The role of androgens on experimental pain sensitivity: a systemic review and meta-analysis.

A comprehensive search was conducted with the assistance of a Washington University Becker Medical Librarian and was based on PRISMA guidelines and standards. 53 CENTRAL, MEDLINE/PubMed, EMBASE, CINAHL, Web of Science, Scopus, ProQuest, OATD, EThOS, conference abstracts/proceeding: GreyLit, GreyNet, OpenGrey, Clinical trials registry platforms were used to search relevant studies published before May 2023. The systemic review was preregistered in PROSPERO (International prospective register of systematic reviews, #410889). Three types of studies were included: (1) studies examining the relationships between testosterone and experimental pain sensitivity (associations), (2) studies comparing group differences in androgen levels or experimental pain sensitivity, and (3) studies examining the effect of androgen medications on experimental pain sensitivity. All studies were required to be conducted in humans, written in English, and have full-text access. Qualitative studies, retrospective studies, case reports, expert opinions, and reviews were excluded. Experimental pain measures included pain thresholds, pain tolerance, pain modulation, and pain ratings of somatosensory stimuli, including pressure, heat, cold, electrical, ischemic, mechanical, and chemical. The androgen measures were measured levels of testosterone, dehydroepiandrosterone (DHEA), dehydroepiandrosterone sulfate (DHEA-s), androstenedione, androstenediol, and dihydrotestosterone (DHT). For androgen intervention studies, the primary mechanism of action of the intervention needed to be through acting on the androgen receptor or primally impacting testosterone, DHEA, DHEA-s, androstenedione, androstenediol, or DHT levels. Interventions in which the effect of the androgens could not be isolated from other treatments were excluded from the review (ie, when multiple interventions with different mechanisms of action were administered). The reference list of identified papers was scanned for additional potentially relevant papers. The complete list of search words is included in Supplementary 1, http://links.lww.com/PR9/A296 . In brief, the androgen search words included androgens, testosterone, androstenedione, DHEA, DHEA-S, dehydroepiandrosterone, or hypogonadism, and the pain search words included pain, experimental pain, quantitative sensory testing, pain measurement, numeric rating scale, nociception, or threshold. Five thousand one hundred thirty studies were identified. Three reviewers (H.N.A., G.B., and E.R.) independently screened the title and abstract to identify potentially relevant studies for inclusion using Covidence (Covidence systematic review software, Veritas Health Innovation, Melbourne, Australia). Each study was screened by at least 2 independent reviewers. The agreement rate was 88%, and conflicts were resolved by the third reviewer. After reviewing the titles and abstracts, 505 studies were included as potentially relevant, and their full text was reviewed. However, due to the broad search terms, most studies assessed clinical pain but not experimental pain or did not assess androgen levels. Thus, after reviewing the full text, 31 papers were identified and are included in this review and meta-analysis. Data are available upon request. To provide some measure of quality for the analyzed studies, we developed a scoring system based on previous similar meta-analyses. 36 , 47 , 49 , 75 The studies were assessed based on study design, hormone collection and analyses, experimental pain assessments, and statistical analysis. In each category, if the study met the criteria, then a score of 1 was given, if not, a score of 0 was given, and if the information was not clear, a score of 0.5 was given. The final score was expressed as a percentage, calculated as the sum of the given points divided by the total points. Data synthesis was conducted separately for the studies (1) examining the relationships between testosterone and experimental pain sensitivity, (2) comparing group differences in androgen levels or experimental pain sensitivity, and (3) examining the effect of androgen medications on experimental pain sensitivity. A synthesis of the data was conducted using a meta-analysis. The meta-analyses were conducted separately for each experimental pain measure. Due to the large heterogeneity between studies, only 2 meta-analyses were conducted to test the relationships between (1) testosterone and pain ratings of heat stimuli and (2) testosterone and electrical pain thresholds, and no other synthesis methods were conducted. 13 For the meta-analyses, the number of participants and correlation coefficient for the relationships were collected from each study. If data were not presented in the manuscript, the authors were contacted in a request for the data. Meta-analyses were performed only when there were 3 or more studies examining the same experimental pain measure and androgen variable. For studies that examined more than 1 group, combined group data were prioritized. One study tested testosterone-pain relationships throughout the menstrual cycle (days 1, 4, 14, 22). 74 In this study, day 1 data were prioritized because it was the only day that showed a significant relationship between testosterone levels and experimental pain measures (electrical pain thresholds). If the meta-analysis of this experimental pain measure was significant, day 1 data would be replaced with other days to test if it impacts the results. For studies with several repetitions of quantitative sensory testing involving an intervention (not androgen related), baseline data were prioritized. All analyses were conducted using R statistical software, 73 the R meta-analysis package “metafor,” 76 and R-Studio. 66 DerSimonian and Laird's 20 random-effects model was used to determine relationships between androgen levels and experimental pain sensitivity. We examined potential publication bias using funnel plots and Egger regression test for funnel plot asymmetry.

