, the vaginal tissue does not provide support to the uterus. The uterosacral ligaments are condensations of endopelvic fascia which provide the primary support, holding the uterus in place.10,11 If prolapse is due to a global increase in collagen degradation as suggested by previous workers, then changes in collagen metabolism similar to those in the vaginal tissue would be seen in the more important supportive structures of the uterus. Recent work has found a direct relationship between collagen III expression in the cardinal ligaments and the presence of prolapse.12

The aim of the present study was to determine if similar changes in collagen metabolism that have been seen in vaginal tissue occur in the uterosacral ligaments of women with prolapse compared with controls. We also sought to assess the suitability of vaginal epithelium as a model for changes in those ligaments and the endopelvic fascia.

Methods

Ethical approval was obtained for the study. Women awaiting hysterectomy for benign conditions were asked to participate in the study. Patients were excluded if they had pelvic malignancy, fibroids, endometriosis, pelvic inflammatory disease or previous pelvic surgery. Patients were matched for age, parity and hormonal status. All patients were examined and their prolapse scored according to the International Continence Society Pelvic Organ Prolapse (POP-Q) Classification.13 The study group had POP-Q prolapse stage 2 or more and the control group had a POP-Q prolapse stage 1 or less.

Samples of vaginal epithelial tissue and uterosacral ligament were obtained at hysterectomy regardless of route. Each biopsy measured approximately a 5-mm cube. The site of the vaginal specimen was 1 cm from the cervix in the midline anterior portion of the pericervical cuff.7,8 The biopsies of the uterosacral ligament were taken 1 cm from their insertion into the cervix. Previous histological examination suggested this location within the uterosacral ligament was the most important in providing uterine support.10 All samples of uterosacral ligament were confirmed by histological analysis. Each tissue sample was snap frozen and stored in liquid nitrogen before analysis for collagen metabolism.

Each sample was thawed, washed with phosphate-buffered saline (PBS), blot dried and measured. The tissue was divided into two portions, each weighing approximately 50 mg.

One portion was used for quantification of hydroxyproline content.14 These samples were hydrolysed in 6 M hydrochloric acid, at 105°C for 24 hours and allowed to evaporate to dryness. One milliliter of citrate buffer (14.7 g trisodium citrate, 9.2 g citric acid, 500 mL distilled water) was added to each sample and incubated for 15 minutes with 250 μl of chloramine T reagent [40 mL reagent added: (28.44 g sodium citrate, 18.82 g trisodium citrate, 2.52 g citric acid, 198 mL distilled water, 309 mL propan-2-ol) to 0.7 g chloramines T + 10 mL distilled water]. Samples were then placed on ice and 2.0 mL of Ehrlich's reagent (6.67 g p-dimethylaminobenzaldehyde, 11 mL concentrated HCl, 89 mL propan-2-ol) was added to each sample and then capped and left overnight at room temperature. The next morning, tubes were mixed on a multi-vortexer and left for 1 hour and the optical density of the samples and standards were read in a Dynex Technologies MRX 96-well plate reader at 560 nm. Hydroxyproline content was measured against standard concentrations and is expressed as micromole per gram of tissue.

The remaining portion of the biopsy was used to measure expression of MMP and TIMP as follows. The biopsy was homogenised using an Ultra-Turrax T8 homogeniser on ice for 4 × 30 seconds, with the appropriate volume of PBS to give a concentration of 100 mg tissue/mL of PBS. The homogenised sample was spun in a centrifuge at 13,000 rpm for 60 seconds and the supernatant removed. Total protein concentration within the supernatant was assessed using Coomassie blue reagent (Pierce Cat No. 2320, Illinois, USA) and colour absorption measured at 630 nm in a plate reader. Total protein content was quantified against albumen standards. Zymography was performed for quantification of MMP-2 and 9 (see 1Fig. 1).15 Quantification of enzyme expression for pro and active MMP-2 and pro-MMP-9 [measured in absorption units (AU) of light transmittance] was determined by transmittance densitometry of the gels on a flatbed scanner (Sharp JX330P at 300 dpi, Sharp Corporation, Japan) using a computerised gel analysis package (Phoretix, Newcastle, UK). TIMP-2 expression was estimated using a standard enzyme-linked immunoadsorbent assay for TIMP-2 (Amersham Pharmacia, Biotrak Cellular Communication Assays, code RPN 2618, Bucks, UK). Band densities of MMP-2 and 9 are expressed as AU/mg protein, and concentrations of TIMP-2 are expressed as nanogram per milligram total protein extracted.

