Further evidence that peritraumatic 17β-estradiol levels influence chronic posttraumatic pain outcomes in women, data from both humans and animals

Introduction

Most individuals experience at least one traumatic stress exposure (TSE) in their lifetime.[52] While recovery following TSEs is common, a substantial subset develop enduring adverse sequelae, including chronic posttraumatic pain (CPTP).[8; 12; 16; 59; 66; 68; 69; 97] Similar to other chronic pain conditions, CPTP can be regional or widespread and is highly morbid, causing lost workdays, increased medical costs, and decreased mental and physical health.[37; 72; 75] CPTP is often co-occurring with neuropsychiatric outcomes of TSE like posttraumatic stress and depressive symptoms, and opioid interventions can lead to abuse or misuse.[7; 10–12; 59; 63; 77; 78; 96]

While all individuals are susceptible to CPTP, women are disproportionately burdened.[61; 101] Women with chronic pain often report increased severity, bodily distribution, and duration of pain compared to men.[14; 70; 92] Unfortunately, biological mechanisms responsible for the increased pain burden in women are poorly understood, precluding the development of novel interventions that decrease sex-associated pain disparities and improve treatment for all individuals.[82; 83]

Sex hormones are a potential driver of biological mechanisms leading to increased female vulnerability to CPTP[44], and modulation of sex hormones serves as a candidate therapeutic approach to reducing CPTP. The influence of sex hormones on various chronic pain conditions has been studied extensively, with a particular focus on the role of 17β-estradiol (E2).[2; 24; 25; 36; 44] E2 is the primary steroid sex hormone in reproductive-aged women and has been shown to have both protective and pathogenic effects on pain processes in humans and animals.[3; 27; 28; 36; 44; 45; 60; 95] Existing literature suggests that the effect of E2 on pain depends upon multiple factors including biological sex, pain type (e.g. widespread pain vs neuropathic pain), timing of E2 levels relative to injury, and other factors. Therefore, further research is warranted to clarify the relationship between E2 and pain in specific populations and animal models.

A previous study from our group showed that higher circulating E2 levels at the time of TSE were associated with lower long-term CPTP symptoms among women TSE survivors enrolled in three independent emergency department (ED)-based longitudinal studies with various TSEs (motor vehicle collision, sexual assault, and major thermal burn injury).[60] In the present study, we aimed to further understand the relationship between circulating E2 levels and CPTP in women and men using participant data and samples from a fourth longitudinal cohort of TSE survivors, the AURORA study.[67] The AURORA study is an ED-based sister cohort to the three cohorts assessed in our previous analyses.[60] However, in AURORA we collected blood samples at three time points following TSE (ED, 2 weeks, and 6 months) vs just one timepoint (ED) in previous studies.[60]

Building from previous results, our primary analyses investigated whether high circulating levels of E2 in the early aftermath of TSE protects against CPTP. In secondary analyses, we examined whether E2 levels were associated with pain at three time points (ED, 2 weeks, and 6 months) cross-sectionally. Further, we assessed the relationship between E2 levels at time of TSE and mechanical hypersensitivity in intact female rats and also performed animal behavioral studies to investigate whether exogenous administration of E2 in ovariectomized (OVX) female and intact male rats reduces mechanical hypersensitivity.

Methods

Human cohort studies and analyses

Study design and participants.

The Advancing Understanding of RecOvery afteR traumA (AURORA) study is a longitudinal cohort that was designed to improve the prevention, diagnosis, and treatment of adverse posttraumatic neuropsychiatric sequelae following trauma.[67] Full eligibility criteria and study procedures have been previously described. Briefly, participants aged 18–75 years who presented to the ED within 72 hours of TSE at one of 23 participating ED sites were screened for study eligibility, enrolled in the ED, and followed serially for one year.[67] Data collected over the course of follow-up and relevant to the current manuscript includes self-report questionnaires and biological samples. The study was approved by the Institutional Review Boards of all participating institutions, and all study participants provided written informed consent.

In total, 389 participants were included in the current analyses. These participants are a subset of the overall AURORA cohort (n=3,829) and were matched to the full cohort with respect to age, sex, race, trauma type and severity, and distribution of adverse outcomes. In total, we measured E2 levels in n=227 ED blood samples, n=155 2-week blood samples, and n=167 6-month blood samples. Most individuals had blood from >1 time point measured (i.e. we included all available blood samples at the time of data analyses, which included 549 total blood samples. These blood samples were collected from a total of 389 participants, thus many participants contributed blood at more than one timepoint). All participants assigned male and female sex at birth identified as men and women, respectively, so a binary gender categorization was used throughout all analyses.

Pain assessments and outcome definition.

Overall musculoskeletal pain severity was assessed in the ED within 72 hours of TSE and at four follow-up time points (2 weeks, 8 weeks, 3 months, and 6 months). Participants were asked to rate their pain severity in the ED, over the past two weeks (at the 2-week time point), or in the past thirty days (at the 8-week, 3-month, and 6-month time points) via the validated 0–10 numeric rating scale (NRS: 0 (no pain) to 10 (maximum possible pain)).[33; 35; 55](1–3)

Blood plasma collection.

