Evidence of compensatory elastin remodeling in non-prolapsed uterosacral ligaments and its divergence in histopathological subtypes of pelvic organ prolapse.

USL biopsies were obtained as previously described from women without POP undergoing hysterectomy for benign gynecological conditions (i.e., abnormal uterine bleeding, symptomatic fibroid uterus; controls) and from women with POP [ 39 ] with approval of the Colorado Multiple Institution Review Board (COMIRB #15–2245) and the Rutgers Institutional Review Board (#Pro20160001312). Written informed consent was obtained. Menopause was defined as no menses within the past year or previous oophorectomy. Using the Pelvic Organ Prolapse Quantification System (POP-Q [ 40 ]), women with greater than or equal to stage II descent were assigned to the POP group and women with less than or equal to stage I descent were assigned to the control group. During surgery, a 0.5 cm segment of each USL was biopsied within 1 cm of their cervical insertion and placed in neutral buffered formalin, paraffin embedded, sectioned, and trichrome stained. A trained histopathologist (Orlicky) utilized the semi-quantitative histopathological POP-HQ scoring system [ 39 ] to assess each USL, the full extent of each USL-section was examined and scored. Exclusion criteria included current pregnancy, previous POP or urinary incontinence surgery, current endometriosis, history of malignancy, or any known genetic alteration or disease of elastin or collagen. A 4-micron section from each USL was Masson’s trichrome stained for POP-HQ subtyping and collagen content, and another was stained to show the elastin fibers using a modified Hart’s-staining method [ 41 ]. All the Hart’s-stained sections were examined at least twice before scoring based on the criteria listed in Supplemental Table 1. All semi-quantitative scoring was performed at least 3 times to ensure reproducibility. Histological images were captured on an Olympus BX51 microscope equipped with a DP73 digital camera (Olympus) using the Cell Sense Application Program; all images were processed identically [ 42 ]. Tissues with crush-injury, cauterization, or inadequate fixation were excluded from analysis. All histology figures are reproduced at a larger size in the supplemental figures. To quantify the elastin or vimentin content following immunohistochemistry, 3–4 representative images were made from each tissue section and imported into SlideBook software (Intelligent Imaging Innovations, Denver, Colorado). Regions of interest were segmented from all images; the negative threshold was determined, and positive pixels were quantified as well as the total pixels of the segmented area. Lateral edges of all images were excluded from segmented areas to obviate any possible lateral (transverse) chromatic aberration effects. Data was expressed as the percentage-of-positive pixels. Importantly, when quantifying the elastin staining, large vessels (both arteries and veins) were excluded from the segmented areas. Elastin IHC. Another 4-micron section was IHC stained using an anti-elastin antibody; Supplemental Table 2 delineates the antibody source and IHC method. Three to five images of each anti-elastin IHC-stained USL section were captured at 100x magnification, and the total USL elastin content was quantified in the images as the percent of pixels positive for the anti-elastin staining using SlideBook software (Intelligent Imaging Innovations, Denver, Colorado). The mean percent-of-positive pixels of anti-elastin IHC staining is reported as Elastin-IHC. Vimentin-IHC. Lastly, another 4-micron section was IHC-stained using an anti-vimentin antibody as detailed in Supplemental Table 2. Anti-vimentin IHC-stained USL sections were imaged, and the staining was quantified in a manner like that for Elastin-IHC. The mean percent-of-positive pixels of anti-vimentin IHC staining is reported as Vimentin-IHC. R software was used for all statistical analyses, with required packages indicated where appropriate [ 43 ]. Welch’s unequal variances t-test was used when the subgroups had unequal variances and were of unequal sample size. Analysis of variance with Tukey’s post-hoc pairwise test was used when making multiple comparisons between continuous variables. Pearson’s Chi-squared test with Yates’ continuity correction for categorical variables was used where required (R package ‘RVAideMemoire’). Significance was set at p < 0.05 for all data, however we understand that performing so many comparisons may lead to a few false positives therefore these P-values should be viewed as exploratory. In the tables, mean values that significantly differ from one another by group are denoted by letters (a, b, c), p < 0.05. Variables sharing the same letter are not significantly different. If two variables have different letters, they are significantly different. Although semi-quantitative scoring was performed in places, those ordinal variables were analyzed using parametric methods for simplicity and this is a limitation of our exploratory study. Where linear regression was performed, we visually inspected data for each specimen group to test assumptions about how representative data are likely to be across the population. Both the limited number of specimens used, and the cross-sectional nature of the study were considered when evaluating linear relationships (statistical significance and R-squared values) between patient and USL tissue parameters. Graphing of data was performed with the program Python [ 44 ], and the linear regression package sklearn (in SciPy) was used for R-squared. POP-subgroup trendlines were compared to Controls using ANCOVA in SciPy. Principal component analysis (PCA) was performed using R packages ‘factoextra’, ‘fviz_pca’, ‘prcomp’, and ‘princomp’ [ 43 ].

Our study cohort included 37 Controls and 63 patients with POP. POP-HQ-subdivided POP patients into 27 POP-A, 19 POP-I, and 17 POP-V. Subjects were chosen based upon the integrity of the USL tissue, the completeness of the subject’s demographic data, and their POP-HQ subgroups (Table 1 ). Table 1 indicates demographic mean ± standard deviation, and intergroup statistics for the categorization employed within this study fully realizing that these groups are heterogeneous for many other demographics like all human studies. The subjects in all three POP-subgroups were older, had more vaginal births, and were more likely to be post-menopausal than the Control group (Table 1 , all p < 0.05). The body mass index (BMI) and smoking status of the Control group and POP-subgroups did not differ. Only the POP-V subgroup had less than a 50% white ethnicity composition. Since estrogen and progesterone are known to regulate genes involved with extracellular matrix synthesis [ 45 , 46 ], estrogen and progesterone pharmaceuticals usage within the past 3 months is indicated. The POP-A and -I subgroups contained a higher percentage of estrogen users than the Control or POP-V subgroups. POP-A and POP-V subgroups contained a lower percentage of progesterone users than the Control or POP-I subgroups. Despite the differences in hormone usage, the incidence of fibroids was similar in the subjects of all 4 subgroups. Table 1 Summary of the Major Demographics; statistical comparisons as indicated in Data Analysis CONTROL POP-A (Adipose) POP-I (Inflammatory) POP-V (Vascular) N 37 27 19 17 Feature Mean ± SD Mean ± SD Mean ± SD Mean ± SD Age 43.3 ± 10.9 a 60.3 ± 12.0 b 60.7 ± 10.8 b 64.0 ± 10.8 b Race 76% White , 3% native Hawaiian , 5% Other , 16% chose not to identify 100% White 95% White , 5% chose not to identify 47% White , 29% Hispanic or Latino , 12% Asian , 6% African American , 6% Other Vaginal Deliveries 1.6 ± 1.6 a 2.6 ± 1.0 b 2.5 ± 1.0 b 3.1 ± 1.5 b BMI 28.0 ± 6.6 a 26.0 ± 5.1 a 27.7 ± 4.5 a 28.9 ± 5.8 a Menopause Status 30 No , 7 Yes (19%) a 5 No , 22 Yes (81%) b 6 No , 13 Yes (68%) b 3 No , 14 Yes (82%) b Smoking status 30 No , 7 Yes (19%) a 23 No , 4 Yes (15%) a 15 No , 4 Yes (21%) a 12 No , 5 Yes (29%) a Estrogen Use 29 No , 6 Yes , 2 not reported (17%) a 