Objective
The mechanical strength of the cervix relies on the cross-linking of the tissue’s collagen network. Clinically, the internal os is functionally distinct from the external os. We sought to detect specific collagen cross-links in human cervical tissue and determine whether cross-link profiles were similar at the internal and external os.
Study Design
Transverse slices of cervical tissue were obtained at the internal and external os from 13 nonpregnant, premenopausal women undergoing a benign hysterectomy. To understand how cross-links were distributed throughout the entire cervix and at the internal and external os, biopsies were obtained from 3 circumferential zones in 4 quadrants from each slice. Biopsies were pulverized, lyophilized, reduced with sodium borohydride, hydrolyzed with hydrochloric acid, and reconstituted in heptafluorobutyric acid buffer. Hydroxyproline was measured by ultraperformance liquid chromatography-electrospray ionization tandem mass spectrometry (UPLC-ESI-MS/MS), converted to total collagen, and normalized by dry weight. Collagen cross-links pyridinoline (PYD), deoxypyridinoline (DPD), dihydroxylysinonorleucine (DHLNL), and the nonenzymatic advanced glycation end product pentosidine (PEN) were measured by UPLC-ESI-MS/MS and reported as cross-link density ratio (cross-link/total collagen). Generalized estimated equation analysis was used to compare results between the internal and external os and to compare quadrants and zones within slices from the internal and external os to determine if cross-link profiles were similar.
Results
A total of 592 samples from 13 patients were analyzed. Collagen cross-links are detectable in the human cervix by UPLC-ESI-MS/MS. When comparing all samples from the internal and external os, similar levels of collagen content, PYD, DHLNL, and DPD were found, but PEN density was higher at the external os (0.005 vs 0.004, P = .001). When comparing all internal os samples, significant heterogeneity was found in collagen content and cross-link densities across zones and quadrants. The external os exhibited heterogeneity only across zones.
Conclusion
Collagen cross-links (PYD, DPD, DHLNL, and PEN) are detectable by UPLC-ESI-MS/MS in the human cervix. The internal os exhibits significant collagen cross-link heterogeneity compared with the external os. Further studies are needed to evaluate how collagen cross-link heterogeneity correlates to the mechanical strength and function of the human cervix.
During pregnancy, the cervix undergoes dramatic changes to allow for successful parturition. Ultrasound studies have shown that cervical ripening at term starts with dilation of the internal os and then progressive funneling toward the external os until total effacement is achieved. A similar sequence of cervical changes occurs in pregnancies complicated by cervical insufficiency. During a sterile vaginal examination, it is also common to note that the external os is open, whereas the internal os remains closed.
Although cervical tissue architecture was traditionally thought to be homogenous, these clinical findings fuel the following questions: is the internal os functionally distinct from the external os? If so, are there differences in the cervical tissue properties at the level of the internal os that influences its mechanical performance? The answers to these questions remain elusive because studies evaluating cervical tissue strength at the internal and external os are lacking.
Human cervical tissue is a hydrated connective soft tissue composed mainly of a collagen-rich extracellular matrix (ECM). Similar to other load-bearing tissues in the body (eg, bone, tendon), the mechanical strength of cervical ECM relies in part on the type and degree of collagen cross-linking in its collagen network. Collagen cross-links stabilize collagen molecules, a vital requirement to building strong, organized collagen networks. During collagen cross-link formation, immature cross-links such as dihydroxylysinonorleucine (DHLNL) are formed between a telopeptide residue and a helical residue. Subsequently, another telopeptide residue is formed between 3 collagen molecules resulting in mature cross-links, deoxypyridinoline (DPD) and pyridinoline (PYD).
Previously the degree of collagen cross-linking was indirectly determined by measuring collagen extractability, in which increased extractability correlated to decreased collagen cross-linking. Studies that used this method showed that collagen extractability significantly increases in a normal mouse cervix during pregnancy. Myers et al demonstrated similar results on human nonpregnant and term pregnant tissue samples. Recently Akins et al measured mature collagen cross-links in mouse cervical tissue by reversed-phase, high-performance liquid chromatography and found cervical ripening in rodents is characterized by a decrease in PYD and DPD.
