Sex differences in brain response to anticipated and experienced visceral pain in healthy subjects

Abstract

Women demonstrate higher pain sensitivity and prevalence of chronic visceral pain conditions such as functional gastrointestinal disorders than men. The role of sex differences in the brain processing of visceral pain is still unclear. In 16 male and 16 female healthy subjects we compared personality, anxiety levels, skin conductance response (SCR), and brain processing using functional MRI during anticipation and pain induced by esophageal distension at pain toleration level. There was no significant difference in personality scores, anxiety levels, SCR, and subjective ratings of pain between sexes. In group analysis, both men and women demonstrated a similar pattern of brain activation and deactivation during anticipation and pain consistent with previous reports. However, during anticipation women showed significantly greater activation in the cuneus, precuneus, and supplementary motor area (SMA) and stronger deactivation in the right amygdala and left parahippocampal gyrus, whereas men demonstrated greater activation in the cerebellum. During pain, women demonstrated greater activation in the midcingulate cortex, anterior insula, premotor cortex, and cerebellum and stronger deactivation in the caudate, whereas men showed increased activity in the SMA. The pattern of brain activity suggests that, during anticipation, women may demonstrate stronger limbic inhibition, which is considered to be a cognitive modulation strategy for impending painful stimulation. During pain, women significantly activate brain areas associated with the affective and motivation components of pain. These responses may underlie the sex differences that exist in pain conditions, whereby women may attribute more emotional importance to painful stimuli compared with men.

sex differences in pain perception and clinical prevalence are well recognized in health and disease (20). Women report more intense, frequent, and longer duration of pain than men, display greater pain sensitivity to experimental stimuli, and respond less well to analgesics (20). They are more likely to suffer from chronic pain conditions, including fibromyalgia, migraine, and functional gastrointestinal disorders (51). Pain arising from the visceral organs is one of the most common forms of pain in clinical setting and one of the most frequent reasons why patients seek medical attention (10). Therefore sex differences in response to visceral pain have important implication for experimental studies. However, in contrast to somatic pain studies, few studies have in detail explored specific sex differences in response to experimental stimulation of selected visceral organs. Previous experimental studies demonstrate conflicting evidence for sex differences in visceral pain perception. For instance, studies using esophageal stimulation have shown increased sensitivity in healthy women (36) whereas those using rectal stimulation have shown no sex-related differences (47). There is still a knowledge gap in understanding the anatomical and physiological organization of pain pathways and modulatory mechanisms that contribute to sex differences in visceral pain perception and prevalence (23).

Brain imaging studies performed to explore the neural mechanisms that might contribute to sex differences in visceral pain in healthy subjects also show conflicting results. In one healthy volunteer study, differences in brain activation patterns were observed between men and women in response to rectal distension, with greater activity in women in the dorsal anterior cingulate cortex (midcingulate cortex), prefrontal cortex, and insula cortex (28). More recently a functional magnetic resonance (fMRI) study has demonstrated greater activity in women in the dorsolateral prefrontal cortex and middle temporal gyrus during anticipation of rectal pain, and in the medial frontal gyrus and cerebellum during painful rectal stimulation (3). Another fMRI study, based on a small number of healthy subjects undergoing rectal distension, found no sex-related differences, although exploratory region of interest analysis suggested possible greater insula activation in men compared with women (5). The only study involving esophageal pain showed no differences in evoked potential response (measured by magnetoencephalography) between sexes in healthy subjects (26). Notably, there are no studies using fMRI or positron emission tomography that have explored sex differences in brain processing of esophageal sensation in health or disease despite the fact that this model is used extensively in brain imaging studies (2, 12) and in models of visceral hyperalgesia (18, 25, 32).

In summary, brain imaging studies examining sex differences in the neural processing of visceral pain are still sparse and somewhat contradictory (33). Interestingly, in these previous studies several psychophysiological factors known to affect visceral pain processing, such as anxiety level (22, 44), personality type (11), and autonomic response, were not measured or compared between men and women. This could account for some of the differences in brain activity reported. Indeed, in several of the studies described above, the perception of pain (measured via pain ratings) was different between men and women; this could arguably explain the disparity in brain processing between sexes (37). In addition sample sizes in most of the healthy subject studies were relatively small, i.e., five to eight volunteers per group (4, 20, 26).

Brain imaging studies of visceral pain in relatively large cohorts of male and female healthy volunteers, during which the above mentioned variability factors are considered, are therefore needed to determine whether there are indeed sex differences in activated brain regions. This information may contribute to our understanding of how women perceive, report, and respond to visceral pain compared with men. The aim of our study was thus to assess the role of sex on the brain processing of esophageal pain and expectation of esophageal pain in groups of psychophysiologically characterized healthy men and women by use of fMRI. In the absence of any previous functional imaging studies that have assessed brain activity in response to esophageal visceral stimulation between healthy men and women using fMRI, it is difficult to hypothesize specific brain regions that may differ between sexes. Therefore, to avoid missing important differences between male and female subjects, brain image acquisition and analysis were not restricted to a priori brain regions, in favor of a whole brain image approach. However, given the abundance of previous work describing brain processing of pain, we restricted our discussion to regions brain previously implicated in pain processing.

MATERIALS AND METHODS

Subjects

Thirty-two volunteers (16 men and 16 women; mean age ± SD = 30.9 ± 7.8 and 27.8 ± 7.1, respectively) participated in this study, approved by the local ethics committee (reference CREC/07/08-7), and gave informed, written consent. Age- and sex-matched subjects were selected from a large cohort of volunteers [n = 120 (68 male)] who had participated in a previous study in our laboratory (19) during which they experienced esophageal intubation and pain stimulation. All subjects were nonsmokers with no history of psychiatric or physiological disorders and not taking any medications including centrally acting or hormonal medicines. All subjects (with the exception of one male) completed the Eysenck Personality Questionnaire-Revised (EPQ-R) (17) to assess personality traits of Extraversion and Neuroticism. The Spielberger State and Trait Anxiety Inventory (STAI) (48) was also administered immediately prior to intubation and scanning to assess state (anxiety level on the day of scanning) and trait anxiety. Before scanning, the pain toleration threshold for esophageal distension was established for each subject, as described below.