14 No , 13 Yes (48%) b 9 No , 10 Yes (53%) b 10 No , 7 Yes (41%) a Progesterone use 21 No , 14 Yes , 2 not reported (40%) a 23 No , 4 Yes (15%) b 15 No , 4 Yes (21%) a 16 No , 1 Yes (6%) b Fibroid status 21 No , 16 Yes (43%) a 17 No , 10 Yes (37%) a 12 No , 7 Yes (37%) a 6 No , 9 Yes , 2 not reported (60%) a Mean ± standard deviation are included for the main demographic features for each subgroup of subjects. The N of each subgroup is indicated. Statistical comparisons of POP-HQ subgroups’ histopathological scoring data were performed using R software with required packages indicated where appropriate [ 43 ]. Welch’s unequal variances t-test was used when the phenotypes had unequal variances and were of unequal sample size. An analysis of variance with Tukey’s post-hoc pairwise test was used when making multiple comparisons between continuous variables and Pearson's Chi-squared test with Yates' continuity correction for categorical variables (R package ‘RVAideMemoire’) Mean values that significantly differ from one another by group are denoted by letters (a, b, c), P <0.05. “a” signifies not different from the first (control) value, “b” indicates different from the first value, “c” indicates different from the second (POP-A) value, “d” indicates different from the third (POP-I) value Summary of the Major Demographics; statistical comparisons as indicated in Data Analysis Mean ± standard deviation are included for the main demographic features for each subgroup of subjects. The N of each subgroup is indicated. Statistical comparisons of POP-HQ subgroups’ histopathological scoring data were performed using R software with required packages indicated where appropriate [ 43 ]. Welch’s unequal variances t-test was used when the phenotypes had unequal variances and were of unequal sample size. An analysis of variance with Tukey’s post-hoc pairwise test was used when making multiple comparisons between continuous variables and Pearson's Chi-squared test with Yates' continuity correction for categorical variables (R package ‘RVAideMemoire’) Mean values that significantly differ from one another by group are denoted by letters (a, b, c), P <0.05. “a” signifies not different from the first (control) value, “b” indicates different from the first value, “c” indicates different from the second (POP-A) value, “d” indicates different from the third (POP-I) value Table 2 summarizes the POP-HQ histopathological scoring for the Control and POP-HQ subgroups. The POP-A-subgroup had less smooth-muscle, more smooth-muscle fiber dropout, more adipose and less NIH, the POP-I-subgroup had more smooth-muscle fiber dropout and inflammation and less NIH, and the POP-V-subgroup had less inflammation and more neointimal hyperplasia, all compared to the Control group ( p < 0.05) and as observed previously [ 39 ]. Importantly, although POP-HQ scoring allows grouping together of samples according to histopathology, patients within each subgroup are still highly heterogeneous according to demographics. Table 2 Summary of the POP-HQ histopathological scoring; statistical comparisons as indicated in Data Analysis N Control POP-A(Adipose) POP-I (Inflammatory) POP-V (Vascular) 37 27 19 17 Histological Feature Mean ± SD Mean ± SD Mean ± SD Mean ± SD Fibrillar Collagen 57.9 ± 16.0 a 60.1 ± 17.4 a 51.9 ± 18.1 a 49.1 ± 22.9 a Total Non-Vascular Smooth Muscle 2.2 ± 0.7 a 1.2 ± 0.7 b 1.9 ± 0.7 a 1.9 ± 0.6 a SmFDO 0.8 ± 0.5 a 1.3 ± 0.5 b 1.3 ± 0.5 b 0.9 ± 0.6 a Muscle Fiber Vesicles 0.1 ± 0.1 a 0.0 ± 0.1 a 0.0 ± 0.1 a 0.2 ± 0.4 a Adipose 0.4 ± 0.8 a 1.2 ± 1.0 b 0.6 ± 0.6 a 0.3 ± 0.7 a PMN-Inflama 0.3 ± 0.4 a 0.4 ± 0.5 a 2.0 ± 0.7 b 0.1 ± 0.1 a PN-Inflam 0.1 ± 0.3 a 0.0 ± 0.1 a 1.0 ± 0.9 b 0.0 ± 0.0 a NIH 0.3 ± 0.7 a 0.0 ± 0.1 b 0.0 ± 0.0 b 2.2 ± 0.7 c Vessel Quantity 1.1 ± 0.8 a 0.7 ± 0.4 b 1.1 ± 0.6 a 1.9 ± 0.5 c Mean ± standard deviation are included for the POP-HQ histopathological scoring features for each subgroup of subjects [ 39 ]. The N of each subgroup is indicated. Statistical comparisons of POP-HQ subgroups’ histopathological scoring data were performed using R software with required packages indicated where appropriate [ 43 ]. Welch’s unequal variances t-test was used when the phenotypes had unequal variances and were of unequal sample size. An analysis of variance with Tukey’s post-hoc pairwise test was used when making multiple comparisons between continuous variables and Pearson's Chi-squared test with Yates' continuity correction for categorical variables (R package ‘RVAideMemoire’) Mean values that significantly differ from one another by group are denoted by letters (a, b, c), P <0.05. “a” signifies not different from the first (here control) value, “b” indicates different from the first value, “c” indicates different from the second (POP-A) value, “d” indicates different from the third (POP-I) value Summary of the POP-HQ histopathological scoring; statistical comparisons as indicated in Data Analysis Mean ± standard deviation are included for the POP-HQ histopathological scoring features for each subgroup of subjects [ 39 ]. The N of each subgroup is indicated. Statistical comparisons of POP-HQ subgroups’ histopathological scoring data were performed using R software with required packages indicated where appropriate [ 43 ]. Welch’s unequal variances t-test was used when the phenotypes had unequal variances and were of unequal sample size. An analysis of variance with Tukey’s post-hoc pairwise test was used when making multiple comparisons between continuous variables and Pearson's Chi-squared test with Yates' continuity correction for categorical variables (R package ‘RVAideMemoire’) Mean values that significantly differ from one another by group are denoted by letters (a, b, c), P <0.05. “a” signifies not different from the first (here control) value, “b” indicates different from the first value, “c” indicates different from the second (POP-A) value, “d” indicates different from the third (POP-I) value Mean ± standard deviation are included for the POP-HQ histopathological scoring features for each subgroup of subjects [39]. The N of each subgroup is indicated. Statistical comparisons of POP-HQ subgroups’ histopathological scoring data were performed using R software with required packages indicated where appropriate [43]. Welch’s unequal variances t-test was used when the phenotypes had unequal variances and were of unequal sample size. An analysis of variance with Tukey’s post-hoc pairwise test was used when making multiple comparisons between continuous variables and Pearson's Chi-squared test with Yates' continuity correction for categorical variables (R package ‘RVAideMemoire’). Mean values that significantly differ from one another by group are denoted by letters (a, b, c), P <0.05. “a” signifies not different from the first (here control) value, “b” indicates different from the first value, “c” indicates different from the second (POP-A) value, “d” indicates different from the third (POP-I) value. In Fig. 1 (all histology figures and inserts are reproduced at a larger size in the supplemental figures) we provide representative lower magnification images of consecutive USL serial sections where elastin fibers (Hart’s elastin-stained USL tissue, panels b, d, f, h) appear dark purple to black and fibrillar collagen bundles appear blue (Masson’s trichrome stained panels a, c, e, g). Comparison of panels a and b demonstrates that elastin fibers with elongated, “wavy” appearances (b) are found at high density in the region of collagen bundles (matching areas in serial sections indicated by arrows). We have shown previously that smooth muscle fascicles in USL specimens from patients with POP can be characterized by smooth muscle cell loss, or “dropout” [ 39 ]. Panels c and d include intact smooth muscle fascicles with minimal dropout (fascicles are bounded by black dashed lines in panels c-f, muscle fibers appear red in panels a, c, e, g). In panels c and d, elastin fibers are tortuous and in higher abundance in the area around/outside smooth-muscle fascicles but are sparse or