Nociceptive Processing in Women With Premenstrual Dysphoric Disorder (PMDD)

Page created by Raymond Mcbride
 
CONTINUE READING
ORIGINAL ARTICLE

       Nociceptive Processing in Women With Premenstrual
                   Dysphoric Disorder (PMDD)
                   The Role of Menstrual Phase and Sex Hormones
                        Emily J. Bartley, PhD, Shreela Palit, BS, Bethany L. Kuhn, BS,
                      Kara L. Kerr, MA, Ellen L. Terry, MA, Jennifer L. DelVentura, MA,
                                           and Jamie L. Rhudy, PhD

Objective: Premenstrual dysphoric disorder (PMDD) is associated
with increased pain, but there has been a lack of well-controlled
                                                                        P   remenstrual dysphoric disorder (PMDD) is a cyclical
                                                                            syndrome characterized by debilitating affective,
                                                                        behavioral, and physical symptoms that occur during the
research assessing pain responsivity, sex hormones, and their           late-luteal (premenstrual) phase of the menstrual cycle.1
relationships in this group. This study was designed to address this
gap in the literature.
                                                                        Interestingly, pain processing may be disrupted in PMDD.
                                                                        Specifically, laboratory studies suggest PMDD is associated
Materials and Methods: Healthy, regularly cycling participants (14      with lower ischemic pain thresholds/tolerances,2–4 in addi-
PMDD, 14 non-PMDD) attended pain testing sessions during the            tion to greater ischemic,4 pressure,5 and cold pressor pain
mid-follicular, ovulatory, and late-luteal phases of the menstrual      intensity ratings,4 when compared with healthy control
cycle (order counterbalanced) and salivary estradiol, progesterone,     women. Moreover, it appears this PMDD-related hyper-
and testosterone were assessed at each testing session. Pain sensi-
tivity was measured from electrocutaneous threshold/tolerance,
                                                                        algesia may not be phase-dependent (ie, not during
ischemic threshold/tolerance, sensory and affective ratings of elec-     late-luteal phase only); rather, it persists throughout the
trocutaneous and ischemic stimuli, and the nociceptive flexion           menstrual cycle.3 These disturbances of pain processing
reflex threshold (NFR, a measure of spinal nociception).                 may explain why clinical pain symptoms (eg, mastalgia,
                                                                        headaches, musculoskeletal pain) are among those that
Results: Women with PMDD had higher sensory pain ratings of             contribute to PMDD-related disability.6–8
electrocutaneous stimuli and trends for lower ischemic thresholds
and higher affective pain ratings of electrocutaneous stimuli.
                                                                              Although the specific mechanisms contributing to
However, there were no group differences observed in NFR                 enhanced pain in PMDD are unknown, abnormalities in
threshold. Testosterone levels were also lower during the mid-fol-      sex hormones may play a role.8 Supporting this is evidence
licular and ovulatory phases in PMDD. Correlations between pain         that premenstrual symptoms are: (1) reduced by oopho-
outcomes and estradiol and testosterone indicated that these hor-       rectomy9,10 and use of ovulation inhibitors11,12; and (2)
mones are hypoalgesic, with estradiol having a greater hypoalgesic      absent during ovulatory cycles.13 Sex hormones are also
effect within the PMDD group.                                            related to pain processing in women without PMDD. For
Discussion: Overall, women with PMDD may have a phase-inde-             instance, a recent study from our laboratory found that
pendent hyperalgesia, with pain amplification likely occurring at        testosterone was associated with hypoalgesia on several
the supraspinal level rather than the spinal level, given the lack of   experimental pain outcomes in healthy women (Bartley
group differences in NFR threshold. Because testosterone was             et al, unpublished data). Several studies have also shown
hypoalgesic and lower in women with PMDD, and there were                that testosterone plays a protective role in pain and
strong associations between pain and estradiol in PMDD, sex             inflammation in animals,14–16 and having androgen levels
hormones may play a role in PMDD-related hyperalgesia.                  that are too low (especially testosterone) may promote
Key Words: pain sensitivity, sex hormones, premenstrual dysphoric       chronic pain in both men and women.17 Although estradiol
disorder, menstrual cycle, nociceptive flexion reflex                     and progesterone appear to also influence pain in healthy
                                                                        women, the results are not consistent across studies. Some
(Clin J Pain 2015;31:304–314)                                           find that they are hypoalgesic and others find they are
                                                                        hyperalgesic.18 Although it is unclear as to why divergence
                                                                        exists across studies, methodological differences may
                                                                        account for these contradictory findings.
                                                                              To our knowledge, only 1 study has examined the
Received for publication September 3, 2013; revised May 14, 2014;
   accepted April 17, 2014.
                                                                        relationship between sex hormones and experimental pain
From the Department of Psychology, University of Tulsa, Tulsa, OK.      in PMDD. Straneva et al3 assessed estradiol and ischemia
Supported by a Grant (HR09-080) from the Oklahoma Center for the        pain thresholds/tolerances during the mid-follicular and
   Advancement of Science and Technology (OCAST) awarded to             late-luteal phases of the menstrual cycle and observed
   J.L.R. The authors declare no conflict of interest.
Present address: Emily J. Bartley, PhD, College of Dentistry, Pain
                                                                        hyperalgesia in PMDD, but found no relationships between
   Research and Intervention Center of Excellence, University of        estradiol and pain outcomes. However, it is important to
   Florida, Gainesville, FL.                                            point out that their study assessed a single pain modality
Reprints: Jamie L. Rhudy, PhD, Department of Psychology, University     (ie, ischemia) and only examined estradiol. Further, they
   of Tulsa, 800 South Tucker Drive, Tulsa, OK 74104 (e-mail: jamie-
   rhudy@utulsa.edu).
                                                                        did not study pain during the ovulatory phase, which is
Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved.       associated with enhanced pain inhibition in women without
DOI: 10.1097/AJP.0000000000000112                                       PMDD.19,20 Thus, it is unclear whether a sex hormone/pain

304 | www.clinicalpain.com                                                        Clin J Pain      Volume 31, Number 4, April 2015
Clin J Pain      Volume 31, Number 4, April 2015                                        Sex Hormones and Pain Processing in PMDD