Blood samples were collected into 10-ml Vacutainer EDTA tubes (Beckton Dickinson, Franklin Lakes, NJ, Catalog #366643) at baseline (in the ED within 72 hours of TSE), 2-weeks post-TSE, and 6-months post-TSE. Within 30 minutes of collection, these tubes were centrifuged (1,500g for 15 minutes), the plasma layer was removed and aliquoted (250μl per aliquot), and the individual aliquots were frozen at −20°C to −80°C at participating study sites. All samples were shipped in batches to the University of North Carolina at Chapel Hill BioSpecimen Processing Facility on dry ice and stored at −80°C until use.

Measurement of 17β-estradiol levels in AURORA participant samples.

Plasma samples were thawed on ice immediately before hormone analysis. E2 levels were measured in duplicates using the Ultrasensitive Estradiol Enzyme-linked Immunosorbent Assay (ALPCO, Salem, NH, Catalog #20-ESTHUU-E01) according to manufacturer’s specifications. Sample absorbance (450nm) was measured using a Synergy HTX Multi-mode Microplate Reader (BioTek Instruments, Santa Clara, CA). Plasma samples with initial E2 measurements greater than 210 pg/μl were diluted using the provided sample diluent, measured again, and diluted hormone concentrations were multiplied by the dilution ratio. E2 concentrations were calculated from a standard curve generated with 4-parameter logistic regression using standards with known concentration. Standards were included on each 96-well plate containing study samples, along with two controls of known concentration and two samples in duplicate that were measured across all plates to control for cross-plate variability. Intra-assay coefficients of variability were below suggested thresholds (CV <20%).

Human cohort statistical analyses.

Sociodemographic characteristics were summarized using standard descriptive statistics. E2 levels were determined to be non-normally distributed via the Shapiro-Wilk test[90]; therefore, a Box-Cox analysis[76] was performed and E2 levels were transformed by taking the square root of raw E2 values. Repeated measures mixed models were used to assess the relationship between peritraumatic E2 levels and CPTP. Separate models were run for women and men. To match our previous study[60], participant age, self-reported race/ethnicity, BMI, education level, baseline pain levels, array, and time since TSE were included in the models as covariates. Participants with outliers that were ±3 standard deviations from the mean for E2 and BMI were excluded (n=6, two women and four men). Marginal means corresponding to CPTP levels for each participant were derived from fully adjusted models and used for data visualization purposes. In such graphs, participant CPTP levels were divided into two equally-sized groups based on the median level of E2 (to create ‘high E2’ and ‘low E2’ groups). Pairwise t-tests were used to assess for statistically significant differences between mean levels of CPTP in high E2 and low E2 groups at each time point. P-values were adjusted for multiple comparisons using Bonferroni adjustment.[13] For analyses assessing whether the relationship between E2 and CPTP depended on BMI, the interaction term E2*BMI was included in the model. For stratified analyses in premenopausal and postmenopausal women, 51.4 years of age was used as the cut-off.[42]

Animal experiments

Animals:

Experiments were performed on gonadally intact adult female (total n=28) and male (total n=8) Sprague Dawley rats that were 8–13 weeks old (males were 290g–450g, females were 170g–280g; Charles River Laboratories, Raleigh, NC). In addition, experiments were also performed on ovariectomized (OVX) female Sprague Dawley rats (n=16) that were purchased with their ovaries removed at postnatal days 36–42 at Charles River Laboratories (Raleigh, NC). Rats were pair-housed under a 12-hour light/dark cycle in the Division of Laboratory Animal Medicine at The University of North Carolina at Chapel Hill. Food and water were provided ad libitum. All animals were acclimated to the housing facility for at least one week prior to the start of handling. Animal care and use conformed to NIH guidelines and was approved by the Institutional Animal Care and Use Committee at The University of North Carolina at Chapel Hill.

Single Prolonged Stress:

Human TSE was modeled in animals using the single prolonged stress (SPS) model. This model lacks a tissue injury component but has been shown across multiple laboratories to result in enduring hypersensitivity.[46; 61; 94; 100; 105] Details of the SPS protocol have been reported previously.[62; 104] Briefly, rats were exposed to serial stressors on one day as follows: restraint for 2 hours, forced swimming for 20 minutes, recovery for 15 minutes, then exposure to diethyl ether until general anesthesia was induced (generally < 7 minutes). Following SPS, rats were single-housed.[53] Modified from the original SPS model, animals did not undergo a 7-day quiescent period following SPS which allowed for daily monitoring of mechanical hypersensitivity starting 24 hours following SPS. Animals not exposed to SPS were brought to a procedure room, left undisturbed for the same length of time as SPS exposed animals, and then moved to individual housing.

von Frey testing:

Paw withdrawal thresholds (PWT) were used to assess mechanical hypersensitivity. PWTs were determined by applying touch test sensory probes, specifically von Frey monofilaments (Stoelting Co., Wood Dale, IL), at a right angle to the plantar surface of the left hind paw until the animal withdrew its paw. The up-down method described by Chaplan et al was used, starting with a monofilament of 8g force.[18]

Before the start of experimental testing, animals were acclimated to the testing environment for 30 minutes each day over the course of four days. Acclimation involved placing the animals in the testing chambers and applying a small filament to the paw such that no withdrawal response was elicited. Baseline PWTs were collected on two separate days following acclimation and the data was averaged to determine a baseline sensitivity for each animal. For the determination of PWT on each day of testing, the up-down method was performed five times with five-minute intervals between each test. These five technical replicates were averaged to find the 50% pain threshold (i.e. the force required to elicit a response 50% of the time). Animals were assigned to groups such that average baseline PWT and average baseline weight were balanced across groups. Mechanical hypersensitivity was assessed daily following SPS or in naïve animals for seven days.