Figure 1.

A representative zymogram. Distinct bands can be seen at 72 kDa and 66 kDa where the pro and active forms of MMP-2 have degraded the gelatin and not taken up the gel stain. A band can be seen at 92 kDa for pro-MMP-9 but no distinct band can be seen at 86 kDa for active MMP-9. Hence, only the well-defined bands were quantified. Lanes 1 and 2 demonstrate dilutions of standard purified human MMP-2, lanes 3 and 4 demonstrate dilutions of purified human MMP-9 and lanes 5–7 represent a tissue sample in triplicate, which show the presence of MMP-2 and MMP-9.

Non-parametric analyses were used for comparisons and correlations as the data were sometimes skewed. The Wilcoxon matched pairs, signed rank test and Spearman rank correlation test were used.

Results

Twenty-eight women (14 with stage 2–4 prolapse, 14 with stage 0–1 prolapse) met the entry criteria and could be appropriately matched. Sufficient tissue was available to perform quantification of hydroxyproline on all 28 patients, but in only 26 was there also enough tissue for MMP and TIMP analysis. There were no differences in the patient demographics between the two groups (1Table 1).

Table 1.  Demographic details of patients
VariableControl group (n= 14)Prolapse group (n= 14)
  1. Age is presented as median (range), and parity as median (range).

  2. All premenopausal women were sampled in the second half of their menstrual cycle (secretory phase).

Age (years)54 (36–84)54 (38–87)
Parity2 (0–6)3 (0–7)
Postmenopausal: noHRT77
Postmenopausal: onHRT11
Premenopausal66

Results of the analyses on vaginal tissue are shown in 2Table 2. Results in 3Tables 2 and 3 have been separated into total population as well as premenopausal and postmenopausal subpopulations. This was initially done to see if any significant differences were present in the expression of MMP and TIMP between the subpopulations; however, this does not appear to be the case. Separation of results into subpopulations also helps to demonstrate whether any significant results in the total population may be skewed by results in a subpopulation. No difference was seen in the hydroxyproline content of vaginal tissue from women with prolapse compared with controls. There was a significant increase in pro-MMP-2 expression in the vaginal tissue of women with prolapse compared with controls (2Fig. 2) but no differences in the expression of activated MMP-2, pro-MMP-9 and TIMP-2 (Table 2). A significant increase in pro-MMP-2 expression was seen in the vaginal tissue of the postmenopausal subpopulation but was less significant than in the population as a whole, which suggests the increase found in the total population is not due to changes in the postmenopausal group alone.

Table 2.  Tissue composition of vaginal tissues of women with prolapse compared with controls
AssayPatientsControlProlapseP
  1. Data are presented as median (range) with P values included for significant data only (Wilcoxon matched paired, signed rank test).

