Pelvic pain symptoms and endometriosis characteristics in relation to oxidative stress among adolescents and adults with and without surgically-confirmed endometriosis

Pain symptom assessment

Detailed information was collected on the baseline questionnaire on the presence, severity, and frequency of dysmenorrhea (pain with periods), acyclic/general pelvic pain (pain not associated with menses), and dyspareunia (pain with sexual vaginal intercourse/penetration). Usual severity of dysmenorrhea was assessed categorically as none, mild (medication never or rarely needed), moderate (medication usually needed), and severe (medication and bed rest needed). Dysmenorrhea frequency within the past 12 months was assessed as never, occasionally, often, usually, and always. For acyclic pelvic pain, participants reported if they had experienced acyclic pelvic pain within the past three months. Among those with acyclic pelvic pain in the past three months, the 11-point numeric rating scale (NRS) was used to assess the acyclic pelvic pain severity during that timeframe, with 0=no pain and 10=worst pain imaginable. Acyclic pelvic pain frequency was assessed as less than monthly, monthly, weekly, and daily. Among participants aged 18 or older, participants reported if they had experienced dyspareunia in the last 12 months. Among those reporting dyspareunia, the 11-point NRS scale was used to assess severity while frequency of dyspareunia during or in the 24 hours after intercourse/penetration in the past 12 months was assessed as never, occasionally, often, usually, and always. Finally, participants reported if they had pain with bowel movements in the past 12 months. Those who reported pain rated the severity of their pain on the 11-point NRS scale.

Endometriotic lesion characteristics

The WERF EPHect surgical form was used to capture information on rASRM score, endometriosis subtype, and endometriosis lesion(s) color and location at the surgery closest to urine collection for endometriosis cases.²² We further categorized endometriosis lesions into colors that are normally observed earlier in the lesion progression (i.e. red, yellow, and clear lesions) and later in the lesion progression (i.e. blue, black/brown, and white lesions).

Urine collection

Urine samples were collected at baseline in compliance with the WERF EPHect standardized fluids tools,²³ with the exception that we did not require clean catch collection of urine samples. Participants completed a biospecimen questionnaire at the time of sample collection on which they reported date of last menstrual period, timing of last foods/beverages consumed, and recent medication and hormone use. All urine samples were aliquoted into cryovials and stored at -80^oC until assayed.

Oxidative stress measurement

8-OHdG was measured in urine using the HT 8-oxo-dG ELISA Kit II (R&D Systems, Inc. Minneapolis, MN, USA) at the Clinical and Epidemiology Laboratory at Boston Children’s Hospital (BCH). Creatinine levels were also measured in the urine samples using an FDA-approved enzymatic assay method on the Roche Cobas 6000 system using Roche Diagnostics reagents (Indianapolis, IN) at the Clinical and Epidemiology Laboratory at BCH. Approximately 2-3 blinded quality control (QC) urine samples were distributed randomly within each batch. The coefficient of variation (CV) in blinded QC samples for 8-OHdG was 13% and for creatinine was 1.5%.

Covariates

Information on covariates was collected on the baseline questionnaire and biospecimen form and included: age at urine collection (continuous), cigarette smoking history (never, former, current), age at menarche (continuous), menstrual cycle phase at time of urine collection (follicular, peri-ovulatory, luteal), hormonal medication use within 30 days from urine collection (yes, no), and pain medication use within 48 hours from urine collection (yes, no). Additionally, participants reported their menstrual period frequency in the past 12 months. We calculated body mass index (BMI) as kg/m² based on self-reported weight and height. For women aged ≥20 years, BMI was categorized according to the World Health Organization Criteria: underweight (BMI < 18.5 kg/m²), normal weight (18.5-24.9 kg/m²), overweight (25-29.9 kg/m²), and obese (≥30 kg/m²). For those <20 years, the age- and sex-specific BMI Z-score was calculated and participants were categorized as underweight (Z-score ≤-2), normal weight (Z-score >-2 to <1), overweight (Z-score 1-2), and obese (Z-score >2). Participants were also asked to report their physical activity levels at baseline. Participants reported the average time per week they spent doing various activities (e.g. walking/hiking, jogging, running, lap swimming, playing various sports) in the past year. The 2011 Compendium of Physical Activities was then used to assign metabolic equivalent hours (MET-hours) per week to each of the activities.²⁴ We multiplied the reported hours per week engaged in each activity by the appropriate MET score for that activity (e.g. 4 for walking, 8.3 for bicycling, 11.7 for running) and summed the values for the individual activities to create MET-h/wk of total activity. For those with surgically-confirmed endometriosis, information on age at first endometriosis symptoms, number of physicians seen until diagnosed, and time between first symptoms and surgical diagnosis also were reported on the baseline questionnaire.