17β-estradiol preparation and administration in animals.

1g E2 (Sigma-Aldrich, St. Louis, MO, Catalog #E2257) was weighed and dissolved in 1 ml of 100% ethanol. E2 was then diluted 50x with sesame oil to reach a working stock of 20μg/ml. E2 was administered at a concentration of 45μg/kg via subcutaneous injection. Vehicle-treated animals received the same volume of sesame oil alone. The dosage was selected based on previous studies showing the biological effects of E2 on neurobiological processes at this concentration and via this delivery route.[65; 73] E2 injections were administered starting either immediately after SPS (i.e. while still anesthetized by ether) or 72 hours after SPS (Figure 1). Injections were administered in the scruff of the neck by trained research assistants at indicated times following SPS and the treatment groups were blinded to the behavioral scientist performing von Frey testing. All animals received two total injections spaced 24 hours apart.

Figure 1.

Schematic of the study design for human (top) and animal (bottom) data collection and analyses. In humans, blood samples to assess circulating E2 levels were collected immediately following traumatic stress exposure (TSE) in the emergency department (ED), at 2 weeks, and at 6 months following TSE. Pain was assessed via the 0–10 numeric rating scale in the ED and at 2 weeks, 8 weeks, 3 months, and 6 months post-TSE. Associations between E2 levels and pain were measured cross-sectionally (red/gold/purple arrows, corresponding to association data analysis results presented in Tables 4/5/6) and longitudinally (blue/green arrows, corresponding to results Tables 2 and 3) in both women and men. In animals, male and female (intact and OVX) Sprague Dawley rats were assessed for baseline mechanical hypersensitivity (von Frey filaments) prior to single prolonged stress (SPS), and on D1–7 following SPS. E2 (or vehicle) was administered twice, starting either immediately after SPS (D0) and on D1 or on D3 and D4.

Animal serum collection and measurement of circulating 17β-estradiol levels.

Whole blood was collected from animals via tail vein clip at two timepoints prior to SPS (2 and 4 days prior) and at multiple timepoints following SPS: within 60 minutes of the first E2 injection, 24 hours following SPS (before the second E2 injection) and at days 3 and 5 following SPS, and serum was separated via centrifugation (3,000g for 15 minutes at 4°C) immediately following blood collection. Serum was separated into 60μl aliquots, and samples were frozen at −80°C until use. On the day of E2 quantitation, a single serum aliquot for each collection time point was thawed on ice and E2 levels were measured in duplicate using a 17β-estradiol ELISA kit (Abcam, Cambridge, UK, Catalog #ab108667) according to the manufacturer’s instructions. Absorbance measurement and standard curve calculations were conducted using the same instrument and methods as described for human samples. To confirm that SPS elicits a stress response, blood serum was collected from a separate set of rats (n=8) prior to SPS and 3 minutes following SPS. Serum corticosterone levels were measured using an ELISA (Invitrogen Cat# EIACORT) according to the manufacturer’s instructions.

Measurement of core body temperature of freely moving rats

Implanted data loggers (SubCue, Calgary, Alberta, Canada) were used to assess core body temperature in freely moving female OVX rats during the E2 administration experiments. Briefly, rats were anesthetized using 2–3% isoflurane. Sterilized implants (1.7cm × 0.7cm in size) were placed within the intra-peritoneal cavity via a small (2.5cm) ventral midline incision. The muscle layer was closed using simple interrupted sutures and the skin was closed using wound clips. All rats received one dose of meloxicam (2mg/kg) prior to surgery and then one dose of meloxicam (2mg/kg) 24 hours post-surgery. Wound clips were removed 7–10 days post-surgery. Temperatures were recorded on the implanted data loggers every 20 minutes starting 10 days post-surgery. At the end of the experiments, implants were collected from sacrificed animals, and the data was downloaded and analyzed using SubCue Analyzer software. Data was visualized using GraphPad Prism (Prism Software Corp. version 10.2.3).

Animal experiment statistical analyses.

Force of PWT data is represented normalized by log transformation per Weber’s law of nociception perception, which states that mechanical sensitivity is perceived on a logarithmic scale.[71] Statistical comparisons between treatment and vehicle groups were performed using RStudio servers (v1.4.1103). The normality assumption was assessed using the Shapiro-Wilk test for each combination of factor levels. Significant differences between groups were evaluated using two-way mixed analysis of variance (ANOVA) to assess group (E2 vs vehicle (OVX animals), or high E2 vs low E2 (naturally cycling animals))*time interactions that incorporate data from all time points assessed in the experiment, starting on the day of SPS until the last day that PWTs were collected. Of note, in animal behavioral data figures from OVX animal experiments, mechanical sensitivity data for intact female animals is shown for context, but these animals were not included in the statistical analyses since they were not included in the study hypotheses. Additionally, group effects at individual time points were compared using one-way ANOVAs or pairwise t-tests followed by Bonferroni post-hoc adjustments to account for multiple comparisons. Statistical significance for individual time point comparisons are indicated in figures using asterisks (p<0.05). All data are expressed as mean ± standard error of the mean (SEM).

Data accessibility

All human and animal data used in the current manuscript are available for download from https://doi.org/10.15139/S3/EHOK47.

