Androgen levels and experimental pain sensitivity in healthy young adolescent girls.

This study was approved by the Institutional Review Board of Washington University in St. Louis. All participants and their parents or guardians provided written informed consent and, where applicable, assent before initiating the study procedures. Data were collected between September 2022 - July 2024. Forty-five healthy girls (11.91 ± 1.35 years) participated in the study. Participants were recruited via the research participant registry at Washington University in St. Louis and Peachjar (a digital platform for flyer distribution from schools to parents of pupils), as well as word-of-mouth. The inclusion criteria were: (1) age between 9 and 16 years, (2) without a diagnosis of chronic pain, psychiatric or neurological disorders, or conditions associated with pubertal maturation, (3) no regular use of medications that could affect pain sensitivity (e.g., anxiolytics/sedatives), (4) not using hormonal contraception (5) neither pregnant nor breastfeeding, (6) capable of understanding basic instructions and communicating in English. Participants were instructed to refrain from taking pain medications 24 hours before the study visit and to avoid activities, such as intense exercise, that might result in persistent pain or soreness in the days preceding the visit. Participants attended a study visit at the School of Medicine, Washington University in St. Louis. Participants were first familiarized with the experimental procedures, stimuli, and pain rating scale (visual analog scale, VAS). Tests included pain thresholds (heat, cold, and pressure in a pseudo-randomized order based on participant ID numbers), temporal summation, suprathreshold pain (heat test stimulus and offset analgesia in a pseudo-randomized order based on participant ID numbers), CPM paradigms (the test stimulus delivered together with the conditioning stimulus for both heat and pressure CPM paradigms, delivered in a pseudo-randomized order based on participant ID numbers) and cold pain tolerance. During the breaks between paradigms, participants completed a survey. Lastly, participants underwent phlebotomy to provide blood samples for the analysis of sex hormone levels. The entire study visit lasted approximately two and a half hours. Each participant was compensated $85 for attending the study visit. The study protocol was based on our previous studies in adolescents, ensuring the feasibility of the study and that participants would be able to tolerate the stimuli [ 42 ; 47 ]. The tests were pseudo-randomized within each pain level to avoid an order effect. Heat and cold pain thresholds (HPT and CPT, respectively) were assessed using a computer-controlled device with a 16×16mm thermode and a trigger button (TSA-2/Pathway, Medoc, Israel). The baseline temperature was 32°C, and the temperature increase and decrease rate was 1.5°C/sec for both heat and cold pain thresholds. The stimulus was delivered to the nondominant volar forearm, and participants were instructed to press the button the first moment they felt pain from the heat or cold stimulus. Four measurements were recorded for each modality (heat and cold), with the first serving as a familiarization trial. The heat/cold pain thresholds were determined by averaging the results of the final three measurements. For pressure pain threshold (PPT), a pressure stimulus was applied at a rate of 60 kPa/s at a 90° angle to the upper trapezius muscle on the dominant side using a digital algometer with a 1 cm 2 circular probe (Algomed, Medoc, Israel). Participants pressed a button to indicate the first moment they felt pain from the stimulus. Four PPT measurements were completed, with the first serving as familiarization. PPT was determined by averaging the results of the final three measurements. A 6.45 Nm von Frey filament was utilized to evaluate the temporal summation value. Participants rated the pain intensity using a mechanical VAS after a single pinprick stimulus and after a series of 10 identical stimuli. Each stimulus was applied at a 90-degree angle to the nondominant volar forearm at a rate of one stimulus per second within a 1 cm 2 area. This procedure was repeated twice. The temporal summation value was calculated as the difference in pain intensity ratings between the tenth stimulus in the series and the single stimulus. The values of the two temporal summation paradigms were averaged to determine the overall temporal summation value. A positive value indicates a facilitatory