"There are two major classes of living organisms-male and female. In many cases, they are so different in form and habit that one might well be excused the thought that males and females are different species."
-Darcy B. Kelley1
(1) Kelley DB. The genesis of male and female brains. Trends Neurosci 1986;9:499-502.
Nonsteroidal antiinflammatory drugs (NSAIDs) are widely used, although clinical response is variable [1]. Because variability in clinical response has often been attributed to the episodic nature of painful disorders, we have previously studied the effects of NSAIDs on stable, experimental inflammation [2] and still found considerable intra- and inter-subject variability. Gender differences were not the explanation, but they could be relevant in the analgesic variability, in the light of studies that reported gender-based differences in nociception [4,5] (2) and actions of opioids [6-9]. Few studies have assessed variability in the analgesic response to NSAIDs, however. Therefore, we focused on the analgesic actions of NSAIDs in seeking a basis for the response variability in the gender of the subjects. Using cutaneous electrical stimulation of the earlobe and the tolerance end point, we were able to discriminate the effects of both ibuprofen and diflunisal from those of placebo [10]. As a result, in the current study, we chose ibuprofen and this proven method of inducing consistent pain sensations. We also investigated the relationship between the analgesic effect and the drug concentration to determine whether there are any gender differences in the pharmacokinetics or pharmacodynamics of this drug.
(2) Rollman GB, Hapidou EG, Jarmain SH. Gender differences in pain responsiveness: contributing factors [abstract]. Pain 1990; (Suppl 5):314.
Methods
The subjects were healthy, pain-free volunteers, 18-30 yr of age. Their written, informed consent was obtained in conformity with our institutions' guidelines on human experimentation. Twenty age-matched subjects (10 male, 61-107 kg [mean 81 kg]; and 10 female, 40-66 kg [mean 57 kg]) participated; they were paid for their participation and were free to withdraw at any time. Before each subject's entry into the study, a full medical history was taken, a clinical examination was performed, a blood sample taken for routine hematology and biochemical screening, and a urine sample was examined for glucose and total protein (all performed at St. Vincent's Hospital). Significant abnormality resulting from these tests led to exclusion of that subject (no subjects were excluded). Other exclusion criteria were pregnancy; regular use of medication, especially analgesics (confirmed by chromatography of the subjects' plasma); abuse of alcohol; participation in a trial of an investigational drug in the preceding 4 wk or during the study; surgery within the previous 3 mo; or symptoms of a clinically important illness within 4 wk of the study (no subjects were excluded). All of the female subjects had normal menstrual cycles, and adequate precautions against pregnancy were part of the ethical approval. No significance was attached to whether subjects were taking oral contraceptives, because, in either case, hormone levels should clearly distinguish the male from the female subjects, and because others have found no influence of the menstrual cycle on sex-related differences in analgesia [6,7] or of oral contraceptives on ibuprofen pharmacokinetics [11]. Furthermore, because the experiment was randomized, the women's replicate exposures to drug and placebo were at different stages of their menstrual cycles, so the possible influences of menstrual timing should have been minimal.
Each subject was familiarized with the analgesic testing apparatus for 1 h 1 wk before the study. Consistency of reporting is crucial for the reliability of the method; therefore, whenever a subject terminates the noxious stimuli, pain perception should always be the same. Changes in stimulus intensity at the end point therefore do not represent changes in pain perception; rather, they reflect altered sensitivity of the system (i.e., an increased voltage at the end point indicates an analgesic effect because a greater stimulus is necessary to reach the same level of perception).
Subjects were requested to have at least 8 h of sleep the night before each experimental session. They were asked to refrain from alcohol, caffeine, and analgesic drugs for 24 h before each experimental day and, in the morning, to eat a light carbohydrate breakfast 2 h before the beginning of the observation period. Testing was conducted in a temperature-controlled, sound-attenuated room between 8 AM and 5 PM. The same trained observer (JSW) supervised the pain measurements during the day. Subjects sat upright in a comfortable chair and were allowed to read between the hourly testing sessions. Food (a light snack or lunch) and orange juice were provided at 2-h intervals starting 1 h postmedication. The insistence on breakfast and the provision of a standardized diet during the experiments were intended to minimize any possible gastrointestinal disturbances or possible influence of plasma glucose concentration on nociception [12].
