Anatomical connections between brain areas activated during rectal distension in healthy volunteers: A visceral pain network

1. Introduction 

Functional neuroimaging studies have been performed in both healthy volunteers and patients with irritable bowel syndrome (IBS) (Derbyshire, 2003a). Although the results were equivocal, they appear to be consistent with the insula, the Anterior Cingulate Cortex (ACC), the thalamus, the frontal cortex and the primary and secondary somatosensory cortices (S1 and S2) being implicated in the perception of visceral painful or non-painful stimuli. Various hypotheses have been put forward with respect to the specific function of theses regions in visceral pain and their putative role in stress and other associated psychosocial variables (Mayer et al., 2006). These studies provided valuable information for the localization of brain areas involved in visceroception, but there is a lack of information regarding their anatomical relationships. Animal studies have described some direct connections between some of these brain areas (Morecraft et al., 1993, Cipolloni and Pandya, 1999, Petrides and Pandya, 2002, Jasmin et al., 2004) but such information is not yet available in humans.

Diffusion Tensor Imaging (DTI) is a novel technique based on the random, diffusion-driven displacements of molecules at the microscopic scale, which can map white matter anatomical connections in the living human brain. Diffusion is a three-dimensional process, and therefore molecular mobility in tissues may be anisotropic, as for example in brain white matter. With DTI, diffusion anisotropy effects can be fully extracted, characterized, and exploited, providing microscopic anatomy in vivo. This specific application in the brain, called ‘fiber tracking’, used in combination with functional MRI, opens up a window on brain structures connectivity. This method was first described in humans by Conturo et al. (1999) and was further validated by Parker et al. (2002) in the macaque brain showing anatomical connection pathways and anatomical connectivity maps consistent with known anatomy.

Here, we combined fMRI and DTI in healthy volunteers to investigate the anatomical relationships of the brain areas activated during rectal distension and determine whether these areas form a network focusing on the role of insula which has a pivotal role in visceral regulations and sensations (Mayer et al., 2006).

2. Material and methods 

Eleven healthy females (mean age 38.4±3.1years) were included. Exclusion criteria were history of abdominal pain, transit dysfunction (either diarrhoea or constipation), major depression according to DSM-IV criteria, history of major psychiatric disease, head trauma, or epilepsy. Analgesics, antispasmodics, laxatives and antidiarrhoeal agents were stopped at least 7days before the start of the experimental protocol. The study was approved by the Local ethics committee and all the volunteers gave informed written consent.

3. Results 

3.1. Changes in brain activity induced by non-painful rectal distension 

The mean volume used to produce a permanent non-painful sensation (score 2) was 97.8±5.7ml, corresponding to a mean pressure of 16.8±0.9mmHg. The BOLD signal during non-painful distension showed a significant and bilateral increase (i.e. ‘activation’) in the PreFrontal Cortex (PFC) [Brodmann Area (BA) 9/45/46/47] and Anterior Cingulate Cortex (ACC) [BA 32]; a unilateral increase was also observed in the anterior part of the right insula (Fig. 1). Other activated areas included the middle cingulate cortex (MCC), the left putamen and the right thalamus and precuneus (see Table 1).

View Large Image Download to PowerPoint

Fig. 1. Changes in brain activity associated with non-painful rectal distension. Non-painful rectal distension induced bilateral activation in the PreFrontal Cortex (PFC), which was slightly stronger on the right side, bilateral activation in the Anterior Cingulate Cortex (ACC), and unilateral activation in the left anterior insula.

Table 1. Brain areas activated during non-painful distension. Z-values>3.09 represent significant activity (P<0.001). The coordinates are expressed in millimetres according to the reference line defined by Talairach and Tournoux: x=medial–lateral: 0 midline, + right; y=anterior–posterior: 0 anterior commissure, + anterior; z=superior–inferior: 0 commissural line, + superior.” Abbreviations: PFC=PreFrontal Cortex, ACC=Anterior Cingulate Cortex; MCC=Medial Cingulate Cortex; BA=Brodmann Areas.