relationship would be found with a more comprehensive                session. Diagnosis was then confirmed at the end of the
assessment of pain and hormones.                                     study after reviewing daily symptoms reported on the
      The purpose of the current study was to examine the            Prospective Record of the Impact and Severity of Men-
relationships between sex hormones (estradiol, progester-            strual Symptoms (PRISM) calendar.23 The PRISM calen-
one, testosterone) and measures of experimental pain in              dar was used to record daily affective (eg, depressed),
women with and without PMDD during the mid-follicular,               behavioral (eg, insomnia), and physical (eg, breast tender-
ovulatory, and late-luteal menstrual cycle phases. Noci-             ness) symptoms; lifestyle impact; life events; basal body
ceptive responding was measured from electrocutaneous                temperature; and luteinizing hormone (LH) test results.
pain threshold/tolerance, ischemia pain threshold/toler-             Participants completed calendars daily for 3 consecutive
ance, sensory and affective ratings of electrocutaneous and           menstrual cycles (during which pain testing occurred) for
ischemia pain, and the nociceptive flexion reflex threshold            verification of cycle regularity and to prospectively diag-
(NFR, a physiological measure of spinal nociception).                nose PMDD. To discourage retrospective reporting, par-
Assessment of NFR in PMDD is innovative because it can               ticipants were asked to mail in calendars on a weekly basis.
be used to determine whether hyperalgesia stems from                       Women in the PMDD group were required to meet the
sensitization at the spinal level. We predicted that PMDD            following criteria: (1) at least a 30% increase in Z5 PMDD
would be associated with phase-independent hyperalgesia              symptoms from the follicular phase (days 5 to 10) PRISM
and that testosterone would be hypoalgesic. Given equiv-             ratings to the luteal phase (1 to 6 d before menses) PRISM
ocal evidence for the effects of estradiol and progesterone,          ratings2; (2) at least one of the symptoms had to be
no directional hypotheses were made.                                 depressed mood, anxiety/tension, mood lability, anger, or
                                                                     irritability, and there had to be a symptom-free period
                  MATERIALS AND METHODS                              during the follicular phase; (3) a 30% increase in functional
                                                                     impairment in work/school or social activities/relationships
Participants                                                         from the follicular to the luteal phase PRISM ratings, with
      Healthy women with PMDD (PMDD; n = 14) and                     lifestyle impact items rated at least “moderate” in intensity.
healthy controls without PMDD (HC; n = 14) were                      Because of the brief nature of the study requirements (ie,
recruited from the surrounding community by radio/                   completion of only 3 cycle calendars), participants were
newspaper advertisements, online advertisements, OB/                 said to have PMDD as long as they met criteria for at least
GYN doctors, flyers, and email distribution. Participants             2 of the 3 monitored cycles (even if they were non-
were initially evaluated for exclusion criteria using a brief        consecutive). Women were considered as controls if they:
phone screen; however, a more thorough evaluation/inter-             (1) did not meet criteria for PMDD; (2) experienced only
view was conducted during their initial session. Participants        mild affective symptoms during the late-luteal phase (pre-
were excluded for: less than 18 years of age; menopausal or          menstrual) days; and (3) experienced only moderate phys-
postmenopausal; use of hormone preparations in the last 6            ical symptoms on 35 (due to diffi-               Instruments, Austin, TX) equipped with dual monitors and
culty getting a nociceptive reflex in persons with high               A/D board (National Instruments, PCI-6036E) controlled
adiposity); history of cardiovascular, neuroendocrine, or            all stimuli, questionnaire presentation, and data acquis-
neurological disorders; Raynaud disease; hypertension;               ition. Physiological signals for NFR and experimental
history of, or current chronic pain; current opioid, anti-           timing were monitored by an experimenter in an adjacent
depressant, or anxiolytic medication use; apparent cogni-            room by use of a 17-inch flat panel monitor. Questionnaires
tive impairment; or current Axis I pathology defined by the           were presented by an LCD projector onto a large screen
DSM-IV-TR1 (assessed from the Structured Clinical                    positioned approximately 2 m in front of the participant,
Interview for DSM-IV Axis I Disorders, Nonpatient Ver-               and sound attenuating headphones and a video camera
sion, SCID-I/NP).21                                                  allowed the experimenter to communicate with and mon-
                                                                     itor the participant from the adjoining room. Electro-
Prospective Diagnoses of PMDD                                        cutaneous stimulations were delivered to the left ankle over
      Confirmation of PMDD was based upon DSM-IV-                     the retromalleolar pathway of the sural nerve by use of a
TR1 diagnostic criteria for the disorder, which is defined as         Digitimer stimulator (model DS7A) and bipolar stimulat-
having Z5 of the following 11 symptoms present for 2                 ing electrode (Nicolet, 019-401400, Madison, WI). A com-
consecutive symptomatic cycles: (1) depressed mood,                  puter controlled the timing of the stimulations and the
hopelessness, or self-deprecating thoughts; (2) anxiety,             maximum stimulation intensity was set at 50 mA to ensure
tension; (3) affective lability; (4) persistent and marked            participant safety. Physiological signals for recording the
anger or irritability or increased interpersonal conflicts; (5)       NFR were amplified and filtered online using Grass
decreased interest in usual activities; (6) difficulty concen-         Instruments Model 15LT amplifiers (with AC Modules
trating; (7) lethargy, easy fatigability, or marked lack of          15A54 and DC Modules 15A12). A Lafayette Instrument
energy; (8) change in appetite, overeating, or specific food          Hand Dynamometer (Models 78010 and 78011, Lafayette,
cravings; (9) hypersomnia or insomnia; (10) overwhelmed              IN) and Prestige Medical Cuff (Northridge, CA) were used
or feeling out of control; or (11) physical symptoms. At             to assess ischemia pain.
least 1 symptom must be mood-related (eg, irritability,
depressed mood) and symptoms must interfere with work,               Hormone Assessment
school, or relationships/usual social activities.                         Urinary LH surge tests (Clearblue Easy; Swiss Pre-
      To provide a preliminary diagnosis of PMDD, a                  cision Diagnostic, Bedford, United Kingdom) were con-
semistructured interview22 was conducted during the initial          ducted at home by participants to verify and identify the

Copyright     r    2014 Wolters Kluwer Health, Inc. All rights reserved.                           www.clinicalpain.com |     305
Bartley et al                                                                        Clin J Pain      Volume 31, Number 4, April 2015

timing of ovulation. The test provides a positive or negative     fossa, with a common ground electrode placed over the
reading for the LH surge and is over 99% accurate. Par-           lateral epicondyle of the left femur. The raw biceps femoris
ticipants were asked to take a digital photo of each positive     signal was amplified, bandpass filtered online (10 to
test and email it to experimenters immediately for ver-           300 Hz), and rectified.
ification of compliance and scheduling of testing sessions.             NFR threshold was determined using validated pro-
      Saliva was collected on days of pain testing during the     cedures.31,32 Trains of five 1 ms rectangular wave pulses at
mid-follicular, ovulatory, and late-luteal phases. Salivary       250 Hz were delivered to the sural nerve with a varying
assessment was chosen over serum-based/plasma-based               intertrain interval of 8 to 12 seconds to reduce stimulus
measurement as it represents a noninvasive and cost-effec-         predictability. The first train began at 0 mA (current) and
tive medium for measuring sex hormone levels, and is              was increased in 2 mA steps until an NFR was detected.
reliable and relatively easy to collect.24–27 To avoid poten-     The mean biceps femoris EMG activity 90 to 150 ms post-
tial salivary contamination that could impact immuno-             stimulation that exceeds mean EMG activity during the
assay, participants were asked to abstain from: (1) alcohol       60 ms prestimulus baseline interval by 1.4 SD was used to
use 12 hours before sample collection; (2) eating a meal          define the presence of the NFR.31 Prior research has shown
60 minutes before testing; and (3) brushing their teeth           that using the 90 to 150 ms timeframe avoids potential
45 minutes before saliva was collected. Approximately             contamination by the non-nociceptive RII reflex, startle
10 minutes before sample collection, participants rinsed          responses, and/or voluntary movements.33,34 The stimulus
their mouths with water to eliminate any potential con-           intensity was decreased in 1-mA steps until an NFR was no
taminants in their saliva. Per specifications of Salimetrics       longer observed. Then, this updown staircase procedure
LLC (State College, PA),28 participants were initially asked      was repeated twice using 1-mA steps. NFR threshold was
to allow saliva to pool in their mouth. Then, they were           defined as the average stimulation intensity (mA) of the last
required to passively salivate down a 2-inch straw that was       2 peaks and troughs of this updown staircase procedure.
placed into a plastic test tube. Samples were refrigerated
within 30 minutes after collection and stored in a freezer        Electrocutaneous Pain Threshold and Tolerance
kept at or below 201C. Samples were shipped on dry ice                  Electrocutaneous pain was assessed by delivering
to Salimetrics LLC for analysis of estradiol, progesterone,       electrical stimulation to the sural nerve. This method
and unbound testosterone levels. On the day the samples           evokes pain through the activation of both A-d and C fibers
were to be assayed, they were thawed to room temperature,         (although it also activates A-b fibers), and produces a
vortexed, and then centrifuged for 15 minutes at approx-          stinging or pricking sensation.35 Electrocutaneous pain
imately 3000 rpm (1500g). Samples were tested for estradiol       threshold was assessed through 3 ascending/descending
using a high-sensitivity enzyme immunoassay (Cat. No.             staircases of electric stimuli. Stimulus parameters were the
1-3702).                                                          same as those used in NFR threshold. Following each
      The test had a lower limit of sensitivity of 0.1 pg/mL, a   electric stimulus, participants rated their sensation using a
standard curve range from 1.0 to 32.0 pg/mL, an average           computer-presented numerical rating scale (NRS) used in
intra-assay coefficient of variation of 7.1%, and an average        numerous prior studies by our laboratory and others.36,37
interassay coefficient of variation 7.5%. The test assessing        The scale was labeled: 0 (no sensation), 1 (just noticeable),
progesterone (Cat. No. 1-1502) had a lower limit of sensi-        25 (uncomfortable), 50 (painful), 75 (very painful), and 100
tivity of 5.0 pg/mL, standard curve range from 10 to              (maximum tolerable). Starting with 0 mA (current), the
2430 pg/mL, an average intra-assay coefficient of variation         current was increased in 4-mA steps, with a variable 8- to
of 6.2%, and an interassay coefficient of variation of 7.6%.        12-second interval, until pain threshold was reached (a
Assessment of unbound testosterone (Cat. No. 1-2402,              rating Z50 on the 0 to 100 NRS). When pain threshold
1-2312) had a lower limit of sensitivity of 1.0 pg/mL,            was reached, the intensity was decreased in 2-mA steps until
standard curve range from 6.1 to 600 pg/mL, an average            the participant rated a stimulus as 40 on the scale. This
intra-assay coefficient of variation of 4.6%, and an average        process was repeated 2 more times in 2-mA steps. Pain
interassay coefficient of variation of 9.8%. Each hormone           threshold was defined as the average of the 4 stimuli (mA)
assay was conducted twice and averaged to obtain more             immediately above and immediately below a rating of 50 on
reliable estimates of the hormone levels. Per Salimetrics         the last 2 ascending/descending staircases.
LLC, acceptance criteria for duplicate hormone results was              Electrocutaneous pain tolerance was assessed by a
a coefficient of variation
Clin J Pain      Volume 31, Number 4, April 2015                                        Sex Hormones and Pain Processing in PMDD