absent inside fascicles. In contrast, representative panels e and f show that smooth-muscle fascicles with greater smooth-muscle fiber dropout possess comparatively elevated intra-fascicle elastin fibers (yellow arrows, f), and that those fibers are both thinner and do not display the elongated pattern seen in elastin outside of the fascicles as in panel d (see also Fig. 2 , panels a-f). Panels g-h demonstrate that elastin fibers are decreased in number and length near adipocyte clusters (indicated by “a”). Indeed, the presence of adipocytes tended to disrupt the cohesive alignment of elastin fibers in the surrounding tissue. Fig. 1 Masson-Trichrome and Hart’s Stain Correlation Four-micron sections of USL tissue were stained with the Masson-Trichrome stain ( a, c, e, g ) or Hart’s stain for elastin ( b, d, f, h ). In panels a and b, images from serial sections of USL tissue composed predominantly of collagen are shown; matching areas are indicated by arrows. In panels c and d, images from serial sections of USL tissue with almost intact smooth muscle fascicles are shown, dashed lines have been drawn around some of these. In panels e and f, images from serial sections of USL tissue with smooth muscle fascicles possessing lots of muscle cell dropout are shown to illustrate the increased elastin within these fascicles. Dashed lines surround muscle fascicles; matched yellow arrows point to areas of fiber dropout. In panels g and h, images from serial sections of USL tissue with an adipocyte (a) infiltrate is shown to illustrate that elastin fibers are disrupted by the adipose tissue. In all panels, the size bar equals 100 microns Fig. 2 Pattern of Hart’s-stained USL Elastin fibers. In panels a-f four-micron sections of USL tissue were stained using Hart’s staining method. Shown are the various patterns of elastin fibers that were observed. Panel a , relatively straight elastin fibers in connective tissue with a small vessel (v). Panel b , tortuous elastin fibers in USL tissue with lots of small fascicles of smooth muscle, longitudinal cut. Panel c , elastin fibers in USL tissue with large no-dropout smooth muscle fascicles (m) in longitudinal section. Panel d , elastin fibers surrounding multiple large no-dropout USL smooth muscle fascicles (m) in cross-section. Panel e , elastin fibers in USL tissue with a high density of collagen and little muscle. Panel f , elastin fibers in USL tissue with adipocytes present (a). In panels a-e, the size bar equals 50 microns. Panels g-j demonstrate serial sections from USL arteries. Panels g and h from a normal artery, or I and J from an artery with neointima hyperplasia were stained with Masson-trichrome stain (g and c) or Hart’s stain (h and j). The layers of the arteries are identified, tunica-intima, tunica-media, and tunica adventitia. The tunica-media of arteries with NIH possess increased quantities of elastin fibers and the tunica-intima of those same arteries is greatly increased in thickness and seen here to have a high elastin content. In panels g - j , the size bars equal 100 microns Masson-Trichrome and Hart’s Stain Correlation Four-micron sections of USL tissue were stained with the Masson-Trichrome stain ( a, c, e, g ) or Hart’s stain for elastin ( b, d, f, h ). In panels a and b, images from serial sections of USL tissue composed predominantly of collagen are shown; matching areas are indicated by arrows. In panels c and d, images from serial sections of USL tissue with almost intact smooth muscle fascicles are shown, dashed lines have been drawn around some of these. In panels e and f, images from serial sections of USL tissue with smooth muscle fascicles possessing lots of muscle cell dropout are shown to illustrate the increased elastin within these fascicles. Dashed lines surround muscle fascicles; matched yellow arrows point to areas of fiber dropout. In panels g and h, images from serial sections of USL tissue with an adipocyte (a) infiltrate is shown to illustrate that elastin fibers are disrupted by the adipose tissue. In all panels, the size bar equals 100 microns Pattern of Hart’s-stained USL Elastin fibers. In panels a-f four-micron sections of USL tissue were stained using Hart’s staining method. Shown are the various patterns of elastin fibers that were observed. Panel a , relatively straight elastin fibers in connective tissue with a small vessel (v). Panel b , tortuous elastin fibers in USL tissue with lots of small fascicles of smooth muscle, longitudinal cut. Panel c , elastin fibers in USL tissue with large no-dropout smooth muscle fascicles (m) in longitudinal section. Panel d , elastin fibers surrounding multiple large no-dropout USL smooth muscle fascicles (m) in cross-section. Panel e , elastin fibers in USL tissue with a high density of collagen and little muscle. Panel f , elastin fibers in USL tissue with adipocytes present (a). In panels a-e, the size bar equals 50 microns. Panels g-j demonstrate serial sections from USL arteries. Panels g and h from a normal artery, or I and J from an artery with neointima hyperplasia were stained with Masson-trichrome stain (g and c) or Hart’s stain (h and j). The layers of the arteries are identified, tunica-intima, tunica-media, and tunica adventitia. The tunica-media of arteries with NIH possess increased quantities of elastin fibers and the tunica-intima of those same arteries is greatly increased in thickness and seen here to have a high elastin content. In panels g - j , the size bars equal 100 microns Higher magnification images from Hart’s-stained USLs are shown in Fig. 2 so that representative and reproducible elastin fiber patterns can be appreciated. Hart’s-stained elastin fibers are easily identified and occur in spread out, relatively straight (panel a) or tortuous patterns (panels b-d), where fibers are found in contiguous patterns extending to hundreds of microns (panels b and e) or are interrupted into fibers of shorter size (panels a, c, and f). It was also notable that clustered, wavy patterns of elastin fibers as seen in panels b and e were absent from USL regions that contained adipose cells (panel f). Near adipose cells (labeled “a” in panel f) straighter and shorter elastin fibers appeared at lower density. Therefore, consistent with images in Fig. 1 panels c-f, USL areas devoid of elastin fibers can be identified as smooth-muscle fascicles lacking fiber dropout (m, panels c and d), adipocytes (a, panel f), or a large vein (v, panel a). In POP-V USLs characterized by vascular neointimal hyperplasia, elastin staining within large arteries differed vastly from the normal pattern seen in Control tissues arteries. For context, NIH was detected in arteries of 4 of the 37 Control USLs, none of the POP-I or POP-A USLs, and all the POP-V USLs. Panels g and h are images from serial sections of a representative Control USL artery lacking NIH (arterial anatomical layers are identified; panels g and h are images from serial sections; panel g, Masson-trichrome-stained for collagen; panel h, Hart’s-stained for elastin). Such “normal” arteries contain lesser amounts of elastin in their tunica-media and tunica-adventitia, and the thick but singular internal elastic lamina-band of the tunica-intima (panel h). In contrast, USL arteries with NIH from POP-V specimens (arterial anatomical layers are identified; panels i and j are images from serial sections; panel i, Masson-trichrome-stained and panel j, Hart’s-stained) contained more elastin in the tunica-media and tunica-adventitia. Elastin was expanded into additional bands in the tunica intima (black, panel j, compare with the singular elastic lamina band in panel h). Similar to the POP-HQ USL scoring system [ 39 ], we devised a strategy to quantify and categorize the morphology of Hart’s-stained elastin fibers in the USL tissue sections (details provided in Supplemental Table 1). Measured elastin fiber length, from measurements on 3–5 images of Hart’s-stained tissue, was decreased in POP-I USLs compared to Controls ( p < 0.05) but did not differ in the POP-A and POP-V subgroups (Table 3 ). Total elastin content, identified by computer-assisted quantification of elastin IHC in 3–5 images (3I, SlideBook, see methods for details), was significantly increased in the POP-I and POP-V USL subgroups compared to Control ( p < 0.05, Table 3 ). Semi-quantitative assessments of fiber features followed. Elastin fiber diameter (Fib-D, 0–2 scale), tortuosity (Tort, 0–2 scale), and clumping (Clmp, 0–2 scale) did not differ between the Control group and the POP-HQ subgroups. However, consistent with the appearance of the artery in histological preparations (Fig. 2 , panel j), the semi-quantitative assessments of elastin quantity in the tunica-intima (Ti-elas, only a 0–1 scale) and tunica-media (Tm-elas, 0–2 scale) were significantly increased in the POP-V subgroup compared to all other subgroups ( p  < 0.05). The relative quantity of elastin found directly around the smooth-muscle fascicles (Peri-sm, 0–2 scale) did not vary between POP-HQ subgroups, but the elastin found within the fascicles (Intra-sm, 0–2 scale) non-significantly trended higher in the POP-A subgroup (Table 3 ). Briefly, a 0–1 scale was used when the differences were profound and a 0–2 scale when they were more subtle. Table 3 Summary of USL elastin fiber morphology and quantification data; Statistical comparisons as indicated in Data Analysis Criteria Measurement CONTROL POP-A(ADIPOSE) POP-I(INFLAMMATORY) POP-V(VASCULAR) N=37 N=27 N=19 N=17 Mean ± SD Mean ± SD Mean ± SD Mean ± SD Measured elastin fiber length From images, In microns 15.6 ± 2.7 a 17.1 ± 4.4 a 14.0 ± 1.0 b 14.6 ± 2.0 ab Elastin–IHC (quantity) Percentage of pixels staining positively 4.1 ± 2.7 a 4.8 ± 2.6 a 7.8 ± 3.7 b 7.1 ± 5.1 b Relative diameter Arbitrary units, 0=thin, 2=thick 1.0 ± 0.1 a 1.0 ± 0.2 a 1.0 ± 0.1 a 1.0 ± 0.1 a Relative tortuosity Arbitrary units, Straight=0, bent=1, curly=2 1.1 ± 0.5 a 0.9 ± 0.4 a 1.1 ± 0.3 a 1.3 ± 0.5 ab Relative clumping Arbitrary units, 0=none, 2=lots 0.8 ± 0.3 a 0.7 ± 0.3 a 0.8 ± 0.3 a 0.9 ± 0.6 a Relative quantity of elastin in tunica intima Arbitrary units, 0=none, 1=lots 0.3 ± 0.4 a 0.0 ± 0.1 b 0.0 ± 0.1 b 1.0 ± 0.1 bc Relative quantity of elastin in tunica media Arbitrary units, 0=none, 2=lots 0.5 ± 0.5 a 0.6 ± 0.3 a 0.6 ± 0.3 a 1.0 ± 0.4 b Relative quantity of elastin around-sm musc fascicles Arbitrary units, 0=none, 2=lots 0.8 ± 0.4 a 0.9 ± 0.5 a 1.1 ± 0.6 b 0.9 ± 0.5 a Relative quantity of elastin inside sm musc fascicles Arbitrary units, 0=none, 2=lots 0.6 ± 0.3 a 0.7 ± 0.4 a 0.6 ± 0.4 a 0.5 ± 0.3 a Each criterion for elastin fiber morphology and how it was scored is indicated. Mean ± standard deviation are included for the results of each criterion for each subgroup of subjects. The N of each subgroup is indicated. Statistical comparisons of data for all subgroups of subjects were performed using R software with required packages indicated where appropriate [ 43 ]. Welch’s unequal variances t-test was used when the phenotypes had unequal variances and were of unequal sample size. An analysis of variance with Tukey’s post-hoc pairwise test was used when making multiple comparisons between continuous variables and Pearson's Chi-squared test with Yates' continuity correction for categorical variables (R package ‘RVAideMemoire’) Mean values that significantly differ from one another by group are denoted by letters (a, b, c), P <0.05. “a” signifies not different from the first (control) value, “b” indicates different from the first value, “c” indicates different from the second (POP-A) value, “d” indicates different from the third (POP-I) value Summary of USL elastin fiber morphology and quantification data; Statistical comparisons as indicated in Data Analysis Each criterion for elastin fiber morphology and how it was scored is indicated. Mean ± standard deviation are included for the results of each criterion for each subgroup of subjects. The N of each subgroup is indicated. Statistical comparisons of data for all subgroups of subjects were performed using R software with required packages indicated where appropriate [ 43 ]. Welch’s unequal variances t-test was used when the phenotypes had unequal variances and were of unequal sample size. An analysis of variance with Tukey’s post-hoc pairwise test was used when making multiple comparisons between continuous variables and Pearson's Chi-squared test with Yates' continuity correction for categorical variables (R package ‘RVAideMemoire’) Mean values that significantly differ from one another by group are denoted by letters (a, b, c), P <0.05. “a” signifies not different from the first (control) value, “b” indicates different from the first value, “c” indicates different from the second (POP-A) value, “d” indicates different from the third (POP-I) value Each criterion for elastin fiber morphology and how it was scored is indicated. Mean ± standard deviation are included for the results of each criterion for each subgroup of subjects. The N of each subgroup is indicated. Statistical comparisons of data for all subgroups of subjects were performed using R software with required packages indicated where appropriate [43]. Welch’s unequal variances t-test was used when the phenotypes had unequal variances and were of unequal sample size. An analysis of variance with Tukey’s post-hoc pairwise test was used when making multiple comparisons between continuous variables and Pearson's Chi-squared test with Yates' continuity correction for categorical variables (R package ‘RVAideMemoire’). Mean values that significantly differ from one another by group are denoted by letters (a, b, c), P<0.05. “a” signifies not different from the first (control) value, “b” indicates different from the first value, “c” indicates different from the second (POP-A) value, “d” indicates different from the third (POP-I) value. We next performed exploratory covariate analysis to examine whether correlations might exist between elastin and collagen fiber data and patient demographic or other USL parameters (Fig. 3 ) in this exploratory study. For each graphical representation in Fig. 3 , R-squared, p-value, significance (yes/no), and for significant trendlines, an indication of whether the slope of the SciPy, sklearn [ 43 ] generated trendline increases or decreases is given (Table 4 ). Visual inspection of data was performed to check model assumptions for regression (see Methods). First, relationships were examined between elastin content and patient demographic and risk factor measures (age, BMI, and number of vaginal deliveries), anatomical measurement (Point C[ 40 ],), and USL inflammatory cell content. For each graphical representation in Fig. 3 , R-squared, p-value, significance (yes/no), and for significant trendlines, an indication of whether the slope of the SciPy, sklearn [ 43 ] generated trendline increases or decreases is given (Table 4 ). Table 4 Summary of the Vimentin-IHC Quantification CONTROL POP-A (ADIPOSE) POP-I (INFLAMMATORY) POP-V (VASCULAR) N 35* 27 17* 16* Histological Feature Mean ± SD Mean ± SD Mean ± SD Mean ± SD Vimentin-IHC 4.77 ± 2.28 a 4.83 ± 1.74 a 6.82 ± 2.12 b 5.10 ± 1.75 a Mean ± standard deviation for each subgroup are included for the quantification of the vimentin-IHC by the program Slidebook. The N of each subgroup is indicated. Statistical comparisons of the data were performed using R software with required packages indicated where appropriate [ 43 ]. Welch’s unequal variances t-test was used when the phenotypes had unequal variances and were of unequal sample size. An analysis of variance with Tukey’s post-hoc pairwise test was used when making multiple comparisons between continuous variables and Pearson's Chi-squared test with Yates' continuity correction for categorical variables (R package ‘RVAideMemoire’) Mean values that significantly differ from one another by group are denoted by letters (a, b, c), P <0.05. “*” Indicates there are 2 less Controls and POP-I, and 1 less POP-V value included here because we ran out of tissue for those samples. The exact samples missing are indicated on the Figshare site Fig. 3 Covariate