Hydroxyproline (μmol/g tissue)Total population (n= 28)96 (51–265)96 (54–176) 
 Premenopausal (n= 12)90 (62–144)96 (65–176) 
 Postmenopausal (n= 16)107 (51–265)90 (54–145) 
Pro-MMP-2 (AU/mg protein)Total population (n= 26)3.2 (1.2–13.0)8.4 (3.2–21.0)0.0034
 Premenopausal (n= 10)3.2 (1.2–13.0)5.2 (3.2–11.0) 
 Postmenopausal (n= 16)3.3 (1.6–13.0)9.4 (5.41–21.0)0.0234
Active-MMP-2 (AU/mg protein)Total population (n= 26)3.9 (1.5–8.1)5.2 (1.1–17.3) 
 Premenopausal (n= 10)2.7 (1.7–5.7)4.3 (1.1–5.2) 
 Postmenopausal (n= 16)4.2 (1.5–8.1)5.7 (2.6–17.3) 
Pro-MMP-9 (AU/mg protein)Total population (n= 26)1.7 (1.0–6.0)2.6 (1.3–7.2) 
 Premenopausal (n= 10)1.7 (1.0–6.0)3.5 (2.5–7.2) 
 Postmenopausal (n= 16)1.5 (1.1–6.0)2.3 (1.3–5.5) 
TIMP-2 (ng/mg protein)Total population (n= 26)6.5 (0.6–33.1)7.3 (0.5–33.8) 
 Premenopausal (n= 10)9.6 (1.8–16.8)5.6 (2.2–7.4) 
 Postmenopausal (n= 16)2.2 (0.6–33.1)13.0 (0.5–33.8) 
Table 3.  Tissue composition of uterosacral ligaments of women with prolapse compared with controls.
AssayPatientsControlProlapse
  1. Data are presented as median (range) with P values included for significant data only (Wilcoxon matched paired, signed rank test).

Hydroxyproline (μmol/g tissue)Total population (n= 28)82 (36–122)79 (36–132)
 Premenopausal (n= 12)92 (62–111)75 (36–132)
 Postmenopausal (n= 16)81 (36–122)70 (53–109)
Pro-MMP-2 (AU/mg protein)Total population (n= 26)4.8 (1.4–10.7)5.6 (1.7–21.9)
 Premenopausal (n= 10)4.8 (1.4–7.9)5.6 (1.7–9.1)
 Postmenopausal (n= 16)4.9 (3.6–10.7)6.0 (2.5–21.9)
Active-MMP-2 (AU/mg protein)Total population (n= 26)4.8 (1.1–8.2)6.2 (0.5–15.0)
 Premenopausal (n= 10)3.2 (1.1–7.1)3.8 (0.5–9.5)
 Postmenopausal (n= 16)5.4 (1.6–8.2)6.4 (3.6–15.0)
Pro-MMP-9 (AU/mg protein)Total population (n = 26)1.6 (0.8–6.3)2.2 (1.1–6.6)
 Premenopausal (n= 10)2.6 (0.8–6.3)2.8 (1.1–5.4)
 Postmenopausal (n= 16)2.0 (1.3–4.7)2.6 (1.1–6.6)
TIMP-2 (ng/mg protein)Total population (n= 26)4.2 (0.7–8.7)7.1 (1.1–30.1)
 Premenopausal (n= 10)6.7 (1.2–8.7)5.4 (1.1–12.9)
 Postmenopausal (n= 16)3.1 (0.7–6.9)8.3 (3.4–30.1)
Figure 2.

Median and distribution of pro-MMP-2 expression (AU/mg total protein) in vaginal tissue from women with prolapse compared with controls.

An attempt was made to assess whether tissue remodelling or degradation was taking place by examining the relationship between MMP-2 expression with TIMP-2 and hydroxyproline expression. A negative relationship between MMP and TIMP or hydroxyproline expression suggests degradation whereas a positive relationship suggests remodelling and repair.16,17 Again analyses of the premenopausal and postmenopausal subpopulations were made to determine if differences exist between the possible mechanisms seen with prolapse in each population.