Results

Study design

Translational studies from both humans and animals were used in the current set of analyses (Figure 1). In humans, blood plasma samples to assess circulating E2 levels were collected in the early aftermath of TSE (in the ED), 2 weeks, and 6 months following TSE. Quantified E2 levels were assessed for association with CPTP both cross-sectionally (i.e., E2 levels and pain were assessed at the same time point) or longitudinally (i.e., E2 levels were used to predict pain outcomes at later time points). In animals, mechanical hypersensitivity was assessed in female OVX and male animals that were administered E2 either immediately following or 3 days post-SPS.

Human study participants

A total of 389 AURORA participants were included in the current study. These participants were selected based on available E2 and CPTP outcome data. Baseline characteristics of these participants are shown in Table 1, with subgroups defined by the post-TSE time point at which E2 levels were measured (i.e. ED, 2-week, and 6-month). Most individuals were less than 40 years of age, non-Hispanic Black, experienced a motor vehicle collision, and had some post-secondary education. Mean circulating E2 levels within the sample were consistent with previously-published levels for normal healthy individuals in both women (100.4 ± 107.9 pg/ml)[4; 50; 86] and men (50.8 ± 58.8 pg/ml)[31; 103]. In the following analyses, women and men were studied separately due to the ~2-fold sex difference in mean E2 levels within the AURORA cohort (Table 1) and known sex differences in the role of E2 in posttraumatic outcome development.[23; 38; 60; 85; 99]

Table 1.

Characteristics of AURORA study participants included in analyses at each time point following traumatic stress exposure.

Characteristic	ED, Baseline (n=227)		2 weeks (n=155)		6 months (n=161)
	Women	Men	Women	Men	Women	Men
Participants, n	164	63	97	58	122	39
Age, years, mean (SD)	40.0 (14.2)	36.5 (14.5)	35.8 (12.9)	36.5 (14.0)	41.8 (14.2)	36.4 (14.9)
Education, n (%)
Less than high school graduate	16 (10)	8 (13)	3 (3)	2 (3)	15 (12)	5 (13)
High school graduate	31 (19)	20 (32)	26 (27)	14 (24)	19 (16)	14 (36)
Some college	73 (45)	25 (40)	38 (39)	31 (53)	53 (43)	14 (36)
College graduate	44 (27)	10 (16)	30 (31)	11 (19)	35 (29)	6 (15)
Ethnicity, n (%)
Non-Hispanic White	53 (32)	28 (44)	42 (43)	24 (41)	35 (29)	16 (41)
Non-Hispanic Black	105 (64)	34 (54)	55 (57)	34 (59)	84 (69)	22 (56)
Hispanic and Non-Hispanic other	6 (4)	1 (2)	0	0	3 (2)	1 (3)
BMI, mean (SD)	31.2 (7.9)	28.2 (6.4)	31 (7.2)	25.8 (14.7)	31.3 (7.7)	30.0 (8.9)
Pain severity in ED # , mean (SD)	6.5 (2.5)	6.2 (2.7)	6.5 (2.2)	6.3 (2.4)	6.5 (2.5)	6.7 (2.3)
Previous traumatic events ^ , mean (SD)	3.3 (2.4)	3.1 (2.8)	3.5 (2.4)	3.4 (3.3)	3.4 (2.5)	3.5 (2.8)
E2 level (pg/ml), mean (SD)	100.4 (107.9)	50.8 (58.8)	92.0 (98.4)	43.2 (51.6)	77.1 (95.0)	48.8 (67.0)

^{# -}Pain severity measured via the 0–10 numeric rating scale

^{^ -}Previous traumatic events assessed via the life events checklist

SD = standard deviation, ED = emergency department, BMI = body mass index, E2 = 17β-estradiol

Circulating peritraumatic ED E2 levels predicted longitudinal CPTP outcomes following TSE in women but not men.

The relationship between blood-plasma ED E2 levels and CPTP outcomes (measured at 8 weeks, 3 months, and 6 months following TSE) were assessed using statistical models and adjustors that matched our previous E2-focused study.[60] In women, we identified an inverse relationship between peritraumatic E2 levels and pain severity over a six-month period following TSE (Table 2 and Supplementary Table 1S; β = −0.137, p = 0.033). Women with high levels of E2 in the ED had lower levels of CPTP over the course of six months compared to women with low levels of E2 in the ED. In men, using the same statistical model and adjustors, no relationship between peritraumatic E2 levels and CPTP was identified (Table 2 and Supplementary Table 1S; β = 0.015, p = 0.902). These results were robust after adjustment for pain medications used in the ED, including the use of both opioids (30% of women and 27% of men) and non-steroidal anti-inflammatory drugs (NSAIDs) (48% of women and 43% of men) (E2 in women with opioid adjustment: β = −0.142, p = 0.033; E2 in women with NSAID adjustment: β = −0.136, p = 0.043).

Table 2.

Repeated measures regression analyses assessing the relationship between circulating 17β-estradiol levels at the time of trauma exposure (Emergency Department, ED) and chronic posttraumatic pain severity over the course of six months (WK8, M3, M6) following trauma exposure in multiethnic women and men (n=227).