Most studies examined the relationships between testosterone and experimental pain in healthy participants. This review focuses primarily on experimental pain measures (eg, pain thresholds rather than detection thresholds). A summary of the studies is presented in Table 1 , and the quality assessment and risk of bias are provided in Supplemental Table 1, http://links.lww.com/PR9/A296 . Relationships between androgen levels and experimental pain sensitivity. DHEA, dehydroepiandrosterone; DHEA-S, dehydroepiandrosterone sulfate. A significant positive correlation was found between heat pain thresholds and free testosterone levels in healthy men (ie, higher levels of testosterone are associated with a higher pain threshold). This supports the antinociceptive effect of testosterone in healthy men. However, this study found no significant correlations between cold/heat detection and free testosterone levels. 5 Another study similarly found an antinociceptive effect such that higher testosterone levels were correlated with lower pain ratings to noxious heat stimuli delivered alone (large effect size) or with transcutaneous electrical nerve stimulation (medium effect size) in healthy men and women. 16 However, higher testosterone levels in healthy women were associated with lower temperatures that induced pain at the intensity of 50 (0–100 scale, medium effect size), suggesting a pronociceptive effect of testosterone. 77 Another study in healthy women found no correlation between testosterone levels and pain thresholds, pain tolerance, and pain intensity ratings to heat stimuli. 24 Interestingly, sex differences in the relationships between testosterone and heat pain ratings were also reported, and while no associations were found when examining the entire cohort when analyzing by sex, a positive medium size correlation was found in men (suggestive of pronociceptive effect) and a negative medium size correlation in women suggestive of antinociceptive effect. 54 One challenge when assessing sex hormone levels in women is the role of the menstrual cycle and medications that can impact sex hormone levels, such as contraceptive pills and hormone replacement therapy. Testosterone levels and experimental pain sensitivity were examined in healthy women on days 1, 4, 14, and 22 of their menstrual cycle. Overall, in any of those days, no correlations were found between free testosterone levels and heat detection and heat pain thresholds. 74 In addition, the change in testosterone levels between the early-mid follicular and mid-luteal phases did not correlate with the change in pain sensitivity to cold stimuli. 84 In women with and without oral contraceptives, testosterone levels did not correlate with heat pain sensitivity (the temperature that induced pain at the intensity of 50 [0–100 scale] and the pain intensity and unpleasantness ratings of this temperature). 64 , 78 In patients with chronic pain, no relationship between testosterone levels and provoked pain has been identified. In men and women with migraine, there was no correlation between testosterone levels and heat pain thresholds or pain ratings to heat stimuli. 54 Similarly, in healthy women without chronic pain other than dysmenorrhea and/or bladder sensitivity, there was no correlation between testosterone levels and pain ratings to cold water stimulus. 32 A meta-analysis was conducted between testosterone levels and pain ratings of heat stimuli. Three studies were included in the meta-analysis. Even though 2 reported on a medium-large antinociceptive effect of testosterone, overall, the meta-analysis found no evidence of a significant correlation between these factors (Fig. 1 A, Table 2 ), and there was no indication of publication bias (Fig. 1 B, Table 2 ). Importantly, 2 of the included studies assessed pain ratings to heat stimuli, which was set to evoke pain at an intensity of 5/10; thus, the variability in pain ratings in these studies is, by design, small. 77 , 78 However, this measure can provide additional insight into pain sensitivity to heat stimuli. Forest and funnel plots for the association between testosterone levels and heat pain. (A) No correlation between testosterone levels and heat pain ratings across the studies. Importantly, in the studies by Vincent et al., participants rated their pain intensity to stimuli that were tailored to evoke an intensity of 50 (0–100 scale); thus, lower variability in these studies is expected for pain ratings. (B) No indication of publication bias. Summary of tests for heterogeneity, grand mean, and publication bias. H 2 , total variability/within-study variance; I 2 , % of total variability due to heterogeneity; t, square root of t 2 ; t 2 , estimate of the total amount of heterogeneity. Several studies reported an antinociceptive effect of testosterone on electrical pain sensitivity. In healthy men, a positive small effect size correlation between testosterone levels and electrical pain thresholds was found, suggesting an antinociceptive effect of testosterone. 