Every participant was trained in a standardized manner by JSW. They were told that a NSAID was involved but, to reduce possible bias, were not informed that a placebo would be administered [2,13]. They were also told, in general terms, that the aim of the study was to establish whether responses to NSAIDs differed among individuals. Each subject was provided with a pro forma for the recording of side effects and intercurrent illness of any kind.
The details and advantages of our method for achieving nociceptive stimulation have been published previously [10]. In particular, the application of the cutaneous electrodes to the earlobe reduces detection thresholds and avoids artifacts and record instability resulting from the muscle contraction that may occur with stimulation sites that overlie muscle (e.g., on the hand), from variations in skin impedance that follow from variation in skin thickness (consequences of differences in gender or in use of the hands), and stress-induced sweating. When stimulation reaches the maximum that subjects will tolerate [commonly called pain tolerance, and having a burning quality most likely caused by activity in unmyelinated C-fiber nociceptors [14]], they report a sensation closely resembling clinical pain. This is a better indicator than the electrically induced pain threshold, the voltage at which they first report that the sensation is painful [10].
Surface electrodes (pregelled, pediatric, disposable silver/silver chloride electrocardiogram electrodes, 34-mm diameter; Medicotest, Olstykke, Denmark) were affixed to each surface of the cleaned earlobe on the subject's nondominant side to avoid dextrality effects [15]; the stimulator was connected to the main power line via a hospital standard (10 mA) current leakage detection unit. The test stimuli were 1-ms square wave pulses delivered continuously at 20 Hz; the amplitude, initially 1 V, was increased by 1 V every 2 s. During their exposure to this stimulus train, the subjects indicated (by pressing a button) three grades of perception: 1 = when the stimulus was first perceived (a "tingling" sensation); 2 = when the stimulus became sharp and painful (pain threshold); 3 = when the sensation had become deep and burning and no further increase in stimulus intensity was accepted by the subject (pain tolerance). The stimulus was terminated by the third button press; with each press, the data were captured by the laboratory computer, which controlled the stimulus delivery.
The determination of these three levels of intensity represented one stimulus train. One stimulus block consisted of four successive stimulus trains and took approximately 5-10 min to complete for each subject. Starting at 8 AM, pain was measured 3 times, at half-hourly intervals, before the administration of the medication (baseline) and at hourly intervals for 8 h thereafter (until 5 PM). Our previous description of the methodology [10] detailed the technical question of whether constant-current or constant-voltage stimulation is preferable; we found that variability was substantially less, and reproducibility therefore greater, when constant-voltage stimulation was used. The present study improved the technique by delivering the stimuli via a calibrated, computer-driven stimulator (designed by Richard Troughgear, Biomedical Engineer, St Vincent's Hospital, Sydney who had provided the expertise in physics during the development of the model).
Subjects were randomly assigned, in a double-blind, cross-over design, to receive either ibuprofen (800 mg: four capsules, each 200 mg; ACT3; Wyeth, Sydney, Australia) on two occasions or matched placebo (four capsules of identical appearance) on two other occasions. The four treatment sequences were ibuprofen-ibuprofen-placebo-placebo, ibuprofen-placebo-ibuprofen-placebo, placebo-ibuprofen-placebo-ibuprofen, or placebo-placebo-ibuprofen-ibuprofen; successive treatments were at weekly intervals. This dose was chosen on the basis of our previous studies, in which 800 mg was effective against experimentally induced pain or inflammation [2,10], but it did not induce any discernible sensation of drug action in the subjects [16]. This dose is smaller than those used in other studies [17], which also reported no side effects. Because there is little evidence of clear dose-effect relationships with these drugs [18], the use of a replicate treatment plan, in preference to different doses, added rigor to the data.