effect. Based on the recommendations for assessing the CPM response, two CPM paradigms, Heat-CPM and Pressure-CPM, were tested in this study [ 71 ]. Both paradigms involved the application of a test stimulus (heat or pressure), administered alone and then together with a conditioning stimulus. These paradigms were also utilized and were well tolerated in our previous studies with adolescents [ 42 ; 46 ; 48 ]. The conditioning stimulus was immersion of the nondominant foot in a temperature-controlled cold-water bath (Cole-Parmer, USA) maintained at 8°C for 60 seconds. During the final 30 seconds of the conditioning stimulus, the test stimulus (heat or pressure) was applied concurrently. At the end of the conditioning stimulus, participants rated their pain intensity using a mechanical VAS. The cold pain intensity was calculated as the mean of the pain ratings to the two conditioning stimuli (for the heat and pressure CPM paradigms). Heat-CPM paradigm: The test stimulus was a continuous 30-second tonic heat stimulus (46°C) applied to the nondominant volar forearm using the 16×16mm computer-controlled thermode. The temperature increase rate was 6°C/s from a baseline temperature of 35°C. Participants rated their pain intensity in real-time using a computerized VAS (COVAS, Medoc, Israel), with endpoints ranging from ‘no pain sensation’ to the ‘most intense pain imaginable.’ The average of heat pain intensity ratings over the 30-second period was calculated. The test stimulus was delivered alone and together with a conditioning stimulus. The Heat-CPM response was calculated as the difference between heat pain intensity ratings obtained with vs. without the conditioning stimulus. A negative CPM value indicates an inhibitory effect. Pressure-CPM paradigm: The test stimulus was the PPT test (see description above). The test stimulus was delivered alone and then together with the conditioning stimulus. The Pressure-CPM response was calculated as the difference between the PPT obtained with vs. without the conditioning stimulus. A negative CPM value indicates an inhibitory effect. The tonic three-temperature stimulus sequence method was used [ 24 ; 36 ; 45 ]. The stimulus was administered to the nondominant volar forearm using the 16×16mm thermode. The first temperature stimulus was 46°C and lasted for 5 seconds, the second was 47°C, also lasted for 5 seconds, and the third stimulus returned to 46°C, lasting for 20 seconds. Throughout the 30-second test, participants rated their pain intensity in real-time using the COVAS. The control paradigm for the OA test was the Heat-CPM test stimulus (i.e., constant 46°C for 30 seconds). As was done in previous studies, the OA response was calculated as the difference between the average heat pain ratings from 13 to 23 seconds for the OA paradigm and the Heat-CPM test stimulus delivered alone, with a negative value indicating an inhibitory effect [ 45 ]. Participants were instructed to immerse their dominant foot in an 8°C cold water bath for as long as possible until they could no longer tolerate the pain. The cut-off time for this test was 120 seconds, which the participants were not informed of. The duration of immersion was recorded as the cold pain tolerance. The pubertal stage was assessed using the Pubertal Development Scale, a validated and reliable self-administered instrument for adolescents [ 53 ]. Pubertal scores were calculated by summing the ratings for body hair growth and breast development, each rated on a four-point scale (1: not yet started, 4: seems completed), and menarche, rated on a two-point scale (1: no, 2: yes) [ 32 ]. As was done in a previous study, the scores were then categorized into early-mid stage (≤7, without menarche) and late stage (any score with menarche or >7, without menarche) in order to examine if the pubertal stage impacts the androgen-pain relationships [ 32 ]. Blood samples were collected following the completion of experimental pain assessments. To minimize pain and discomfort associated with the blood draw, participants were offered a topical numbing spray and/or cream. Approximately 20 mL of blood was drawn into serum separation tubes. The samples were kept upright at room temperature (approximately 22°C) for 30 minutes to allow clotting. The clotted samples were centrifuged at 3000 rpm for 10 minutes, after which 1 mL of serum was pipetted into small plastic transport tubes. The serum samples were stored at −20°C until they were transported to Labcorp (Laboratory Corporation of America ® Holdings, USA) for