Plasma proteins and persistent post-surgical pelvic pain among adolescents and young adults with endometriosis

Abstract

Background:

Non-invasive biomarkers that predict surgical treatment response would inform personalized treatments and provide insight into potential biological pathways underlying endometriosis-associated pain and symptom progression. Thus, we evaluated plasma proteins in relation to persistence of pelvic pain following laparoscopic surgery in predominantly adolescents and young adults with endometriosis using a multiplex aptamer-based proteomics biomarker discovery platform.

Methods:

We conducted a prospective analysis including 142 participants with laparoscopically-confirmed endometriosis from the Women’s Health Study: From Adolescence to Adulthood (A2A) observational longitudinal cohort with study enrollment from 2012–2018. Biologic samples and patient data were collected with modified World Endometriosis Research Foundation Endometriosis Phenome and Biobanking Harmonization Project (EPHect) tools. In blood collected before laparoscopic ablation or excision of endometriosis, we simultaneously measured 1,305 plasma protein levels including markers for immunity, angiogenesis and inflammation using SomaScan. Worsening or persistent post-surgical pelvic pain was defined as having newly developed, persistent (i.e., stable), or worsening severity, frequency, or persistent life-interference of dysmenorrhea or acyclic pelvic pain at one-year post-surgery compared to pre-surgery. We calculated odds ratios (OR) and 95% confidence intervals (CI) using logistic regression adjusted for age, body mass index, and fasting status and hormone use at blood draw. We applied Ingenuity Pathway Analysis and STRING analysis to identify pathophysiologic pathways and protein interactions.

Results:

Median age at blood draw was 17 years (interquartile range 15–19), and most participants were white race (90%). All had superficial peritoneal lesions only and were treated by excision or ablation. One-year post-surgery, pelvic pain worsened or persisted for 76 (54%) of these participants with endometriosis, while pelvic pain improved for 66 (46%). We identified 83 proteins associated with worsening or persistent pelvic pain one-year post-surgery (nominal p<0.05). Compared to those with improved pelvic pain one year post-surgery, those with worsening or persistent pelvic pain had higher plasma levels of CD63 antigen (OR=2.98, 95% CI:1.44–6.19) and CD47 (OR=2.68, 95%CI=1.28–5.61), but lower levels of Sonic Hedgehog protein (SHH; OR=0.55, 95%CI=0.36–0.84) in pre-surgical blood. Pathways related to cell migration were upregulated and pathways related to angiogenesis were downregulated in those with worsening/persistent post-surgical pelvic pain compared to those with improved pain. When we examined change in proteins levels from pre- to post-surgery and its subsequent risk of worsening/persistent post-surgical pain at one-year follow-up, we observed increasing levels of SHH from pre- to post-surgery was associated with four-fold increase in risk of post-surgical pain (OR _{quartile 4 vs. 1}=3.86, 1.04–14.33).

Conclusion:

Using an aptamer-based proteomics platform, we identified plasma proteins and pathways associated with worsening or persistent pelvic pain post-surgical treatment of endometriosis among adolescents and young adults that may aid in risk stratification of individuals with endometriosis.

Tweetable statement:

Among adolescents and young adults with endometriosis, we identified plasma proteins and biological pathways dysregulated in pre-surgical blood associated with increased risk of worsening/persistent pelvic pain at one-year follow-up.

INTRODUCTION

Endometriosis is a chronic inflammatory condition with growth of endometrial-like tissue outside the uterus that often presents with severe pelvic pain, affecting about 10% of reproductive-aged women.¹ While current conventional treatment for endometriosis is hormone therapy and surgical resection of endometriotic lesions,² we and others have reported that about 30% of endometriosis patients suffer from chronic pelvic pain after standard of care treatment.^3–6 Emerging evidence suggests central sensitization⁷ may play a role in chronic or persistent endometriosis-associated pain. For these women excision of the endometriotic lesions in the pelvis does not improve their pain. Currently there are no methods to predict response to surgical treatment prior to surgical intervention and the biological mechanisms driving surgical treatment resistance are not known.

For approximately two-thirds of women diagnosed with endometriosis in adulthood, their pelvic pain symptoms started during adolescence.^8–10 Considering the chronic pain experience at a younger age increases risk of mental health disorders¹¹ and opioid use disorder,¹² predicting who will and will not respond to conventional treatments could provide new opportunities for improved personalized treatment and faster symptom improvement. Furthermore, defining the pathophysiology driving worsening/persistent pelvic pain after endometriosis-related surgery would provide new targets for treatment development. Thus, we sought to discover plasma proteomic profiles in pre-surgical blood samples associated with risk of worsening/persistent pelvic pain after surgery among adolescents and young adults with endometriosis.