to the right biceps of the arm and inflated to 220 mm/Hg to           electrocutaneous tolerance, ischemic pain threshold and tol-
occlude blood flow and induce ischemia pain. Participants             erance assessment, and SF-MPQ rating of ischemia tolerance.
were then asked to continuously (in real-time) rate the              Data from the emotional controls of nociception and con-
intensity of the blood pressure cuff on the NRS described             ditioned pain modulation procedures will be reported else-
above. The time (seconds) until they reached pain threshold          where. Although the order of procedures was fixed, partic-
(rating Z50 on the NRS) and tolerance (rating of 100 on              ipants were given scheduled rest breaks in between tasks to
the NRS) was assessed. To ensure participant safety, a               allow them time to recover before the next task and to
maximum cutoff of 25 minutes was used.39                              minimize any possible carryover effects. The length of these
                                                                     scheduled breaks did not vary across participants. At the
McGill Pain Questionnaire-Short Form (SF-MPQ)                        culmination of the session, each participant was reminded to
     The SF-MPQ40 is a reliable and valid pain rating scale          continue menstrual phase monitoring until 3 cycles were
commonly used in pain research that allows quantitative,             completed. Participants were provided an honorarium at the
multidimensional pain ratings to be obtained in a brief              end of the experiment (or upon withdrawal from the study)
period. Participants were asked to rate 11 sensory (eg,              with the amount dependent upon the amount of the study
throbbing, shooting) and 4 affective (eg, sickening, fearful)         completed (up to $375).
pain descriptors on a scale ranging from 0 (none) to 3
(severe). The SF-MPQ was administered twice, once                    Data Analysis
immediately following electrocutaneous pain tolerance and                  Significance was set at Pr0.05 (2-tailed). Group dif-
once immediately following ischemia pain tolerance. A sum            ferences on participant characteristics and background
of sensory words and affective words was used to compute              variables were conducted using independent samples t tests
sensory and affective pain rating scores, respectively, for           or w2 analyses. Group and menstrual phase differences in
each stimulus modality.                                              hormones and pain outcomes were analyzed using linear
                                                                     mixed models (MIXED procedure, SPSS 21.0) to increase
Procedure                                                            statistical power and ensure cases with missing data were
      All procedures were approved by the University of              not excluded.41 The data file was in “long form,” meaning
Tulsa ethics review board. Before study enrollment, a brief          that each participant contributed up to 3 rows of data (1
phone health screening was conducted to evaluate inclusion/          row for each testing session) to each analysis. Participant
exclusion criteria; however, a more comprehensive assess-            ID was used as the grouping variable to address the issue of
ment was conducted during an initial laboratory session with         nonindependence. The repeated-measures error structure
those individuals who met inclusion criteria via the phone           was modeled as a first-order autoregressive structure (AR1)
screen. During this initial laboratory session, participants         and models included a random intercept to allow outcomes
were given a complete overview of the study and informed             to vary across individuals. Menstrual phase was entered as
consent was obtained. Then, a brief demographics/health              a nominal within-subjects variable and group (PMDD vs.
status form was used to assess inclusion/exclusion criteria and      HC) was entered as a nominal between-subjects variable.
attain relevant background information. Afterwards, height           Dependent variables were NFR threshold, electrocutaneous
and weight were assessed to determine body mass index. To            pain threshold and tolerance, ischemia pain threshold and
evaluate potential PMDD status, a semistructured inter-              tolerance, and McGill sensory and affective pain ratings to
view22 was conducted, and then the SCID-I was administered           electrocutaneous and ischemic stimuli. For analyses of pain
to assess and exclude for the presence of current psychiatric        outcomes, testing order (ie, session 1 to 3) was entered as a
conditions (other than PMDD). If deemed eligible to par-             variable to determine whether testing order influenced the
ticipate, participants were given instructions on menstrual          pain outcomes independently of the menstrual phases.
cycle monitoring and then randomly assigned to one of 6              Controlling for order improves statistical power and
testing orders: (1) mid-follicular—ovulatory—late-luteal; (2)        improves the validity of the statistical models by removing
mid-follicular—late-luteal—ovulatory; (3) ovulatory—mid-             potential variance due to exposure effects. Follow-up mean
follicular—late-luteal; (4) ovulatory—late-luteal—mid-fol-           comparisons to significant F tests were conducted using
licular; (5) late-luteal—mid-follicular—ovulatory; and (6)           Fisher Least Significant Difference tests. The SPSS MIXED
late-luteal—ovulatory—mid-follicular.                                procedure uses Satterthwaite estimation for the denomi-
      Pain testing sessions were scheduled within the mid-           nator degrees of freedom (df) which produced noninteger
follicular, ovulatory, and late-luteal phases of the partic-         values that vary from analysis to analysis. These dfs were
ipant’s menstrual cycle, with testing order counterbalanced          rounded to the nearest integer for ease of reporting.
between-subjects. The mid-follicular phase was defined as             Cohen’s d was reported as the effect size for mean
days 5 to 8 following menses onset, the ovulatory phase was          comparisons.
defined as the 2 days following a positive LH surge, and the                To determine whether hormone levels were associated
late-luteal phase was defined as days 1 to 6 preceding                with pain outcomes, Pearson correlations were conducted
menses (or 9 to 11 d postpositive LH surge).                         separately for each group, and for the combined sample.
      Upon arrival at each pain testing session, a complete          Fisher r-to-z analyses were conducted to determine whether
overview of the procedures was provided again, followed by           there were group differences in the correlations.
review of informed consent and health status. Participants
were provided instruction for filling out the NRS for pain                                     RESULTS
and then physiological sensors were applied. Participants
were next asked to sit quietly for 5 minutes to acclimate.           Participant Characteristics
The rest of the session included pain testing in the following            Table 1 reports demographic as well as other relevant
order: NFR threshold, electrocutaneous pain threshold,               menstrual cycle-related variables for PMDD and HC par-
emotional controls of nociception, conditioned pain mod-             ticipants. Results revealed no significant differences
ulation, electrocutaneous pain tolerance, SF-MPQ rating of           between the 2 groups on any of the variables. Overall, 28

Copyright     r    2014 Wolters Kluwer Health, Inc. All rights reserved.                           www.clinicalpain.com |    307
Bartley et al                                                                                         Clin J Pain        Volume 31, Number 4, April 2015

TABLE 1. Demographic and Clinical Characteristics of PMDD and HC Participants
                                                     PMDD                                             HC                             Group Comparison
Characteristics                       M (or %)          SD          Range            M (or %)          SD             Range            P              d
Age                                      31.1           8.6         18-46              29.3            7.4            20-39           0.56           0.22
Ethnicity (% non-Hispanic)               92.9%                                         85.7%                                          0.33
Race (% white)                           64.3%                                         64.3%                                          0.60
Years of education                       15.7           2.1         12-20              15.5            2.7            12-21           0.84           0.08
Marital status (% married)               28.6%                                         28.6%                                          0.99
Employment (% employed)                  42.9%                                         57.1%                                          0.61
Body mass index (kg/m2)                  24.2           4.8       18.0-33.1            25.7            5.0           18.9-34.2        0.42           0.31
Cycle length (d)                         29.6           4.8       21.0-48.0            28.5            3.4           22.0-42.0        0.49           0.02
Luteal phase length (d)                  14.2           4.8       6.0-38.0             14.8            1.8           11.0-23.0        0.65           0.17
   HC indicates healthy control; PMDD, premenstrual dysphoric disorder.