subsequent analyses. Total testosterone (ng/dL), DHEA (ng/dL), DHEA-S (ug/dL), DHT (ng/dL), and A4 (ng/dL) levels were measured using high-pressure liquid chromatography-tandem mass spectrometry (LC-MS/MS), while estradiol (pg/mL) was quantified using two-dimensional high-pressure LC-MS/MS. LC-MS/MS techniques are highly precise and reproducible assays with high specificity and sensitivity, particularly at low concentrations, which are typically observed in adolescents and women [ 6 ; 26 ; 55 ]. The main outcome variable of interest in this study was testosterone, therefore, the power calculation was based on testosterone. Based on our preliminary data from n=14 healthy adolescent girls at both early and late pubertal stages (unpublished data from Cincinnati Children’s Hospital), 35 participants were required to identify a significant association between testosterone levels and pain intensity rating to heat stimuli (α = 0.05 with 80% power). The analyses involving the other androgens are considered exploratory. For the present study, the data were stored in the Research Electronic Data Capture system (REDCap, version 14.0.7), and statistical analyses were conducted using JMP ® Pro (Version 16.0.0, SAS Institute Inc., Cary, NC). Descriptive statistics were generated for demographic variables, and the measures of experimental pain sensitivity were inspected for their appropriateness for parametric analysis. First, bivariate Pearson correlations were performed to determine the associations between androgen levels (testosterone, DHEA, DHEA-S, DHT, and A4, independent variables, fitted separately) and the experimental pain sensitivity measures. The experimental pain sensitivity (dependent variables, fitted separately) included heat pain threshold, cold pain threshold, pressure pain threshold, heat pain intensity rating, cold pain intensity rating, cold pain tolerance duration, temporal summation value, offset analgesia magnitude, Heat-CPM response, and Pressure-CPM response. Participants who failed to complete a sensory test were excluded from the test’s analyses (the numbers of participants included in each analysis are provided in Tables 1 – 5 ). Androgens follow a diurnal secretion pattern [ 4 ], thus it was tested whether the timing of each participant’s blood draw was related to androgens levels. Testosterone, DHEA, DHEA-S, and DHT levels were not correlated with the time of blood draw. Therefore, the blood draw time was not controlled for in the analyses involving these hormones. A4 was related to the time of blood draw, and thus, blood draw time was included as a covariate in the A4-related analyses. To examine the potential impact of pubertal stage on androgens levels and their relationships with experimental pain sensitivity, t-tests were conducted to assess differences in the levels of testosterone, DHEA, DHEA-S, DHT, and A4 between participants in early-mid and late stages. In addition, to determine whether the relationships between androgens and experimental pain sensitivity vary based on pubertal stages, models were fitted with the androgens (testosterone, DHEA, DHEA-S, DHT, or A4) x pubertal stage (early-mid vs. late) interaction. In addition, since estradiol may impact the testosterone-experimental pain relationship [ 66 ], we also examined the effect of the interaction between testosterone and estradiol levels (low vs. high based on a median split) on experimental pain sensitivity. To further examine the impact of estradiol on the testosterone-experimental pain relationships, we analyzed the data using the following division: pre-menarche levels ≤30 pg/mL, post-menarche (menstrual phase) levels 31–300 pg/mL, and higher levels >300 pg/mL [ 58 ; 65 ]. However, since the highest value in our sample was 230 pg/mL, we had only 2 groups: the pre-menarche group (n=15, mean=10.6 pg/mL) and the post-menarche group (n=30, mean= 72.3 pg/mL). Thus, we used a median split in the post-menarche group to create 3 groups: pre-menarche (≤30 pg/mL, n=15, mean=10.6 pg/mL), post-menarche low (31–59 pg/mL, n=15, mean=40.2 pg/mL), and post-menarche high (≥60 pg/mL, n=15, mean=104.3 pg/mL). A two-tailed p-value of <0.05 was considered statistically significant. Estimated regression coefficients (β) with 95% confidence intervals (CI) were reported to assess the influence of the independent variables. Corrections for multiple comparisons were performed only for the main outcome analysis involving total testosterone and not for the pre-planned exploratory analyses involving DHEA, DHEA-S, DHT, and A4.