MATERIALS AND METHODS

Study population

The Women’s Health Study: From Adolescence to Adulthood (A2A) is an ongoing observational longitudinal cohort that enrolled participants from 2012–2018, as described previously.^{6, 13–15} In brief, those with laparoscopically-confirmed endometriosis were identified from two tertiary academic hospitals. Of the 1,839 eligible participants approached, 1,569 (85%) enrolled. At baseline and Year 1 follow-up, an expanded version of the World Endometriosis Research Foundation Endometriosis Phenome and Biobanking Harmonization Project (WERF EPHect) standard clinical questionnaire¹⁶ was used to collect information including demographic characteristics, details on pelvic pain, treatment regimen, and medication use. Detailed clinical information was collected using an EPHect compliant surgical format time of endometriosis-related surgeries, including revised American Society for Reproductive Medicine (rASRM) score and medication use.¹⁷ All endometriosis patients had visualization and removal of endometriotic lesions by either excision or ablations techniques at surgery.¹⁸

Blood samples were collected at baseline and post-surgical follow-up [i.e. 5 weeks to 6 months after surgery, with the median for timing of post-surgical blood work being 16.3 weeks (IQR 12.3–19.7 weeks)] were collected, processed, and stored at ≤−80°C following the standard protocols of WERF EPHect fluid biospecimen collection¹⁹ with one exception - bloods were centrifuged at 1790 × g for 10 minutes. Detailed participant information at time of blood collection, such as fasting status and recent medication use, was collected using a biospecimen questionnaire.

Institutional Review Board (IRB) approval was provided by the Boston Children’s Hospital (BCH) and Brigham and Women’s Hospital IRBs. All participants provided written consent, with additional parental consent for participants <18 years old.

Defining pelvic pain worsening/persistence from pre-surgery to one-year post-surgery

Changes in pelvic pain symptoms from baseline to one-year post-surgery were calculated based on five acyclic pelvic pain variables and six dysmenorrhea variables assessing pain severity, frequency, and life interference assessed at the two timepoints (Supplemental Table 1).¹⁵ Among the 11 pain variables, change was dichotomized as pain symptoms worsened or persisted if any of the 11 pain variables worsened or if all symptoms persisted from baseline to Year 1 (i.e., >2 points in NRS score for severity and ≥ 1 category worsening for frequency and life interference). The changes were classified as pain symptoms improved if at least one of the pain symptoms improved and none of the other pain symptoms worsened from baseline to Year 1.

Covariates

Demographic and behavioral characteristics were collected at baseline, including age (years), self-reported race/ethnicity (white non-Hispanic, other race/ethnicity), analgesic use (never or regular use; regular use >1 dose per week for three months or longer), hormone use (no, yes), fasting status (no, yes), and rASRM stage (I/II, III/IV). Body mass index (BMI) was calculated from baseline height and weight and categorized as underweight (BMI <18.5 kg/m²), normal weight (BMI 18.5–24.9 kg/m²), overweight (BMI 25–29.9 kg/m²), or obese (BMI ≥30 kg/m²) according to World Health Organization criteria for those ≥ 20 years old.²⁰ For participants <20 years old, age specific BMI Z-score was calculated and categorized as underweight (Z-score ≤ −2), normal weight (>−2 to <1), overweight (1–2), or obese (>2).²¹ Hormone use and fasting status were obtained from the biospecimen questionnaire and reflect the status at time of blood draw. rASRM stage was assessed at laparoscopic surgery.

SomaScan Proteomics Assay

The 1.3k SomaScan Assay Kit was used to quantify relative protein levels in plasma (50μl)following the manufacturer’s standard protocol (SomaLogic; Boulder, CO), as described previously.^22–24 Simultaneous measurement of 1,305 proteins was achieved using highly selective single-stranded Slow Off-rate Modified DNA Aptamers (SOMAmer), which was demonstrated to have high reproducibility and within-person stability over time.²⁵ Most proteins included on this panel were circulating proteins involved in inflammatory/immune response, angiogenesis, and extracellular matrix. In parallel with the study participants’ plasma samples, we ran five pooled human plasma controls and one no-protein buffer control. Several hybridization spike-in controls were used to adjust for between sample variability. We conducted quality control, calibration, and normalization of the data as previously described.²⁶ The 30 quality control (QC) samples which were blinded to the laboratory and were randomly distributed among the participants’ samples showed that 98% of proteins had intra-batch coefficient of variation (CVs) <25%, 99% had inter-batch CV <25%; there were no missing protein values.