	Women (n=164)			Men (n=63)
	β	S.E.	p value	β	S.E.	p value
Intercept	0.990	1.904	0.603	−10.455	6.694	0.121
Time	−0.260	0.071	<0.001	−0.280	0.122	0.023
17β-estradiol	−0.137	0.064	0.033	0.015	0.118	0.902
Age	0.019	0.017	0.283	0.054	0.028	0.056
BMI	0.022	0.029	0.459	0.047	0.067	0.481
Baseline pain	0.343	0.091	0.002	0.444	0.167	0.009

Chronic posttraumatic pain severity was measured via a 0–10 numeric rating scale (NRS). Participants’ education level, race/ethnicity, and the plate that the 17β-estradiol was run on were also included as a categorical variable in the model, but for simplicity (due to multiple categories), were not included in the table. Out of the covariates not included in this table, the only ones that were statistically significant were some assay plates, as shown in Supplementary Table 1S.

To better understand and visualize the relationship between peritraumatic E2 and longitudinal CPTP in women, we identified subgroups of high and low peritraumatic E2 based on a cutoff at the median E2 value of women participants (Supplementary Figure 1S). Women with high ED E2 levels had lower levels of CPTP over the full follow-up period (Figure 2) and at each individual time point assessed (8 weeks, 3 months, and 6 months following TSE; Supplementary Table 2S).

Figure 2.

Relationship between peritraumatic circulating levels of 17β-estradiol (E2) and chronic posttraumatic pain (CPTP) severity in multiethnic women (n=164) in the AURORA study. For visualization, data was split into two groups based on median E2 levels. Women with low E2 levels (mean E2 = 19.49pg/ml ± 14.04pg/ml) are represented by open circles and a dotted line; women with high E2 levels (mean E2 = 181.20pg/ml ± 99.94pg/ml) are represented by black circles and a solid line. CPTP levels were measured in the emergency department (t0), and at 8-weeks, 3-months, and 6-months following traumatic stress exposure using the 0–10 numeric rating scale (the y-axis is abbreviated to show detail). Statistical analyses were performed using repeated measures mixed models (including all follow-up time points) and for each time point individually. Asterisks indicate p<0.05.

In extension to these findings, because circulating E2 levels fluctuate substantially during and following perimenopause, and many studies focus exclusively on either pre- or post- menopausal women[15], we stratified our population of women based on presumed menopausal status (stratified at 51.4 years of age[41]; premenopausal: ≤51.4 years old, n = 122, mean circulating E2 = 117.19 ± 108.52 pg/mL and postmenopausal: >51.4 years old, n = 42, mean circulating E2 = 51.44 ± 90.69 pg/mL). In premenopausal women, an inverse relationship between peritraumatic E2 levels and CPTP was present (β = −0.175, p = 0.02, Supplementary Table 3S). However, among postmenopausal women, peritraumatic E2 levels were not associated with longitudinal CPTP outcomes (β = −0.010, p = 0.524).

Contrary to our previous study, no statistically significant interaction was identified between E2 and BMI in predicting longitudinal CPTP outcomes in men.

Our previous study showed that the relationship between peritraumatic E2 levels and longitudinal CPTP outcomes in men following TSE depends upon body mass index (BMI), such that obese men with higher E2 levels were at greater risk of developing CPTP.[60] To assess whether this finding generalizes to AURORA study participants, we utilized the same statistical model and adjustors as before to test whether an interaction between E2 and BMI predicts CPTP in men. In this analysis, we did not identify a statistically significant interaction between peritraumatic E2 levels and BMI in predicting CPTP (β = −0.018, p = 0.39; Supplementary Table 4S).

E2 levels measured 2 weeks following TSE did not predict CPTP outcomes in women or men.

After replicating ED E2 prediction of CPTP outcomes in women following TSE, we were curious whether this relationship was unique to ED E2 levels or whether E2 levels 2 weeks following TSE would also predict longitudinal CPTP outcomes. Therefore, using the same statistical models as for ED E2 levels, we assessed the relationship between 2-week E2 levels and CPTP. No statistically significant association was observed for either sex (Table 3, women: β = −0.022, p = 0.714; men: β = −0.192, p = 0.192).

Table 3.

Repeated measures regression analyses assessing the relationship between circulating 17β-estradiol levels 2 weeks after trauma exposure and chronic posttraumatic pain severity over the course of six months (WK8, M3, M6) following trauma exposure in multiethnic women and men (n = 152).

	Women (n=94)			Men (n=58)
	β	S.E.	p value	β	S.E.	p value
Intercept	5.442	2.990	0.070	−7.006	4.896	0.155
Time	−0.180	0.091	0.048	−0.244	0.118	0.040
17β-estradiol	−0.022	0.060	0.714	−0.192	0.146	0.192
Age	0.027	0.023	0.244	0.097	0.029	0.001
BMI	−0.003	0.041	0.948	−0.002	0.083	0.985
Baseline pain	0.412	0.134	0.002	0.230	0.191	0.232

Chronic posttraumatic pain severity was measured via a 0–10 numeric rating scale (NRS). Participants’ education level, race/ethnicity, and the plate that the 17β-estradiol was run on were also included as a categorical variable in the model, but for simplicity (due to multiple categories), were not included in the table and were not statistically significant aside from some assay plates.

E2 levels measured at the same time point as pain levels showed no statistically significant associations between E2 and pain in women.