Given that (1) mature collagen cross-links provide stability and thus strength to the ECM, (2) collagen cross-link profiles are altered in cervical remodeling/softening, and (3) the internal os appears to soften or weaken first during parturition and premature cervical remodeling, the goal of this study was to determine whether specific collagen cross-links are detectable in the human cervix and whether the internal os has a different collagen cross-link profile compared with the external os.
Materials and Methods
This study was approved by the Columbia University Medical Center Institutional Review Board (institutional review board no. AAAI0337). Nonpregnant, premenopausal women undergoing a total hysterectomy for benign indications were identified and consented to participate. Women were excluded if they were older than 50 years of age, had an abnormal Papanicolaou smear, or had prior cervical surgery. Demographic information (age, race, body mass index, obstetric history) and indication for procedure were collected. Uterine weight was obtained from pathology reports ( Table 1 ).
| Patient number | Age, y | Race | BMI | Parity | Obstetric history | Type of hysterectomy | Uterine plus cervix weight, g |
|---|---|---|---|---|---|---|---|
| 1 | 44 | H | 32.6 | 0 | None | TAH | 2994 |
| 2 | 49 | H | 23.3 | 0 | None | TAH | 1150 |
| 3 | 46 | AA | 34.9 | 0 | None | TAH | 6100 |
| 4 | 40 | AA | 37.9 | 0 | None | TAH | 1005 |
| 5 | 49 | C | 28.3 | 1 | NSVD | TRH | 267 |
| 6 | 48 | H | 27.8 | 1 | NSVD | TLH | 113 |
| 7 | 42 | C | 25.3 | 1 | NSVD | TRH | 225 |
| 8 | 41 | AA | 25.1 | 1 | NSVD | TLH | 192 |
| 9 | 42 | C | 28.1 | 1 | NSVD | TLH | 75 |
| 10 | 49 | C | 29.9 | 2 | NSVD × 2 | TVH | 175 |
| 11 | 44 | H | 21.2 | 4 | NSVD × 4 | TLH | 223 |
| 12 | 46 | H | 30.9 | 4 | NSVD × 3, CD × 1 | TRH | 178 |
| 13 | 48 | H | 30.2 | 5 | NSVD × 5 | LAVH | 250 |
Immediately following the hysterectomy, 2-3 mm transverse slices of the cervix were obtained at the level of the internal and external os. Anterior/posterior orientation of the tissue was maintained. One slice from the internal and external os was frozen and stored at –80°C until collagen content and cross-links analysis were performed. An adjacent slice from each area was fixed in 10% neutral buffered formalin for 24 hours, paraffin embedded, and sectioned for hematoxylin and eosin staining. The hematoxylin and eosin–stained slides were reviewed with a pathologist to confirm the location in the cervix and exclude pathology. The formal pathology report from the surgical specimen was also reviewed, and we proceeded with further tissue processing if the findings were benign.
Because early studies evaluating collagen structure in the cervix reported that the stroma contains zones of preferentially aligned collagen ( Figure , A), our goal was to evaluate whether collagen content and collagen cross-link densities differed, depending on the zone or quadrant of the cervix. Using a 2 mm punch biopsy (number 33-31; Miltex, York, PA), biopsies were obtained in duplicate from each zone from 4 distinct quadrants ( Figure , B). Samples were lyophilized and dry weights recorded. Dehydrated samples were reduced in sodium borohydride (NaBH 4 ) and then hydrolyzed in 12 M hydrochloric acid in vaccuo at 110°C for 18-24 hours. The hydrolysate was lyophilized and resuspended in heptaflurobutyric acid buffer.