females completed mid-follicular phase testing (1 HC par-                          (P < 0.001, d = 0.66) and late-luteal (P = 0.01, d = 0.39)
ticipant completed only partial testing), 27 completed                             phases, whereas progesterone levels were lowest during the
ovulatory phase testing, and 28 completed late-luteal phase                        mid-follicular phase relative to ovulatory (P < 0.001,
testing. Eight females (4 HC, 4 PMDD) attended their first                          d = 1.02) and late-luteal (P < 0.001, d = 1.29) phases. The
testing session during their mid-follicular phase, 9 females                       main effect of group and the Menstrual phase Group
(4 HC, 5 PMDD) attended their first testing session during                          interaction for estradiol and progesterone were not
the ovulatory phase, and 11 females (6 HC, 5 PMDD)                                 significant (Ps > 0.05). The main effect of group was sig-
attended their first testing session during their late-luteal                       nificant for testosterone, but this was qualified by a
phase. Of the 28 women who completed the mid-follicular                            significant Menstrual phase Group interaction indicating
testing session, 27 (96.4%) were tested during days 5 to 8                         that the PMDD group had lower testosterone levels than
(M = 7.25, SD = 1.14) and all were tested before ovulation.                        the HC group during the mid-follicular (P = 0.04, d = 0.71)
Of the 28 women who completed the late-luteal testing                              and ovulatory (P = 0.006, d = 1.17) phases, but not the
session, 25 (89.2%) were tested 1 to 6 days preceding                              late-luteal phase (P = 0.65, d = 0.17). The main effect of
menses (M = 4.04, SD = 3.32) and all were tested after                             menstrual phase was nonsignificant for testosterone
ovulation and during an ovulatory cycle. Reasons for not                           (P > 0.05).
testing within the designated timeframe include inability to
predict the onset of menses from prior cycles and LH
                                                                                   Measures of Experimental Pain Sensitivity
surges, as well as menses lasting >8 days. All 27 women
tested during the ovulatory phase were tested in the 48                                 Descriptive and inferential statistics for measures of
hours immediately following a positive ovulation test.                             experimental pain sensitivity are reported in Table 3. There
                                                                                   were no significant main effects of menstrual phase for any
                                                                                   of the pain outcomes (Ps > 0.05), although there was a
Sex Hormone Levels                                                                 nonsignificant trend for ischemia pain threshold (P = 0.06).
     Descriptive and inferential statistics for estradiol,                         Pairwise comparisons indicated that thresholds were lower
progesterone, and testosterone are presented in Table 2.                           during the late-luteal phase relative to the mid-follicular
The main effect of menstrual phase for estradiol and pro-                           phase (P = 0.02, d = 0.41).
gesterone was significant. Specifically, estradiol levels were                            For the main effect of group, all measures of pain
higher during the ovulatory phase than mid-follicular                              sensitivity were nonsignificant (Ps > 0.05), with the

TABLE 2. Descriptive and Inferential Statistics for Sex Hormone Levels Across Menstrual Phase and Group
                                                                                                                            F
                                                                                                                                              Phase Main
                    Mid-Follicular                  Ovulation                      Late-Luteal         Phase (P) Group (G) P¾G
                                                                                                                                             Effect Pairwise
Hormones        M      SD      Range         M       SD       Range         M       SD      Range             F                 F     F        Contrast
E2 (pg/mL)
  PMDD      3.3        1.3     1.8-7.0        4.3      2.2 0.77-10.5         3.6     1.1    1.7-5.9
  HC        3.8        0.95    1.6-5.1        5.2      2.4 1.8-9.9           4.4     1.5    1.5-6.9          7.71*          1.92     0.44    O > F, O > L
Prog (pg/mL)
  PMDD 63.9           45.5    21.7-173.1 159.4      111.2 22.7-336.2 156.7 97.7 31.8-398.9
  HC       70.3       38.8    16.7-148.3 115.3       46.6 48.8-176.5 165.5 94.6 55.9-414.8               14.22*             0.21     1.82    F < O, F < L
Testo (pg/mL)
  PMDD 55.4w          18.4    15.9-78.4     48.3w    16.6 22.9-82.0         54.7 15.7 27.0-73.4
  HC       70.9w      24.5    32.2-130.9    69.2w    20.3 22.1-87.0         58.0 21.4 32.8-106.9             2.26           4.21*    5.24*
    *Pr0.05.
    wGroup simple effect, Pr0.05.
    E2 indicates estradiol; F, mid-follicular; HC, healthy control; L, late-luteal; O, ovulation; PMDD, premenstrual dysphoric disorder; Prog, progesterone;
Testo, testosterone.

308 | www.clinicalpain.com                                              Copyright      r   2014 Wolters Kluwer Health, Inc. All rights reserved.
Clin J Pain      Volume 31, Number 4, April 2015                                                          Sex Hormones and Pain Processing in PMDD

TABLE 3. Descriptive and Inferential Statistics for Measures of Experimental Pain Sensitivity Across Menstrual Phase and Group
                               Mid-Follicular                        Ovulation                  Late-Luteal              Phase (P)   Group (G)      P¾G
Pain Outcomes                M             SD      Range      M         SD      Range       M       SD       Range          F             F           F
NFR Thr (0-50 mA)
   PMDD               13.7                 11.5   2.5-35.8 15.2        12.9    1.5-46.5     13.5   11.9      2.3-46.5
   HC                 15.6                  9.3   2.5-30.5 16.0         7.9    3.5-27.3     16.2    9.9      5.0-35.3      0.21         0.26        0.21
Elec Thr (0-50 mA)
   PMDD               13.1                  5.9    4-27.5 16.0         10.6    3.0-39.5     13.5     7.5     4.5-30.0
   HC                 16.1                  8.4    7-39.5 16.2         10.9    5.5-39.5     16.4     9.3     5.0-39.5      0.99         0.47        0.98
Elec Tol (0-50 mA)
   PMDD               22.8                 12.8    8-50     23.3       14.5     8-50        22.7   12.9       8-50
   HC                 25.8                 10.1    10-50    30.7       11.9     14-50       28.9   11.4       14-50        2.16         1.73        1.64
Isch Thr (0-1200 s)
   PMDD               51.8                 31.9     9-97 44.7          19.9     16-76       35.6   23.3      10-75
   HC                 63.3                 38.1    13-159 56.3         38.3     15-159      52.4   35.2      10-159        2.92w        3.09w       0.16
Isch Tol (0-1200 s)
   PMDD              118.0                 64.7    32-227 140.7        82.5     32-315     115.8   75.0      16-282
   HC                161.8                 94.3    39-389 160.3        98.0     76-389     145.0   72.5      76-328        1.89         1.41        0.70
MPQ Sens Elec (0-33)
   PMDD               17.4                  5.7     8-24    16.3         5.2     7-24       16.9     6.4      8-28
   HC                 11.1                  6.5     2-23    10.9         5.6     4-23       11.3     5.8      4-27         0.29         9.47*       0.12
MPQ Aff Elec (0-12)
   PMDD                4.4                  3.0     0-8       5.4        4.0     0-12        5.1     4.0      0-12
   HC                  3.0                  3.6     0-10      2.2        3.2     0-10        2.7     3.3      0-9          0.13         3.70w       2.29
MPQ Sens Isch (0-33)
   PMDD               17.9                  7.0     4-28    18.4         7.6     5-30       19.1     7.3      4-29
   HC                 14.1                  7.4     7-28    15.1         6.9     6-27       15.4     5.6      6-26         1.18         2.36        0.07
MPQ Aff Isch (0-12)
   PMDD                5.0                  3.8     0-11      5.6        4.0     0-12        4.8     3.7      0-12
   HC                  3.1                  3.0     0-9       3.6        3.3     0-9         3.5     3.5      0-11         0.80         2.06        0.36
    *Pr0.05.
    wPr0.10.
    Aff indicates affective; Elec, electrocutaneous; HC, healthy control; Isch, ischemia; MPQ, McGill Pain Questionnaire-Short Form; NFR, nociceptive flexion
reflex; PMDD, premenstrual dysphoric disorder; Sens, sensory; Thr, threshold; Tol, tolerance.