Visceral Stimulation

Painful phasic esophageal stimulation was delivered by distending a 2-cm-long silicone balloon with air, as previously described (11), with each distension lasting 1 s. The balloon was mounted 15 cm from the tip of a 4-mm-diameter standard manometry catheter. The balloon was passed transnasally into the esophagus, and the center of the balloon was positioned 5 cm proximal to the lower esophageal sphincter. To determine pain toleration threshold in each volunteer, the volume of distension was increased in steps of 1 ml (from zero) until the subjects reached the point at which they could no longer tolerate an increase in stimulus. The pain toleration threshold level was recorded and used as the level of stimulation during scanning.

Experimental Design

An event-related design was employed. Figure 1 demonstrates the time course of a single trial. To measure pain and anticipation of pain we employed the use of several visual cues, the details of which subjects were informed of prior to scanning. Each event commenced with the presentation of a visual warning cue (a yellow square) signaling a painful visceral stimulation was imminent. The next event was a painful visceral stimulation lasting 1 s. This was followed by a second visual cue (a blue square) which signaled a “safe” condition during which there was no risk of stimulation. Following each stimulation a visual analog scale (VAS) was presented and subjects rated the intensity of stimulation using an MRI-compatible button box, positioned in the right hand, to move the cursor: from 0 (no sensation) to 100 (maximal pain). The timing of each condition (anticipation and pain) of the event-related design was pseudo-randomized and jittered to the TR (repetition time) to avoid habituation and enable a representative sample of the brain response during each condition; the anticipation phase signaled the start of a new trial and lasted between 3 and 12 s, whereas the “safe” phase lasted between 28 and 35 s including the VAS, which commenced 9–15 s after the onset of painful stimulation. The safe phase included a null event that served as the baseline condition. The use of colored squares as signals was also pseudo-randomized such that half the subjects received the blue square as a warning signal and yellow as “safe” and the other half received the yellow square as a warning and blue as rest.

Fig. 1.

Time course of a single trial. Each trial consisted of 3 events. The anticipation period commenced with a visual warning cue (a yellow square), signaling that a painful visceral stimulation was imminent. The next event was painful visceral stimulation lasting 1 s, which was followed by a second visual cue (a blue square) indicating a “safe” condition during which there was no risk of stimulation. A visual analog scale (VAS) was presented for 6 s, 9–15 s after onset of painful stimulation, where a rating of 0 indicated “no sensation” and a rating of 100 indicated “maximum pain.” After VAS rating, subjects relaxed until the onset of next trial. The use of colored squares as signals was pseudo-randomized.

Skin Conductance Response

Skin conductance response (SCR) during anticipation, pain, and rest was measured throughout the task by silver-silver chloride electrodes attached to the middle phalange of the middle and index fingers of the left hand. SCR was measured in microsiemens by a constant-voltage method (0.5 V), and the signal was digitized at 25-ms intervals (Contact Precision Instruments, Boston, MA).

Magnetic Resonance Image Acquisition

fMRI data (T2*-weighted images) were collected on a General Electric Signa Excite II 3.0 T scanner based at the Centre for Neuroimaging Sciences, Institute of Psychiatry, Kings College London, London, UK.

While inside the scanner, participants could view a screen, via a prism attached to the head coil, upon which visual cues and an electronic version of a VAS were projected.

Prior to the fMRI experiment, a high-resolution Gradient Echo structural scan [43 × 3-mm slices, 0.3 interslice gap, echo time (TE) 30 ms, TR 3,000 ms, flip angle 90°, matrix size 128², voxel size 1.875 × 1.875 × 3.3 mm] was acquired in each volunteer to be used for normalization (cf. below in Individual analysis). During fMRI, a total of 480 T2* weighted images (40 × 3-mm slices, 0.3 interslice gap, TE 30 ms, TR 2,500 ms, flip angle 80°, matrix size 64², voxel size 3.75 × 3.75 × 3.3 mm) depicting blood oxygen level-dependent (BOLD) contrast were collected while participants periodically received painful phasic distensions to the esophagus.

Image and Statistical Analysis

Overview of general principles of fMRI analysis.

fMRI has come to dominate research studying brain activity since the early 1990s and generally has good spatial resolution with ability to localize activity to within millimeters. In a typical fMRI session, brain images are taken while stimulation (e.g., esophageal distension) is applied and during periods or rest or a control condition. Because the images are taken using an MR sequence which is sensitive to changes in local blood oxygenation level (BOLD level), images taken during stimulation show increased intensity of BOLD response compared with those taken while at rest, and thus the difference between these two sets of images corresponds to the brain areas that are activated by the stimulation. The goal of fMRI analysis is to detect, in a robust, sensitive, and valid way, those parts of the brain that show increased intensity at the points in time that stimulation was applied. A single image is made up of individual cuboid elements called voxels. After preprocessing steps (as below in Individual analysis), statistical analysis is carried out to determine which voxels are activated by the stimulation. To combine statistics across different sessions or subjects, the first necessary step is to align the brain images from all sessions into a common space (e.g., Talairach template). Once all the images are aligned, a variety of statistical methods are used for combining results across sessions or subjects, to either create a single result for a group of subjects or to compare different groups of subjects. In this study, we first conducted group analysis in women and men and then compared the brain activation between women and men. Details of methods we used in the present study are described below.

Statistical analysis of fMRI data.

All MRI data were analyzed by use of XBAM (http://brainmap.co.uk/), an fMRI analysis software package developed at the Institute of Psychiatry, King's College London, which implements permutation-based methods to minimize the number of assumptions used in making statistical inference.