Correlations Data from the USL tissue examined is graphed to show the relative relationships between the subgroups. R-squared values and p-values for each trendline in each panel are found in Table 4 . Panel a , the Age of the patients was graphed versus the elastin content of the USLs. Panel b , the BMI of the patients was graphed versus the elastin content of the USLs. Panel c , the Number of Vaginal Births of the subjects was graphed versus the elastin content of the USLs. Panel d , the Point C values of the subjects’ USLs were graphed versus the elastin content of the USLs. Panel e , the Point C values of the subjects’ USLs were graphed versus the collagen content (from the POP-HQ scoring) of the USLs. Panel f , the subjects’ USL collagen content was graphed versus the USL elastin content. Panel g , the density of T-lymphocytes in the USLs was graphed versus the elastin content of the USL. Panel h , the density of mast cells in the USLs was graphed versus the elastin content of the USL. Graphing of data was performed with the program Python [ 44 ], and the linear regression package sklearn (in SciPy [ 44 ]) was used for R-squared Summary of the Vimentin-IHC Quantification Mean ± standard deviation for each subgroup are included for the quantification of the vimentin-IHC by the program Slidebook. The N of each subgroup is indicated. Statistical comparisons of the data were performed using R software with required packages indicated where appropriate [ 43 ]. Welch’s unequal variances t-test was used when the phenotypes had unequal variances and were of unequal sample size. An analysis of variance with Tukey’s post-hoc pairwise test was used when making multiple comparisons between continuous variables and Pearson's Chi-squared test with Yates' continuity correction for categorical variables (R package ‘RVAideMemoire’) Mean values that significantly differ from one another by group are denoted by letters (a, b, c), P <0.05. “*” Indicates there are 2 less Controls and POP-I, and 1 less POP-V value included here because we ran out of tissue for those samples. The exact samples missing are indicated on the Figshare site Covariate Correlations Data from the USL tissue examined is graphed to show the relative relationships between the subgroups. R-squared values and p-values for each trendline in each panel are found in Table 4 . Panel a , the Age of the patients was graphed versus the elastin content of the USLs. Panel b , the BMI of the patients was graphed versus the elastin content of the USLs. Panel c , the Number of Vaginal Births of the subjects was graphed versus the elastin content of the USLs. Panel d , the Point C values of the subjects’ USLs were graphed versus the elastin content of the USLs. Panel e , the Point C values of the subjects’ USLs were graphed versus the collagen content (from the POP-HQ scoring) of the USLs. Panel f , the subjects’ USL collagen content was graphed versus the USL elastin content. Panel g , the density of T-lymphocytes in the USLs was graphed versus the elastin content of the USL. Panel h , the density of mast cells in the USLs was graphed versus the elastin content of the USL. Graphing of data was performed with the program Python [ 44 ], and the linear regression package sklearn (in SciPy [ 44 ]) was used for R-squared In Panels a, b, and c of Fig. 3 , we examined the effect of known prolapse risk factors upon the quantity of elastin. In panel a, none of the trendlines reach significance however the Control, POP-A and POP-I subgroups trended towards possessing quantities of elastin invariant of the subject’s age and the POP-V trendline increases with ageing (Table 4 ). In panel b, none of the trendlines reach significance however all four groups trended towards constant quantities of elastin invariant of BMI. POP-I and POP-V exhibited increased elastin compared to Control and POP-A as previously indicated in Table 3 . In panel c, the Control trendline reached significance (R-squared = 0.2931, p  < 0.0005, Table 4 ) and indicated increased elastin quantity with increased vaginal parity, but none of the POP-HQ trendlines reached significance. The POP-A trendline trended towards a non-significant increase in elastin content with increasing parity similar to the control trendline, whereas the POP-I and POP-V trendlines trended towards a constant quantity of elastin as their vaginal parity increased. In panel d, USL elastin content is displayed relative to the USL POP-Q stage; these are indicative of the amount of uterine descent, and indirectly, stretch and tension on the USL [ 46 ]. Because multiple morphological criteria contribute to only a few stages of prolapse extent, use of the POP-Q stage was discontinued. Instead, the value for Point C was used as a marker of descent (panel e). Trendlines for the Control, and POP-V subgroup data reached significance, and both displayed a negative slope indicating a decreased content of elastin in those uterus-vaginal complexes that had descended further. The trendlines for the POP-A and POP-I subgroups did not reach significance however the present data appear to support a POP-A trendline that differs from the trendlines for the control and POP-V subgroups. Due to the unexpected POP-A findings in panel e, we examined whether collagen content varied in USLs that had descended various amounts, Fig. 3 , panel f. With progressive uterine descent, none of the trendlines for the Control, POP-A, POP-I, or POP-V quantities of collagen reached significance although the POP-A group was close to significance ( p  = 0.08) and had a positive slope indicating increasing collagen content in those uterus-vagina complexes that had descended further. In Fig. 3 , panel g, the POP-A quantity of elastin increased significantly with increasing collagen whereas the changes in elastin content were not significant for the control, POP-I or POP-V subgroups. We then examined whether key immune cells affected the elastin content of the USL. In Fig. 3 , panel h, the elastin content of USLs was plotted against their density of T-lymphocytes and in panel i the elastin content was plotted against their density of mast cells. This data relied upon T-lymphocyte and mast cell quantification previously published [ 47 ]. None of the trendlines in panels h and i reached significance (Table 4 ), nor were any of the p-values less than 0.10. Clearly, many more samples would help. In summary, Fig. 3 indicates that there is still tremendous heterogeneity within the POP-HQ subgroups regarding elastin and collagen fiber content even after controlling for histopathology. The data here support the hypothesis that the connective tissue fibers differ between the Control and POP-HQ subgroup USLs. Beyond that, these data only suggest trends such as the POP-A subgroup’s connective tissue fiber content differs from that of the POP-I and POP-V subgroups. The results from Table 4 and Fig. 3 are shown in summary form in Fig. 4 wherein significant results are indicated by a large red arrow in the orientation of the change and the small blue arrows indicate a trend ( p  < 0.15) in the orientation indicated. Fig. 4 Summary of Table 4 and Fig 3 data. A summary of the data is shown in simple Fig form. “C” means did not change, a large red arrow indicates significant (p<0.05) change in the orientation indicated by the arrow, and a small blue arrow indicates a trend (p<0.15) in the orientation indicated Summary of Table 4 and Fig 3 data. A summary of the data is shown in simple Fig form. “C” means did not change, a large red arrow indicates significant (p<0.05) change in the orientation indicated by the arrow, and a small blue arrow indicates a trend (p<0.15) in the orientation indicated In some prolapsed women, elastin was found in USL sites where it was not observed in younger individuals. These sites included the arterial tunica-intima of POP-V USL tissue and within smooth-muscle fascicles undergoing loss of muscle fibers (Figs. 1 and 2 ). As a correlative check, we examined whether the cell type that synthesizes elastin during injury repair