Also at baseline, participants completed a semi-quantitative Food Frequency Questionnaire (FFQ), which included over 130 items on the consumption of a range of foods and beverages. The FFQ was utilized to calculate the Alternative Healthy Eating Index (AHEI) score for all participants. Details on this scoring method can be found elsewhere.^25,26 Briefly, AHEI scoring is based on the consumption of fruit, vegetables, whole grains, sugar-sweetened beverages and fruit juices, nuts and legumes, red/processed meat, trans fat, long-chain (n-3) fats (EPA + DHA), polyunsaturated fat, sodium, and alcohol. Given the younger age of this cohort (37% <21 years old at baseline), we omitted alcohol from the AHEI calculation. The AHEI score is summed across all of the components and ranges from 0 to 100, with higher scores representing dietary patterns that are more aligned with healthy eating.

Statistical analysis

Of the 1549 participants enrolled in the A2A cohort, 1209 provided a urine sample at baseline and all 549 endometriosis participants and 660 participants without endometriosis had 8-OHdG and creatinine measured in their baseline urine sample. Of these 1209 participants, we excluded participants who did not complete the questionnaire at baseline (5 endometriosis, 3 without endometriosis) or completed the questionnaire at baseline more than 60 days before/after their urine collection (105 endometriosis, 36 without endometriosis). We also excluded incident endometriosis participants (3 participants without endometriosis at enrollment diagnosed with endometriosis up to 3 years after enrollment) and those who were premenarchal or never cycled (5 endometriosis, 12 without endometriosis) for a final analytic sample size of 434 surgically diagnosed endometriosis participants and 606 participants without endometriosis. Dysmenorrhea analyses were restricted to participants who reported having menstrual periods in the past three months (264 endometriosis and 517 without endometriosis). Analyses of acyclic pelvic pain severity and frequency were restricted to participants who reported having acyclic pelvic pain in the past 3 months (268 endometriosis and 78 without endometriosis). Additionally, analyses of dyspareunia severity and frequency were restricted to participants age ≥18 who reported experiencing dyspareunia in the past 12 months (132 endometriosis and 136 without endometriosis). Analyses of endometriotic lesion characteristics were restricted to the 380 participants with a WERF EPHect surgical form completed at their most recent endometriosis surgery.

8-OHdG and creatinine levels were log-transformed to improve normality. To adjust for differences in urine volume, we divided the 8-OHdG measurement by the creatinine measurement to calculate creatinine-corrected 8-OHdG levels (ng/mg). We then used the generalized extreme studentized deviate many-outlier detection approach to identify statistical outliers for creatinine-corrected 8-OHdG.²⁷ For batch adjustment, levels of creatinine-corrected 8-OHdG were recalibrated to have a comparable distribution to an average batch according to the methods described by Rosner and colleagues.²⁸

After exclusions, we used linear regression to calculate geometric means (GM) and 95% confidence intervals (CI) for creatinine-corrected 8-OHdG levels, adjusting for age (continuous), hormone use within the 30 days prior to urine collection (yes, no), any pain medication use within 48 hours prior to urine collection (yes, no), Alternative Healthy Eating Index (quartiles), body mass index (underweight, normal weight, overweight, obese), and physical activity (quartiles of MET-hrs/week). Analyses were conducted among all participants and separately among those with and without endometriosis. Participants were excluded from analyses if they were missing the main pain symptom exposure variable. We calculated trend tests by modeling the categorical pain exposure variable as ordinal, adjusting for the same variables listed above. Pairwise comparisons between levels of categorical variables with three categories or more were performed using the Tukey adjustment for multiple testing. To evaluate if associations between pain symptoms and 8-OHdG levels differed between participants with and without endometriosis, we included an interaction term for endometriosis diagnosis status (endometriosis vs. no endometriosis) and each pain symptom in the linear regression models. The Wald statistic was used to calculate the two-sided p-value for interaction.