15 Similarly, in healthy women, higher levels of testosterone were associated with lower electric sensory and affective pain ratings. 9 The same group also found a small antinociceptive effect for a combined group of women with premenstrual dysphoric disorder and healthy controls, in which testosterone levels were positively related to electrocutaneous pain thresholds and electrocutaneous pain tolerance and negatively correlated with affective pain ratings of electrocutaneous stimuli. 10 Contradictory to this, free testosterone levels were not related to electrical pain thresholds in healthy postmenopausal women receiving and or not receiving hormone replacement therapy. 57 In addition, another study assessing a different type of electrical stimulus, which is nonpainful and represents sensory detection (not pain), found that free testosterone levels were not related to electrical detection thresholds in healthy women tested on days 1, 4, 14, and 22 of their menstrual cycle. 74 However, this study found an antinociceptive medium size effect of testosterone on electrical pain thresholds only on day 1 of the menstrual cycle. 74 Four studies were included in the meta-analysis. Even though 3 reported on a small-medium antinociceptive effect of testosterone, overall, the meta-analysis revealed no evidence of a significant correlation between testosterone levels and electrical pain thresholds across the studies (Fig. 2 A, Table 2 ). There was no indication of publication bias (Fig. 2 B, Table 2 ). Forest and funnel plots for the association between testosterone levels and electrical pain thresholds. (A) No correlation between testosterone levels and electrical pain thresholds across the studies. (B) No indication of publication bias. In healthy women, higher levels of testosterone were associated with higher ischemia tolerance and lower ischemia sensory and affective pain ratings. 9 Similarly, in women with premenstrual dysphoric disorder and healthy controls, testosterone levels were positively correlated to ischemia pain tolerance, which suggested an antinociceptive effect, although the effect size was small. 10 However, other studies found no correlations between testosterone levels and ischemic pain sensitivity for women with and without oral contraceptives in the follicular or luteal phases. 64 Testosterone was also not related to pain thresholds, pain tolerance, and pain intensity ratings of ischemic stimuli. 24 , 52 In addition, the changes in testosterone levels between the early-mid follicular and mid-luteal phases were not related to the changes in ischemic pain sensitivity. 84 In patients with fibromyalgia, a positive relationship of small size effect between total testosterone levels and ischemic pain thresholds was found, suggesting an antinociceptive effect of testosterone. However, this relationship was found only in the mid-luteal phase and not during the late follicular or perimenstrual phases. 52 A meta-analysis was not performed because there were <3 studies with complete data for the same experimental ischemic pain measure and androgen variable. Overall, no relationships were found between testosterone levels and pressure pain sensitivity. In both healthy men and women, testosterone levels were not related to pressure pain thresholds (PPTs). 4 Similar results were found in healthy women tested 4 times across their menstrual cycle 74 and healthy women with and without contraceptive pills. 40 , 57 , 64 No correlations were found between the change in testosterone levels and the change in pain sensitivity to pressure stimuli between the early-mid follicular and the mid-luteal phase. 84 Notably, 1 study found a pronociceptive effect (medium size effect) of testosterone, which was negatively related to PPT and pressure pain tolerance in healthy women in the follicular phase. 35 Only 1 study tested these associations in patients with chronic pain conditions. No relationships between testosterone levels and pressure pain thresholds were found for a combined group, which included women with dysmenorrhea, women diagnosed with bladder pain syndrome, women with dysmenorrhea and increased bladder sensitivity, and healthy pain-free controls. 32 A meta-analysis was not performed because there were <3 studies with complete data for the same experimental pressure pain measure and androgen variable. For mechanical pinprick stimuli, testosterone levels were negatively correlated to mean pain ratings (medium antinociceptive effect) but were not related to mechanical pain thresholds, incision pain, or mechanical temporal summation. 