exception of the PMDD group having higher MPQ sensory                          was primarily due to medium and large correlations in the
ratings for electrocutaneous stimuli relative to the HC                        PMDD group (all these correlations were significant in
group (P = 0.005, d = 1.00). Further, there was a non-                         PMDD group, but nonsignificant in the HC group). The
significant trend for ischemia pain threshold (P = 0.09,                        Fisher r-to-z analysis indicated that the groups significantly
d = 0.43) and MPQ affective electrocutaneous ratings                            differed on correlations with ischemia pain threshold and
(P = 0.06, d = 0.64) indicating that the PMDD group had                        ischemia pain tolerance (Ps < 0.05). Thus, estradiol
lower thresholds and higher affective pain ratings than the                     appears to have been hypoalgesic in the PMDD group.
HC group. The Menstrual phase Group interaction was                                 No correlations between progesterone and pain out-
nonsignificant for all of the pain sensitivity outcomes (all                    comes were significant in the combined group, although
Ps > 0.05).                                                                    there was a nonsignificant trend for the relationship with
      Results also indicated main effects of testing order (ie,                 electrocutaneous pain tolerance (P < 0.07). This was pri-
pain varied across the 3 testing sessions, but independent of                  marily because of a medium-sized and significant correla-
menstrual phase) for MPQ sensory ischemia ratings                              tion in the PMDD group, but the Fisher r-to-z failed to
(F1,25 = 5.79, P = 0.02) and MPQ affective ischemia ratings                     show a group difference (P > 0.05). There was also a
(F1,25 = 5.20, P = 0.03). Specifically, sensory ischemia ratings                nonsignificant trend for progesterone to be related to elec-
(P = 0.04, d = 0.28) and affective ischemia ratings (P = 0.02,                  trocutaneous pain threshold in the PMDD group
d = 0.29) were higher during the participant’s third exper-                    (P = 0.10). Thus, it appears that progesterone has minimal
imental testing session relative to their first experimental                    relations to these pain outcomes.
session, suggesting sensitization across testing sessions.                           Testosterone was significantly related to electro-
                                                                               cutaneous pain threshold, electrocutaneous pain tolerance,
Relationship Between Sex Hormones and Pain                                     ischemia pain tolerance, and affective pain ratings of electro-
Sensitivity                                                                    cutaneous stimuli in the combined group (all Ps < 0.05;
     Correlations between sex hormones and pain out-                           Fig. 2). These correlations were of similar magnitude in both
comes are presented in Table 4. Estradiol was significantly                     the groups, as confirmed by the Fisher r-to-z analyses (all
related to electrocutaneous pain threshold, electro-                           Ps > 0.05). But, when subgroup analyses were conducted,
cutaneous pain tolerance, ischemia pain threshold, and                         only the relationship between testosterone and electro-
ischemia pain tolerance in the combined group (all Ps <                        cutaneous pain threshold in the PMDD group was significant
 0.05). However, as can be seen in Table 4 and Figure 1, this                  (P < 0.05). According to the Fisher r-to-z analysis, there was

Copyright     r    2014 Wolters Kluwer Health, Inc. All rights reserved.                                                www.clinicalpain.com |      309
Bartley et al                                                                                    Clin J Pain      Volume 31, Number 4, April 2015

TABLE 4. Correlations Between Sex Hormones and Pain Outcomes by Group
                               Correlations With Estradiol          Correlations With Progesterone                 Correlations With Testosterone
Pain Outcomes              HC       PMDD        ALL         z       HC       PMDD          ALL         z           HC       PMDD     All       z
NFR Thr (0-50 mA)           0.074    0.228      0.105      1.33      0.060       0.128     0.100      0.30         0.117    0.252w   0.084    1.64w
Elec Thr (0-50 mA)          0.090    0.373*     0.243*     1.32      0.092       0.251w    0.085      1.52         0.258w   0.303*   0.300*   0.22
Elec Tol (0-50 mA)          0.145    0.339*     0.275*     0.90      0.035       0.332*    0.200w     1.34         0.200    0.159    0.222*   0.18
Isch Thr (0-1200 s)         0.111    0.532*     0.300*     2.09*     0.150       0.020     0.070      0.57         0.063    0.239    0.175    0.79
Isch Tol (0-1200 s)         0.018    0.503*     0.257*     2.32*     0.011       0.060     0.046      0.21         0.167    0.205    0.223*   0.17
MPQ Sens Elec (0-33)        0.165    0.045      0.002      0.53      0.076       0.006     0.004      0.36         0.130    0.054    0.185    0.80
MPQ Aff Elec (0-12)          0.051    0.237      0.153      1.27      0.084       0.217     0.135      0.59         0.289w   0.229    0.327*   0.28
MPQ Sens Isch (0-33)        0.063    0.096      0.063      0.69      0.117       0.040     0.039      0.68         0.246    0.203    0.115    1.99*
MPQ Aff Isch (0-12)          0.030    0.241      0.150      1.20      0.098       0.093     0.078      0.02         0.231    0.003    0.180    1.04
    *Pr0.05.
    wPr0.10.
    ALL indicates all participants combined; HC, healthy control; PMDD, premenstrual dysphoric disorder; z, Fisher r-to-z value that tests the group
differences in correlations.

a significant group difference in the correlation between tes-                 group (P = 0.10), implying that testosterone might be asso-
tosterone and sensory pain ratings of ischemia stimuli                       ciated with dampened spinal nociception in this group.
(P < 0.05), but neither of the correlations were significant in               Therefore, it appears that testosterone has a small to moder-
the subgroups. There was also a nonsignificant trend for tes-                 ate relationship with some pain outcomes in both groups
tosterone to be correlated with NFR threshold in the PMDD                    suggesting it has a hypoalgesic effect.

                                                                         HC
                                                                         PMDD

                           A                                                 B

                           C                                                 D

FIGURE 1. Significant relationships between estradiol and pain outcomes. Estradiol was significantly associated with electrocutaneous
pain threshold (A), electrocutaneous pain tolerance (B), ischemia pain threshold (C), and ischemia pain tolerance (D) in the combined
group. The magnitudes of these correlations were larger in the PMDD group, but only the correlations between estradiol and ischemia
pain threshold (C) and estradiol and ischemia pain tolerance (D) were significantly different according to Fisher r-to-z analyses. Thus,
estradiol was hypoalgesic, primarily so in the PMDD group. HC indicates healthy control; PMDD, premenstrual dysphoric disorder.

310 | www.clinicalpain.com                                          Copyright      r   2014 Wolters Kluwer Health, Inc. All rights reserved.
Clin J Pain      Volume 31, Number 4, April 2015                                           Sex Hormones and Pain Processing in PMDD

                                                                    HC
                                                                    PMDD

                            A                                          B

                            C                                          D

FIGURE 2. Significant relationships between testosterone and pain outcomes. Testosterone was significantly associated with electro-
cutaneous pain threshold (A), electrocutanous pain tolerance (B), ischemia pain threshold (C), and affective pain ratings of electro-
cutaneous stimuli (D). None of the correlations significantly differed by group; therefore, testosterone was associated with hypoalgesia
in both of the groups. HC indicates healthy control; PMDD, premenstrual dysphoric disorder.