Table 1 presents demographic and hormonal data of the sample. Figure 1 presents the heat map of the associations between androgen levels and experimental pain sensitivity. No significant associations were observed between total testosterone levels and any measures of experimental pain sensitivity, even without correcting for multiple comparisons [ Table 2 , Models (a)], [ Fig. 2 ]. Overall, the variances explained by total testosterone levels were low in these models (< 5.9%). When the impact of pubertal stage on testosterone and its relationships with experimental pain sensitivity was examined, higher testosterone levels were found in young adolescent girls in the late pubertal group compared to the early-mid group (29.45 ± 17.74 vs. 11.91 ± 10.07, p < .001) [ Fig. 3A ]. however, pubertal stage did not moderate the relationships between testosterone levels and experimental pain sensitivity, and the effect of total testosterone on experimental pain sensitivity measures may not vary by pubertal stage models [ Table 2 , Models (b)]. Similarly, although testosterone levels were significantly lower in the low estradiol group than in the high estradiol group (14.33 ± 12.17 vs. 28.66 ± 18.43, p = 0.004) [ Fig. 3B ], no interactions between testosterone and estradiol were observed for any of the experimental pain measures [ Table 2 , Models (c)], indicating that estradiol levels may not modify the effect of testosterone on experimental pain sensitivity. Similarly, using the division into groups based on pre-menarche and post-menarche levels, no significant interaction between testosterone and estradiol on experimental pain was found, suggesting no different relationships between testosterone and experimental pain based on estradiol levels ( Supplementary Table 1 ). Supplementary Figure 1 presents the associations between testosterone levels and experimental pain in the pre-menarchal, post-menarche low, and post-menarche high groups. No relationships between DHEA levels and experimental pain sensitivity measures were observed [ Table 3 , Models (a), Supp Fig. 2 ]. Overall, the variances explained by DHEA levels were low in these models (< 8.8%). In addition, DHEA levels were significantly higher in the late pubertal group compared to the early-mid group (315.60 ± 158.10 vs. 156.90 ± 89.71, p < .001) [ Fig. 3C ]. However, no interaction effects between DHEA levels and pubertal stage were observed for any of the experimental pain sensitivity measures [ Table 3 , Models (b)], suggesting that pubertal stage does not moderate the relationships between DHEA levels and experimental pain sensitivity. An inverse association was observed between DHEA-S levels and cold pain threshold (R 2 = 0.093, p = 0.049), while a positive relationship was observed between DHEA-S and cold pain tolerance (R 2 = 0.118, p = 0.026), suggestive of an antinociceptive effect ( Table 4 , Models (a), Suppl Fig. 3 ). Similar to the other androgens, DHEA-S levels were significantly higher in the late pubertal group compared to the early-mid pubertal group (98.75 ± 52.61 vs. 53.61 ± 30.64, p = 0.002, Fig. 3D ). A DHEA-S * puberty interaction effect was observed for cold pain tolerance (R 2 = 0.113, p = 0.030, Table 4 , Models (b)), indicating that the relationship between DHEA-S and cold pain tolerance is different based on the pubertal stage. In young adolescents in the late pubertal stage, DHEA-S levels were significantly related to cold pain tolerance (R 2 = 0.205, p = 0.027) however, no association was found for young adolescents in the early-mid pubertal stage (R 2 = 0.055, p = 0.348, Suppl Fig. 4 ). No other interaction effects were observed [ Table 4 , Models (b)]. No associations were observed between DHT levels and any experimental pain sensitivity measures ( Table 5 , Models (a), Suppl Fig. 5 ). Overall, the variances explained by DHT levels were low in these models (< 9.3%). Although DHT levels were significantly higher in the late pubertal group compared to the early-mid group (10.46 ± 4.85 vs. 3.82 ± 2.09, p < .001, Fig. 3E ), no interaction effects between DHT and pubertal stage on experimental pain sensitivity were observed ( Table 5 , Models (b)). Higher A4 levels were significantly related to higher pressure pain thresholds (R 2 = 0.164, p = 0.034, Table 6 , Models (a), Suppl Fig. 6 ), suggesting an antinociceptive effect. No other associations were observed. As observed with other androgens, A4 levels were significantly higher in the late pubertal group compared to the early-mid group (86.24 ± 42.91 vs. 36.78 ± 29.73, p = 0.004, Fig. 3F ). However, no A4 x pubertal interaction effects on experimental pain sensitivity were observed ( Table 6 , Models (b)).