The molar content of hydroxyproline, PYD, DPD, DHLNL, and pentosidine (PEN) was measured using ultraperformance liquid chromatography–electrospray ionization tandem mass spectrometry as previously described. Total collagen content was determined by using a mass ratio of collagen to hydroxyproline (molecular weight, 131 g/mol) of 7.14:1, and the concentration of collagen was determined by dividing by the dry weight. Collagen cross-link density was determined for each type of collagen cross-link by dividing by the collagen content of the tissue sample. Therefore, all collagen cross-link densities are reported on a mole-per-mole basis with collagen content.
Analysis
Statistical methods
An assumption of regression models is that observations or in this case the biopsy samples are independent of one another. Because we obtained multiple samples from the same patient, this violates the statistical assumption of independence. Failure to adjust for the intracluster dependence will result in biased variance estimation. Therefore, to account for this intracluster dependence, we fit all regression models based on the method of generalized estimating equations procedure.
Collagen content and cross-link analysis
A priori contrast procedures were performed. Using a linear regression model based on the methods of generalized estimating equations, we initially established the collagen content and collagen cross-link densities in the total cervix (internal os samples plus external os samples from all 13 women). We then evaluated for any heterogeneity with respect to location (quadrants and zones). Next, we evaluated heterogeneity in collagen content and collagen cross-links between the samples from the internal os and the external os. We also analyzed the differences in quadrants and zones between the internal and external os (ie, quadrant or zone 1 from internal os to quadrant or zone 1 of external os). Lastly, we analyzed the tissue heterogeneity within the internal and external os (comparing quadrants and zones within each slice from the internal and external os). Because all comparisons were planned a priori before the beginning of the experiments, we did not correct for multiple testing. All regression models were fit based on the GENMOD procedure in SAS version 9.4 (SAS Institute, Cary, NC).
Results
Thirteen patients were consented and cervical tissue collected. A total of 624 biopsies were collected. Biopsies were excluded if an error was made during processing or if the collagen content was higher than 100%. This left a total of 592 samples for analysis. The average age was 45.2 years, average body mass index was 28.8 kg/m 2 , and the mean uterine weight was 1011 g ( Table 1 ).
Total cervix
The total cervix collagen content and specific collagen cross-link values reported in Table 2 represent a combination of all internal and external os samples from the 13 patients. PYD was the most common cross-link followed by DHLNL and then DPD, with average densities of 0.123, 0.094, and 0.045 mol/mol, respectively. PEN was present in the cervix but in small amounts compared with PYD ( Table 2 ). DHLNL was the only cross-link density affected by parity, with higher levels in nulliparous women (0.126 vs 0.080, P = .045). There was no difference in the collagen content, PEN, DPD, and PYD densities between nulliparous and multiparous subjects ( Table 3 ).
| Characteristics | Mean (n = 592) a | SD |
|---|---|---|
| Collagen content, mg/mg dry weight | 0.344 | 0.163 |
| PEN, mol/mol | 0.005 | 0.003 |
| DPD, mol/mol | 0.045 | 0.042 |
| DHLNL, mol/mol | 0.094 | 0.072 |
| PYD, mol/mol | 0.123 | 0.073 |
| Characteristics | Nulliparous (n = 173) a (95% CI) | Multiparous (n = 419) a (95% CI) | P value |
|---|---|---|---|
| Collagen content, mg/mg dry weight | 0.391 (0.366–0.416) | 0.325 (0.31–0.34) | NS |
| PEN, mol/mol | 0.004 (0.004–0.005) | 0.005 (0.005–0.005) | NS |
| DPD, mol/mol | 0.053 (0.048–0.058) | 0.041 (0.037–0.046) | NS |
| DHLNL, mol/mol | 0.126 (0.112–0.140) | 0.080 (0.075–0.086) | .045 |
| PYD, mol/mol | 0.145 (0.131–0.158) | 0.114 (0.108–0.120) | NS |
When comparing collagen content by quadrants ( Table 4 and Figure , C), there was a significantly lower collagen content in quadrant 4 (Q4) compared with quadrant 1 (Q1) (32.7% vs 34.7%, P = .034). There were no significant differences in collagen cross-link densities between quadrants ( Table 4 ). When analyzing the zones in the total cervix ( Table 5 and Figure , D), significant heterogeneity was noted. The midstromal zone (Z2) had higher levels of collagen content, PYD, DPD, and DHLNL densities and lower PEN density when compared with the inner zone (Z1). The outer zone (Z3) had higher amounts of all collagen cross-links compared with the inner zone (Z1) ( Table 5 ).