Individual analysis.

fMRI data preprocessing (which included coregistration, spin-excitation history correction, global normalizing, slice timing correction, high-pass filtering, and Gaussian smoothing), and individual brain activation mapping were performed according to methods previously described (11).

Group analysis.

The individual maps were normalized into standard Talairach space, beginning with rigid body transformation of the fMRI data to the structural image of the same subject and followed by affine transformation onto a Talairach template (49). A group brain activation map for the task was then produced by calculating the median observed statistic over all subjects at each voxel in standard space and testing them against the null distribution of median statistics computed from the identically transformed wavelet resampled data (8).

Analysis of variance.

Analysis of variance (ANOVA) was undertaken to compare responses between groups (in this case men vs. women), by fitting an ANOVA model to the standardized BOLD effect size at each intracerebral voxel as follows: Y = a + bX + e where Y is the vector of BOLD effect sizes for each individual, X is the contrast matrix for the particular intercondition/group contrasts required, a is the mean effect across all individuals in the various conditions/groups, b is the computed group/condition difference, and e is a vector of residual errors. At each voxel the null hypothesis was tested by comparing the real (observed) value b against critical values of its nonparametrically obtained null distribution (9). Critical values for a two-tailed test of size α (alpha can be set to any desired type I error rate for the test) are the 100*(α/2)th and 100*(1−α/2)th percentile values of this distribution.

Both group and ANOVA analyses were then extended from the voxel to the 3D cluster level by the methods described in detail in Ref. 9. The statistical thresholds (voxel and cluster level ones) were set in such a way as to obtain less than one false positive 3D cluster per brain. Since we used whole brain cluster statistics, all tests were intrinsically corrected for multiple comparisons.

We then extracted each subject's mean BOLD signal response from the significant clusters and graphed these values.

Analysis of Psychophysiological Data

The SCR amplitude to each stimulus event was analyzed by use of a custom-made software program, SC-ANALYZE (Department of Neuroimaging, Institute of Psychiatry, King's College London). SCR (in μS) was defined as the maximum increase occurring 1 to 4 s after onset of condition and therefore represents the maximum arousal to each stimulus. This time window is variable owing to the uncertain characteristics of SCR response and has been previously validated (30, 56). The mean SCR level for each condition was calculated per individual and then averaged across each group (men and women) to produce mean SCR levels for anticipation, pain, and baseline. Repeated-measures ANOVA analyses with post hoc t-tests were used to examine differences in mean SCR level during rest, anticipation and pain within and between groups. All behavioral data, including personality questionnaire scores and pain intensity ratings, were processed with SPSS version 18.0 (SPSS 2009, Chicago, IL).

RESULTS

Psychophysiological and Behavioral Responses

EPQ-R scores.

There were no significant differences between men and women in neuroticism [men 8.6 ± 6.7 (mean ± SD) and women 4.1 ± 2.7; P = 0.6] and extraversion (men 17.5 ± 6 and women 17.4 ± 3.5; P = 0.97) scores (Fig. 2A).

Fig. 2.

A: mean personality score of men and women. There was no significant difference between men and women in terms of neuroticism (P = 0.6) and extraversion (P = 0.97). B: mean anxiety level between men and women. There was no significant difference between male and female in both trait (P = 0.4) and state (P = 0.08) Spielberger State and Trait Anxiety Inventory (STAI) scores. C: average balloon volume to reach pain toleration threshold and average pain scores (VAS) for men and women. The average balloon volume required to deliver painful stimulation at pain toleration was not significantly different between men and women (P = 0.2). There was also no significant difference between sexes in the pain reporting [average VAS over 20 painful stimulations during scanning (P = 0.6)]. Errors bars indicate standard error of the mean. EPQ-r, Eysenck Personality Questionnaire-revised.

STAI scores.

There were also no differences in anxiety levels between men and women (trait: men 34.7 ± 11.9 and women 34.1 ± 9; P = 0.8, state: men 31.4 ± 6.7 and women 29.3 ± 7.2; P = 0.4) (Fig. 2B).

Threshold level of stimulation.

There was no difference between men and women in the average balloon volume required to deliver stimulation at pain toleration (men 21 ± 7 ml and women 24 ± 7 ml; P = 0.2) (Fig. 2C).

Subjective pain perception.

All volunteers rated the esophageal stimulus as painful. There was no significant difference (P = 0.9) in average VAS rating of pain intensity between sexes (62 ± 12 for men and 64 ± 11 for women; P = 0.9) (Fig. 2C).

SCR.

One-way ANOVA of each sex showed significant differences in the SCR during baseline, anticipation, and pain for men (mean SCR ± SE; baseline 3.04 ± 0.65, anticipation 3.09 ± 0.66, pain 3.18 ± 0.66, P = 0.001) and women (mean SCR ± SE; baseline 2.01 ± 0.39, anticipation 2.02 ± 0.4, pain 2.19 ± 0.4, P = 0.002). Men showed increased SCR during anticipation (P = 0.02) and visceral pain (P = 0.0002) compared with baseline, whereas women only showed significantly increased SCR during pain (P = 0.0006) compared with baseline and not during anticipation (P = 0.2). However, a two-way ANOVA demonstrated no sex-related effect on SCR value among the conditions.

Group Brain Activation During Anticipation and Pain

Anticipation of pain.

Both men and women demonstrated an increase in brain activation during anticipation of pain in the superior and middle frontal gyrus, supplementary motor area (SMA), caudate, anterior insula, parahippocampal gyrus, cingulate gyrus, precuneus, cuneus, lingual gyrus, middle occipital gyrus, and cerebellum (Table 1). Both men and female showed deactivation in the medial and middle frontal gyrus, SMA, amygdala, parahippocampal gyrus, middle and posterior insula, thalamus, temporal gyrus, inferior parietal lobule, midbrain, and cerebellum (Table 2).