A positive relationship was found between the expression of pro-MMP-2 with the expression (r= 0.85, P < 0.01), of TIMP-2 and also between activated MMP-2 and the expression of TIMP-2 in samples of vaginal skin from postmenopausal women with prolapse (r= 0.9, P < 0.01). There was also a positive relationship between pro-MMP-2 and hydroxyproline content of vaginal tissue (r= 0.88, P < 0.01), and between activated MMP-2 expression and the hydroxyproline content (r= 0.73, P < 0.05). There was no significant relationship between proMMP-2 and activated MMP-2 expression with TIMP-2 and hydroxyproline expression in the vaginal tissue of premenopausal women.

Results of the analyses on uterosacral ligaments are shown in Table 3. There were no significant differences in the expression of pro MMP-2, active MMP-2, pro MMP-9, TIMP-2 or hydroxyproline content between control and prolapse tissues.

Levels of the assayed proteins were compared between vaginal skin and uterosacral ligaments using Spearman's rank correlation. Significant correlations were found between the vaginal tissue and uterosacral ligaments for all assays (3Fig. 3a–d).

Figure 3.

Figure 3.

Non-parametric correlations between expression of proteins in the vaginal skin with their expression in the uterosacral ligaments. Solid lines represent the regression line, with 95% confidence intervals shown by the dotted lines. Spearman's rank correlation: (a) r= 0.69, P≤ 0.0001; (b) r= 0.64, P= 0.0005; (c) r= 0.42, P= 0.038; (d) r= 0.43, P= 0.031.

Figure 3.

Figure 3.

Non-parametric correlations between expression of proteins in the vaginal skin with their expression in the uterosacral ligaments. Solid lines represent the regression line, with 95% confidence intervals shown by the dotted lines. Spearman's rank correlation: (a) r= 0.69, P≤ 0.0001; (b) r= 0.64, P= 0.0005; (c) r= 0.42, P= 0.038; (d) r= 0.43, P= 0.031.

Figure 3.

Figure 3.

Non-parametric correlations between expression of proteins in the vaginal skin with their expression in the uterosacral ligaments. Solid lines represent the regression line, with 95% confidence intervals shown by the dotted lines. Spearman's rank correlation: (a) r= 0.69, P≤ 0.0001; (b) r= 0.64, P= 0.0005; (c) r= 0.42, P= 0.038; (d) r= 0.43, P= 0.031.

Figure 3.

Figure 3.

Non-parametric correlations between expression of proteins in the vaginal skin with their expression in the uterosacral ligaments. Solid lines represent the regression line, with 95% confidence intervals shown by the dotted lines. Spearman's rank correlation: (a) r= 0.69, P≤ 0.0001; (b) r= 0.64, P= 0.0005; (c) r= 0.42, P= 0.038; (d) r= 0.43, P= 0.031.

Discussion

This study has shown that the expression of pro-MMP-2, active MMP-2, pro-MMP-9 and TIMP-2 in vaginal tissue does reflect the expression of these proteins in uterosacral ligaments. Some MMP-2 expression is elevated in vaginal tissue of women with prolapse compared with controls, but significant differences were not found for all parameters measured in the vaginal tissue or uterosacral ligaments. Changes in collagen metabolism may reflect a degree of tissue remodelling rather than degradation.

This study has strengths and weaknesses when compared with the existing literature. The first benefit is the tissue sampling from both the vaginal tissue and supporting parametrium. Secondly, the relationships between MMP and TIMP expression were assessed rather than just addressing MMP expression alone. Previous studies have looked at one tissue type or the other but our data are the first to analyse the relationship between MMP and TIMP expression in both the vaginal tissue and supporting ligaments.8,9,12,18 The significant correlations that existed between the levels of hydroxyproline, MMPs and TIMP-2 in vaginal tissue and uterosacral ligaments confirm the findings in each tissue and demonstrated a sound laboratory technique. Patients were matched more strictly for hormonal status and phase of cycle in this study than in other studies. This reduced any error that may have be introduced by sex hormones on MMP expression, because these have been shown to fluctuate in the endometrium, vaginal tissue and parametrium during the menstrual cycle.19–21 Furthermore, in this study MMP and TIMP were expressed in relation to total protein content rather than weight,8,18 which improved accuracy by accounting for any variation in the amount of enzyme extracted at tissue homogenisation.