For determination of ibuprofen concentrations, blood was obtained from a vein in the antecubital fossa (by JJC). The first (zero time) was taken before drug administration; subsequent samples were obtained 1, 4, 5, 6, 7, and 8 h after drug administration, although technical difficulties sometimes precluded the successful collection of all samples. This collection protocol was chosen to minimize disturbance of the subjects in the early part of the experiment. The total ibuprofen concentration was determined by using high-performance liquid chromatography as described previously [10]. The limit of detection was 1 [micro sign]g/mL, and repeatability and reproducibility in terms of the coefficients of variation were between 3.5% and 10% for concentrations of 100 [micro sign]g/mL and 1 [micro sign]g/mL, respectively.
Data are expressed as mean +/- SEM. The baseline values for each pain intensity (threshold and tolerance) were determined from the average of the three pretreatment blocks of stimuli. Analgesia was calculated as an increase in the voltage required to produce the pain threshold or pain tolerance sensations, i.e., postdrug minus predrug stimulus magnitude [10]. Possible gender differences in baseline pain were analyzed by using one-way analysis of variance. The effects of treatment were analyzed using two-factor, repeated-measures analysis of covariance that incorporated baseline pain as a covariate; the factors (sequence of treatment and gender) were considered fixed. Post hoc analyses were performed on preplanned comparisons using Fisher's least squares difference multiple comparison tests. These comparisons were: (i) placebo 1 with placebo 2, ibuprofen 1 and ibuprofen 2; placebo 2 with ibuprofen 1 and ibuprofen 2; ibuprofen 1 with ibuprofen 2 (within-gender); and (ii) placebo 1; placebo 2; ibuprofen 1; and ibuprofen 2 or placebo (between-gender).
Blood concentration-time data for each subject were used to derive pharmacokinetic parameters using noncompartmental methods. Total plasma clearance was calculated as the dose divided by the area under the plasma concentration-time curve extrapolated to infinity. The apparent volume of distribution at steady state and the terminal half-life of ibuprofen were calculated by using moment analysis and linear regression of the terminal plasma concentration data, respectively. Linear regression analysis (Pearson's correlation) was used to investigate possible correlations between the area under the serum concentration-time curve and the analgesic effect, with values of P < 0.05 considered statistically significant.
All statistical computations were performed by using the Number Crunching Statistical System (NCSS, Kaysville, Utah); values of P < 0.05 were considered statistically significant.
Results
No subject reported any side effects after the administration of ibuprofen or placebo; further, no subjects were excluded from the investigation, although one failed to complete one arm of the study (placebo 2).
Neither ibuprofen nor placebo had any effect on pain detection (pain threshold; response 2), although ibuprofen not placebo, had a significant effect on pain tolerance (response 3; Table 1). Pain tolerance is thus the relevant measure, and the experimental dose was appropriate. To test our principal hypothesis, the data were partitioned according to gender. Both pain detection and tolerance were significantly higher (P < 0.05) in the male subjects than in the female subjects (Table 2). Further, after the drug treatment, only men showed any response (Figure 1): the analgesia measure (posttreatment stimulus voltage - pretreatment voltage) after ibuprofen was 2.80 +/- 0.33 V, whereas it was -0.18 +/- 0.34 V after placebo (P < 0.05; n = 10).
;)
Table 1: Mean Analgesic Response
Table 2: Baseline Values
Figure 1: Mean analgesic response after either placebo or ibuprofen (800 mg, single dose) in male and female healthy volunteers. Data were pooled from both exposures to ibuprofen or placebo. The mean analgesic response (volts) is the difference between the mean predrug pain tolerance voltage (baseline) and the postdrug values at each exposure.
(Figure 2) shows the time course of the pain tolerance measurements. As expected, such values show some variation over the course of a day, but they are always positive in the men receiving ibuprofen, whereas the results are not different from zero in the female subjects; there is no net effect during the placebo treatment in either gender.