Inclusion and exclusion

We enrolled 785 participants with surgically-confirmed endometriosis.⁶ Of the 785 participants with surgically-confirmed endometriosis, we excluded those who did not complete the baseline questionnaire (N=156), did not have a surgery at baseline (N=172), those who had an endometrioma(s) or deep lesion(s) (N=9), did not provide blood at baseline (N=180) or blood not collected within 90 days prior to the surgery (N=13), those whose blood was drawn more than 6 months from questionnaire completion (N=19) and those missing data on dysmenorrhea or acyclic pelvic pain (N=29). We further excluded those who did not complete the Year 1 questionnaire (N=54). Among the remaining 153 cases, we included those with post-surgical blood samples (N=89, collected 5 weeks to 6 months post-surgery) and randomly selected 53 additional cases, for a total of 142 participants with endometriosis (Supplemental Figure 1). There were only 20 (2.5%) participants with either endometrioma or deep endometriosis in the A2A study. The vast majority had superficial peritoneal lesions only, which is consistent with clinical presentation of endometriosis diagnosed in adolescents and young adults.^{6, 27, 28} We excluded patients with deep endometriosis a priori, because with such a small number of patients with anything other than superficial peritoneal only presentation, it was not possible in this population to compare and contrast the influence of different endometriosis macro phenotypes on the proteins predictive of persistent pelvic pain. At the sample size for this analysis, our power analysis estimate for type-I error at 0.05 the power estimate was 0.85 to detect an odd ratio of at least 1.7. When type-I error was set at 0.000038, the power estimate was only 0.13. We note that from our main results, most of the estimated odds ratio estimates were about 1.6 or above.

Statistical analysis

Multivariable logistic regression models adjusting for age at blood draw (years), BMI (underweight or normal, overweight or obese), hormone use at blood draw (no, yes), and fasting at blood draw (no, yes) were used to calculate odds ratios (ORs) and 95% confidence intervals (CIs) per one standard deviation (SD) increase in protein levels. False discovery rate (FDR)²⁹ was used due to the large number of hypotheses being tested, each of which was based on the 1,305 SomaScan proteins. FDR-adjusted p-values were calculated to account for multiple testing. Among endometriosis patients with blood collected at two timepoints (pre-surgery and 5 weeks to 6 months post-surgery), we created quartiles based on relative percent change [i.e., 100*(post-surgery − pre-surgery) / pre-surgery] in protein levels at two timepoints for each protein and evaluated the association between the relative percent change in protein levels at two timepoints and the odds (risk) of worsening/persistent post-surgical pelvic pain at one-year follow-up.

Ingenuity Pathway Analysis (IPA) and functional category analysis of all dysregulated proteins with a p-value<0.05 from the logistic regression analysis (QIAGEN, Redwood City, CA) was used to identify biological pathways, as previously described. ^{22, 23, 30} The STRING database version 11.5 was used to identify protein-protein functional and physical interactions. These results were displayed as an interactive network³¹, with interactions considered with a STRING confidence score of ≥0.7 (out of 1.0) garnered from the “experimental” and “databases” categories. We removed proteins without associations to other proteins in the network. A k-means clustering algorithm was performed to select connected proteins (k-means/number of clusters=6). We conducted a manually curated evaluation of enriched KEGG pathway, Gene Ontology (GO), Reactome, STRING local network clusters terms, and PubMed literature search to assign functional description of the identified clusters.

All statistical analyses were performed using SAS/STAT version 9.4 (SAS Institute Inc., Cary, NC); systems biology analyses were performed using the Ingenuity Pathways Knowledge Base (Qiagen, Redwood City, CA).

RESULTS

Of the 142 participants with endometriosis, 76 (54%) reported their pelvic pain worsened or persisted (51 worsened and 25 persisted) and 66 (46%) reported their pelvic pain improved from pre-surgery (baseline) to one year follow-up (Year 1) (Table 1). Median age at blood draw was 17 years in both groups and most (>88%) were white race and on hormonal medications at time of blood draw. As expected, the majority (97%) were diagnosed with rASRM stage I or II. Of the 142 patients, 105 (74%) used hormones at one year follow-up and reported having no periods.

Table 1.

Baseline characteristics of 142 endometriosis patients in the A2A study.