	Brain areas	BA	Side	Z-score	Coordinates
					x	y	z
	PFC	21, 45, 46, 47	Left	5.53	−40	42	28
	PFC	9, 45, 46	Right	4.86	40	30	34
	Anterior Insula		Right	4.85	34	30	0
	ACC/MCC	32	Right	4.84	10	38	32
	ACC	32	Left	3.78	−10	56	−2
	MCC		Left	4.31	−4	−20	30
	Putamen		Left	4.41	−22	−2	12
	Precuneus	29, 30	Right	4.28	6	−46	10
	Thalamus		Right	3.69	8	−18	6

3.2. Changes in brain activity induced by painful rectal distension 

The mean volume used to produce a moderate pain sensation (score 5) was 268.7±23.3ml, corresponding to a mean pressure of 32.3±3.1mmHg. This painful stimulus induced bilateral activation in the primary (S1) [BA 2] and more strongly in the secondary (S2) [BA 42/48] somatosensory cortices; in the motor cortex [BA 4, 6, 44]; in the frontal cortex [BA 44 to 48]; in the insula (anterior and posterior portions on the right side; only the posterior portion on the left side); in the cerebellum; in the striatum (both in the putamen and in the caudate), and in the thalamus (see Fig. 2). The major activated areas, are listed in the Table 2.

View Large Image Download to PowerPoint

Fig. 2. Changes in brain activity associated with painful rectal distension. The painful rectal distension induced activations in the insula, primary (S1) and secondary (S2) somatosensory cortices, in the cerebellum, in the thalamus, in the middle frontal gyrus, in the caudate nucleus and in the primary motor (M1) and pre motor (M2) cortices.

Table 2. Brain areas activated during painful distension. Z-values>3.09 represent significant activity (P<0.001). The coordinates are expressed in millimetres according to the reference line defined by Talairach and Tournoux: x=medial–lateral: 0 midline, + right; y=anterior–posterior: 0 anterior commissure, + anterior; z=superior–inferior: 0 commissural line, + superior. Abbreviation: S1=primary somatosensory cortex, S2=secondary somatosensory cortex, M1=primary motor cortex, SMA=supplementary motor area, BA: Brodmann Areas, PFC= Prefrontal cortex.

	Brain areas	BA	Side	Z-score	Coordinates
					x	y	z
	S1+S2	2, 5, 40, 42, 48	Right	6.80	58	−26	30
	S1+S2	2, 22, 42,48	Left	6.19	−60	−32	18
	Cerebellum	19	Left	6.70	−30	−76	−30
	Thalamus		Right	4.02	16	−19	2
	Thalamus		Left	3.80	−15	26	3
	SMA	6	Right	5.00	6	−12	68
	Insula post		Right	3.82	38	−10	−1
	Insula ant		Right	5.28	32	36	6
	Insula ant		Left	3.68	−32	20	8
	Caudate		Right	6.19	18	−6	24
	Caudate		Left	4.45	−20	−6	20
	Putamen		Left	4.73	−20	8	12
	Putamen		Right	5.27	25	−13	−7
	Frontal Mid	21, 22, 37	Right	5.60	62	−56	−4
	PFC	9, 32, 46	Left	4.68	−18	50	28
	Pre Motor	6, 44	Left	5.41	−38	4	34
	Pre Motor	6, 44	Right	5.16	40	10	32
	Paracentral (S2/M1)	4, 5, 7		4.95	0	−38	70
	MCC		Right	5.21	6	−8	30
	Parietal Sup	7	Right	4.44	32	−62	62
	Parietal Sup	5	Left	5.09	−22	−56	68

3.3. Anatomical connections between the areas functionally associated with visceral pain perception 

We analysed the connections between the insula and the main areas activated during non-painful or painful visceral stimuli, on the basis of the functional data described above. The main areas included the thalamus, somatosensory cortex (S1/S2), ACC and PFC. As illustrated in Fig. 3, direct inter-connections were observed between the PFC and insula (Fig. 3A), the insula and thalamus (Fig. 3B), the insula and ACC (Fig. 3C), the insula and S1/S2 (Fig. 3D). FA, ADC, mean number of fibers and the mean length of fibers for each bundle are given in Table 3. These connections were present in all subjects, although minor non-significant inter-individual quantitative variations were observed (Z score, P>0.05).