Secondary Analyses: Examining Testosterone as a                                               DISCUSSION
Mediator of Diagnostic Group and Sensory Pain
      As testosterone was associated with hypoalgesia and             PMDD-related Differences in Pain Processing
PMDD had lower overall levels of testosterone, it is plausible             The current study assessed the relationships between
that sex hormones may contribute to pain in PMDD. This                sex hormones (estradiol, progesterone, testosterone) and
may be especially true for the enhanced sensory pain ratings          experimental pain sensitivity across 3 menstrual phases in
observed within the PMDD group. To address this, a post hoc           women with and without PMDD. The only statistically
mediation analysis was conducted to examine whether testos-           significant difference that emerged was higher sensory pain
terone significantly mediated the relationship between-group           ratings for electrocutaneous stimuli relative to healthy
(PMDD vs. HC) and MPQ sensory ratings for electro-                    women; however, inspection of the means for pain outcomes
cutaneous stimuli. Follow-up analyses of mediation were               (Table 3) suggest that, compared with control women,
conducted using a mixed-effects regression model entering              women with PMDD had lower thresholds/tolerances and
group (PMDD vs. HC) as the predictor, sensory pain ratings            rated noxious stimuli as more painful. This pattern of results
as the outcome, and testosterone as the hypothesized mediator.        is consistent with 2 prior studies that found severe pre-
To be considered a mediator, the relationship of the predictor        menstrual symptomatology was associated with no differ-
to the outcome must be reduced (or become nonsignificant)              ence in measures of thresholds or tolerances, but higher pain
once the mediator (testosterone) is entered into the model.42         ratings.5,43 Specifically, Kuczmierczyk et al5 found that
      Initial analyses examining the effect of group on MPQ            women with severe premenstrual syndrome (PMS, a less
ratings was significant (F1,28 = 9.47, P = 0.005, b = 5.63).           severe form of PMDD) exhibited higher ratings of pressure
When the analysis was repeated including testosterone as a            pain relative to healthy women even though there were no
predictor variable, the group effect was no longer significant          significant differences for pressure pain thresholds and tol-
(F1,78 = 3.23, P = 0.076, b = 6.26), suggesting that testo-           erances. Similarly, Klatzkin et al43 found that women with
sterone may partially mediate group differences in MPQ                 current PMDD (plus a history of major depressive disorder)
sensory ratings for electrocutaneous stimuli.                         experienced greater pain unpleasantness to a cold pressor

Copyright     r    2014 Wolters Kluwer Health, Inc. All rights reserved.                               www.clinicalpain.com |     311
Bartley et al                                                                       Clin J Pain      Volume 31, Number 4, April 2015

task than women without PMDD (but with a history of              testosterone appears to exert an antinociceptive effect on
major depressive disorder). This was noted even though           pain in both animals14,16,53,54 and humans,55 and testoster-
there were no group differences on ischemic or cold pressor       one therapy in women can improve libido, mood, and well-
pain thresholds and tolerances. By contrast, 2 studies did       being.56 To our knowledge, the use of testosterone as a
find PMDD-related hyperalgesia on measures of threshold/          therapy in women with PMDD-related pain has not been
tolerance, but the effect sizes for group differences on pain      investigated, likely due to the potential for treatment-related
ratings tended to be even larger than the effect sizes for        side effects (ie, masculinization).56,57 Therefore, this may
threshold/tolerance measures.2,3 This suggests that PMDD         require further research to determine whether testosterone
may affect the retrospective report of pain more so than          is, in fact, pathophysiologically relevant to PMDD and to
threshold/tolerance measures, perhaps due to PMDD-               assess whether testosterone treatment (eg, transdermal tes-
related affect influencing memory for pain more so than            tosterone) might help PMDD-related symptoms.
real-time evaluation of painful stimuli.44                             Interestingly, groups differed in the degree to which sex
      Two additional pain-related observations from the          hormones modulated experimental pain outcomes. For
present study are worth noting. First, we found no evidence      instance, estradiol had a hypoalgesic effect on thresholds and
that enhanced experimental pain sensitivity in PMDD was          tolerances for electrocutaneous and ischemic stimuli; how-
specific to the premenstrual (late-luteal) phase, because all     ever, only the associations between estradiol and ischemia
group by phase interactions for pain outcomes were non-          thresholds/tolerances varied across groups. This suggests
significant (Table 3). This is consistent with a prior study      that estradiol had a more robust hypoalgesic effect in women
that assessed pain across multiple menstrual phases in           with PMDD. These results are in contrast to a study that
PMDD3 and found no changes in PMDD-related hyper-                failed to observe an association between estradiol and
algesia across menstrual phases. Second, NFR threshold, a        ischemia pain thresholds/tolerances in PMDD.3 Although it
physiological measure of spinal nociceptive processing, did      is unclear why our results differ, methodological differences
not differ between groups (and Cohen’s d effect sizes were         may have contributed (eg, serum vs. salivary assays, assess-
small, ranging from 0.08 to 0.25 across phases) or across        ment during ovulatory phase). Progesterone was not related
menstrual phases. The current findings corroborate a recent       to any of our pain outcomes, with the exception of electro-
study that found NFR does not vary across the mid-fol-           cutaneous pain tolerances in PMDD. However, this rela-
licular and late-luteal phases of the menstrual cycle in         tionship did not differ when comparing across groups. This
healthy women45 and extends those findings to show that it        is interesting as it suggests that progesterone may have
also does not vary during ovulation or by PMDD status.           minimal influence on experimental pain. Further, as pre-
Because the spinally mediated NFR was not significantly           dicted, testosterone appeared to have a hypoalgesic effect on
correlated with estradiol, progesterone, or testosterone, it     electrocutaneous thresholds/tolerances, ischemia tolerances,
also appears that these sex hormones may not affect spinal        and affective ratings for electrocutaneous stimuli. These
nociceptive processing. Therefore, findings from the current      relationships did not differ across groups, suggesting tes-
study and others2,3,5,43 suggest that enhanced pain in           tosterone was associated with hypoalgesia in both PMDD
PMDD may be: (1) most pronounced on measures of pain             and HC. Although there have been few investigations of
report; (2) phase-independent; and (3) more likely to stem       testosterone’s effects on pain in women, this is in line with
from amplification of pain at the supraspinal level (eg,          research suggesting testosterone attenuates pain sensi-
corticocortical mechanisms, report bias) rather than the         tivity58,59 and protects against clinical pain.17,60
spinal level. However, given our small sample size, caution            Our findings also suggest that sex hormones may con-
is warranted in interpreting our null findings.                   tribute to pain in PMDD because testosterone was hypo-
                                                                 algesic, and women with PMDD had lower levels of testos-
Hormones and Pain Processing                                     terone. To examine this further, a post hoc mediation analysis
      Estradiol and progesterone levels indicated that testing   was conducted to assess whether testosterone significantly
sessions were conducted during the correct phases. Although      mediated the relationship between diagnostic group and MPQ
the expected phase-related changes in these hormones were        sensory ratings for electrocutaneous stimuli, given group dif-
noted, there were no group differences. In contrast, unbound      ferences in this variable (B = 5.63, P = 0.005). When anal-
(bioavailable) testosterone did not show phase-related           yses were repeated using testosterone as a predictor, the group
changes, but did show group differences. Testosterone was         effect was no longer significant (B = 6.26, P = 0.076) indi-
significantly lower in women with PMDD, particularly dur-         cating that testosterone may partially mediate group differ-
ing the mid-follicular and ovulatory phases. Unfortunately,      ences in sensory pain ratings for electrocutaneous stimuli. This
data on sex hormone levels in women with menstrual cycle-        suggests testosterone may play a role in group differences in
related disorders is equivocal. Three studies found PMS was      pain and could be a potential target for treatment.
not associated with differences in total (bound + unbound)              When taken together, results indicate estradiol and
testosterone.46–48 Two studies found that PMS was not            testosterone were antinociceptive, but that progesterone
associated with differences in total testosterone49,50 but that   had a minimal effect on experimental pain sensitivity.
unbound testosterone was higher during follicular,49 ovula-      Further, some relationships between sex hormones and
tory,49 and luteal49,50 phases. And finally, 1 study found that   pain outcomes varied across groups (particularly with
PMS was associated with lower plasma levels of total and         estradiol). Although the current findings suggest sex hor-
unbound testosterone during the ovulatory phase.51               mones may contribute to PMDD, it is important to note
      Only 1 study has examined testosterone in women            that some evidence suggests that PMDD may be related to
prospectively diagnosed with PMDD with the authors               individual sensitivity to normal cyclical fluctuation in sex
finding a nonsignificant trend for unbound, but not total,         hormones (rather than absolute hormone levels)—effects
testosterone to be lower during the ovulatory phase.52 If our    which may influence neurotransmitter responsiveness to
results can be replicated, this suggests that low testosterone   trigger symptomatology.61 Regardless of the etiology, our
could potentially contribute to PMDD symptoms. Indeed,           findings highlight the complexity of sex hormone influences