Characteristics	Pain worsened or persisted (N=76)	Pain improved (N=66)
Age at blood draw, years, median (IQR)	17 (15, 19)	17 (15, 19)
White race, n(%)	67 (88)	61 (92)
Body mass index1, n(%)
Normal	48 (63)	44 (67)
Overweight	22 (29)	15 (23)
Obese	6 (8)	7 (11)
Taking hormonal medications at blood draw	67 (88)	60 (91)
Fasting status
Fasting	21 (28)	21 (32)
Non-fasting/unknown	55 (72)	45 (68)
Regular analgesic use2	44 (59)	32 (51)
rASRM stage
Stage I/II	74 (97)	64 (97)
Stage III/IV	2 (3)	2 (3)

Abbreviations: A2A: The Women’s Health Study: From Adolescence to Adulthood

¹For women aged ≥20 years: underweight (BMI < 18.5 kg/m2), normal weight (BMI 18.5–24.9 kg/m2), overweight (BMI 25–29.9 kg/m2), or obese (BMI ≥ 30 kg/m2) according to World Health Organization criteria; For those <20 years, the age- and gender-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), or obese (Z-score > 2).

²Regular use of analgesic medications defined as use at least once a week for a period of three months or longer. Analgesic medications include aspirin, acetaminophen, nonsteroidal anti-inflammatory drugs, and narcotics.

Examining individual proteins, we identified 83 proteins significantly associated with worsening/persistent pelvic pain one-year post-surgery (unadjusted p<0.05; Table 2, Supplemental Table 2). In pre-surgical bloods, those with higher plasma levels of proteins suggesting activation of immune response [i.e., elevated CD63 (OR=2.98, 95% CI=1.44–6.19), CD47 (OR=2.68, 95%CI=1.28–5.61), and macrophage migration inhibitory factor (MIF) (OR=3.99, 95%CI=1.33–12.03)] had higher likelihood of experiencing worsening/persistent pelvic pain. Conversely, those with higher levels of parathyroid hormone (PTH) (OR=0.55, 95% CI: 0.37–0.80) and proteins that promote angiogenesis [e.g., SHH (OR=0.55, 95%CI=0.36–0.84), stromal cell-derived factor 1 (CXCL12) (OC=0.62, 95%CI=0.43–0.89)], and vascular endothelial growth factor A (VEGFA) (OR=0.61, 95%CI=0.43–0.89) were associated with lower likelihood of worsening/persistent pelvic pain one-year after surgery. When we examined the protein-protein interactions, several clusters of proteins were identified including proteins involved in angiogenesis, cell migration, activation of leukocytes, and neuropeptides (Figure 1). Consistent with these findings, pathway analysis revealed that those with worsening/persistent post-surgical pelvic pain had upregulation of pathways related to cell migration and downregulation of pathways related to angiogenesis compared to those with improved pain (Figure 2, Supplemental Table 3).

Table 2.

Individual proteins at baseline associated with worsening or persistent pelvic pain one-year after surgery

Top 15 proteins positively associated with worsening/persistent pain
Protein	Entrez Gene Symbol	Odds Ratio (95%CI)1	p-value
CD63 antigen	CD63	2.98 (1.44–6.19)	0.003
Leukocyte surface antigen CD47	CD47	2.68 (1.28–5.61)	0.009
N-acetyl-D-glucosamine kinase	NAGK	2.00 (1.19–3.37)	0.009
Seizure 6-like protein 2	SEZ6L2	1.94 (1.17–3.20)	0.01
15-hydroxyprostaglandin dehydrogenase [NAD(+)]	HPGD	1.63 (1.11–2.38)	0.012
Angiopoietin-4	ANGPT4	1.67 (1.11–2.50)	0.013
Dickkopf-like protein 1	DKKL1	2.46 (1.21–5.01)	0.013
Macrophage migration inhibitory factor	MIF	3.99 (1.33–12.03)	0.014
Proteasome activator complex subunit 1	PSME1	1.83 (1.10–3.04)	0.019
Elongation factor 1-beta	EEF1B2	1.88 (1.11–3.20)	0.02
14-3-3 protein theta	YWHAQ	1.95 (1.11–3.43)	0.02
Signal transducer and activator of transcription 6	STAT6	1.54 (1.07–2.21)	0.02
Cytokine receptor-like factor 2	CRLF2	4.44 (1.25–15.77)	0.02
cGMP-dependent 3’,5’-cyclic phosphodiesterase	PDE2A	1.90 (1.09–3.28)	0.02
Fibronectin Fragment 3	FN1	1.63 (1.07–2.48)	0.02
Top 15 proteins negatively associated with worsening/persistent pain
Protein	Entrez Gene Symbol	Odds Ratio (95%CI)a	Unadjusted p-value
Parathyroid hormone	PTH	0.55 (0.37–0.80)	0.002
WAP, Kazal, immunoglobulin, Kunitz and NTR domain-containing protein 2	WFIKKN2	0.57 (0.39–0.84)	0.004
Ciliary neurotrophic factor receptor subunit alpha	CNTFR	0.59 (0.41–0.85)	0.005
Sonic hedgehog protein	SHH	0.55 (0.36–0.84)	0.006
Ectodysplasin-A, secreted form	EDA	0.24 (0.09–0.66)	0.006
Lactadherin	MFGE8	0.61 (0.42–0.88)	0.007
Mediator of RNA polymerase II transcription subunit 1	MED1	0.61 (0.43–0.88)	0.008
Netrin receptor UNC5D	UNC5D	0.57 (0.38–0.87)	0.009
Stromal cell-derived factor 1	CXCL12	0.62 (0.43–0.89)	0.009
Agouti-related protein	AGRP	0.62 (0.43–0.89)	0.01
Vascular endothelial growth factor A	VEGFA	0.61 (0.43–0.89)	0.01
Wnt inhibitory factor 1	WIF1	0.60 (0.41–0.88)	0.01
Plasma serine protease inhibitor	SERPINA5	0.49 (0.28–0.85)	0.011
Ephrin-A4	EFNA4	0.65 (0.46–0.92)	0.016
Follistatin-related protein 3	FSTL3	0.64 (0.44–0.92)	0.017