This study is the first to examine the relationships between experimental pain sensitivity and androgen levels, including testosterone, DHEA, DHEA-S, DHT, and A4 in young adolescents. Contrary to our hypotheses, overall, no significant correlations were found between androgen levels and experimental pain sensitivity, and pubertal stage did not impact these relationships in healthy young adolescent girls. The results of this study advance the understanding of hormonal mechanisms underlying individual differences in pain sensitivity and can guide future pediatric pain studies. Overall, no relationships between testosterone levels and experimental pain sensitivity were found, which contrast the previously reported antinociceptive effects of testosterone observed in animals and adult men and women. Animal studies have mostly demonstrated an inverse relationship between testosterone levels and nociceptive behavior [ 2 ; 3 ; 11 ; 19 – 21 ; 35 ]. For instance, gonadectomized male rats with reduced testosterone levels exhibited a greater nociceptive behavior in response to formalin-induced noxious stimulation compared to non-gonadectomized rats [ 2 ; 11 ]. Moreover, testosterone administration in studies involving male and female rats, with or without gonadectomy, has been linked to pain modulation and a protective role against the development of conditions such as temporomandibular joint pain [ 3 ; 19 – 21 ; 35 ]. The antinociceptive effect of testosterone observed in animal studies may be due to the extreme manipulation of testosterone levels involving gonadectomy and administration of exogenous testosterone for under- or supra-physiological levels [ 20 ]. Thus, the findings from animal studies may not be directly comparable to human experimental studies investigating mostly natural physiological variations in testosterone levels and their correlations with pain sensitivity. Similar to testosterone, overall, no relationships were found between experimental pain sensitivity and DHEA, DHEA-S, DHT, and A4, even without correction for multiple comparisons. A few associations were found between DHEA-S and cold pain sensitivity and A4 and PPT. It is not clear why these associations were found and if androgens have a specific role in cold or pressure sensitivity. Recent studies suggest that androgens modulate the expression of cold-sensitive TRPM8 receptors in sensory neurons [ 5 ]. Although data on DHEA are lacking, animal models indicate that testosterone inhibits TRPM8-mediated cold sensitivity via androgen receptors [ 22 ]. However, the effect of testosterone on TRPM8 was found to be sex-dependent with a greater antinociceptive effect observed in males [ 1 ; 39 ], which may explain the lack of relationships between testosterone and cold pain found in the present study in girls. Studies examining the relationship between androgens and experimental pain sensitivity in healthy adults have produced inconsistent findings [ 7 ; 8 ; 13 ; 15 ; 50 ; 56 ; 63 ; 69 ; 72 ]. Some studies in healthy women and men have shown that higher testosterone levels are linked to higher electrical pain thresholds and electrical and ischemic pain tolerance, as well as lower pain ratings to noxious heat, ischemic, and electric stimuli [ 7 ; 13 ; 15 ; 63 ] indicating an antinociceptive effect of testosterone. Conversely, other studies in healthy women have found no associations [ 72 ] or reported relationships with only some of the experimental pain measures that were tested [ 56 ]. Additionally, while one study reported a link between testosterone and heat-CPM [ 67 ], another study found no association with pressure-CPM [ 57 ]. The conflicting findings between studies could be attributed to variations in experimental methodologies, including differences in testosterone sampling methods (e.g., salivary, plasma, or serum), assay techniques (e.g., LC-MS vs. immunoassays), study sample (e.g., men vs. women; premenopausal vs. post-menopausal women), and experimental pain models, including the type, duration, and site of stimulation. Additionally, the study population is a key distinction between previous studies and the current study. Findings from adult participants may not be directly comparable to those from adolescents. During adolescence, there are various biopsychosocial changes that may impact experimental pain sensitivity [ 44 ]. Moreover, during adolescence, experimental pain sensitivity gradually decreases in healthy boys and girls [ 28 ; 43 ]. Importantly, testosterone and the other androgens may have a different impact on brain plasticity before vs. after puberty [ 33 ; 34 ; 49 ; 54 ]. Therefore, the impact of sex hormones on pain may be different in adults vs. adolescents, which is why it is critical to determine the relationships between sex hormone levels and experimental pain sensitivity in young adolescents. The results of the present study suggest that biopsychosocial factors other than androgens may impact experimental pain sensitivity in healthy young adolescent girls. Puberty is a dynamic period characterized