Figure 2: Mean time course of analgesic effect in healthy volunteers after two oral treatments with either placebo (open symbols) or ibuprofen (closed symbols) (800 mg, single dose). For reasons of clarity, error bars are not shown; overall SEM was calculated from residual variance in the analysis of variance matrix and was 0.38 for the male subjects' data and 0.15 for the female subjects' data. The squares represent first treatments and the circles the second. Data are presented separately according to gender. The mean analgesic response (volts) was calculated as for
Figure 1.
Peak plasma concentrations were achieved by the first hour in all subjects and were not affected by treatment sequence (ibuprofen 1 versus ibuprofen 2: 39 +/- 5 vs 42 +/- 6 [micro sign]g/mL) or gender (male versus female: 45 +/- 6 vs 37 +/- 6 [micro sign]g/mL; Figure 3); the time courses of the plasma concentrations also show no gender differences (Figure 3). The pharmacokinetic variables of ibuprofen were not different among treatment sequences; therefore, the data for both active phases were pooled (Table 3). On an absolute basis, there were no gender differences in the pharmacokinetic variables of ibuprofen (Table 3), but after body weight was taken into account in the calculations, the volume of drug distribution (VD) was twice as high in women compared with men (P < 0.05; Table 3). In the male subjects, there was a consistent level of analgesia throughout the observation period (Figure 2), but there was no correlation between this analgesia and the area under the plasma concentration-time curve (r = 0.27).
Figure 3: Mean plasma concentrations of ibuprofen after drug treatment (800 mg, single dose) in male (--[square]--) and female ([horizontal bar][black square][horizontal bar]) subjects. Pooled data from both exposures to ibuprofen are presented for clarity.
Table 3: Mean Pharmacokinetic Parameters of Ibuprofen According to Gender
Discussion
Our technique is a development of the pioneering work of Hallin and Torebjork [19], which distinguished the early and late components of electrically induced pain, which correspond to A delta- and C-fiber activation, respectively. This noninvasive method, by which we seek to achieve a form of clinically relevant pain, is simple and has reinforced its utility in the present study by showing that it is well accepted by the subjects and yields consistent results for periods as long as 8 h (see [10] for details of the validation of this method). The present data also confirm previous reports that pain tolerance measures are sensitive to various analgesics, both opioid and nonopioid [10,14]. This is most likely because electrical stimulation at intensities that elicit pain tolerance sensations can produce a pain experience that closely resembles that of clinical pain by activating unmyelinated nociceptive C fibers, whereas pain detection (threshold) probably involves insensitive A delta activation [10,14,19].
There are, however, fundamental differences between experimental pain (in the absence of inflammation) and clinical pain (in which inflammation is usually involved, especially the types of pain for which NSAIDs are prescribed). Inflammation activates a class of "sleeping nociceptors" [20], which are quite different from physiologically operational polymodal C-nociceptors [e.g., they have lower conduction velocities [21]].
There are disparate reports about gender differences in sensitivity to thermal pain in humans [4,5] and in animals [9]. Our results concur with those of Lautenbacher and Rollman [4], who found significant gender differences in baseline electrically induced nociception. Of the several studies that indicate that women have a lower tolerance of pressure pain than men, the results of Ellermeier and Westphal [22] are especially convincing because they used an autonomic response (pupillary reaction) combined with psychophysical measures and found diverging stimulus response plots, with women being more sensitive to noxious pressure stimulation, the disparity increasing with stimulus intensity.
A striking and unexpected result was the lack of any analgesic effect of ibuprofen in our female subjects. Similar results were obtained both of the times ibuprofen was administered. Male and female subjects had similar plasma concentrations (Figure 3), but only the men showed an analgesic effect (Figure 1). The gender difference could not be caused by an inadequate dose of ibuprofen and, therefore, not by the baseline differences, because analgesia is measured as the change in stimulus magnitude. Animal studies support our results: morphine analgesia is significantly greater in male than in female rats [8], and N-methyl-D-aspartate antagonists are more potent in male than in female mice [9].