Table 1.

Activation during anticipation of pain

Brain Regions of Activation	Men						Women
	Brodmann area	Talairach coordinates			N of voxels	Cluster P value	Brodmann area	Talairach coordinates			N of voxels	Cluster P value
		X	Y	Z				X	Y	Z
Inferior frontal gyrus								51	19	10	61	0.0014
Superior frontal gyrus	9	−31	34	31	79	0.0019	9	−31	42	27	40	0.0016
	6	−18	−4	63	34	0.0017
Middle frontal gyrus	6	−36	−7	56	115	0.0002	6	−36	−7	46	20	0.0024
	46	41	35	15	13	0.0015
SMA	6	5	−8	64	78	0.0004	6	4	−4	53	305	<0.0001
Midcingulate cortex	24	−4	0	46	204	<0.0001
Caudate head								11	11	3	213	0.0011
Caudate body		14	19	7	171	0.0017
Caudate tail								25	−38	13	178	0.0016
								−18	−28	21
Anterior insula	13	−31	22	13	17	0.0020	13	−45	12	5	12	0.0009
Inferior parietal lobule	40	50	−30	25	32	0.0016
Parahippocampal gyrus	28	25	−20	−9	20	0.0007	36	22	−41	−10	13	0.0007
Precentral gyrus	6	40	0	36	74	0.0007
Superior temporal gyrus	41	−50	−32	14	37	0.0009
	41	49	−34	16	59	0.0019	41	46	−34	13	113	0.0016
Inferior temporal gyrus	19	48	−57	−1	126	0.0002
Middle temporal gyrus	37	−47	−59	0	107	0.0002
Thalamus								18	−22	16	104	0.0005
Cingulate gyrus	31	20	−43	37	58	0.0019	31	15	−39	35	156	0.0008
	31	−4	−44	36	229	0.0008	23	−2	−12	26	117	0.0008
	7	25	−60	43	165	0.0003	7	22	−62	30	166	<0.0001
Precuneus	7	−5	−66	54	148	0.0010	7	−3	−51	54	181	0.0008
	7	−27	−52	52	50	0.0015	7	−18	−67	46	108	0.0005
	7	−21	−63	33	141	0.0014
							31	20	−72	18	196	<0.0001
Cuneus	30	25	−70	13	248	<0.0001	19	14	−78	33	125	<0.0001
	18	−20	−72	18	154	0.0009	18	−11	−74	26	196	<0.0001
Lingual gyrus	18	11	−81	−7	307	0.0001	19	25	−71	−3	340	<0.0001
	18	−16	−50	2	23	0.0018	18	−7	−63	3	270	0.0001
Middle occipital gyrus	18	−27	−84	3	163	0.0001	19	−32	−82	7	347	<0.0001
Anterior cerebellum		32	−48	−16	162	<0.0001		29	−56	−20	228	0.0001
		−25	−56	−20	123	0.0008
		−40	−48	−33	76	0.0016		−22	−67	−16	266	0.0011
Posterior cerebellum		25	−67	−16	214	<0.0001
		11	−70	−33	154	0.0014
		−40	−74	−20	107	0.0006

N voxels = number of voxels in the 3D clusters; SMA, supplementary motor area.

Table 2.

Deactivation during anticipation of pain

Brain regions of activation	Men						Women
	Brodmann area	Talairach coordinates			N of voxels	Cluster P value	Brodmann area	Talairach coordinates			N of voxels	Cluster P value
		X	Y	Z				X	Y	Z
Medial frontal gyrus							9	11	48	33	88	0.0011
							9	−14	41	33	157	0.0012
							32	11	30	33	303	0.0005
							6	−7	−26	56	217	0.0006
	6	22	−11	53	161	0.0012	6	7	−33	56	119	0.0005
	47	47	41	−7	102	0.0007
Middle frontal gyrus							46	43	33	17	78	0.0021
Inferior frontal gyrus							46	40	41	3	74	0.0007
	47	−40	33	−3	87	0.0002	47	−47	41	−10	65	0.0007
							47	51	33	−17	14	<0.0001
							47	−54	26	−17	88	<0.0001
SMA							6	−7	19	56	98	0.0013
	6	4	−22	50	319	0.0010
Premotor cortex							6	−40	−11	33	480	0.0001
Cingulate gyrus							24/32	−14	7	30	212	0.0008
Amygdala								25	0	−17	252	0.0001
		−25	−7	−20	70	0.0007		−27	−2	−11	126	0.0004
Parahippocampal gyrus	36	25	−4	−26	30	0.0046		−36	−4	−13	404	0.0004
	34	22	−11	−17	94	0.0014	36	−36	−41	−7	282	0.0001
	35	−22	−22	−23	40	0.0014
Middle insula							13	33	−4	7	296	0.0003
Posterior insula	13	36	−19	0	114	0.0012
	13	−36	−11	13	393	0.0020	13	−36	−11	7	242	0.0002
	13	−43	−22	0	163	0.0013	13	−51	−11	10	242	0.0001
Thalamus		−11	−15	0	92	0.0011		−14	−15	7	141	0.0011
Precentral gyrus	6	54	−15	40	120	0.0004	6	51	−4	20	319	<0.0001
Superior temporal gyrus	22	72	−22	0	10	0.0012	38	54	19	−13	10	0.0001
							38	51	15	−20	88	0.0001
	21	−58	−30	−7	96	0.0026
Inferior parietal lobule	40	54	−30	40	157	0.0009	40	51	−26	33	165	0.0001
							40	33	−33	43	238	0.0008
							40	43	−56	36	75	0.0017
Fusiform gyrus							37	47	−33	−13	209	0.0003
Midbrain								4	−7	−10	118	0.0010
		5	−21	−8	73	0.0011		4	−19	−23	104	0.0013
								4	−41	−13	398	0.0005
Posterior Cingulate							30	11	−52	10	36	0.0006
Superior parietal lobule							7	−29	−52	43	129	0.0011
Angular gyrus	39	40	−59	30	14	0.0007
Anterior cerebellum		−14	−41	−23	130	0.0002
		4	−48	−17	222	0.0001
Posterior cerebellum		−25	−63	−40	25	0.0005		−11	−78	−33	99	0.0023

Pain.