FDR-adjusted p-values were non-statistically significant (>0.05) for all proteins.

¹Adjusted for age, BMI, fasting status, hormone use at blood draw.

Figure 1. Protein-protein interaction clusters and relevant pathways associated with worsening/persistent pelvic pain at Year 1 using STRING analysis (k-means = 6 clusters indicated by node color).

Protein-protein interaction network was created based on the 83 proteins that were significantly associated with worsening/persistent pelvic pain at Year 1 with unadjusted p-value <0.05. Related functional categories are labeled based on proteins with their reported functional involvement in the pathways of angiogenesis (red node), activation of leukocytes (green node), cell migration (light-green node), and neuropeptides (blue node). Solid line represents within-cluster, dashed gray line represents between-cluster interactions. Line thickness indicates strength of data support.

Figure 2. Biological pathways associated with risk of worsening/persistent pelvic pain one-year after surgery (p-value <1.3×10⁻¹²).

Biological pathways associated with risk of worsening/persistent pelvic pain are grouped by the direction of association (i.e., activation z-score) and presented within group ordered by p-value. Upregulated pathways are denoted by red bubbles and downregulated pathways are denoted by blue bubbles. Of the top 20 statistically significant pathways, three pathways (morphology of cardiovascular system, morphology of vasculature, and morphology of body cavity pathways) were removed due to activation z-score of 0.0.

Next, we examined the association between change in protein levels in blood samples collected before and after surgery and risk of worsening/persistent pelvic pain one-year after surgery in 89 with endometriosis. When comparing the lowest to highest quartile of the relative percent change in protein levels from pre- to post-surgery (or when comparing those whose protein levels decreased from pre- to post-surgery to those whose protein levels increased from pre- to post-surgery), there were 12 nominally significantly proteins associated with risk of worsening/persistent pelvic pain one-year after surgery, with 9 being associated with increased risk and 3 associated with decreased risk comparing extreme quartiles (Table 3). Of these, 9 proteins had significant linear trend (p-trend across 4 quartiles <0.05). Interestingly, only SHH was significantly associated with risk of worsening/persistent post-surgical pain (one year after surgery) in both pre-surgical blood and change from pre- to post-surgical blood. Compared to those who had ≤ −35.4% decreasing change in blood levels of SHH from pre- to post-surgery, those who had an increasing change of >21.6% had about four times greater risk of having post-surgical worsening/persistent pain at Year 1 (OR=3.86, 1.04–14.33; p-trend 0.03) (Figure 3). When comparing to those who had minimal change in SHH levels from pre- to post-surgery (percent change ranging from −10.8% to 21.6% from pre- to post-surgery), those who had an increasing change of >21.6% had about three times greater risk of having post-surgical worsening/persistent pain at Year 1 (OR=3.22, 95%CI=0.87–11.93).

Table 3.