312 | www.clinicalpain.com                                 Copyright   r   2014 Wolters Kluwer Health, Inc. All rights reserved.
Clin J Pain      Volume 31, Number 4, April 2015                                          Sex Hormones and Pain Processing in PMDD

in the modulation of pain and suggest that sex hormones              Future studies may consider counterbalancing pain proce-
may impact experimental pain differently in PMDD.                     dures to circumvent this issue.
Future studies are needed to clarify the specific effect that
sex hormones have on pain and PMDD, and to investigate               Summary
whether testosterone and estradiol therapy, or a combina-                  In sum, the current results indicate that women with
tion of both, may be a therapeutic option when treating              PMDD have phase-independent hyperalgesia on sensory
pain-related symptoms in PMDD.                                       pain ratings of electrocutaneous stimuli. This hyperalgesia
                                                                     is likely to occur at the supraspinal level, because there were
                                                                     no group differences in NFR (and between-group effect
Strengths and Limitations
                                                                     sizes were very small). Further, sex hormones may con-
      There are several strengths of this study that merit           tribute to pain in PMDD because testosterone was hypo-
acknowledgment including the prospective diagnosis of                algesic and women with PMDD had lower testosterone. In
PMDD, measurement of hormones, verification of cycle                  addition, stronger relationships between pain and estradiol
regularity, and assessment of multiple pain outcomes                 were found in women with PMDD. However, future studies
(including NFR). Despite these strengths, some limitations           are needed to explore whether exogenous administration of
are worth noting. First, our failure to detect significant            sex hormones influence pain in PMDD.
group differences (with the exception of sensory ratings for
electrocutaneous stimuli) may have been impacted by low                                  ACKNOWLEDGMENTS
power. Thus, some meaningful group differences may have
been missed. For example, post hoc power analyses                         The authors thank the members of the Psychophysiology
revealed that group sizes between 22 and 148 would have              Laboratory for Affective Neuroscience (PLAN) at the Uni-
resulted in power equal to 0.80 for outcomes that were               versity of Tulsa, Department of Psychology, Tulsa, OK, for
marginally significant (ie, MPQ affective electrocutaneous             their assistance with data collection. Research team members
ratings, ischemia threshold), but at least 50 or more (up to         include Satin Martin, MA; Lauren Guderian, BA; Abby
43,485 per group) would have been needed for other non-              Greenhaw, BA, and Yvette Guereca, BA
significant outcomes. So, caution is warranted when inter-
preting the nonsignificant findings given our small sample                                       REFERENCES
size. Second, even though several experimental measures               1. APA. Diagnostic and Statistical Manual of Mental Disorders,
were used to assess pain sensitivity in the present study, it is         4th Edition Text Revision. Arlington, VA: American Psychi-
possible that the stimuli used were not adequate to assess               atric Association; 2000.
the clinical phenomenon of interest in PMDD. For                      2. Fillingim RB, Girdler SS, Booker DK, et al. Pain sensitivity in
instance, pressure stimulation evokes a natural pain sen-                women with premenstrual dysphoric disorder: a preliminary
                                                                         report. J Women Health. 1995;4:367–374.
sation, activating both A and C fibers, as well as noci-
                                                                      3. Straneva PA, Maixner W, Light KC, et al. Menstrual cycle,
ceptors of the muscles and tendons.35 Thus pressure may be               beta-endorphins, and pain sensitivity in premenstrual dys-
more suitable for evoking pain similar to PMDD-related                   phoric disorder. Health Psychol. 2002;21:358–367.
symptomatology (eg, mastalgia, headache, musculoskeletal              4. Fleischman DS, Bunevicius A, Leserman J, et al. Menstrually
pain). For this reason future studies might consider other               related mood disorders and a history of abuse: moderators of pain
pain-induction methods (eg, mechanical pressure pain,                    sensitivity. Health Psychol. 2014;33:147–154.
heat/cold) and/or symptoms (eg, emotion dysregulation) to             5. Kuczmierczyk AR, Adams HE, Calhoun KS, et al. Pain
study in this population.35,62 Third, the women who par-                 responsivity in women with premenstrual syndrome across the
ticipated in our study may not be representative of other                menstrual cycle. Percept Mot Skills. 1986;63:387–393.
women with PMDD. The study required significant par-                   6. Steiner M, Born L. Advances in the diagnosis and treatment of
                                                                         premenstrual dysphoria. CNS Drugs. 2000;13:287–304.
ticipant involvement to track menstrual cycles, attend                7. Di Giulio G, Reissing ED. Premenstrual dysphoric disorder:
multiple testing sessions, and endure several pain tests.                prevalence, diagnostic considerations, and controversies.
Further, our classification of PMDD was based upon a                      J Psychosom Obstet Gynaecol. 2006;27:201–210.
modified version of the diagnosis (ie, criteria did not have           8. Halbreich U. The etiology, biology, and evolving pathology of
to be met for 2 consecutive cycles). As a result, our sample             premenstrual syndromes. Psychoneuroendocrinology. 2003;28:55–99.
may have been more resilient or less symptomatic than                 9. Casper RF, Hearn MT. The effect of hysterectomy and
other women with PMDD. Fourth, we chose to measure                       bilateral oophorectomy in women with severe premenstrual
hormone levels from saliva rather than blood, because                    syndrome. Am J Obstet Gynecol. 1990;162:105–109.
                                                                     10. Cronje WH, Vashisht A, Studd JWW. Hysterectomy and
saliva-based assays are noninvasive, inexpensive, and easy
                                                                         bilateral oophorectomy for severe premenstrual syndrome.
to collect.24–27 Further, they measure the biologically                  Hum Reprod. 2004;19:2152–2155.
available fraction of hormones in the body. For this reason,         11. Bancroft J, Boyle H, Warner P, et al. The use of an LHRH
they may provide a more valid estimate of hormones                       agonist, buserelin, in the long-term management of premen-
available to affect target tissues (and thus pain process-                strual syndromes. Clin Endocrinol. 1987;27:171–182.
ing).24,25,63 Nonetheless, a limitation of saliva-based assays       12. Hammarback S, Backstrom T. Induced anovulation as treat-
is their difficulty detecting low hormone levels, especially in            ment of premenstrual tension syndrome. A double-blind cross-
nonreproductive females. As a result, our study may not                  over study with GnRH-agonist versus placebo. Acta Obstet
accurately characterize the relationships between hormones               Gynecol Scand. 1988;67:159–166.
and pain in women with very low hormone levels. And                  13. Hammarback S, Ekholm UB, Backstrom T. Spontaneous
                                                                         anovulation causing disappearance of cyclical symptoms in
finally, we chose not to counterbalance the order of our                  women with the premenstrual syndrome. Acta Endocrinol.
pain tasks within the testing sessions as this could have                1991;125:132–137.
potentially resulted in carryover effects. We attempted to            14. Aloisi AM, Ceccarelli I, Fiorenzani P, et al. Testosterone
minimize this problem by building in mandatory rest peri-                affects formalin-induced responses differently in male and
ods and placing the most intense tolerance tests at the end.             female rats. Neurosci Lett. 2004;361:262–264.