was present and in the correct histologic locale to afford these increases. Elastin is normally made by smooth-muscle cells, fibroblasts, and endothelial cells in non-injured tissue [ 26 ], and by myofibroblasts during injury repair [ 48 ]. Vimentin is a convenient immunohistochemical marker of fibroblasts/myofibroblasts, but this assessment requires careful histological co-confirmation since it is also found in endothelial cells and inflammatory cells. Figure 5 shows images of vimentin-IHC-stained USL arteries without (a) and with (b) neointima hyperplasia. In panel b vimentin expression was observed in the newly formed POP-V tunica-intima within cells whose phenotype was identical to that of fibroblasts/myofibroblasts (inset, from red box); also, these vimentin positive cells were not at a surface suggesting they were not epithelial and hence not endothelial cells and they did not have the morphology of inflammatory cells. This tunica intima corresponds to the site of abundant new elastin noted in Fig. 2 , panel j. In Fig. 5 , panels c and d, a USL smooth-muscle fascicle in longitudinal orientation is presented in serial cut sections following Masson-trichrome staining (c) or vimentin-IHC (d). In panel c the smooth-muscle fascicle (red) was relatively intact with a minor loss of muscle cells on the right-hand side of the image. In panel d, the vimentin-IHC stained serial section showed that there is only very low expression of vimentin found inside the smooth-muscle cells present (inset, from red box). In contrast, panels e and f demonstrated serial cut sections of USL smooth-muscle fascicles with muscle cell loss. In panel e two smooth muscle fascicles are circled, the upper one has abundant muscle cell dropout. Following anti-vimentin IHC-staining (panel f), an increase in the intra-fascicle vimentin-IHC staining was observed that has the fibroblast/myofibroblast phenotype (inset, from red box). Thus, panels b and f demonstrate the presence of elastin synthesizing cells with the fibroblast/myofibroblast phenotype at sites where only very low levels of elastin was found until neointimal hyperplasia (panel b) or smooth muscle cell loss (panel f) had occurred, thus both appeared to be attempts at increasing connective tissue fibers in damaged tissue, i.e. repair. Fig. 5 Vimentin-IHC highlights the presence of fibroblasts/myofibroblasts in sites of new elastin. Panels a (Control) and b (neointima hyperplasia); Images of arteries from USL tissue show the elongated form of vimentin-positive cells in the hyperplastic tunica-intima. The inset image in panel b depicts myofibroblasts in the neointimal tissue taken from the area of the dashed red box. Panels c (Masson-trichrome-stained) and d (Hart’s-stained) images of serial sections from a USL showing a smooth muscle fascicle with little loss of muscle fibers. The inset image in panel d depicts the normal low-level vimentin immunohistochemical staining pattern of smooth muscle fibers. Only small quantities of vimentin-positive staining are observed within the fascicle. Panels e (trichrome-stained) and f (Hart’s-stained) images from serial sections of USL smooth-muscle fascicles that possess increased smooth-muscle fiber dropout. The inset image in panel f depicts a couple of myofibroblasts from inside the smooth muscle fascicle taken from the area of the dashed red box. Larger quantities of vimentin staining are observed within the muscle fascicle following dropout. In panels a-f , the size bar equals 100 microns Vimentin-IHC highlights the presence of fibroblasts/myofibroblasts in sites of new elastin. Panels a (Control) and b (neointima hyperplasia); Images of arteries from USL tissue show the elongated form of vimentin-positive cells in the hyperplastic tunica-intima. The inset image in panel b depicts myofibroblasts in the neointimal tissue taken from the area of the dashed red box. Panels c (Masson-trichrome-stained) and d (Hart’s-stained) images of serial sections from a USL showing a smooth muscle fascicle with little loss of muscle fibers. The inset image in panel d depicts the normal low-level vimentin immunohistochemical staining pattern of smooth muscle fibers. Only small quantities of vimentin-positive staining are observed within the fascicle. Panels e (trichrome-stained) and f (Hart’s-stained) images from serial sections of USL smooth-muscle fascicles that possess increased smooth-muscle fiber dropout. The inset image in panel f depicts a couple of myofibroblasts from inside the smooth muscle fascicle taken from the area of the dashed red box. Larger quantities of vimentin staining are observed within the muscle fascicle following dropout. In panels a-f , the size bar equals 100 microns Vimentin-IHC staining was also quantified across each whole USL section (Table 5 ). Only the POP-I subgroup differed significantly (increased, p < 0.05) from the Control group. Table 5 Summary of the Vimentin-IHC Quantification Panel Control ( n  = 37) Control Control POP-A ( n  = 27) POP-A POP-A POP-I ( n =19) POP-I POP-I POP-V ( n =17) POP-V POP-V R -Squared P -value Signif & trend R -Squared P-value Signif & trend R-Squared P-value Signif & trend R-Squared P-value Signif & trend Age v Elastin 0.0004 0.9141 N 0.0350 0.3403 N 0.0005 0.9300 N 0.0590 0.3647 N BMI v Elastin 0.0015 0.8188 N 0.0260 0.4119 N 0.0056 0.7601 N 0.0033 0.8319 N Vag births v Elastin 0.2931 0.0005 Y , ↑ 0.0609 0.2054 N 0.0013 0.8844 N 0.0038 0.8207 N Point C v Elastin 0.2032 0.0403 Y , ↓ 0.0186 0.4978 N 0.0428 0.3957 N 0.7978 0.0000 Y, ↓ Point C v Collagen 0.0400 0.3868 N 0.1176 0.0800 N 0.1552 0.0951 N 0.0069 0.7873 N Collagen v Elastin 0.0442 0.2118 N 0.1628 0.0333 Y, ↑ 0.0615 0.3060 N 0.0000 0.9962 N T-cells v Elastin 0.1106 0.1520 N 0.0050 0.7887 N 0.1364 0.1196 N 0.0229 0.5902 N Mast cells v Elastin 0.0838 0.2157 N 0.1210 0.1712 N 0.0132 0.6390 N 0.0778 0.3141 N N No, Y Yes ↑, increasing; ↓, decreasing R-squared values, p-values whether the trendline is significant, and the direction of its trend for each of the 4 trendlines found in each panel in Fig 4 are presented. Graphing of data was performed with the program Python [ 43 ], and the linear regression package sklearn (in SciPy [ 44 ]) was used for R-squared and p -values Graphing of data was performed with the program Python [44], and the linear regression package sklearn (in SciPy [ 44 ]) was used for R-squared and p-value and slope assessment. Summary of the Vimentin-IHC Quantification N No, Y Yes ↑, increasing; ↓, decreasing R-squared values, p-values whether the trendline is significant, and the direction of its trend for each of the 4 trendlines found in each panel in Fig 4 are presented. Graphing of data was performed with the program Python [ 43 ], and the linear regression package sklearn (in SciPy [ 44 ]) was used for R-squared and p -values Graphing of data was performed with the program Python [44], and the linear regression package sklearn (in SciPy [ 44 ]) was used for R-squared and p-value and slope assessment. Because the elastin quantity was increased in the POP-I and -V subgroups, and the POP-HQ-subgroups trended towards different responses to increases in Point C, we tested whether addition of either of these two metrics (elastin content and Point C value) to the POP-HQ histopathologic criteria would yield patterns with better separation of the POP-HQ-subgroups following principal component analysis (PCA). Briefly, this involved the addition of the data of either the elastin IHC, or the data for Point C to those data already used during the PCA to group the histopathology results from the USLs of prolapsed women [ 39 ]. Supplemental Fig. 4 a-c (respectively) illustrates the patterns produced in PC1 versus PC2 two-dimensional-plots using either the POP-HQ criteria alone, or when either elastin-IHC or the Point C value was added to the criteria already employed to separate prolapsed USLs using only histopathologic criteria. The addition of either of those two criteria had only very minor effects upon the distribution of subject data points in the PC1 versus PC2 plots.