This gender difference is important because although there is interindividual variability in the analgesic response to NSAIDs [to the extent that patients have been classified as responders and nonresponders [23]], gender has not been considered a possible basis for this. Development of tolerance after multiple doses could also be an explanation for the unsatisfactory response to NSAIDs [16,24], but in the present study, with active drug administered at intervals of 1-3 wk, tolerance hardly seems relevant. Therefore, the fact that there was significant analgesia in the male subjects but not the female subjects, highlights the importance of gender to explain variability in response and the phenomenon of nonresponders.
Our analysis showed no absolute pharmacokinetic differences between male and female subjects, but after taking body weight into account, VD was twofold larger in women (Table 3), consistent with the data of Knights et al. [11]. The gender difference in the magnitude of VD might be a consequence of differences in respective proportions of body water or because of gender-based differences in binding of the drug to plasma proteins [25]; ibuprofen is 99% bound, and a change to 98% would produce the differences in VD that we found. The crucial matter is to achieve equivalent plasma levels; for that reason, women received a larger dose in the present study (14 mg/kg vs 9.8 mg/kg). Few studies have unequivocally related analgesic effects to NSAID concentrations in plasma [1]; we did not find any obvious relationship between plasma concentration and analgesic effect; indeed, analgesia persisted despite declining drug concentrations. That finding supports the hypothesis that an endogenous substance is involved in the analgesic effect, its release being triggered by the drug [1], i.e., drug concentration and analgesia necessarily have different time courses, a phenomenon familiar from the action of steroid hormones.
Although the traditional view of the mode of action of NSAIDs is peripheral inhibition of prostaglandin (PG) synthesis, a recent literature survey revealed little relationship between the ability of NSAIDs to inhibit PG synthesis in vitro and analgesic activity [26]. NSAIDs also exert effects at central sites, e.g., the spinal cord [1]. As we have noted previously [10], electrical stimulation of the skin evokes a flare and increases local skin temperature by more than 1[degree sign]C, but no hyperalgesia results. This effect is probably mediated by the release of vasoactive substances such as substance P, which, in turn, activate PGs-in other words, neurogenic inflammation. Because NSAID suppression of such inflammatory responses shows no gender difference [2], our results suggest that the analgesic and antiinflammatory actions of NSAIDs are achieved at quite distinct sites.
Regardless of the site of action of NSAIDs, the analgesic and antiinflammatory actions seem to be distinct because the gender dependence of analgesia is striking in the present experiments, whereas there is no such disparity in the antiinflammatory action of these drugs.2 The fact that clinicians have not remarked on such differences with NSAID use (and, thus, a gender contribution to the response irregularity has not been canvassed) may indicate that the antiinflammatory action is the more clinically important component of NSAID action. Nevertheless, our finding that NSAIDs are less effective in premenopausal women than in men may have clinical significance.
We thank Richard Troughgear, Biomedical Engineer, St. Vincent's Hospital, Sydney, for his expertise and dedication in designing the electronically controlled stimulator. We also thank Dr. Jeffrey Mogil and Professors Gerhard Levy and Hermann Handwerker for their incisive and informed comments on various drafts of the manuscript. We also thank Wyeth Pharmaceuticals Australia for supplying ibuprofen and matched placebo capsules.
REFERENCES
1. Walker JS. NSAIDs: an update on their analgesic effects. Clin Exp Physiol Pharmacol 1995;22:855-60.
2. Walker JS, Nguyen TV, Day RO. Clinical response to nonsteroidal anti-inflammatory drugs in urate-crystal induced inflammation: a simultaneous study of intersubject and intrasubject variability. Br J Clin Pharmacol 1994;38:1-5.
4. Lautenbacher S, Rollman GB. Sex differences in responsiveness to painful and non-painful stimuli are dependent upon the stimulation method. Pain 1993;53:255-64.
5. Feine JS, Bushnell MC, Miron D, Duncan GH. Sex differences in the perception of noxious heat stimuli. Pain 1991;44:255-62.