56 In addition, the change in testosterone levels was not related to the change in pain sensitivity to pinprick stimuli between the early-mid follicular and mid-luteal phases. 84 Testosterone levels were not related to pain intensity ratings using a paradigm of glutamate injections into the temporomandibular joint. 4 Finally, no relationships were found between testosterone levels and bladder pain evoked by drinking water in a combined group of healthy women without chronic pain other than dysmenorrhea and/or bladder pain syndrome. 32 Similar to experimental pain sensitivity, overall, testosterone did not affect inhibitory pain modulation capabilities using various pain modulation paradigms. One of the pain modulation paradigms that were tested was the conditioned pain modulation (CPM) paradigm, which assesses the “pain inhibits pain” phenomenon. In this paradigm, 1 noxious stimulus is used to inhibit the pain evoked by another noxious stimulus delivered to a remote area in the body. 45 , 48 , 50 , 82 The 2 stimuli can be delivered simultaneously (parallel) or one after the other (sequential). 46 The exact mechanism of CPM is not known but may involve descending modulation, propriospinal inhibition and, supraspinal mechanisms. 48 , 50 In women in early follicular, ovulation, or mid-luteal phases, no correlations between testosterone levels and CPM responses were found. 61 In healthy men and women, testosterone had no effect on the sequential CPM response or offset analgesia but a significant effect on the parallel CPM response in a multiple regression model correcting for sex, catastrophizing, and stimulus temperature. However, in this study, testosterone levels had a very small effect and explained less than 6% of the variance in CPM responses (parallel or sequential, conducted separately for men and women, and for women divided based on luteal, follicle, and ovulation). 79 Another study in healthy women used a different pain modulation paradigm of emotional pain modulation. 63 The inhibitory mechanisms underlying emotional pain modulation may differ from the mechanisms of CPM and include supraspinal regions involved in emotional or cognitive processing. 63 Testosterone levels and the interaction between testosterone and menstrual phases were not related to the emotional modulation of pain ratings or nociceptive flexion reflex to electrical stimuli during the viewing of pictures containing mutilation, neutral, or erotic contents. 63 A few studies focused on DHEA and DHEA-S, which are involved in testosterone synthesis. A summary of these studies is presented in Table 1 B. In patients with trauma but no chronic pain and patients with chronic pain/posttraumatic stress disorder, a greater increase in DHEA after exercise was correlated with a decrease in cold pain tolerance, suggesting a pronociceptive medium size effect of DHEA. 70 In women with anorexia nervosa, heat pain threshold latency correlated negatively with DHEA (higher DHEA associated with higher pain, medium size effect), supporting a pronociceptive effect of DHEA. 81 However, this was found only in the patient group and not in the control group. When patients and controls were combined, a significant pronociceptive medium size effect was found (higher DHEA-S was associated with higher heat pain threshold latency). 81 By contrast, in a study of patients with fibromyalgia, a positive correlation was found between DHEA-S levels and pressure pain thresholds and tolerance and, although the effect size was small, it indicates an antinociceptive effect of DHEA-S. 26 Thus, DHEA and DHEA-S have been studied less, and published studies have shown contradictory results on their role in experimental pain sensitivity. A summary of the studies is presented in Table 1 C. Healthy men were divided based on their normal variability in testosterone levels. Men with low testosterone levels had higher pain intensity and unpleasantness ratings to a heat stimulus compared with men with high testosterone levels (large effect size for both pain intensity and unpleasantness ratings). 17 Using a different approach, healthy women were divided into women who displayed/did not display pain behaviors during a cold pressor test (based on the experimenter's judgment, which noted any physical or vocal display of discomfort and divided the participants accordingly). Women who displayed pain behaviors had higher pain scores on the cold pressor test compared with women who did not display pain behaviors. However, no significant differences in testosterone levels were found between the groups at baseline, after the cold pressor test, or in the change from baseline to after the cold pressor test. 6 All studies have examined testosterone-related interventions. A summary of the studies is presented in Table 3 . In healthy men, the effect on the experimental pain sensitivity of a one-time testosterone gel application was compared with placebo. Men who received the testosterone had higher pain intensity and unpleasantness ratings to noxious and nonnoxious electrocutaneous stimuli compared with placebo, suggesting a pronociceptive effect of testosterone gel on both pain and somatosensory stimuli in general. 