Risk of worsening/persistent pelvic pain one-year after surgery with increasing protein levels in quartiles

		Odds ratio (95%CI) of having worsening/persistent pelvic pain one-year after surgery by increasing change in proteins levels from pre- to post-surgery1
Protein	Entrez Gene Symbol	Quartile 1	Quartile 2	Quartile 3	Quartile 4	p-trend
Interleukin-17 receptor B	IL17RB	ref	2.73 (0.80–9.31)	1.88 (0.54–6.58)	4.74 (1.32–04.04)	0.03
Inter-alpha-trypsin inhibitor heavy chain H4	ITIH4	ref	1.46 (0.40–5.33)	1.04 (0.29–3.71)	4.86 (1.31–17.98)	0.03
Cathepsin F	CTSF	ref	0.58 (0.17–2.00)	1.22 (0.34–4.43)	0.21 (0.06–0.80)	0.08
S-formylglutathione hydrolase	ESD	ref	0.66 (0.19–2.30)	0.56 (0.16–1.90)	0.23 (0.06–0.81)	0.01
Protein Wnt-7a	WNT7A	ref	1.27 (0.38–4.23)	2.35 (0.69–7.97)	4.09 (1.12–14.95)	0.02
Trypsin-3	PRSS3	ref	1.49 (0.43–5.11)	1.36 (0.40–4.60)	3.86 (1.08–13.76)	0.04
Sonic hedgehog protein	SHH	ref	0.98 (0.30–3.23)	1.20 (0.36–3.95)	3.86 (1.04–14.33)	0.03
Persulfide dioxygenase ETHE1, mitochondrial	ETHE1	ref	0.90 (0.27–3.03)	1.90 (0.57–6.32)	3.73 (1.04–13.43)	0.03
Macrophage-capping protein	CAPG	ref	0.43 (0.12–1.49)	0.81 (0.23–2.89)	0.28 (0.08–0.97)	0.08
Leucine-rich repeat transmembrane neuronal protein 1	LRRTM1	ref	1.19 (0.34–4.12)	4.13 (1.16–14.70)	3.57 (1.03–12.32)	0.01
Glial fibrillary acidic protein	GFAP	ref	2.52 (0.73–8.70)	3.01 (0.88–10.38)	3.60 (1.01–12.86)	0.04
Thioredoxin domain-containing protein 12	TXNDC12	ref	2.28 (0.66–7.91)	1.68 (0.50–5.63)	3.56 (1.00–12.63)	0.09

¹25%tile, 50%tile, 75%tile cutoffs for increase in protein levels from pre-to post-surgery: IL17RB (−16.7%, −1.0%, 19.2%), ITIH4 (−6.2%, −0.1%, 7.4%), CTSF (−19.8%, −7.9%, 11.0%), ESD (−35.9%, −17.8%, 11.7%), WNT7A (−11.2%, 1.1%, 14.9%), PRSS3 (−6.1%, 0.8%, 12.0%), SHH (−35.4%, −1.08%, 21.6%), ETHE1 (−13.1%, −5.7%, 14.3%), CAPG (−11.6%, −1.7%, 13.0%), LRRTM1 (−8.9%, 0.8%, 14.8%), GFAP (−7.1%, 0.2%, 7.2%), TXNDC12 (−5.2%, 0.7%, 7.2%)

Figure 3. Relative percent change in Sonic hedgehog (SHH) protein levels across pre- and post-surgical blood collections and its association with subsequent risk of having persistent/worsening pain at 1-year post-surgery.

The table is showing the odds ratios (95% CIs) of having worsening/persistent pelvic pain one-year after surgery by increasing change in SHH levels in quartiles from pre- to post-surgery. The figure below is the plotting the change in SHH levels from pre- and post-surgery in each quartiles.

COMMENT

Principal Findings

Among adolescents and young adults with endometriosis, we discovered dysregulation of plasma proteins and biological pathways related to immune response, cell migration, and angiogenesis in pre-surgical bloods being associated with risk of worsening/persistent pelvic pain at one-year follow-up, providing evidence that there are differences in the systemic milieu prior to surgery indicative of response to surgical therapy of endometriosis. Furthermore, we examined the association between change in blood proteins levels from pre- to post-surgery with subsequent risk of having worsening/persistent pelvic pain at one-year follow-up and identified SHH as being significant in both analyses.

Results in the Context of What is Known

Emerging evidence supports that some women with endometriosis have pain central sensitization, therefore, conventional treatment that targets the endometriotic lesion fails to alleviate pain.³² To our knowledge, this is the first analysis that have comprehensively investigated pre-surgical plasma to discover risk biomarkers that are predictive of postoperative pain among endometriosis patients. Our results identified potential candidate biomarkers founding the basis for further investigation to develop blood-based biomarker signatures that will inform treatment response and provide biological support that inflammatory/immune signaling contributes to persistent pain after surgical treatment for endometriosis.