6. Gear RW, Gordon NC, Heller PH, et al. Gender difference in analgesic response to the kappa-opioid pentazocine. Neurosci Lett 1996;205:207-9.
7. Gear RW, Miaskowski C, Gordon NC, et al. Kappa-opioids produce significantly greater analgesia in women than in men. Nat Med 1996;2:1248-50.
8. Cicero TJ, Nock B, Meyer ER. Gender-related differences in the antinociceptive properties of morphine. J Pharmacol Exp Ther 1996;279:767-73.
9. Mogil JS, Sternberg WF, Kest B, et al. Sex differences in the antagonism of swim stress-induced analgesia: effects of gonadectomy and estrogen replacement. Pain 1993;53:17-25.
10. Walker JS, Arroyo JF, Nguyen T, Day RO. Analgesic efficacy of non-steroidal anti-inflammatory drugs in experimental pain in humans. Br J Clin Pharmacol 1993;36:417-25.
11. Knights KM, McLean CF, Tonkin AL, Miners JO. Lack of effect of gender and oral contraceptive steroids on the pharmacokinetics of (R)-ibuprofen in humans. Br J Clin Pharmacol 1995;40:153-6.
12. Morley GK, Mooradian AD, Levine AS, Morley JE. Mechanism of pain in diabetic peripheral neuropathy: effect of glucose on pain perception in humans. Am J Med 1984;77:79-82.
13. Wall PD. The placebo response In: Wall PD, Melzack R, eds. Textbook of pain. Edinburgh: Churchill-Livingstone, 1994:1297-308.
14. Cooper BY, Vierck CJ, Yeomans DC. Selective reduction of second pain sensations by systemic morphine in humans. Pain 1986;24:93-116.
15. Haslam D. Lateral dominance in the perception of size and pain. Q J Exp Psychol 1970;22:503-7.
16. Walker JS, Lockton A, Nguyen NV, Day RO. Effect of single and multiple doses of ibuprofen in chronic spinal pain patients. Analgesia 1996;2:93-101.
17. Evans AM, Nation RL, Sansom LN, et al. The relationship between the pharmacokinetics of ibuprofen enantiomers and the dose of racemic ibuprofen in humans. Biopharmacol Drug Dispos 1990;11:507-18.
18. Neilsen JC, Bjerring P, Arendt-Nielsen L, Petterson K-J. A double-blind, placebo controlled, cross-over comparison of the analgesic effects of ibuprofen 400 and 800 mg on laser-induced pain. Br J Clin Pharmacol 1990;30:711-5.
19. Hallin RG, Torebjork HE. Electrically induced A and C fibre responses in intact in human skin nerves. Exp Brain Res 1973;16:309-20.
20. Schaible H-G, Schmidt RF. Time course of mechanosensitivity changes in articular afferents during a developing experimental arthritis. J Neurophysiol 1988;60:2180-95.
21. Schmidt R, Schmelz M, Forster C, et al. Novel classes of responsive and unresponsive C nociceptors in human skin. J Neurosci 1995;15:333-41.
22. Ellermeier W, Westphal W. Gender differences in pain ratings and pupil reactions to painful pressure stimuli. Pain 1995;61:435-9.
23. Walker JS, Sheather-Reid R, Carmody JJ, et al. Non steroidal anti-inflammatory drugs (NSAIDs) in rheumatoid and osteoarthritis: support for the concept of "responders" and "non-responders." Arthritis Rheumat 1997;40:1944-54.
24. Walker JS, Levy G. Effect of multiple dosing on the analgesic action of diflunisal in rats. Life Sci 1990;46:737-42.
25. Greenblatt DJ, Shader RI, Franke K, et al. Kinetics of intravenous chlordiazepoxide: sex differences in drug distribution. Clin Pharmacol Ther 1977;22:893-903.
26. McCormack K, Brune K. Dissociation between the anti-nociceptive and the anti-inflammatory effects of non-steroidal anti-inflammatory drugs: a survey of their analgesic efficacy. Drugs 1991;41:533-47.