During pain, both men and women demonstrated significantly increased brain activity in several brain regions known to process pain including the inferior frontal gyrus, midcingulate, anterior insula, amygdala, thalamus, and postcentral gyrus among others (see Table 3 for a complete list of regions). Both men and women demonstrated deactivation in several regions including the inferior and medial frontal gyrus, caudate, parahippocampal gyrus, precuneus, and cerebellum (see Table 4 for a complete list of regions).

Table 3.

Activation during pain

Brain regions of activation	Men						Women
	Brodmann area	Talairach coordinates			N of voxels	Cluster P value	Brodmann area	Talairach coordinates			N of voxels	Cluster P value
		X	Y	Z				X	Y	Z
Inferior frontal gyrus	47	52	43	−11	20	0.0015	47	47	20	−2	83	<0.0001
							44	53	3	18	514	<0.0001
							47	−52	34	−12	16	0.0007
Middle frontal gyrus	10	43	41	10	19	0.0002
SMA	6	8	−10	51	275	<0.0001	6	11	−7	53	439	<0.0001
Midcingulate cortex	24	−5	2	38	236	<0.0001
							32	12	16	33	348	<0.0001
Anterior insula	13	33	20	2	121	0.0020	13	27	21	9	225	<0.0001
							13	−31	22	13	60	<0.0001
Middle insula	13	−31	7	8	298	0.0001	13	−32	0	16	289	<0.0001
	13	43	0	13	815	<0.0001
	13	−35	−6	14	207	0.0001
Posterior insula	13	46	−16	17	425	<0.0001
Amygdala		23	−3	−22	32	0.0020		25	−1	−22	35	0.0020
Parahippocampal gyrus								−24	−10	−22	273	0.0004
Thalamus		16	−25	−5	200	0.0012		8	−12	−3	275	0.0001
								−8	−14	−3	221	0.0003
Premotor cortex	6	40	−7	36	269	<0.0001
	6	−54	0	20	120	<0.0001	6	−40	−7	50	118	<0.0001
							6	−36	−8	33	250	<0.0001
Precentral gyrus	4	−41	−17	38	170	0.0001	4	−20	−23	53	288	0.0001
Postcentral gyrus							3	22	−26	56	60	0.0001
	43	−51	−16	17	192	0.0001
	3	−22	−26	56	152	0.0004
Paracentral Lobule	5	18	−37	50	95	0.0004
Superior temporal gyrus							22	−52	10	−1	121	0.0003
							38	25	7	−26	201	0.0002
							41	−57	−18	13	276	<0.0001
							41	51	−30	16	336	<0.0001
Middle Temporal gyrus							37	50	−45	−9	134	0.0005
	21	−55	−22	−6	20	0.0009	37	−47	−44	−7	25	0.0002
Midbrain								10	−16	−10	288	0.0004
		−4	−19	−7	273	0.0014
Inferior parietal lobule	40	−43	−33	46	78	0.0008
	40	38	−46	52	162	0.0003
Cingulate gyrus							31	20	−28	42	216	0.0001
Posterior Cingulate	29	−2	−57	12	14	0.0012	30	−4	−56	7	12	0.0006
							30	−14	−63	13	19	0.0003
Precuneus							7	−18	−62	30	29	0.0019
Cuneus							7	4	−67	33	14	<0.0001
Anterior cerebellum		−13	−37	−38	15	0.0011
		32	−41	−33	15	0.0012
		−32	−48	−33	25	0.0017
Posterior cerebellum		−11	−52	−16	164	0.0014		−22	−52	−26	168	0.0002
		4	−56	−16	103	0.0011		−4	−59	−13	438	<0.0001

Table 4.