85 Other studies examined the effect of testosterone on experimental pain in patients who experienced changes in testosterone levels due to medications. For example, a side effect of opioids is hypogonadism (reduced levels of testosterone), which can be managed with testosterone treatment. In patients with chronic pain using opioids, a reduction in punctate mechanical pain ratings and an increase in pressure pain threshold at the thumb were found in patients receiving testosterone treatment compared with placebo. Cold pain tolerance, cold pain ratings, and pressure pain thresholds at the trapezius were not different between the groups. 11 In another study with the same patient population, testosterone injections and placebo had similar effects, and no differences were found in the change from baseline to 6-month follow-up in pressure pain thresholds, heat pain thresholds, cold pain thresholds, pressure pain tolerance, temporal summation, and the conditioned pain modulation response. 29 Another group of patients who have significant changes in testosterone levels are patients with prostate cancer, which is treated with androgen deprivation therapy to reduce testosterone levels. A comparison between patients with androgen deprivation therapy and a control group of patients with prostate cancer in remission found no significant group differences in pressure pain thresholds, pain ratings of cuff pressure algometry and cold pain, mechanical temporal summation, and conditioned pain modulation responses. 27 Overall, these interventional studies found no effect of testosterone manipulations on experimental pain sensitivity. Effect of androgen nterventions on experimental pain sensitivity.

The present review and meta-analysis explored the impact of androgens on experimental pain sensitivity assessed using quantitative sensory testing, which is a widely used, standardized psychophysical method employing controlled stimuli to measure pain perception and specific mechanistic processes (eg, peripheral and central sensitization, inhibitory modulation, etc.) in humans. 7 Overall, some studies suggest a small antinociceptive effect of androgens, others report a small pronociceptive effect, and some find no significant effects at all. Thus, testosterone may have only a minor impact on experimental pain sensitivity in men and women with or without chronic pain. This is contradictory to animal studies that found an antinociceptive and protective effect of testosterone. In animal models, a reduction in testosterone levels after orchiectomy increases nociceptive behavior, while testosterone treatment reduces nociceptive behavior. 2 , 3 , 14 , 25 , 28 , 39 , 72 Importantly, these manipulations cause extreme changes in testosterone levels dramatically outside of the physiologic range. Thus, animal studies may not be comparable to human studies, which mostly assessed the natural variability of testosterone levels in healthy participants and correlated them with experimental pain sensitivity. Testosterone could impact pain through several mechanisms, including affecting sensory neurons, interacting with immune cells, and impacting brain areas involved in pain processing, stress, and mood, such as anxiety and depression. 21 Thus, it is surprising that no relationships were found, although only 2 meta-analyses were performed due to the large variability in the methodology of the tests. Both meta-analyses between testosterone levels and electrical pain thresholds, as well as pain ratings to heat stimuli, did not find significant relationships. There are many confounding factors that could impact testosterone levels and their relationships with pain. These factors include sex, age, menstrual phase, sample type (saliva, blood), analysis method (radioimmunoassay, liquid chromatography-tandem mass spectrometry), hormone analysis (free, total), and time of sample collection. 