View Large Image Download to PowerPoint

Fig. 3. Connections between the main areas activated during visceral perception. A, B: coronal sections showing the bundles of fibers between the insula and the prefrontal cortex (PFC) (red), and the insula and the thalamus (purple). C, D: oblique sagittal sections showing the bundles of fibers between the insula and the cingulate cortex (ACC) (green) and the insula and the primary and secondary somatosensory cortices (S1/S2) (blue).

Table 3. Connections between main activated brain areas during rectal distension Abbreviations: PFC: prefrontal cortex, ACC: Anterior Cingulate Cortex, S1/S2: primary and secondary somatosensory cortices; FA: fractional anisotropy; ADC: apparent diffusion coefficient.

	Name of bundle	FA mean±SD (max)	ADC mean±SD (max)	Number of fibers mean±SD	Length of fibers (mm) mean±SD
	Insula-PFC	0.39±0.12 (0.74)	0.76±0.08 (1.58)	20±21	72.98±19.99
	Insula-thalamus	0.39±0.12 (0.77)	0.77±0.30 (4.79)	80±7	97.01±22.4
	Insula-ACC	0.39±0.11 (0.69)	0.71±0.07 (2.03)	290±31	17.76±6.44
	Insula-S1/S2	0.40±0.11 (0.79)	0.71±0.10 (1.91)	101±9	74.4±10.12

4. Discussion 

The combination of fMRI and DTI allowed us to identify a network of brain structures involved in visceral sensations, including the insula, thalamus, ACC, PFC and somatosensory cortices.

The results from the first part of our study were consistent with previous data regarding the changes in brain activity induced by painful and non-painful rectal distensions. The first study using Positron Emission Tomography (PET) (Rothstein et al., 1996) for functional imaging of visceral pain was published ten years ago; since then several studies with PET or fMRI have been performed in healthy subjects and/or IBS patients using painful and non-painful stimulation (Derbyshire, 2003a, Derbyshire, 2003b). However, their results are controversial. For example, Kwan et al. (2005) described the activation of the right anterior insula and right ACC during painful rectal distension, whereas another group reported right anterior insula activation and bilateral ACC deactivation (Dunckley et al., 2005).

In fact, most early studies (Silverman et al., 1997, Binkofski et al., 1998, Baciu et al., 1999, Mertz et al., 2000) emphasized the specific role of the Anterior Cingulate Cortex (ACC) in visceral nociception; however several more recent studies (Aziz et al., 2000, Hobday et al., 2001, Yuan et al., 2003) reported clear ACC activation during non-painful visceral distension. The activation of this structure may be more related to the control of autonomic visceral responses than to pain perception per se. Our results, showing higher ACC activation during non-painful distension, are consistent with this hypothesis. Our results confirmed the key role of the insula (most notably its anterior and middle sectors) in visceral perception. The involvement of the insula in visceral regulation and more generally in homeostatic regulation, is well documented (Mayer et al., 2006). This structure has consistently been found activated in imaging studies related to visceral or somatic pain processing (Derbyshire, 2003a, Derbyshire, 2003b, Peyron et al., 2000). This has also been underscored in studies using other approaches. Thus, electrical stimulation of the insula in animals and in humans modifies autonomic regulations including blood pressure, heart rate, respiration or intestinal peristalsis (Derbyshire, 2003b). Direct electrical stimulation of the insular cortex also induces intense painful sensations in humans (Ostrowsky et al., 2002) and intensity-dependent responses to nociceptive stimuli have been directly recorded in this area (Frot et al., 2007).

The role of the insula in visceral pain perception is also supported by neurophysiological studies. The earliest painful cortical evoked potentials recorded after sigmoid colon stimulation were described by dipoles in the insula and in the ACC, while S2 areas are activated later (Drewes et al., 2004, Drewes et al., 2006).