Deactivation during pain

Brain regions of activation	Men						Women
	Brodmann area	Talairach coordinates			N of voxels	Cluster P value	Brodmann area	Talairach coordinates			N of voxels	Cluster P value
		X	Y	Z				X	Y	Z
Superior frontal gyrus	10	33	56	13	80	0.0011	10	33	59	13	114	0.0001
	10	22	56	26	111	0.0018
							9	4	56	26	137	0.0001
	8	7	33	50	114	0.0008	8	4	33	56	21	<0.0001
	8	36	19	50	53	0.0011	8	18	30	56	98	0.0016
							8	−4	41	43	15	0.0014
Medial frontal gyrus	10	14	48	7	466	0.0002
							9	7	52	40	10	0.0014
	9	−11	41	30	421	0.0005
							10	−7	52	10	455	<0.0001
	6	−25	15	53	412	0.0001
Middle frontal gyrus							8	−25	22	46	164	0.0005
Inferior frontal gyrus							46	−47	26	13	165	0.0003
							45	61	22	10	51	0.0006
SMA	6	11	−22	63	11	0.0004
Anterior Cingulate cortex	32	−18	37	0	456	0.0001
Cingulate gyrus	32	22	19	30	307	0.0008
Caudate Body							−7	−4	20	262	<0.0001
Caudate tail		29	−41	13	255	0.0004
							−33	−41	3	253	<0.0001
								−22	−30	26	262	<0.0001
Parahippocampal gyrus	36	−29	−30	−10	135	0.0004		−24	−52	5	506	0.002
		29	−41	0	216	0.0005
Precentral gyrus	3	36	−22	46	31	0.0016	4	36	−19	46	23	0.0015
Paracentral gyrus							5	0	−37	63	276	<0.0001
Superior temporal gyrus							38	40	19	−30	13	0.0009
							38	−47	19	−30	20	0.0011
							39	47	−56	26	140	0.0003
	21	58	−22	0	10	0.0013	22	54	−4	−10	55	0.0020
	21	69	−4	−3	20	0.0008
	22	−51	−15	3	11	0.0016
Middle temporal gyrus	39	43	−63	13	201	0.0018	39	−43	−63	26	256	<0.0001
	39	−43	−59	13	105	0.0004
	39	−43	−67	26	69	0.0005
Posterior Cingulate							31	18	−44	30	265	0.0001
								14	−37	13	460	<0.0001
	31	−4	−41	26	445	<0.0001	31	−4	−33	36	202	0.0002
Inferior parietal lobule							40	43	−52	40	24	<0.0001
							40	−54	−44	30	55	0.0007
Angular gyrus	39	−33	−59	40	58	0.0008
Fusiform gyrus	19	−33	−67	−13	151	0.0002
Precuneus							7	22	−70	46	81	0.0002
							7	4	−74	40	197	<0.0001
	7	−7	−63	53	156	0.0016
Cuneus	19	−18	−78	33	92	0.0012
Lingual gyrus	18	25	−78	−3	241	<0.0001	19	25	−67	3	316	<0.0001
	18	0	−89	0	283	0.0001	18	4	−74	3	188	<0.0001
	18	−25	−74	0	159	<0.0001	19	−29	−63	0	203	<0.0001
Middle occipital gyrus							18	−25	−81	3	151	<0.0001
Anterior cerebellum		11	−41	−40	45	0.0019
		−29	−52	−13	116	0.0004
Posteriror cerebellum							40	−67	−23	113	0.0003
		33	−59	−40	56	0.0004		18	−59	−40	39	0.0007
		36	−63	−26	140	0.0006		4	−78	−20	136	0.0001
		14	−67	−36	96	0.0004		−36	−59	−23	155	0.0003
		−7	−78	−30	135	0.0005

Comparison of Brain Activity Between Men and Women

Anticipation of pain.

Brain areas where men demonstrated greater brain activity compared with women were the right amygdala, left parahippocampal gyrus, and anterior cerebellum (Table 5 and Fig. 3A). Averaged BOLD response plots showed these differences to be driven by a greater decrease in brain activity in women in the right amygdala, an increase in men and decrease in women in the left parahippocampal gyrus, and a greater increase in brain activity in the cerebellum in men (Fig. 3, A and C). In contrast, women produced a greater increase in brain activity than men in the right supplementary motor area (SMA, BA6), cuneus, and precuneus (Table 5B and Fig. 3, B and D).

Table 5.

Sex difference

Phase and Sex	Brain Regions of Activation	Brodmann Area	Talairach Coordinates			N of Voxels	Cluster P Value
			X	Y	Z
Anticipation
Men>Women	Amygdala		29	−7	−17	51	0.005
	Parahippocampal gyrus	36	−31	−32	−18	66	0.004
	Anterior cerebellum		32	−48	−16	48	0.005
Women>Men	SMA	6	4	0	50	31	0.007
	Precuneus	31	20	−72	18	87	0.002
	Cuneus	17	−19	−79	6	51	0.002
Pain
Men>Women	Caudate tail		−32	−39	5	82	0.002
	SMA	6	4	−11	50	45	0.008
Women>Men	Anterior insula	13	−31	22	13	27	0.006
	Midingulate gyrus	32	12	16	33	91	0.002
	Premotor cortex	6	47	0	30	30	0.005
	Premotor cortex	6	−35	−3	40	23	0.005
	Cingulate gyrus	24	−20	−15	42	35	0.005
	Posterior cerebellum		−18	−63	−13	118	0.001

Fig. 3.

Brain regions where there were significant differences between men and women during anticipation. A: brain areas where men demonstrated greater brain activity compared with women. B: brain areas where women produced greater brain activity than men. C: plot of mean blood oxygenation level-dependent signal extracted from each significant cluster in A. D: plot of mean blood oxygenation level-dependent signal extracted from each significant cluster in B. AMYG, amygdala; PHG, parahippocampal gyrus; SMA, supplementary motor area; PCUN, precuneus; CUN, cuneus; SSQ, the sum of squares ratio; Rt, right; Lt, left.

Pain.

During pain, men demonstrated greater activation compared with women in the SMA (BA6) and left caudate (Table 5 and Fig. 4, A and C). Averaged BOLD response plot showed women decreased brain activity more than men in the left caudate, and men increased brain activity in the SMA (Fig. 4C). In comparison, women showed a greater brain activity in the midcingulate cortex (BA32), left insula, bilateral premotor cortex, and posterior cerebellum that was driven by increased brain activity in women and a decrease or lack of difference in brain activity from the control condition in men (Table 5 and Fig. 4, B and D).

Fig. 4.

Brain regions where there were significant differences between men and women during pain. A: brain areas where men demonstrated greater brain activity compared with women in the SMA (BA6) and left caudate. B: brain areas where women produced greater brain activity than men in the midcingulate cortex (BA32) and left insula. C: plot of mean blood oxygenation level-dependent signal extracted from each significant cluster in A. D: plot of mean blood oxygenation level-dependent signal extracted from each significant cluster where women produced greater brain activity than men. Ant., anterior; CD, caudate; MMC, midcingulate cortex; INS, insula.

DISCUSSION

The present study compared the brain processing of visceral pain in psychophysiologically characterized male and female healthy volunteers. There was no significant difference between sexes in psychophysiological factors known to influence the brain processing of visceral pain including trait and state anxiety, personality type, autonomic response to pain, pain threshold, and pain perception levels. Both men and women demonstrated a similar brain response during anticipation of pain, including activity in the anterior insula, SMA, and caudate and deactivation in the amygdala and midbrain. During visceral pain, brain activity occurred in both sexes in regions of the “pain matrix” including midcingulate cortex, insula, amygdala, thalamus, and sensory cortex, and deactivation was seen in areas such as the medial prefrontal and medial parietal cortex. Interestingly, despite no psychophysiological differences between sexes, a direct comparison of brain activity revealed that women demonstrated a greater decrease of brain activity in the amygdala during anticipation and increased brain activity in the midcingulate cortex and anterior insula during pain compared with men.