65 Notably, not all studies controlled for these factors (Supp Table 1, http://links.lww.com/PR9/A296 ), which can affect their results. However, due to the small number of studies, our meta-analyses included all available studies with the same androgen and experimental pain measures, and conducting separate meta-analyses based on the above-mentioned factors was not possible. It is possible that the lack of significant associations in the meta-analyses is due to the small number of included studies, the heterogeneity of the individual studies, and their small sample and methodological limitations. Thus, testosterone may have a subtle small effect on pain, which could not be detected in our analyses. Conducting highly rigorous studies on the relationships between androgens and experimental pain is highly needed. In addition, more studies are needed to determine the role of other androgens on experimental pain in specific populations. For example, androgens may impact pain in some populations, such as in patients with chronic pain conditions in which the underlying mechanism of the condition is related to sex hormones. These conditions, such as endometriosis, vaginal pain, visceral pain, and headache, may be more hormone-responsive compared with other chronic pain conditions or experimental pain measures. 30 , 31 Indeed, several studies found that testosterone or DHEA treatments improve pain in patients with endometriosis and vaginal pain. 22 , 33 , 37 , 38 , 60 Thus, patients with these conditions may have a specific nervous system organization, allowing testosterone to induce its antinociceptive activational effect. Testosterone is the most studied androgen, and only a few studies have focused on the other hormones involved in testosterone synthesis, mainly DHEA and DHEA-S. Dehydroepiandrosterone and dehydroepiandrosterone sulfate are secreted primarily by the adrenal cortex and are precursors to testosterone. 42 , 44 In addition, testosterone can be converted to DHT, which has a greater affinity for the androgen receptor. 44 Although less studied, there are more consistent results for a pronociceptive effect of DHEA on experimental pain sensitivity. 70 , 81 An important next research step is a more in-depth examination of the various less-studied androgens, including DHEA, DHEA-S, DHT, and androstenedione. In addition, examining the ratio or interaction between testosterone and other sex hormones can be of interest as this interplay may have a greater impact on pain sensitivity than an isolated androgen level. For example, the ratio between testosterone and cortisol was correlated (medium effect size) to heat pain ratings in healthy men and women. 16 In addition, while no differences in testosterone levels were found in patients with migraine compared with healthy controls, the testosterone/cortisol ratios were significantly lower in patients with migraine as compared with those without. 55 The lack of associations between androgen levels and experimental pain sensitivity may suggest that a relatively small individual variability in the normative ranges of androgen levels is insufficient to reveal the impact of androgens on pain. It is possible that only extremely low/high values of androgen levels, such as induced following interventions, can impact experimental pain sensitivity. The few human studies that examined the impact of testosterone supplementation on experimental pain sensitivity found contradicting results. These interventions include testosterone supplementation to achieve a normal physiologic range in patients with opioid-induced hypogonadism 41 , 51 and androgen deprivation therapy to reduce testosterone levels in patients with prostate cancer. 62 , 83 Thus, the extreme testosterone manipulations in these populations may be more like the manipulation used in animal studies. Most of the human manipulation studies are of patients with conditions such as chronic pain requiring chronic opioids or cancer, which likely impact testosterone levels, pain sensitivity, and the relationships between them. In addition, it is also possible that androgens and testosterone may be more involved in acute or chronic pain severity rather than experimental pain sensitivity. In patients with fibromyalgia, while no relationships were found between experimental pain sensitivity and testosterone levels, 52 lower testosterone levels were related to higher clinical pain, 67 and testosterone treatment reduced muscle pain and tender points pain. 80 Moreover, testosterone supplementation has been found to reduce clinical pain in various other conditions, including endometriosis. 22 , 33 , 60 Thus, testosterone may have a greater impact on chronic pain rather than experimental pain. Androgens and testosterone have organizational and activational effects. An activational effect depends on the presence of the hormone at a specific moment to impact behavior, while an organizational effect indicates a permanent change in the organization of the nervous system, which occurs when the hormone is present during specific critical periods in life such as perinatal and puberty. 68 , 71 The organizational changes can impact the activational response and behavior later in life (ie, in adulthood). 