| Characteristics | Q1 (n = 142) a (95% CI) | Q2 (n = 151) a (95% CI) | Q3 (n = 150) a (95% CI) | Q4 (n = 149) a (95% CI) | P value |
|---|---|---|---|---|---|
| Collagen content, mg/mg dry weight | 0.347 (0.323–0.371) | 0.345 (0.318–0.372) | 0.357 (0.329–0.386) | 0.327 (0.302–0.353) | .034 for Q1 vs Q4 |
| PEN, mol/mol | 0.005 (0.005–0.006) | 0.005 (0.004–0.005) | 0.005 (0.005–0.006) | 0.005 (0.004–0.006) | NS |
| DPD, mol/mol | 0.045 (0.038–0.052) | 0.042 (0.036–0.049) | 0.049 (0.042–0.056) | 0.043 (0.037–0.050) | NS |
| DHLNL, mol/mol | 0.090 (0.079–0.100) | 0.091 (0.079–0.103) | 0.106 (0.093–0.119) | 0.088 (0.077–0.099) | NS |
| PYD, mol/mol | 0.122 (0.110–0.134) | 0.118 (0.106–0.131) | 0.131 (0.120–0.143) | 0.119 (0.108–0.131) | NS |
| Characteristics | Zone 1 (n = 197) a (95% CI) | Zone 2 (n = 196) a (95% CI) | Zone 3 (n = 199) a (95% CI) | P value |
|---|---|---|---|---|
| Collagen content, mg/mg dry weight | 0.297 (0.276–0.317) | 0.434 (0.412–0.456) | 0.303 (0.282–0.324) | < .001 for Z1 vs Z2, NS for Z1 vs Z3 |
| PEN, mol/mol | 0.005 (0.005–0.005) | 0.004 (0.004–0.004) | 0.006 (0.005–0.006) | .002 for Z1 vs Z2, .008 for Z1 vs Z3 |
| DPD, mol/mol | 0.020 (0.018–0.022) | 0.079 (0.072–0.087) | 0.035 (0.032–0.038) | < .001 for Z1 vs Z2 and Z2 vs Z3 |
| DHLNL, mol/mol | 0.069 (0.060–0.077) | 0.099 (0.090–0.108) | 0.113 (0.101–0.125) | < .001 for Z1 vs Z2 and Z1 vs Z3 |
| PYD, mol/mol | 0.091 (0.082–0.099) | 0.117 (0.109–0.124) | 0.160 (0.149–0.172) | .020 for Z1 vs Z2, < .001 for Z1 vs Z3 |
Comparison of internal os with external os
To determine whether the internal and external os were similar with respect to collagen content and cross-link densities, we compared collagen content and cross-link densities from all the internal os samples with all of the external os samples. There were no significant differences in collagen content, PYD, DPD, and DHLNL densities between the internal and external os. PEN density was significantly higher at the external os than the internal os (0.004 vs 0.005, P = .001) ( Table 6 ).
| Characteristics | Internal os (n = 290) a (95% CI) | External os (n = 302) a (95% CI) | P value |
|---|---|---|---|
| Collagen content, mg/mg per dry weight | 0.345 (0.326–0.364) | 0.343 (0.325–0.362) | NS |
| PEN, mol/mol | 0.004 (0.004–0.005) | 0.005 (0.005–0.006) | .001 |
| DPD, mol/mol | 0.045 (0.040–0.050) | 0.044 (0.040–0.049) | NS |
| DHLNL, mol/mol | 0.090 (0.083–0.096) | 0.098 (0.088–0.107) | NS |
| PYD, mol/mol | 0.123 (0.115–0.131) | 0.122 (0.114–0.131) | NS |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