One possible weakness in our study was sample size. With a larger population, perhaps more statistically significant results may have been seen in the uterosacral ligaments. However, the initial power calculation was based on similar work and similar studies from other centres were of similar size to our study.8,9,18

There was no significant difference in the hydroxyproline content of vaginal tissue of both premenopausal and postmenopausal women when comparing women with prolapse to controls without prolapse. There is contradictory evidence in the literature regarding the quantity and quality of collagen and its association with the development of prolapse or urodynamic stress incontinence. Two papers found reduced hydroxyproline in association with prolapse and stress incontinence,7,8 whereas others found no difference in hydroxyproline expression.22 In contrast to Jackson et al.,8 Falconer et al.23 showed an increase in the hydroxyproline content in premenopausal women with incontinence, compared with controls and suggested that in women with incontinence the connective tissue was more rigid with an impaired mechanical function compared with continent controls. One recent study has shown that collagen fibril diameter is 25% larger in the uterosacral ligaments of women with prolapse and stress urinary incontinence compared with controls.24

Numerous workers have focussed on the properties and metabolism of collagen in the vaginal epithelium and in particular the fascia immediately deep to the vaginal epithelium.7,8,16,24 This is probably due to the relative ease with which these biopsies can be harvested. However, the histology of the paracolpium and uterosacral and cardinal ligaments demonstrates that these structures have a different composition to the vaginal epithelium. The uterosacral ligaments possess a higher proportion of smooth muscle within their connective tissue compared with vaginal tissue.25–27 This difference in composition is reflected in differences in the physical properties of the two tissues. Tension load testing has shown that the load and distension parameters for vaginal skin are significantly different to that of the uterosacral ligament.28 Takano et al.29 found reduced amounts of collagen in the parametrium of women with prolapse compared with controls, but no difference in the biopsies of apical vaginal tissue. Our data showed significant correlations between all assayed protein levels in the vaginal tissue compared with the uterosacral ligaments, suggesting that vaginal tissue does reflect the changes in collagen metabolism within the endopelvic fascia but that these are more pronounced in the vaginal tissue. We hypothesise that this is due to the reduced ability of vaginal tissue to resist tensile forces in comparison with the uterosacral ligaments. We suggest that the differences seen in MMP-2 and -9 and TIMP-2 expression may partially reflect the effects of tissue stretching rather than the cause of tissue weakness. A similar phenomenon has been demonstrated in cervical tissue and fetal membranes.30 Recent work from our department has addressed the effects of stretching on different areas of vaginal skin in the same patient and shown that MMP expression is greater in non-rugose vaginal tissue that has been stretched by prolapse compared with non-stretched vaginal tissue (without rugae) taken from the same patient. Furthermore, a correlation was seen between the severity of prolapse and the elevation in MMP-2 expression in the areas of vaginal tissue that were stretched.31 We suggest these data reflect an inherent abnormality in collagen metabolism in certain individuals, which causes them to react in an up-regulated fashion to tissue stretching to produce a greater expression of metalloproteinases than in non-susceptible individuals.

In conclusion, there is elevated MMP activity in the vaginal skin of women with prolapse. No significant differences were seen in the uterosacral ligaments of these women. Strong correlations were seen for almost all markers of collagen metabolism between vaginal skin and uterosacral ligaments, with an apparent exaggerated effect in the vaginal tissue. This may represent a resistance to stretching within the uterosacral ligaments and the exaggerated results seen in the vagina may partly arise as a result of stretching and not just due to changes in collagen metabolism preceding the prolapse. In view of this, caution should be employed in interpreting results obtained from vaginal tissue and extrapolating these to the endopelvic fascia. Further work is needed to elucidate whether these changes are primary or secondary within the endopelvic fascia and how they may be rectified.

Ancillary