Pelvic floor disorders (PFDs) affect approximately one-quarter of the women in the United States, and this prevalence is anticipated to rise as the population ages [ 1 , 2 ]. Among these disorders, pelvic organ prolapse (POP) is characterized by the descent of pelvic organs into or beyond the vaginal canal, typically due to failure of supportive connective tissues such as uterosacral ligaments (USLs) [ 3 – 5 ]. Pelvic floor supporting tissues include the suspensory USL and cardinal ligaments, the paravaginal attachments to the levator ani muscles, and the skeletal pelvic floor muscles. Failure of pelvic floor supporting tissues only occurs in a subset of individuals who experience similar tissue damage as those who do not fail. For example, the insult of uncomplicated vaginal delivery results in essentially the same mechanical deformation and damage to pelvic support structures across women. In most women, significant regeneration of pelvic floor support occurs after delivery and pelvic organs return to their approximate preconception positions [ 6 ]. However, with aging and repetitive mechanical strain, repair mechanisms may become inadequate, leading to POP. Understanding how risk factors and connective tissue changes contribute to prolapse is essential to improving treatment strategies. The primary risk factors for POP include advancing age, vaginal deliveries, and elevated body mass index (BMI) [ 7 , 8 ]. Specific clinical associations point towards failure of connective tissue extracellular matrix (ECM) in the prolapse etiology [ 9 , 10 ]. For instance, the incidence of POP is increased in individuals with Marfan syndrome [ 11 ], a genetic defect in a major structural element of elastin microfibrils and in women with cutis laxa, a group of disorders displaying decreased elastin or generalized elastosis [ 12 – 14 ]. Similarly, the incidence of POP is increased in women with certain types of Ehlers-Danlos Syndrome [ 11 , 15 ], a hereditary connective tissue disorder with defects in processing of, or the structure of, collagen. Both quantitative and qualitative deficiencies in ECM molecules have been proposed as causative in women with POP [ 16 , 17 ]. There is also a weaker, but positive, association with other connective tissue diseases like systemic lupus erythematosus [ 18 ], rheumatoid arthritis [ 19 ], scleroderma [ 20 ], and Sjogren’s syndrome [ 21 ]. These connections have led to deeper investigation into the role of elastin and collagen in pelvic support. The ECM protein elastin (encoded by the human ELN gene, Ensembl ID# ENSG00000049540) assembles into fibers that have a remarkably long half-life, estimated at nearly 70 years in humans, and is synthesized primarily during fetal life and early adulthood [ 22 – 25 ] after which elastogenesis declines [ 26 , 27 ]. One exception is the uterus in which elastogenesis commences with each new pregnancy followed by removal of elastin following parturition [ 28 , 29 ]. Myofibroblasts synthesize much of the elastin and collagen fibers and other ECM components during wound repair and remodeling throughout the body [ 30 ], therefore, they may play a role in the response to pelvic organ support tissue injury. Animal models have demonstrated the importance of elastin biosynthesis and crosslinking enzymes in pelvic support. Lysyl-oxidase-like-1 (human LOXL1) is one such elastin crosslinking enzyme, and mice lacking LOXL1 develop postpartum prolapse and show partial remodeling within weeks, but with worsening support as they age and with a higher incidence of urinary incontinence [ 31 , 32 ]. Importantly, biomechanical studies identified weaker connective tissue in the prolapsing LOXL1-KO animal [ 33 ]. Similarly, when the gene for the elastin-assembly protein fibulin-5 was knocked out in mice, decreased elastin fiber quantity and POP ensued [ 34 ]. Elastin synthesis and assembly and LOX expression are observed to decrease as mice age and have been postulated to contribute to POP [ 24 , 32 ]. In wild-type mice, pregnancy and delivery significantly reduced the elastic modulus and increased relaxation in young adult USLs; these USL properties recovered to nulliparous levels following pregnancy and delivery [ 35 ]. In sum, these models reveal that pelvic floor support depends upon proper elastin and ECM function, and that lasting mechanical and structural changes in the USL can persist after pregnancy and worsen with age. In contrast to elastin, the ECM protein collagen is synthesized and remodeled throughout an individual’s lifetime although its turnover decreases with aging [ 36 ]. Collagen is essential during the inflammatory, fibroblast activity, and tissue maturation phases of wound healing [ 37 , 38 ]. Despite robust animal studies especially in murine vaginal and pelvic floor tissues [ 31 – 35 ], how elastin and collagen turnover contributes to POP progression in human tissues remains unclear. Our group has previously shown that USLs from POP patients can be classified into three histopathologic subgroups using the Pelvic Organ Prolapse-Histologic Quantification (POP-HQ) system: adipose-rich (POP-A), inflammatory (POP-I), and vascular/neointimal hyperplasia (POP-V) [ 39 ]. POP-A tissues are characterized by increased adipose content along with loss of smooth muscle and replacement by trichrome-positive connective tissue. POP-I samples demonstrate an influx of inflammatory cells [ 39 ], while POP-V tissues exhibit vascular remodeling characterized by neointimal hyperplasia (NIH). We hypothesized that these structural changes result from different etiologies of connective tissue failure and repair. In the present study, we examined elastin and collagen content, distribution, and histomorphology changes in USLs from 63 patients undergoing surgery for POP and 37 Control patients treated for benign gynecologic conditions without prolapse, to better understand the elastin and collagen fiber changes associated with the three POP-HQ phenotypes.

General Considerations. The uterosacral-cardinal complex, including the USL, is responsible for the apical support of the vagina and pelvic organs, and loss of apical support is associated with advanced prolapse. The USL is composed of multiple types of connective tissues including smooth muscle, dense fibrous connective tissue, vasculature, adipose tissue, and neural tissue. The specific cells and fibers of each of these types of connective tissue play a role in the overall structure and physiology. In many organs throughout the body, altered fibers have been found to contribute to disease [ 49 ] and not surprisingly, alterations in fiber quantity/quality have been hypothesized to lead to POP. Reports that patients with Marfan’s syndrome (MS, elastin) and Ehlers-Danlos syndrome (EDS, collagen) have an increased incidence of POP are consistent with this hypothesis but also raise the question of whether altered fiber quantity/quality contribute to prolapse in non-MS and non-EDS individuals. The correlation between structural fibers and the strength of the human uterus-vagina support system has yet to be fully defined although animal experiments to address this issue have been initiated [ 33 , 35 ]. In wild-type mice, pregnancy and delivery were observed to significantly reduce the elastic modulus and to increase the relaxation in young adult USLs. Importantly, and germane to our work here, these USL properties recovered to nulliparous levels following pregnancy and delivery [ 35 ]. Further, these models revealed that, in rodents, pelvic floor support depends upon proper elastin and ECM function, and that lasting mechanical and structural changes in the USL can persist after pregnancy and worsen with age. Similar research is needed utilizing human tissues. Recent studies aimed at identifying optimal surgical methods for repair of pelvic prolapse have been undertaken [ 50 ], those studies might be enhanced by knowledge of the patient’s fiber quantity/quality and overall histopathologic assessment. We hypothesize that the patient’s operative success and post-operative healing may be dependent upon the structural integrity that we are describing herein. Our previous histopathologic analysis divided prolapsed USL tissues into POP-A (Adipose), POP-I (Inflammation), and POP-V (Vascular) subgroups that contained, respectively, increased adipose tissue, increased inflammation, and neointima hyperplasia, all of which are connective tissue changes. Even though we have subgrouped the patients by histopathology, these subgroups still contain considerable heterogeneity [ 39 , 47 ]. Here we looked specifically for changes in the elastin and collagen fibers in these Control and POP-HQ subgroup USLs; our initial hypothesis was that the connective tissue fiber contents would differ between the POP-HQ subgroups and these differences could affect the resulting POP event. First, and unexpectedly, we found that Control levels of elastin significantly increased with increasing vaginal parity suggesting an ongoing attempt at repair in Control non-prolapsed USLs. Although increases in Control levels of collagen were only close to significant, a plot of elastin versus collagen also attained significance. None of the POP-HQ subgroup USLs demonstrated significant covariate elastin content changes, they only demonstrated trends. While the POP-I and POP-V subgroup USLs trended (but did not reach significance) towards invariant or decreasing levels of elastin with increasing age, BMI, vaginal parity, or uterine descent, many of the POP-A trendlines were positive indicating increasing quantities of elastin and demonstrating a difference from the POP-I and POP-V subgroups. The major take home message from all this