by the maturation of the hypothalamic-pituitary-gonadal axis and fluctuating sex hormonal levels [ 29 ; 44 ]. As expected, more developmentally mature adolescents in the late pubertal stage had higher androgen levels than those in the early-mid stage. Notably, there was a large overlap in androgen levels between the early-mid and late groups, emphasizing the wide range of normal androgen levels. Importantly, this large individual variability explains why sex hormone levels alone cannot be used to determine pubertal stage in adolescents [ 61 ]. Still, we hypothesized that pubertal stage would moderate the relationships between experimental pain and androgen levels, assuming that short-term fluctuations in testosterone might not yet exert a stable or measurable influence on pain sensitivity in early-mid pubertal stage adolescent girls. Nonetheless, overall the associations between androgen and experimental pain sensitivity did not vary by pubertal stage. Only the association between DHEA-S and cold pain tolerance, which was also significant for the whole group, was moderated by pubertal stage such that it was significant only in young adolescents at the late pubertal stage. Thus, it is possible the effects of DHEA-S on cold pain tolerance do not fully manifest during adolescence but become more pronounced in early adulthood when the endocrine axes have fully matured. Notably, the age range of our participants was 10–13, thus, even though some participants were at the late pubertal stage, they were relatively young, and it is possible that in older participants in the late pubertal stage, androgens would have a greater impact on pain. A correlation between testosterone levels and brain activity in the amygdala and rostral ventromedial medulla regions during noxious heat stimuli was found, but only in healthy adult women with low estradiol levels [ 66 ]. Thus, the role of estrogen in moderating the testosterone-pain relationships was also examined. The present study found that the associations between testosterone and experimental pain sensitivity did not vary between participants based on their estradiol levels. Thus, in adults, testosterone may impact pain sensitivity through its effect on brain areas involved in pain processing, however, in young adolescents, testosterone may not have a similar impact on these brain areas. In addition, the interactions between testosterone and other hormones (e.g., cortisol) could influence pain sensitivity and should be examined in future studies. For instance, the ratios between cortisol and the androgens testosterone, DHEA, and DHEA-S were related to experimental pain sensitivity [ 14 ; 70 ]. Furthermore, the testosterone/cortisol ratio, rather than testosterone levels alone, was reported to be significantly lower in individuals with migraine than healthy controls [ 52 ]. To our knowledge, this study is the first to examine the relationships between androgen levels and experimental pain sensitivity in young adolescents. Furthermore, we assessed the moderating role of pubertal stage in these relationships. To ensure high-quality data, the investigator was blinded to the participant’s pubertal stage, and hormonal values were analyzed by an outside company that had no information regarding the pain sensitivity of the participants. In addition, we included a familiarization part in order to familiarize participants with the stimuli and the rating scales, the instructions before each test were read via written scripts to ensure participants received exactly the same instructions, and 8-minute breaks were kept between each of the pain modulation tests to avoid carry-over effects [ 41 ]. We also used highly precise and reproducible serum assays to quantify androgen levels, which is critical when measuring relatively low levels of androgens that are typically found in young adolescent girls. Nevertheless, an important limitation of this study is that the exploratory analyses of DHEA, DHEA-S, DHT, and A4 were not adjusted for multiple comparisons, and the power calculation was based only on testosterone and heat pain intensity ratings. Thus, although overall, no significant relationships were found even without correction for multiple comparisons, the findings should be interpreted with caution and should be used to guide future studies. In addition, we focused on healthy girls, and our findings are not applicable to boys. In addition, androgens may have a more pronounced effect on chronic pain rather than on experimental pain [ 9 ; 18 ; 27 ; 30 ; 31 ; 59 ; 62 ; 68 ], especially if the chronic pain pathophysiology is related to sex hormones, such as in migraine and endometriosis [ 23 ; 25 ; 37 ; 51 ]. Overall, natural physiological variations in androgen levels among healthy adolescent girls may not be associated with experimental pain sensitivity, nor do these relationships appear to differ by pubertal stage. These observations advance the understanding of pediatric pain and can guide future research on the hormonal impacts of pain in young adolescents.