Psychophysiological Factors

Psychological factors are considered to play a major role in mediating the differences in pain perception observed between men and women (23) and also influence the brain processing of pain (11, 22, 42–44, 54). For instance, it has been reported that personality, in particular neuroticism, modulates the brain processing of visceral pain and anticipation of pain (11). In our study personality scores and anxiety levels were not different between men and women and therefore the differences in brain processing that we have observed cannot be explained by these psychological factors. In addition, there were no significant sex differences in other psychophysiological factors also known to affect the brain processing of pain including balloon distension volumes or pain intensity ratings. Similarly, both groups also showed increased skin conductance during visceral pain compared with baseline, indicating that all subjects felt pain subjectively and experienced sympathetic arousal. These data provide evidence that both sexes were not significantly different in many factors known to affect the brain processing of pain and therefore any disparity in brain processing between men and women seen in the present study cannot be explained by variation between sexes in these factors. Interestingly, the only difference of note was that men showed significantly greater SCR activity during anticipation compared with baseline whereas women did not. This suggests higher sympathetic arousal during anticipation, relative to baseline, in men than in women.

Brain Processing

Anticipation of visceral pain.

Methods using visual cues to define anticipation have been widely used in not only pain anticipation (6), but also, for example, in reward anticipation using fMRI (16). During the anticipation phase, both men and women demonstrated activation in areas attributed to the “pain matrix,” suggesting that the subjects were anticipating pain. It has previously been reported that pain anticipation produces a neural response similar to that which occurs during pain, encompassing both the cognitive (e.g., anticipation, attention, etc.) and sensory aspects of pain processing (58). Decreased brain activity during anticipation of pain has often been reported in experimental studies (6, 15, 39–41). For instance, decreased activity in the amygdala, insula, and dorsal brain stem has been reported during anticipation of visceral pain induced by rectal distension (6).

Visceral pain.

During the pain phase, both men and women demonstrated increased brain activation in the SMA, midcingulate cortex, anterior, middle, and posterior insula, amygdala, thalamus, premotor cortex, midbrain, and cerebellum. These regions are considered to be part of the “visceral pain neuromatrix” and their activation has been consistently observed in previous visceral pain studies (4, 50, 52, 58). On the other hand, deactivation in the medial frontal superior and middle temporal gyrus and medial parietal cortex was observed in both sexes. Deactivation in these areas has also been reported as a suspension of default mode activity during visceral pain (53).

Overall, both sexes demonstrated a brain response during anticipation and visceral pain that is consistent with previous reports. However, when brain activity was compared statistically between men and women, several differences were observed during anticipation and pain.

Sex differences in brain activity during anticipation.

When differences between sexes were examined during anticipation, the deactivation in the right amygdala and left parahippocampal gyrus was stronger in women than men. The amygdala is considered a central constituent in the evaluation of potential threats as well as fear processing (38, 39) and has also been implicated in pain processing since the central nucleus of the amygdala contains nociceptive-specific neurons (7). Amygdala has been reported to be activated or deactivated depending on the context of painful stimulation (38). Deactivation in limbic brain areas, including amygdala, has often been reported during anticipation of pain, when healthy subjects are asked to tolerate maximal pain but know they will not sustain serious damage and can terminate pain at will (15, 27, 40). The limbic downregulation during anticipation is suggested as a cognitive coping strategy, thus a compensatory or modulating brain mechanism that attenuates the perceived distress caused by an aversive situation (38, 40). A similar reduction in amygdala activity in healthy women compared with irritable bowel syndrome patients during the anticipation of visceral pain has previously been observed (6), which suggests that limbic downregulation could be one of the strategies involved in preparing for impending pain. In the context of our results, it could therefore be speculated that female subjects used this cognitive coping strategy more than men during the anticipation of visceral pain.

It has been reported that there are sex differences in the stress-responsive neuroendocrine system (31). For instance, in a previous study men demonstrated elevated cortisol levels in anticipation of psychological stress but cortisol concentration was unchanged in women (29). Amygdala plays a crucial role in cortisol regulatory networks in the central nervous system, and increased activity in the amygdala has been associated with increased cortisol secretion (14). Although we did not measure neuroendocrine response in the present study, the sex difference in brain activity observed in our study may be connected with sex differences in stress response systems including hypothalamus pituitary adrenocortical axis. Our observation that there was an increase in SCR response during the anticipation phase compared with baseline only in men supports the possibility of less engagement of cognitive coping strategy by men and thus greater engagement of the physiological stress response systems.

There was increased activity of the SMA in women during anticipation. The SMA is implicated in the planning of motor actions, especially in voluntary behaviors such as self-initiated movement (35), and it is commonly reported in pain imaging studies, suggesting motor preparation for responding to an impending painful event (45). Therefore, the increase in SMA activity may indicate that women prepare a motor response (e.g., avoidance behavior) in anticipation of the impending painful stimulus. We observed greater activity in occipital visual areas in women during anticipation phase. This activity is likely to represent modulation by emotional attention of areas involved in processing the visual warning cue indicating an imminent painful event.

Sex difference in brain activity during pain.