Elevation of several immune activation proteins in pre-surgery bloods were associated with increased risk of worsening/persistent post-surgical pain. CD63 is critical in leukocyte recruitment and an initiating inflammatory response.^{33, 34} CD47 is involved in transendothelial migration of neutrophils, monocytes, and dendritic cells, which are also important processes for inflammatory response.³⁵ MIF is a pro-inflammatory cytokine involved in regulation of inflammatory responses through various functions including leukocyte recruitment and immune response with a previously described association with both endometriosis and pain.^{36, 37, 38} Moreover, MIF is a neuroendocrine modulator of inflammation and neuropathic pain and has been demonstrated to be essential for the development of neuropathic and inflammatory pain.^{39, 40} For instance, MIF levels are chronically and acutely elevated in patients with spinal cord injury, resulting in hyperactivation of nociceptors and neuropathic pain.⁴¹ These results together with the pathway analysis results showing upregulation of cell migration pathways and leukopoiesis suggest systemic immune dysfunction and promotion of neuropathic pain may play a role in the emergence and persistence of chronic pain.^{42, 43}

In contrast, PTH, which was the top protein negatively associated with risk of worsening/persistent post-surgical pain, increases calcium levels in blood.⁴⁴ Calcium is involved in neuronal plasticity,⁴⁵ and plasmalemmal calcium signaling is involved in pain pathophysiology.⁴⁶ Calcium signaling is also known to increase production of reactive oxygen species and ATP by mitochondria, and therefore may serve as regulator between these molecules and development of nociceptor sensitization.⁴⁷ Nociceptors are sensory neurons that respond to potentially tissue-damaging stimuli. Nociceptor sensitization causes primary hyperalgesia, or increased pain produced by stimulation at the site of tissue insult. This is important because nociceptor sensitization is the trigger for initiation of an increase in excitability of central neurons in the nociceptive pathway or central sensitization.

Surprisingly, several proteins and pathways related to angiogenesis were downregulated in pre-surgical bloods of endometriosis patients with worsening/persistent post-surgical pain. Although studies suggest angiogenesis in the central nervous system is critical for developing central sensitization,^48–50 diabetic neuropathy results due to destruction of vasculature supplying the peripheral nerves.⁵¹ A similar mechanism may play a role in endometriosis. Alternatively, those with angiogenic response may be those with more locally active endometriosis, as angiogenesis is required for persistence of endometriotic lesions.⁵² If pain is mostly due to the local endometriotic lesions, these patients may more benefit from surgical intervention, leading to less likelihood of suffering from worsening/persistent post-surgical pain.

Interestingly, the increase in SHH levels from pre- to post-surgery was associated with greater risk of post-surgical worsening/persistent pain at Year 1 and is consistent with studies reporting that the SHH signaling pathway is dysregulated in various chronic pain conditions, including cancer pain and neuropathic pain.⁵³ The SHH signaling pathway activation has been reported to play a role in developing nociceptor sensitization.⁵⁴ Another study has reported that SHH signaling activation contributed to chronic post-thoracotomy pain development via activating BDNF pathways in vivo.⁵⁵ These results suggest that women with endometriosis who have increased SHH levels post-surgery may be developing nociceptor sensitization, leading to central sensitization and worsening/persistent pain one-year after their endometriosis surgery.

Clinical and research Implications

Pain is an unpleasant sensory and emotional sensation related with or similar to actual or potential tissue injury.⁵⁶ Current treatment strategies focus on the endometriosis lesions and include surgical therapies aimed at excising the lesions or medical therapies that hormonally suppress them.⁷ However, at least 25% of afflicted women remain symptomatic and suffer persistent pain despite using these treatments,^{1, 57} leading to decreased quality of life and altered life course trajectories.⁵⁸ Furthermore, women with endometriosis are at four times greater risk of chronic opioid use, dependence, and overdose compared to women without endometriosis.⁵⁹ In fact, adolescents and young adult endometriosis patients (age<25 years) have a higher risk of becoming chronic opioid users compared to adult endometriosis patients.⁵⁹ Therefore, optimal pain management in endometriosis patients, especially in adolescence, is critical and will have significant positive impact on clinical outcomes of endometriosis.