Copyright     r    2014 Wolters Kluwer Health, Inc. All rights reserved.                               www.clinicalpain.com |       313
Bartley et al                                                                            Clin J Pain      Volume 31, Number 4, April 2015

15. Ceccarelli I, Scaramuzzino A, Massafra C, et al. The              40. Melzack R. The short-form McGill Pain Questionnaire. Pain.
    behavioral and neuronal effects induced by repetitive nocicep-        1987;30:191–197.
    tive stimulation are affected by gonadal hormones in male rats.   41. Hox JJ. Multilevel Analysis. Mahwah, NJ: Lawrence Erlbaum
    Pain. 2003;104:35–47.                                                 Associates; 2002.
16. Fischer L, Clemente JT, Tambeli CH. The protective role of        42. Baron RM, Kenny DA. The moderator–mediator variable
    testosterone in the development of temporomandibular                  distinction in social psychological research: conceptual, strate-
    joint pain. J Pain. 2007;8:437–442.                                   gic, and statistical considerations. J Pers Soc Psychol. 1986;51:
17. Cairns BE, Gazerani P. Sex-related differences in pain.               1173–1182.
    Maturitas. 2009;63:292–296.                                       43. Klatzkin RR, Lindgren ME, Forneris CA, et al. Histories of
18. Craft RM. Modulation of pain by estrogens. Pain. 2007;132:            major depression and premenstrual dysphoric disorder: evidence
    S3–S12.                                                               for phenotypic differences. Biol Psychol. 2010;84:235–247.
19. Tousignant-Laflamme Y, Marchand S. Excitatory and inhib-          44. Gedney JJ, Logan H. Memory for stress-associated acute pain.
    itory pain mechanisms during the menstrual cycle in                   J Pain. 2004;5:83–91.
    healthy women. Pain. 2009;146:47–55.                              45. Bartley EJ, Rhudy JL. Comparing pain sensitivity and the
20. Rezaii T, Hirschberg AL, Carlström K, et al. The influence           nociceptive flexion reflex threshold across the mid-follicular
    of menstrual phases on pain modulation in healthy women.              and late-luteal menstrual phases in healthy women. Clin J Pain.
    J Pain. 2012;13:646–655.                                              2013;29:154–161.
21. First MB, Spitzer RL, Gibbons M, et al. Structured Clinical       46. Rubinow DR, Hoban MC, Grover GN, et al. Changes in
    Interview for DSM-IV Axis I Disorders. New York, NY:                  plasma hormones across the menstrual cycle in patients with
    Biometrics Research Department, New York State Psychiatric            menstrually related mood disorder and in control subjects. Am
    Institute; 2002.                                                      J Obstet Gynecol. 1988;158:5–11.
22. Endicott J, Nee J, Harrison W. Daily Record of Severity of        47. Backstrom T, Aakvaag A. Plasma prolactin and testosterone
    Problems (DRSP): reliability and validity. Arch Womens Ment           during the luteal phase in women with premenstrual tension
    Health. 2006;9:41–49.                                                 syndrome. Psychoneuroendocrinology. 1981;6:245–251.
23. Reid RL. Premenstrual syndrome. In: Leventhal JM, Hoffman JJ,     48. Watts JF, Butt WR, Logan Edwards R, et al. Hormonal
    Keith LG, eds. Current Problems in Obstetrics, Gynecology, and        studies in women with premenstrual tension. Br J Obstet
    Fertility. Chicago: Year Book Medical Publishers; 1985:345–353.       Gynaecol. 1985;92:247–255.
24. Gann PH, Giovanazzi S, Van Horn L, et al. Saliva as a             49. Eriksson E, Sundblad C, Lisjo P, et al. Serum levels of
    medium for investigating intra- and interindividual differences       androgens are higher in women with premenstrual irritability
    in sex hormone levels in premenopausal women. Cancer                  and dysphoria than in controls. Psychoneuroendocrinology.
    Epidemiol Biomarkers and Prev. 2001;10:59–64.                         1992;17:195–204.
25. Bellem A, Meiyappan S, Romans S, et al. Measuring estrogens       50. Lombardi I, Luisi S, Monteleone P, et al. Adrenal response
    and progestagens in humans: an overview of methods. Gend              to adrenocorticotropic hormone stimulation in patients with
    Med. 2011;8:283–299.                                                  premenstrual syndrome. Gynecol Endocrinol. 2004;18:79–87.
26. Greenspan JD, Craft RM, LeResche L, et al. Studying sex and       51. Bloch M, Schmidt PJ, Su TP, et al. Pituitary-adrenal hormones and
    gender differences in pain and analgesia: a consensus report.         testosterone across the menstrual cycle in women with premen-
    Pain. 2007;132:S26–S45.                                               strual syndrome and controls. Biol Psychiatry. 1998;43:897–903.
27. Quissell DO. Steroid hormone analysis in human saliva. Ann N      52. Eriksson O, Landén M, Sundblad C, et al. Ovarian morphol-
    Y Acad Sci. 1993;694:143–145.                                         ogy in premenstrual dysphoria. Psychoneuroendocrinology.
28. Salimetrics. Saliva Collection and Handling Advice. 2009; LLC:        2012;37:742–751.
    Salimetrics.                                                      53. Gaumond I, Arsenault P, Marchand S. Specificity of female and
29. Skljarevski V, Ramadan NM. The nociceptive flexion reflex in          male sex hormones on excitatory and inhibitory phases of formalin-
    humans. Pain. 2002;96:3–8.                                            induced nociceptive responses. Brain Res. 2005;1052:105–111.
30. Sandrini G, Serrao M, Rossi P, et al. The lower limb flexion      54. Hau M, Dominguez OA, Evrard HC. Testosterone reduces
    reflex in humans. Prog Neurobiol. 2005;77:353–395.                    responsiveness to nociceptive stimuli in a wild bird. Horm
31. Rhudy JL, France CR. Defining the nociceptive flexion reflex          Behav. 2004;46:165–170.
    (NFR) threshold in human participants: a comparison of            55. Stillman MJ. Testosterone replacement therapy for treatment
    different scoring criteria. Pain. 2007;128:244–253.                   refractory cluster headache. Headache. 2006;46:925–933.
32. France CR, Rhudy JL, McGlone S. Using normalized EMG to           56. Davis SR, Tran J. Testosterone influences libido and well being
    define the nociceptive flexion reflex (NFR) threshold: further        in women. Trend Endocrinol Metab. 2001;12:33–37.
    evaluation of standardized NFR scoring criteria. Pain. 2009;      57. Radestad AF. Testosterone treatment in women—an overview.
    145:211–218.                                                          Curr Women Health Rev. 2009;5:29–43.
33. Dowman R. Possible startle response contamination of the          58. Okifuji A, Turk DC. Sex hormones and pain in regularly
    spinal nociceptive withdrawal reflex. Pain. 1992;49:187–197.          menstruating women with fibromyalgia syndrome. J Pain.
34. France CR, Suchowiecki S. Assessing supraspinal modulation            2006;7:851–859.
    of pain perception in individuals at risk for hypertension.       59. Teepker M, Peters M, Vedder H, et al. Menstrual variation in
    Psychophysiology. 2001;38:107–113.                                    experimental pain: correlation with gonadal hormones. Neuro-
35. Arendt-Nielsen L, Lautenbacher S. Assessment of pain percep-          psychobiology. 2010;61:131–140.
    tion. In: Lautenbacher S, Fillingim RB, eds. Pathophysiology of   60. Kaergaard A, Hansen AM, Rasmussen K, et al. Association
    Pain Perception. New York, NY: Plenum Publishers; 2004:25–42.         between plasma testosterone and work-related neck and
36. France CR, France JL, al’Absi M, et al. Catastrophizing is            shoulder disorders among female workers. Scand J Work
    related to pain ratings, but not nociceptive flexion reflex           Environ Health. 2000;26:292–298.
    threshold. Pain. 2002;99:459–463.                                 61. Cunningham J, Yonkers JA, O’Brien S, et al. Update on
37. Rhudy JL, Williams AE, McCabe K, et al. Affective                     research and treatment of premenstrual dysphoric disorder.
    modulation of nociception at spinal and supraspinal levels.           Harv Rev Psychiatry. 2009;17:120–137.
    Psychophysiology. 2005;42:579–587.                                62. Gingnell M, Morell A, Bannbers E, et al. Menstrual cycle effects
38. Fillingim RB, Maixner W. The influence of resting blood               on amygdala reactivity to emotional stimulation in premenstrual
    pressure and gender on pain responses. Psychosom Med.                 dysphoric disorder. Horm Behav. 2012;62:400–406.
    1996;58:326–332.                                                  63. Lu Y, Bentley GR, Gann PH, et al. Salivary estradiol and
39. Fillingim RB, Maixner W, Girdler SS, et al. Ischemic but not          progesterone levels in conception and nonconception cycles in
    thermal pain sensitivity varies across the menstrual cycle.           women: evaluation of a new assay for salivary estradiol. Fertil
    Psychosom Med. 1997;559:512–520.                                      Steril. 1999;71:863–868.

314 | www.clinicalpain.com                                      Copyright   r   2014 Wolters Kluwer Health, Inc. All rights reserved.
You can also read