Previous studies have shown that E2 levels can influence acute pain processing.[24] Therefore, we assessed whether E2 levels measured at any of the three time points in which blood was collected were associated with pain levels at that same time point. In such analyses (using statistical models and adjustors as described above), we did not observe statistically significant relationships between any of the models tested in women (i.e. baseline E2 levels and baseline pain (Table 4), 2-week E2 levels with 2-week pain (Table 5), and 6-month E2 levels with 6-month pain (Table 6)). However, we did observe a relationship between 2-week E2 levels and 2-week pain in men (Table 5; β = −0.338, p = 0.045), but this association did not survive adjustment for multiple hypothesis testing, and no statistically significant relationship for the other two time points in men (Tables 4 and 6).

Table 4.

The relationship between circulating 17β-estradiol levels at the baseline (ED) time point following traumatic stress exposure and ED/baseline pain levels in AURORA women and men (n=227)

	Women (n=164)			Men (n=63)
	β	S.E.	p value	β	S.E.	p value
Intercept	5.019	1.720	0.004	8.528	6.228	0.179
17β-estradiol	−0.012	0.060	0.840	0.022	0.119	0.853
Age	0.002	0.016	0.918	0.004	0.028	0.893
BMI	0.070	0.027	0.010	−0.008	0.068	0.912

Chronic posttraumatic pain severity was measured via a 0–10 numeric rating scale (NRS). Participants’ education level, race/ethnicity, and the plate that the 17β-estradiol was run on were also included as categorical variables in the model, but for simplicity (due to multiple categories), was not included in the table and were not statistically significant aside from some assay plates.

Table 5.

The relationship between circulating 17β-estradiol levels at the 2 week time point following traumatic stress exposure and 2 week pain levels in AURORA women and men (n=155)

	Women (n=97)			Men (n=58)
	β	S.E.	p value	β	S.E.	p value
Intercept	7.153	2.717	0.010	−1.424	5.167	0.784
17β-estradiol	−0.050	0.066	0.453	−0.338	0.163	0.045
Age	−0.015	0.022	0.487	0.071	0.098	0.436
BMI	−0.045	0.038	0.247	0.077	0.098	0.436
Baseline pain	0.411	0.124	0.001	0.443	0.211	0.042

Chronic posttraumatic pain severity was measured via a 0–10 numeric rating scale (NRS). Participants’ education level, race/ethnicity, and the plate that the 17β-estradiol was run on were also included as a categorical variable in the model, but for simplicity (due to multiple categories), were not included in the table and were not statistically significant aside from some assay plates.

Table 6.

The relationship between circulating 17β-estradiol levels at the 6 month time point following traumatic stress exposure and 6 month pain levels in AURORA women and men (n=167)

	Women (n=122)			Men (n=39)
	β	S.E.	p value	β	S.E.	p value
Intercept	1.535	3.386	0.651	−0.416	7.482	0.956
17β-estradiol	−0.115	0.090	0.206	0.190	0.357	0.602
Age	0.027	0.026	0.298	0.059	0.078	0.456
BMI	0.011	0.045	0.808	−0.076	0.112	0.507
Baseline pain	0.335	0.131	0.012	0.667	0.373	0.093

Chronic posttraumatic pain severity was measured via a 0–10 numeric rating scale (NRS). Participants’ education level, race/ethnicity, and the plate that the 17β-estradiol was run on were also included as a categorical variable in the model, but for simplicity (due to multiple categories), were not included in the table and were not statistically significant aside from some assay plates.

SPS elicits an acute stress response in Sprague Dawley rats

Following the above observations from human cohort data analyses, we next performed supportive studies in a rat model of TSE called the single prolonged stress model. Other labs have shown this model to be a reliable model of TSE by eliciting an acute stress response[62], and we and others have shown that SPS elicits enduring mechanical hypersensitivity[46; 61; 94; 100; 105]. To further confirm that SPS induces a stress response, we measured circulating corticosterone levels in the early period following stress exposure. Consistent with previous reports, we found that SPS substantially increases serum corticosterone levels 3-minutes following SPS (6.5x fold increase relative to SPS-unexposed controls, t₍₂₀₎=15.25, p<0.0001, Supplementary Figure 2S).

Administration of E2 immediately following SPS in female OVX rats causes a prolonged reduction in post-SPS hypersensitivity.

The above observational data from humans, along with our previously published human cohort data[60], suggested that exogenous administration of E2 following TSE could reduce CPTP in women with naturally low levels of E2. Therefore, to initially explore this possibility, we designed experiments in which we treated animals with experimentally induced low levels of E2 (i.e. OVX animals) with exogenous E2. However, before conducting this experiment, we first assessed whether high levels of E2 at the time of TSE (i.e. single prolonged stress, SPS) in naturally cycling female animals (similar to our human observational studies) were enough to reduce mechanical hypersensitivity in the days following SPS. In this experiment, E2 levels were measured in the collected serum samples (n=12) and animal PWT data was grouped based on a median split, resulting in high circulating E2 (83.8pg/ml) and low circulating E2 (29.2pg/ml) levels (Supplementary Figure 3S). We found that female rats with high E2 on the day of SPS had, on average, less mechanical hypersensitivity over the seven days following SPS compared to female rats with low E2 on the day of SPS (Figure 3; F(1,10)=3.225, p=0.103). However, this relationship was not statistically significant, indicating that high E2 levels in naturally cycling animals were not sufficient to fully reduce post-TSE mechanical hypersensitivity.

Figure 3.