analysis is that many more samples per group will be needed to attain significant results. However, the trend in Control, and to some degree the POP-A, subgroups’ connective fibers suggest the interesting hypothesis that structural repair may occur within the USL. This possibility could help explain why all women do not present with pelvic prolapse. Since neither Control nor POP-A USL attempts at repair have been previously recognized, their incidence and success rate are presently unknown. We hypothesize that the identification of a non-invasive marker for the USL repair process(es) would allow a better understanding of this phenomenon including the incidence and success rate and might allow experiments designed to enhance this phenomenon. Clinically, knowledge of ongoing repair would be a major step forward. We identified a 10% shortening of the elastin fibers in the POP-I USLs (Table 3 , p < 0.05), but no change in the elastin fiber length or various morphologies of the other POP-HQ USLs. A decrease in the length of elastin fibers in the presence of inflammation has been reported by others examining vasculature and skin [ 51 ]. We did not observe a significant correlation between the length of the elastin fibers and descent of the uterus. Future work will look for a correlation between elastin fiber length and biomechanical strength of the USL. Within the tunica-intima and -media of POP-V USL arteries there was a significant increase in the elastin content and in the presence of vimentin-positive cells with size, shape, and phenotype expected for fibroblasts or myofibroblasts, not inflammatory cells. Myofibroblasts are the stem cell that, along with fibroblasts, is required for repair of damaged connective tissue [ 51 , 52 ]. Therefore, their presence in the tunica intima and -media, along with the increased elastin observed indicates that this neointimal hyperplasia (NIH) is an attempt to repair damaged vasculature; other NIH researchers have come to a similar conclusion [ 52 ]. Unfortunately, our studies did not suggest a reason for the observed NIH in this anatomical locale. Vasculature researchers have hypothesized and successfully demonstrated that common reasons for NIH include ageing, diet, and relative fitness [ 52 ]. The POP-V subgroup have the oldest average age of the POP-HQ subgroups however whether these individuals have poor diets and fitness was not documented. Clinically, this suggests that older patients need to be made aware that their diet and fitness level may be important with respect to the possibility of prolapsing. However, both the POP-A and POP-I subgroups contain older prolapsing patients who did not show signs of NIH again emphasizing the heterogeneity of the prolapsing patients. A second site of new elastin content was also noted. As USL smooth muscle fascicles underwent loss of muscle cells (dropout [ 39 ]),, a heterogeneous increase in vimentin-positive cells with the phenotype expected for myofibroblasts occurred within the fascicles along with increased quantities of both collagen and elastin fibers. The heterogeneity of the response resulted in a lack of significance (Table 3 ). None-the-less we interpreted this as an attempt at muscle fascicle repair. More smooth muscle fiber dropout has been identified in the POP-A and -I USLs ( p < 0.05) but not in the POP-V USLs [ 39 ]. Therefore, a conundrum arises, the POP-A, POP-I, and POP-V USL tissues examined here have failed in their support function since they were taken from women with prolapse, however we saw evidence that all three have attempted to repair some of their damaged connective tissue prior to failure. The POP-A and -I USLs have new collagen and elastin fibers inside of smooth muscle fascicles and POP-V USLs have abundant new elastin in their arterial tunica intima, and in both sites myofibroblasts are now found where they were not observed previously. Again, we conclude that repair is possible but that the incidence of this is quite variable. Further, we believe these data support future experimentation that examines the hypothesis that these repair processes are important for the health of the USL tissues. Inflammatory cells have multiple roles in the regulation of elastin and collagen levels. Mast cells are reported to promote elastin deposition in vessel grafts in mice [ 53 ] and are commonly found in normal tissue undergoing remodeling, for instance in the mammary gland following weaning and the cessation of lactation [ 54 ]. Mast cells are a known stimulator of myofibroblasts and fibrosis [ 55 ]. Similarly, macrophages play a key role in remodeling ECM through secretion of proteases and degradation of the matrix; this is well documented in some types of breast cancer [ 56 ]. In a liver fibrosis model, macrophages were a key source of metalloelastase [ 57 ]. T-lymphocytes are hypothesized to be important in wound repair for their release of TGF-β [ 58 ], myofibroblasts respond to TGF-β by secreting ECM [ 59 , 60 ]. Our previous work [ 47 ] demonstrated that USL tissue from POP-I subgroup subjects possessed increased numbers of neutrophils, mast cells, T-lymphocytes, and resolving macrophages compared to the other subgroup USLs. However, here we were unable to show any direct correlation between mast cell or T-cell numbers and changes in elastin levels. As noted above, we observed increased collagen and elastin, variably, in smooth muscle fascicles. Others have reported similar increases in collagen and elastin in bladder wall smooth muscle fascicles and have attributed this to prior neurogenic damage and muscle cell loss [ 61 , 62 ]. In the bladder wall, an overall increase in smooth muscle accompanied the increase in elastin and collagen, whereas in the prolapsing USL, a decrease in smooth muscle content occurred. During our study we looked for but were unable to find histological evidence of neurogenic injury. Our study adds to and extends the results of prior studies that have quantified collagen and elastin levels in the context of POP where mouse and human vagina, cardinal ligament, or cervical tissue [ 31 – 33 , 45 , 46 , 63 – 65 ], or in vitro studies of cells in tissue culture models [ 41 , 66 , 67 ] were variously assessed. Our histological observations of new sites of elastin and collagen within USL tissues similarly extend and provide visual context for the conclusions from prior work examining mRNA transcripts [ 24 , 68 ] that suggest prolapsing support structures may be trying to repair themselves prior to or during the prolapse event. The first major strength of our study is its utilization of histologically defined prolapse subgroups through use of the POP-Q and POP-HQ scoring systems. A better definition of the prolapse tissue then allowed us to test for differences between prolapsed subgroups following elastin and collagen quantification. A second strength is examination of elastin and collagen protein levels in distinct histological locales within the USL tissue through use of special stains and immunohistochemistry, all of which provided some visual context for the prolapse event. There are weaknesses to acknowledge with this study. First, our study tissue was collected at an OB/GYN clinic, and all subjects were experiencing some type of medical problem, therefore the Control group is not representative of the general population. Second, the age distribution of our subjects mirrors the typical clinical population seeking treatment which includes young control and, predominantly, older POP patients. A more representative study would include more old controls and young POP patients. Third, the detected changes in elastin and collagen suggest but do not prove a functional consequence. Biomechanical experiments are required to verify the histological analysis conclusions. Biomechanical experiments on animal tissues are in progress [ 35 ] and should allow examination of clinical specimens in the future. Fourth, we acknowledge that the USL provides only one of the three levels of pelvic organ support, and changes in the USL are not expected to be solely responsible for the development of POP. Whether other apical support tissues exhibit similar alterations in connective tissue fiber content remains to be examined. Lastly, our study suffers from insufficient subjects; our use of a 100 subject cohort proved insufficient due to unforeseen heterogeneity even after our various attempts at prior stratification. Clinically actionable results will require many more subjects. Nonetheless, results here portend a promising future.

We utilized a histological approach to examine the hypothesis that USL failure and the ensuing prolapse were due to alterations in USL connective tissue fiber content. Separation of USL-tissues into histopathologically distinct POP-HQ subgroups followed by fiber quantification demonstrated differing trends in elastin and collagen content by the Control and POP-HQ subgroups. The data here suggest that Control USLs can increase connective tissue fibers in response to certain stresses while the responses observed in the POP-HQ was more variable and often negative. These data support a model where distinctly different responses to damage can occur within the USL but also indicate attempts at tissue repair can occur leading us to hypothesize there may be ways to facilitate the success of these repair attempts. Due to the differences we have documented, we further hypothesize that women within the different POP-HQ subgroups may have different risks for complications and recurrence after POP repair and suggest personalized approaches will be needed to address the different etiologies.

Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2