Sex hormones can influence pain, although research has mostly focused on ovarian sex hormones rather than androgens [ 38 ]. Among androgens, testosterone is the most studied and has been suggested to have an antinociceptive effect [ 17 ]. In animal studies, male rats with gonadectomy, which reduces testosterone levels, have higher nociceptive behavior compared to rats without gonadectomy [ 2 ; 11 ]. In addition, male and female rats with testosterone administration demonstrate lower nociceptive behavior [ 3 ; 19 – 21 ; 35 ]. In adult men and women, higher testosterone levels have been associated with higher pain thresholds and tolerance, as well as reduced pain intensity ratings [ 7 ; 13 ; 15 ; 56 ]. Additionally, testosterone administration in hypogonadal men has been linked to a reduction in experimental pain sensitivity [ 9 ]. Interestingly, the influence of testosterone on experimental pain sensitivity may be moderated by estradiol levels, as a significant relationship between testosterone and experimental pain sensitivity is found only in women with low, but not high, estradiol levels [ 66 ]. Other androgens, such as dehydroepiandrosterone (DHEA), dehydroepiandrosterone sulfate (DHEA-S), dihydrotestosterone (DHT) and androstenedione (A4), are involved in testosterone metabolism [ 17 ], but have been less studied in relation to pain. One study reported positive associations between DHEA-S levels and pressure pain threshold and tolerance in postmenopausal women with fibromyalgia, suggesting that DHEA-S may have an antinociceptive effect [ 16 ]. Still, the role of these other androgens on experimental pain is still unclear. Adolescence is a critical period involving significant puberty-related biopsychosocial changes [ 44 ], including increased androgen levels in both boys and girls [ 40 ; 61 ]. This period also involves alterations in clinical and experimental pain sensitivity and the emergence of sex differences in pain with an increase in chronic pain prevalence and greater experimental pain sensitivity in adolescent girls compared to boys [ 10 ; 12 ; 43 ; 60 ; 64 ]. One explanation for the increase in chronic pain prevalence in girls is that androgens may not have the same antinociceptive effect in adolescent girls as in adults. Thus, this study focused on healthy young adolescent girls and aimed to investigate the relationships between androgen levels and experimental pain sensitivity. We hypothesized that similar to the effects observed in adults, higher testosterone levels would be related to lower experimental pain sensitivity and greater pain modulation capabilities. In addition, we explored whether pubertal maturation moderates these relationships, hypothesizing that stronger associations would be found in young adolescents who are at late compared to early pubertal stage.

Supplementary Figure 1. Associations between testosterone and experimental pain in pre-pubertal, post-pubertal low, and post-pubertal high estradiol levels. Pre-menarche levels: <30 pg/mL, n=15, mean=10.6 pg/mL, post-menarche low: 60 pg/mL, n=15, mean=104.3 pg/mL. All associations are not significant (p-value > 0.05) except for the correlation between testosterone levels and offset analgesia in the post-menarche low group (r 2 =0.403, p=0.049). CPM- conditioned pain modulation Supplementary Figure 2. Associations between DHEA levels and experimental pain sensitivity. A. Heat pain threshold. B. Cold pain threshold. C. Pressure pain threshold. D. Pain intensity ratings to a heat stimulus delivered for 30 seconds at 46°C. E. Pain intensity ratings to a cold stimulus delivered for 60 seconds at 8°C. F. Cold pain tolerance to foot immersion in cold water (8°C), with a cut-off time of 120 seconds. G. Temporal summation which was calculated as the difference in pain ratings evoked by a series of 10 pinprick stimuli vs. a single pinprick stimulus. Positive values indicate a facilitatory response. H. Offset analgesia which was calculated as the difference in pain ratings between the mean pain ratings of 13–23 seconds of a 46-47-46°C heat stimulus paradigm and a constant 46°C heat stimulus. Negative values indicate efficient inhibitory response. I. Heat conditioned pain modulation (Heat-CPM), which was calculated as the difference in mean pain ratings of a heat stimulus (46°C, 30 seconds) delivered with and without a conditioning stimulus (8°C water bath, 60 seconds). Negative values indicate efficient inhibitory response. J. Pressure conditioned pain modulation (Pressure-CPM), which was calculated as the difference between pressure pain thresholds without and with a conditioning