Posttraumatic mechanical hypersensitivity in intact female rats with naturally low and naturally high circulating 17β-estradiol (E2) levels on the day of traumatic stress exposure (single prolonged stress, SPS). A) Intact normally cycling female Sprague Dawley rats (n=12) were grouped based on their naturally occurring levels of serum E2 on the day of single prolonged stress (SPS). Female animals in the low E2 group (n=7) had average circulating E2 levels of 29.2 pg/ml and female animals in the high E2 group (n=5) had average circulating E2 levels of 83.8 pg/mL. B) Paw withdrawal thresholds were measured via mechanical von Frey in both groups of animals starting 24 hours following SPS and every subsequent 24 hours, for a duration of seven days. Solid lines indicate daily group mean +/− SEM while dashed lines show the average change in PWT across days 1–7 following SPS. *p=0.103, two-way ANOVA.

We then assessed exogenous administration of E2 in female OVX experiemnts. In the first of these experiments, we administered two subcutaneous injections of E2 (45ug/kg), spaced 24 hours apart, starting immediately following SPS. In these animals, we observed a reversal in PWTs compared to vehicle treated animals, indicating reduced SPS-induced hypersensitivity (Figure 4a; F_1,6 = 12.56, p = 0.01). This reduced hypersensitivity persisted for the seven-day duration of the experiment. In the second experiment, we again administered two subcutaneous injections of E2 to OVX animals, but waited until 72 hours post-SPS to do the first injection. In this experiment, we saw no statistically significant differences in PWTs when compared to vehicle-treated animals (Figure 4b; F_1,6 = 0.17, p = 0.70). Finally, in the third experiment, we repeated the first experiment but in intact male animals. We observed no statistically significant difference in E2- vs vehicle-treated animals, with all animals developing post-SPS hypersensitivity (Figure 4c; F_1,6 =0.80, p = 0.40). In control experiments, to confirm that E2 administration led to an increase in circulating E2 levels, we performed ELISAs on serum collected from animals before and after E2 administration. In each assay, we detected a substantial increase in circulating E2 following injection (Supplementary Figure 3S). In a parallel control experiment, in a duplicate cohort to the first E2 administration experiment described above, we used implanted data loggers to record core body temperatures in female OVX animals before and after E2 (or vehicle) injection. In this experiment, consistent with known effects of E2 on core body temperature[19; 54], we observed a statistically significant decrease in core body temperature for at least two hours following each injection (1^st injection: t₅₆=6.726, p<0.0001; 2^nd injection: t₅₆=8.353, p<0.0001; Supplementary Figure 4S).

Figure 4.

Exogenous administration of 17β-estradiol (E2) in ovariectomized (OVX) Sprague Dawley female rats and intact male rats and examination of the effect of E2 on mechanical hypersensitivity following traumatic stress exposure. A) Female intact and OVX animals were exposed to traumatic stress (single prolonged stress, SPS) and OVX animals were administered E2 (45μg/kg) or vehicle via two subcutaneous injections, one injection immediately following SPS and the next one 24 hours later. Mechanical hypersensitivity was measured daily via mechanical von Frey and paw-withdrawal thresholds calculated via standard methods. B) Female OVX rats were administered E2 (45μg/kg) on days 3 and 4 following SPS (injections were separated by 24 hours and were administered at the same time of day as injections in panel A) and PWTs measured daily via mechanical von Frey. C) Male rats received two E2 injections at the same dose and time interval following SPS as female OVX animals in panel A. *p<0.05.

Discussion

In this translational study of analogous human and animal studies, we validated our previously published results[60] by showing that peritraumatic E2 levels in TSE survivors predict CPTP outcomes in women. Consistent with our previous results, higher circulating E2 levels in women in the immediate aftermath of TSE were associated with less CPTP over the subsequent six months. We extended these findings to show that E2 levels measured at 2-week and 6-month time points following TSE were not associated with future or concurrent CPTP levels in women. Parallel experiments in female rats showed that administration of E2 immediately following TSE protected against mechanical hypersensitivity in female OVX rats, while delayed administration of E2 72 hours after TSE had little effect on mechanical hypersensitivity. We did not observe similar protective effects of E2 on CPTP and mechanical hypersensitivity in men and male rats. Together, these results build on previous findings indicating that high peritraumatic E2 levels reduce long-term CPTP outcomes, but only in women and female OVX animals that were administered E2 exogenously.

The design of the AURORA study, in which we collected longitudinal blood samples 2 weeks and 6 months following TSE, enabled us to examine whether E2 is associated with reduced pain levels at time points past the early peritraumatic period. Based on our preliminary results presented here, we showed that, in contrast to the validated effect of E2 on long-term CPTP outcomes, E2 has little effect on reducing CPTP once it has already established. Thus, E2 levels in the peritraumatic time may be of clinical interest since this time period establishes a trajectory of future CPTP outcomes. These time-dependent effects are consistent with growing evidence indicating that potential interventions for CPTP and related neuropsychiatric outcomes are most effective if administered early following TSE.[6; 9; 17; 21; 58; 100; 106]