In sensitivity analyses, we restricted analyses to never smokers as cigarette smoking is known to have a significant effect on oxidative stress levels. Additionally, we excluded endometriosis participants who had an endometriosis-related surgery before they completed their baseline questionnaire and/or their baseline urine collection to remove the affects surgery may have had on oxidative stress levels and pain symptoms. All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC) and all p-values are two-sided.

Study population characteristics

We included 434 surgically confirmed endometriosis participants and 606 participants without endometriosis in our analyses. On average, baseline questionnaires were completed 6.8 days (standard deviation=17.5) prior to urine collection. At enrollment, those with endometriosis were younger (median age 18 vs. 24 years), and a higher proportion were overweight (25% vs. 20%) and identify as White race (91% vs. 71%) compared to those never diagnosed with endometriosis (Table 1). Additionally, participants without endometriosis were more likely to have had a period in the last three months (85% vs. 61%) compared to endometriosis participants, while endometriosis participants were more likely to have taken hormonal medications within 30 days of urine collection (87% vs. 54%) and pain medication within 48 hours of urine collection (23% vs. 18%) compared to those without endometriosis. Participants without endometriosis also reported higher physical activity and AHEI scores compared to endometriosis participants.

Pelvic pain symptoms

For pelvic pain symptoms, we observed no associations between 8-OHdG and dysmenorrhea (severity, frequency), acyclic pelvic pain (presence, frequency, or severity), dyspareunia (presence, frequency, or severity) as well as severity of pain with bowel movements (Table 2). Results between pain symptoms and 8-OHdG were similar when restricted to never cigarette smokers (Table 3). Additionally, results were similar when restricted to participants with urine collection and questionnaire completion before their baseline surgery (Table 4).

Endometriosis lesion characteristics

Among the 380 endometriosis participants who had a WERF EPHect surgical form, the median time between surgery and urine collection was 13 days with an interquartile range of 0 days to 41 days. The vast majority of endometriosis participants had rASRM stage I/II disease (95%) and superficial peritoneal lesions only (96%; Table 5). There was a suggestion of lower 8-OHdG levels for participants with rASRM stage III/IV disease (GM=68.0; CI=55.4-83.5 ng/mg) compared to participants with rASRM stage I/II disease (GM=81.0; CI=77.3-84.9 ng/mg; p=0.10), although this was based on a small sample size of stage III/IV disease and thus limited power. These results remained similar when analyses were restricted to participants with urine collection prior to a baseline surgery (Table 6).

For superficial peritoneal endometriotic lesion color, the highest 8-OHdG levels were observed for participants with any yellow lesions compared to those without (GM_yes=95.3; CI=77.6-116.9 ng/mg vs. GM_no=78.9; CI=75.3-82.7 ng/mg; p=0.08; Table 5). Additionally, endometriosis participants with white, blue/black or brown lesions had lower 8-OHdG levels compared to participants without (GM_yes=76.7; CI=72.0-81.8 ng/mg vs. GM_no=82.9; CI=77.6-88.7 ng/mg; p=0.10). This difference appeared to be driven by lower 8-OHdG levels among participants with any blue/black lesions compared to those without (GM_yes=72.8; CI=65.9-80.5 ng/mg vs. GM_no=81.6; CI=77.2-85.9 ng/mg; p=0.05) and similarly low levels among participants with brown lesions (p=0.10). Results for brown lesions were attenuated when analyses were restricted to participants with urine collection before their baseline surgery; however, results for having any yellow lesions, any blue/black lesions, and any white, blue/black or brown lesions remained similar (Table 6). We did not observe any differences in 8-OHdG levels by lesion location.