71 For example, in hamsters, testosterone administration before or during puberty but not after led to changes in mating behavior in adulthood. 69 Thus, based on the early life organizational changes, there is an individual variability in the response to sex hormones, which can explain the large variability in the effect of sex hormones on behavior and pain. 71 It is possible that testosterone has an activational antinociceptive effect only in individuals who had a specific organizational change, making them more responsive to testosterone effects later in life. This review has several strengths: (1) the protocol was preregistered in PROSPERO (#410889); (2) a comprehensive search strategy was developed with the assistance of a medical librarian and experts in pain, endocrinology, and gynecology; (3) a comprehensive literature search was conducted; (4) risk of bias was assessed; (5) screening of studies was completed by several investigators with high agreement level; and (6) results were synthesized, and meta-analyses were performed. Notably, there are several limitations: (1) the studies are heterogeneous in their methods and studied population, and thus, only 2 meta-analyses were performed; (2) a small number of studies were included in each meta-analysis; (3) an independent information specialist did not review the search strategy; and (4) many studies have biases and limitations. These limitations may explain the lack of associations found in the meta-analyses. However, overall, most individual studies reported a lack of association or no impact of androgen intervention on experimental pain. Even if only the high-quality studies were included, no consistent finding of the effect of androgens on experimental pain would be found, as conflicting results between and within studies are reported.

This review and meta-analysis summarize the current knowledge on the role of androgen on experimental pain, point out the existing gaps, and propose future directions. Overall, physiologic variability in testosterone levels may have a minimal impact on experimental pain sensitivity in adult humans with and without chronic pain. This is contradictory to the findings in animal studies. Manipulation of testosterone to extremely high or low levels may have a greater impact on pain, but only in some patient populations. Future studies should focus on examining the interactions between androgens and other hormones in specific subgroups of patients.

Sex differences in chronic and experimental pain are well documented, with higher experimental pain sensitivity in women. 8 , 12 , 23 , 34 , 43 , 58 , 59 Many possible explanations for the sex differences have been suggested, including the difference in sex hormone levels. 1 , 18 , 19 Interestingly, most studies have focused on the ovarian hormones, estrogen and progesterone, while the role of androgens in pain has been less studied, especially in humans. Androgens are a group of sex hormones that play a key role in sexual differentiation, reproductive health and behavior, and body development and maintenance by binding to androgen receptors. 44 Androgen secretion is under neuronal control, primarily by the hypothalamus and the pituitary gland. 44 Testosterone is the most studied androgen and animal studies have found that lower testosterone levels are associated with higher nociceptive responses in male rats after gonadectomy, while in female rats, lower nociceptive responses are found after testosterone treatment. 2 , 3 , 14 , 25 , 28 , 72 In addition, studies that used androgens as a treatment for pain showed a reduction in chronic pain severity. 22 , 33 , 37 , 38 , 60 Thus, the current notion is that testosterone has an antinociceptive effect on pain, including experimental pain sensitivity, and there is a critical need to assess this notion. Experimental pain sensitivity is assessed using quantitative sensory testing, which is a standardized, psychophysical approach employing controlled stimuli (eg, thermal, mechanical, vibratory) and potentially providing insights into mechanisms of pain processing and modulation. Thus, the present systemic review and meta-analysis could advance the understanding of how androgens modulate pain. The study aims were to systemically examine the findings on the role of androgens in experimental pain sensitivity in humans by focusing on (1) examining the relationships between testosterone and experimental pain sensitivity, (2) comparing group differences in androgen levels or experimental pain sensitivity, and (3) examining the effect of androgen medications on experimental pain sensitivity.

Supplemental digital content associated with this article can be found online at http://links.lww.com/PR9/A296 .

The authors declare no competing interests in relation to this work.