The mechanism through which E2 influences CPTP likely includes the effect that E2 has on the substantial neurologic, stress, and immune changes that occur in response to TSE, and the multiple mechanisms through which E2 has been proposed to influence chronic pain conditions. For example, previous research has indicated that E2 influences neurotransmission via modulation of serotonin, noradrenaline, and endogenous opioids.[47; 49; 91] In addition, E2 has been shown to have anti-inflammatory properties via its effect on cytokine and antibody production[29; 56]. (Of note, seemingly contradictorily, E2 has also been shown to have pro-inflammatory effects[64; 93]. However, such pro-inflammatory influence might help promote the resolution of stress and pain, as proposed by other groups[32; 81].) Further, E2 influences the musculoskeletal system by impacting collagen production, cartilage and bone growth, and muscle cell proliferation.[20; 44] Finally, E2 has been shown to affect the activity of the stress response, including the hypothalamic-pituitary adrenal (HPA) axis (and vice versa[5; 84]), leading to robust changes in the body’s response to stress.[1; 48; 80; 88] Further research is needed to fully understand whether these mechanisms and/or other mechanisms are driving the relationship between peritraumatic E2 and CPTP in women.

The current study validated the finding that E2 had little association with long-term CPTP in men and male animals. While we hypothesized little direct association between E2 and CPTP in men based on previous results, we anticipated observing protective effects of E2 when administering it to male animals immediately following TSE. However, E2 provided no protective effect in male animals. This result suggests that additional factors beyond E2 levels influence the effect of E2 on CPTP outcomes in men vs women. Such factors could include differences in E2 receptor expression levels across cell types between sexes,[57; 102; 108] sex-dependent pharmacokinetic properties of E2,[89; 107] and transcriptional patterns activated by E2 that differ across sex.[51; 61] These, as well as additional potential differences in other sex hormones (e.g. progesterone and testosterone) and other sex characteristics, should be explored further such that future therapeutic opportunities based on E2 treatment could be extended equitably to all genders.

While the current study did not assess whether E2 could serve as a preventative therapeutic in a human cohort population via a randomized clinical trial, it did provide indirect evidence via observational human and interventional animal data to suggest that administration of E2 to women in the early aftermath of TSE could reduce CPTP. Of note, the circulating E2 levels reached following E2 administration in animals peaked at high levels, and such levels might not be feasible to administer in humans. However, our human cohort data and previous research from other fields[34; 39; 40] suggests that exogenous administration of E2 in well-tolerated doses could raise plasma levels enough to substantially reduce adverse outcomes of TSE. One example of this previous research is a small study of sexual assault TSE survivors who were administered E2 in the form of an emergency contraceptive.[34] In this study, women who received the emergency contraceptive Ogestrel (containing synthetic E2) early after sexual assault were most protected against a neuropsychiatric outcome of TSE that commonly co-occurs with CPTP compared to women who received either Plan B (containing synthetic progesterone) or no emergency contraceptives.[34] Such data, in combination with our current results and the fact that many E2-containing medications are FDA-approved, inexpensive, and safe for short durations of time, indicates that randomized clinical trials assessing the feasibility and efficacy of E2 interventions in TSE survivors are warranted.

Strengths of our study include the translational research approach in which we used both longitudinal multiethnic human cohort study data and a parallel set of animal experiments. Several limitations should also be considered when interpreting our study results. First, our analyses did not adjust for the stage of the menstrual cycle or contraceptive use, which are known to impact E2 levels.[74; 79] Additionally, since stress has been shown to reduce E2 levels[5; 84], it is not clear whether the TSE experienced by participants in our study influenced circulating E2 levels in the immediate aftermath of TSE. However, since the early peritraumatic period appears to be a crucial window for predicting long-term pain outcomes, the exact source of circulating E2 levels is likely to be less critical. Second, the sample sizes of two subgroups of the AURORA cohort, the men and the post-menopausal women subgroups, were smaller than that of pre-menopausal women, so the lack of associations between E2 levels and CPTP in those two groups could be due to smaller sample size rather than lack of biological effect. Likewise, the sample size of our naturally cycling female animals was also small, limiting our power to determine whether naturally high levels of E2 could reduce mechanical hypersensitivity. Third, we did not assess other sex hormones in the current study. This is important because women have high progesterone levels in the luteal phase of their menstrual cycle, at a time when they also have high E2 levels. Because progesterone has also been shown to have anti-nociceptive properties[22; 26; 43; 98], it is possible that progesterone is also playing a role in the observed protective effects of E2. Therefore, future studies should aim to disentangle the influence of each sex hormone on CPTP outcomes in women and men. Fourth, the differences in CPTP between women with high vs low E2 were less than established minimum cutoffs that are needed to imply clinical significance, so the clinical impacts of E2 levels on CPTP should be further evaluated. Finally, animal behavioral experiments utilized only a single measure of mechanical hypersensitivity, von Frey. While this type of measurement is one of the most commonly used methods of assessing mechanical hypersensitivity in rodent models[30], recent studies have shown that additional measures of mechanical hypersensitivity, such as naturalistic or operant behaviors, provide a more thorough assessment of animal mechanical hypersensitivity.[87]

Together with previous findings, our study data indicate that high peritraumatic E2 levels are associated with lower long-term pain outcomes following TSE and suggest that immediate post-TSE administration of E2 in women could be a promising therapeutic strategy for reducing development of CPTP. Subsequent research should further validate these results among different study populations with larger sample sizes and broader distributions of sex and gender subgroups. In addition, future research should aim to identify mechanisms through which E2 influences CPTP and should assess via a randomized clinical trial whether E2 is an effective preventative intervention against CPTP for women TSE survivors.