Interestingly, the SMA activation in the men was higher than women during pain. As previously described, the SMA regulates motor function (35), suggesting that the activity in SMA and caudate in men may represent planning for a motor response to pain. The activity in the caudate tail was decreased in the female subjects. The caudate nucleus plays an important role not only in motor response to pain but also in the modulation of the pain experience such as pain suppression (57). Recent meta-analysis of placebo analgesia indicates placebo-related deactivation in the caudate during noxious stimulation (1). The stronger deactivation in women in the present study therefore could suggest that this region may be less engaged in suppression of visceral pain in women compared with men. Increased activity in women in the midcingulate and insula during esophageal pain in our study is consistent with the only other visceral pain study showing sex differences in brain processing of visceral stimulation in health. Kern et al. (28) reported that the midcingulate region and insula were activated in women but not in men in response to rectal pain. The midcingulate cortex is commonly activated in studies of visceral and somatic pain (21, 50), and recent review emphasizes the role of midcingulate in the integration of negative affect, pain, and cognitive control (46). The midcingulate constitutes a hub where information can be linked to motor centers responsible for expressing affect and executing behavior (46). As for the insula, visceral pain from the periphery is conveyed via the spinothalamic pathway to different parts of this structure. The posterior insula processes information about the physiological condition of the body, whereas the anterior insula cortex processes subjective feelings from the body and emotional awareness through interoception (13). Activity in the anterior insula increases when visceral pain is delivered during negative emotional arousal (12, 42). Mayer et al. (34) have proposed a concept of homeostatic emotions, which are the motivations and feelings associated with changes in the body's physiological condition and are the background emotions that affect our mood and disposition. They indicate that the anterior insula and midcingulate are the responsible areas in the brain for homeostatic emotions engendered by visceral information (34). Given the current evidence, the greater activation shown here in the midcingulate and anterior insula in women during pain may therefore represent greater processing of the homeostatic emotions accompanied by visceral pain. In rats, noxious visceral stimulation did not demonstrate differences in behavioral response between male and female rats but did show greater activation of the ventromedial prefrontal cortex and broader limbic/paralimbic changes in the females than males, also suggesting greater engagement of affective brain regions during visceral pain (55). Interestingly, another study, in which muscle pain was induced by hypertonic saline injections, also demonstrated increased fMRI signal intensity in the midcingulate in women compared with men (24). Women also demonstrate stronger activity in the premotor areas than men. Lateral premotor areas are associated with orientation and preparedness for action, which can take place without any actual movement occurring (15), suggesting that women may use this strategy in response to visceral pain more than men. Biologically endogenous pain modulatory systems, including opioid, dopamine, serotonin, and NMDA receptor function have been reported to differ between male and female subjects (20). For example, greater μ-opioid receptor binding has been shown to occur in the amygdala in women compared with men at rest (i.e., no stimulation). Furthermore, in response to sustained pain, a greater magnitude of μ-opioid system activation occurs in men than women in the anterior thalamus, ventral basal ganglia, and amygdala (59). It is likely that these biological pain modulatory circuits also underlie some of the sex difference in BOLD contrast brain activity reported in the present study and previous fMRI studies.

Limitations

One of the potential limitations of this investigation is that we did not determine the phase of the menstrual cycle our female participants were in. Pain sensitivity changes during the menstrual cycle, with greatest sensitivity displayed during the luteal phase (20). However, given that the pain intensities tolerated in female and male groups were similar, it is unlikely that any sex-based differences are related to pain intensity. A recent pain consensus report suggests that, in the absence of previous evidence for large menstrual/estrous cycle-related variations in the measure of interest, it is not absolutely necessary to test female subjects in specific stages (23).

The psychophysiological factors we investigated in our study were chosen because these are the key factors consistently reported in the literature as important in influencing pain behavior and brain activity in response to pain even in healthy subjects (20). However, additional behavioral measures, especially affective, subjective, and physiological responses, would be helpful to investigate other potential relations between different sex-related factors and brain activation during anticipation and experience of painful stimulation. For example, our measurement of the psychological factors was all self-reported questionnaires, and pain threshold data was based on self-rating VAS. Some researchers have questioned the use of self-reported measures of psychological state and have recommended multiple methods including structured interviews.

In summary, our data suggest that even in the absence of differences in the psychophysical experience of visceral pain, men and women differ in their brain processing during expectation and visceral pain. During anticipation, women demonstrate stronger deactivation in the amygdala and parahippocampal gyrus, suggesting engagement of a cognitive coping strategy in preparation for processing an impending pain experience. During esophageal pain, women preferentially activate brain areas associated with the affective and motivational component of pain, indicative of a stronger emotional response to visceral pain. Our data suggest that women may attribute more emotional importance to painful stimuli, which may in turn influence how they perceive, report, and respond to pain compared with men. Indeed it is well known that negative emotion exacerbates pain, resulting in lower pain thresholds and greater pain reporting, which, although not seen in our healthy population, may be important in exacerbating symptoms of pain or discomfort. This type of response may therefore represent a risk factor for functional pain disorders that are more prevalent in women and where emotional processing is thought to be important. Designing future studies to include men and women will enable the identification of possible sex differences in all aspects of pain from the mechanisms of pain inhibition and facilitation, to variations in treatment effects that may be sex specific.

GRANTS

This study was jointly funded by a Medical Research Council grant held by Q. Aziz and by a British Academy grant held by S. Coen.

DISCLOSURES

Q. Aziz has received educational grants from GSK, Pfizer, and Novartis pharmaceutical companies, none of which are relevant to the work described in this paper. The other authors of this paper have no conflict of interest to disclose.

AUTHOR CONTRIBUTIONS

M.K., V.G., M.J.B., and S.J.C. conception and design of research; M.K., A.D.F., and S.J.C. performed experiments; M.K., V.G., and M.J.B. analyzed data; M.K., Q.A., and S.J.C. interpreted results of experiments; M.K. prepared figures; M.K. drafted manuscript; M.K., A.D.F., Q.A., V.G., M.J.B., S.C.W., S.F., and S.J.C. edited and revised manuscript; M.K., A.D.F., Q.A., V.G., M.J.B., S.C.W., S.F., and S.J.C. approved final version of manuscript.