Our prospective study, comprehensively investigating the plasma proteome in relation to worsening/persistent post-surgical pain, is a critical step forward in developing a non-invasive tool that will inform surgical treatment response in these young endometriosis patients and could personalize treatment for endometriosis-associated pain. After replicating our findings in independent datasets, the ideal clinical assay should be based on multi-plex ELISA including <20 biomarkers, which would be more cost effective to predict post-surgical pelvic pain outcome. Furthermore, our findings together with mechanistic studies provide insight into the underlying biological pathways that may be involved in transitioning from acute to chronic pain and identify new therapeutic targets for persistent pain management. Future research is needed to validate our findings in independent datasets and mechanistic studies to elucidate the function of the identified proteins in development of post-surgical worsening/persistent pain in adolescents and young adults with endometriosis.

Strengths and Limitations

There are several strengths to this study. First, we applied a state-of-the art proteomics technology with high reproducibility, and comprehensively evaluated the plasma proteome to identify proteins associated with increased risk of worsening/persistent post-surgical pain in adolescents and young adults with endometriosis. Second, we focused on our unique study population of adolescents and young adults with deep phenotyping including detailed assessment of pain symptoms and prospective longitudinal study design. Third, we investigated changes in blood protein levels from pre- to post-surgery using bloods collected at two timepoints. However, our study population consisted mainly of white race and therefore generalizability may be limited. While this study is the most comprehensive proteomics analysis assessing proteins related to worsening/persistent post-surgical pain in endometriosis to our knowledge, there may be additional proteins that we were not able to evaluate on the SomaScan 1.3k platform and therefore were unable to detect their association with risk of persistent post-surgical pain. Validation of these results in an independent dataset is needed, although we believe this is a first critical step towards biomarker discovery for tailoring personalized treatment for endometriosis. Given that none of the individual proteins were statistically significantly associated with persistent post-surgical pain after multiple testing correction (all FDR>0.05), caution is warranted in interpretation of these results. However, we believe that this study provides initial evidence that there are likely differences in the blood proteome at pre-surgery that are predictive of post-surgical pain outcome among women with endometriosis. While the timing of the first post-surgical blood collection reflects the real-world and typical range of timing of first clinic return visit following surgical discharge, it is wide. This may drive toward the null the prediction of persistent pain for markers that are more dynamic during this window. A step toward establishment of clinically translational protein markers predictive of persistent pain should include studies that quantify change in these markers from week to week post-surgery and better define the optimal window for blood sample collection. Our results are valid and applicable to those patients with superficial peritoneal lesions only – the most common endometriosis subphenotype at all ages.^{60, 61} However, these results are not generalizable to patients with endometriomas or deep endometriosis; replication studies in populations oversampled for those endometriosis presentations are warranted to confirm similar or differing proteomic predictive pathways. Future studies are needed in populations deliberately designed to compare and contrast the proteome-defined pathways associated with persistent pain in those with superficial peritoneal endometriosis only and in those with deep endometriosis. While the surgical technique did not differ among these participants, post-surgical treatment was not standardized. However, this would not alter the pre-surgical proteome prediction of post-surgical persistence in pain, given that the clinician could not have considered the protein expression status in their treatment recommendations. From an epidemiologic methodological perspective, rather post-surgical treatment could act as a mediator (not a confounder) of the pre-surgical protein milieu and prediction of pain persistence one year after surgery.⁶² This does not alter or invalidate the interpretation of the present results but rather could be explored in studies intentionally designed to quantify mediating factors.

Conclusions

Our findings suggest that there are differences in the plasma proteome that could inform response to surgical treatment in endometriosis and provide novel biological evidence that inflammatory/immune signaling is playing a role in endometriosis patients developing worsening/persistent post-surgical pain. We observed that immune dysregulation prior to surgery and increase in blood SHH levels from pre- to post-surgery may be associated with increased risk of post-surgical worsening/persistent pain in adolescents and young adults with endometriosis. Further research is needed to validate in independent cohorts whether these proteomic markers are predictive of surgical treatment response in endometriosis patients.

Supplementary Material

1

Supplemental Figure 1. Overview of study design.

AJOG at a Glance:

A. Why was this study conducted?

To identify plasma proteins indicating pathophysiological mechanisms associated with persistent pelvic pain following laparoscopic surgery in adolescents and young adults with endometriosis using a multiplex aptamer-based proteomics biomarker discovery platform.

B. What are the key findings?

Among adolescents and young adults with endometriosis, we discovered plasma proteins and biological pathways dysregulated in pre-surgical blood samples that were associated with an increased risk of worsening or persistent pelvic pain at one-year follow-up.

C. What does this study add to what is already known?

These results provide evidence that there may be differences in the systemic milieu that could predict response to surgical treatment of endometriosis.