stimulus (8°C water bath, 60 seconds). Negative values indicate efficient inhibitory response. Supplementary Figure 3. Associations between DHEA-S levels and experimental pain sensitivity. A. Heat pain threshold. B. Cold pain threshold. C. Pressure pain threshold. D. Pain intensity ratings to a heat stimulus delivered for 30 seconds at 46°C. E. Pain intensity ratings to a cold stimulus delivered for 60 seconds at 8°C. F. Cold pain tolerance to foot immersion in cold water (8°C), with a cut-off time of 120 seconds. G. Temporal summation which was calculated as the difference in pain ratings evoked by a series of 10 pinprick stimuli vs. a single pinprick stimulus. Positive values indicate a facilitatory response. H. Offset analgesia which was calculated as the difference in pain ratings between the mean pain ratings of 13–23 seconds of a 46-47-46°C heat stimulus paradigm and a constant 46°C heat stimulus. Negative values indicate efficient inhibitory response. I. Heat conditioned pain modulation (Heat-CPM), which was calculated as the difference in mean pain ratings of a heat stimulus (46°C, 30 seconds) delivered with and without a conditioning stimulus (8°C water bath, 60 seconds). Negative values indicate efficient inhibitory response. J. Pressure conditioned pain modulation (Pressure-CPM), which was calculated as the difference between pressure pain thresholds without and with a conditioning stimulus (8°C water bath, 60 seconds). Negative values indicate efficient inhibitory response. Supplementary Figure 4. Pubertal stage moderates the relationship between DHEA-S levels and cold pain tolerance. DHEA-S levels were significantly correlated with cold pain tolerance in girls that are in the late pubertal stage (r = 0.453, p = 0.027) but not in the early-mid stage (r= 0.235, p = 0.348). Supplementary Figure 5. Associations between DHT levels and experimental pain sensitivity. A. Heat pain threshold. B. Cold pain threshold. C. Pressure pain threshold. D. Pain intensity ratings to a heat stimulus delivered for 30 seconds at 46°C. E. Pain intensity ratings to a cold stimulus delivered for 60 seconds at 8°C. F. Cold pain tolerance to foot immersion in cold water (8°C), with a cut-off time of 120 seconds. G. Temporal summation which was calculated as the difference in pain ratings evoked by a series of 10 pinprick stimuli vs. a single pinprick stimulus. Positive values indicate a facilitatory response. H. Offset analgesia which was calculated as the difference in pain ratings between the mean pain ratings of 13–23 seconds of a 46-47-46°C heat stimulus paradigm and a constant 46°C heat stimulus. Negative values indicate efficient inhibitory response. I. Heat conditioned pain modulation (Heat-CPM), which was calculated as the difference in mean pain ratings of a heat stimulus (46°C, 30 seconds) delivered with and without a conditioning stimulus (8°C water bath, 60 seconds). Negative values indicate efficient inhibitory response. J. Pressure conditioned pain modulation (Pressure-CPM), which was calculated as the difference between pressure pain thresholds without and with a conditioning stimulus (8°C water bath, 60 seconds). Negative values indicate efficient inhibitory response. Supplementary Figure 6. Associations between A4 levels and experimental pain sensitivity. A. Heat pain threshold. B. Cold pain threshold. C. Pressure pain threshold. D. Pain intensity ratings to a heat stimulus delivered for 30 seconds at 46°C. E. Pain intensity ratings to a cold stimulus delivered for 60 seconds at 8°C. F. Cold pain tolerance to foot immersion in cold water (8°C), with a cut-off time of 120 seconds. G. Temporal summation which was calculated as the difference in pain ratings evoked by a series of 10 pinprick stimuli vs. a single pinprick stimulus. Positive values indicate a facilitatory response. H. Offset analgesia which was calculated as the difference in pain ratings between the mean pain ratings of 13–23 seconds of a 46-47-46°C heat stimulus paradigm and a constant 46°C heat stimulus. Negative values indicate efficient inhibitory response. I. Heat conditioned pain modulation (Heat-CPM), which was calculated as the difference in mean pain ratings of a heat stimulus (46°C, 30 seconds) delivered with and without a conditioning stimulus (8°C water bath, 60 seconds). Negative values indicate efficient inhibitory response. J. Pressure conditioned pain modulation (Pressure-CPM), which was calculated as the difference between pressure pain thresholds without and with a conditioning stimulus (8°C water bath, 60 seconds). Negative values indicate efficient inhibitory response.