Pelvic pain

The majority of the previous studies on oxidative stress and pelvic pain symptoms have focused on dysmenorrhea with most observing an association between higher oxidative stress levels among those with dysmenorrhea.^10–15 The largest study to date with 897 adolescents observed that the serum pro-oxidant/antioxidant balance was shifted more towards the pro-oxidant side among participants with primary dysmenorrhea compared to those without dysmenorrhea when adjusting for age and BMI.¹⁰ Conversely, Konishi et al. (2018) noted that severity of menstrual pain was not associated with urinary 8-OHdG levels among 188 female university students after adjusting for age and BMI. Similar to the results of Konishi et al. (2018), we observed that dysmenorrhea severity was not associated with 8-OHdG. Further, we observed that neither acyclic pelvic pain nor dyspareunia were associated with urinary 8-OHdG among participants with and without endometriosis. We may not have observed associations between 8-OHdG and pelvic pain symptoms in our study due to (1) assessing the DNA oxidative product of 8-OHdG given lipid or protein oxidation may have been more important for endometriosis-associated pelvic pain, (2) it may be that the interplay between oxidative stress and other molecules in the peritoneal cavity, such as inflammatory molecules, may be important for pelvic pain as opposed to oxidative stress on its own, or (3) the younger age of our study population (37% <21 years old) if associations between oxidative stress and pelvic pain are more apparent at older ages. Therefore, future research on other types of oxidative stress and the interplay between oxidative stress and the peritoneal microenvironment in relation to pelvic pain is needed.

Endometriosis lesion characteristics

Oxidative stress has been implicated in the onset and progression of endometriosis and higher levels of oxidative products have been observed in the peritoneal fluid of individuals with endometriosis compared to control participants.²⁹ However, limited studies have assessed differences in oxidative stress levels among subsets of endometriosis patients and those that have, mainly focused on endometriosis patients presenting with infertility. In our study of mainly pain presenting endometriosis participants, we observed a suggestion of lower 8-OHdG levels for rASRM stage III/IV endometriosis compared to stage I/II; however, these results were based on a small sample of rASRM stage III/IV endometriosis participants. Contrary to our results, previous studies have observed increased oxidative stress with higher rASRM stage,^{11,14,30–32} while two small studies among infertile endometriosis patients observed no association between endometriosis stage and lipid oxidation.^33,34 Differences between our results and the previous studies may be due to differences in the study populations with the younger, mostly pain presenting population in the A2A for which the association between oxidative stress and disease stage may be different from endometriosis patients who present with infertility. Additionally, differences between oxidative stress markers measured and biological sample types (e.g. blood, urine) utilized, as well as a lack of adjustment for potential confounders in the previous studies, may have led to differences between our results and previous studies. Finally, we noted that endometriosis participants with any white, blue/black or brown lesions had lower 8-OHdG levels compared to endometriosis participants with no white, blue/black or brown lesions, which appeared to be driven by the presence of blue/black lesions. To our knowledge, no other study has looked at lesion color and oxidative stress levels; therefore, these results warrant further exploration in other studies of endometriosis.

Strengths and limitations

This study had some limitations including that we had only one marker of oxidative stress and thus may have missed associations between pelvic pain and protein and/or lipid oxidation. Additionally, some of the participants that have never been diagnosed with endometriosis within our study may have undiagnosed endometriosis; however, it is estimated that the community prevalence of undiagnosed endometriosis is <2%. Up to 10 years on since enrollment began in 2012, only four participants who at enrollment into our study had not been diagnosed with endometriosis, were subsequently diagnosed; the three participants who would have been eligible for these analyses were excluded. Effects of the characteristics of this small proportion of undiagnosed cases will be diluted among the true endometriosis-free participants. Further, as these analyses were cross-sectional, we cannot directly elucidate the cause and effect relationship between endometriosis lesion characteristics and 8-OHdG. Finally, the A2A population is predominately White, particularly among endometriosis participants; however, the population reflects the patients treated at the two participating hospitals and the general population of the people referred to those hospitals. Future research involving a more diverse population is needed.

Our study also had several strengths. It is one of the largest studies to date to assess oxidative stress and pelvic pain symptoms, and it included a predominately young population including endometriosis participants who are more proximal to their endometriosis symptom onset compared to previous studies. Although we only included one measure of DNA damage due to oxidative stress, urinary measurement of 8-OHdG has been validated as a reliable biomarker of oxidative stress in previous studies; and in comparison to blood samples, is less affected by potential misclassification due to oxidative processes that occur during sample collection and storage. Finally, we assessed multiple dimensions of pelvic pain, including presence, severity and frequency, which provided a more nuanced assessment of the relationship between pelvic pain symptoms and oxidative stress.