The role of the somatosensory cortices (S1, S2) in visceral perception has been debated, but several studies reported clear activation of S1 and/or S2 if the distension was sufficient to induce a sensation described as a deep-seated, poorly localized pelvic sensation, a strong feeling of discomfort or an urge to defecate (Lotze et al., 2001). Consistent with these results, we observed that non-painful stimulation produce no activation in the somatosensory and motor cortices. By contrast, painful distension induced a bilateral activation in the primary (S1) and more strongly secondary (S2) somatosensory cortices, as well as in the thalamus. We also detected the activation of the primary motor cortex (M1) and supplementary motor region (M2), which may reflect a reflex motor reaction (i.e. abdominal wall contraction, anal sphincter contraction) (Bittorf et al., 2006).

Visceral spinal afferents are usually classified as low-threshold, high-threshold, or silent mechanoreceptors (Gebhart, 2000, Furness et al., 2003). Low-threshold afferents respond to physiologic levels of distention and continue to encode excessive levels of distention that evoke pain in humans and pain behavior in animals. High-threshold afferents respond to higher levels of distention that are in the noxious range. Silent nociceptors do not respond at all in the normal intestine but become responsive to distention when the intestine is injured or inflamed. The anal canal is innervated with mechanoreceptors, thermoreceptors and nociceptors, comprising both visceral (A delta and C fibers) and somatic Abeta fibers. Therefore, one can hypothesize that, while a weak distension activates only the small A delta and C visceral fibers of the rectum, the increased distension to obtain a painful stimulation may also activate the somatic A beta fibers present in the anal canal and be responsible for S1 activation.

Painful and non-painful distension resulted in different levels of activation in the PFC. The activation was mostly located in the dorsolateral part of the PFC (DLPFC), and was lower during painful distension than that observed with non-painful stimulation. The activation of the DLPFC has been shown to be negatively associated with pain affect. It has been proposed that the DLPFC exerts active control on pain perception by modulating corticosubcortical and corticocortical pathways (Lorenz et al., 2003). Thus, the decrease in PFC activation reported here might reflect the painful and unpleasant features of distension.

In our study we only included healthy subject of female gender to have an homogeneous group because gender differences have been described both in healthy and in IBS patients (Kern et al., 2001, Berman et al., 2006).

The main objective of our study was to look for direct anatomical relationships between the main brain areas activated during visceral perception (i.e. insula, thalamus, somatosensory cortex, PFC and ACC). The DPTools software has a sufficient spatial resolution to visualize direct relations between various brain areas in a single subject. The combination of DTI with functional data from classical fMRI was necessary, as tractography needs to determine, a priori, the regions of interest to visualize the bundles of axons. To date, relatively few studies concerning DTI and pain have been published (Hadjipavlou et al., 2006, Seghier et al., 2005, Sundgren et al., 2007). Hadjipavlou et al. (2006) used DTI in eight healthy volunteers to investigate the white matter connections originating from brainstem nuclei: the periaquaductal grey (PAG) and the nucleus cuneiformis (NCF). On the basis of a group analysis, they showed direct connections between the PAG and separately the NCF, the PFC, amygdala, thalamus, hypothalamus and rostroventral medial medulla bilaterally. However, in contrast to our study, functional and anatomical data were not acquired in these subjects and cortico-cortical connectivity was not explored. We focused fiber tracking analysis on connexions with insula because of the prominent role of the insula in pain processing, the insula being the most frequently activated structure in fMRI (Apkarian et al., 2005). In our study, group DTI analysis and fiber tracking were confirmed by individual analysis, as the connections were present in all the subjects. There was some slight inter-individual differences regarding the quantity of bundles of fibers and their anisotropic state, but the connections were present in each subject.

Thus, our data indicates that there is a visceral pain network centered on the insula, which is directly connected to the thalamus, PFC, S2, and ACC. The anatomical relationships between these area have been studied in rats (Jasmin et al., 2004) and in primates (Morecraft et al., 1993, Cipolloni and Pandya, 1999; Petrides et al., 2006) using tracer techniques, but have not been studied yet in humans. Studies in animals showed massive reciprocal connections between the posterior insula and orbitofrontal cortex, the posterior insula and ACC, and the posterior insula and somatosensory cortex. Our data are consistent with the conclusions of Mayer et al. (2006) regarding the key role of the right anterior insula in visceral perception. Future studies in IBS patients combining activation imaging during visceral stimulation and DTI with fiber tracking will determine if there are any morphological or functional abnormalities of the network described above in healthy subjects.