Research Paper

Brain activations during pain: a neuroimaging meta-

analysis of patients with pain and healthy controls
Karin B. Jensena,*, Christina Regenbogena, Margarete C. Ohseb, Johannes Frasnellic,d,e, Jessica Freiherrb,c,
Johan N. Lundstroma,b,f

Abstract
In response to recent publications from pain neuroimaging experiments, there has been a debate about the existence of a primary
pain region in the brain. Yet, there are few meta-analyses providing assessments of the minimum cerebral denominators of pain.
Here, we used a statistical meta-analysis method, called activation likelihood estimation, to define (1) core brain regions activated by
pain per se, irrelevant of pain modality, paradigm, or participants and (2) activation likelihood estimation commonalities and
differences between patients with chronic pain and healthy individuals. A subtraction analysis of 138 independent data sets revealed
that the minimum denominator for activation across pain modalities and paradigms included the right insula, secondary sensory
cortex, and right anterior cingulate cortex (ACC). Common activations for healthy subjects and patients with pain alike included the
thalamus, ACC, insula, and cerebellum. A comparative analysis revealed that healthy individuals were more likely to activate the
cingulum, thalamus, and insula. Our results point toward the central role of the insular cortex and ACC in pain processing, irrelevant
of modality, body part, or clinical experience; thus, furthering the importance of ACC and insular activation as key regions for the
human experience of pain.
Keywords: Pain, Meta-analysis, Chronic pain, Neuroimaging, Thalamus, Cingulate cortex

1. Introduction Literature reviews of the neural correlates to pain3,12,25,49,61

The complexity of the subjective experience of pain entails
involvement of multiple different brain regions.25,40 The commu-
nication between these different regions, and their reciprocal
modulatory effects, is believed to largely account for the individual
pain experience.61 Yet, a recent debate has discussed the
existence of a primary pain cortex that would be comparable to,
for example, the primary visual cortex.13 Because pain is defined
as a subjective experience modulated by cognitive, affective, and
contextual factors, scientists have long considered it unlikely that
one primary brain area would respond to pain. Still, recent data
suggest that the posterior insula plays a fundamental role in pain
perception,58 a notion that finds support in earlier studies
proposing that the insula is required for encoding and processing
of nociceptive input.8,39
Sponsorships or competing interests that may be relevant to content are disclosed
at the end of this article.
a
Department of Clinical Neuroscience, Karolinska Institutet, Stockholm, Sweden,
b analyses was developed.62 Activation likelihood estimation allows
Clinic of Diagnostic and Interventional Neuroradiology, RWTH Aachen University,
Aachen, Germany, c Monell Chemical Senses Center, University of Pennsylvania,
Philadelphia, PA, USA, d CogNAC, Department of Anatomy, Universite du Quebec a
Trois-Rivieres, Trois-Rivieres, QC, Canada, e CEAMS, Research Center, Sacre-
Coeur Hospital, Montreal, QC, Canada, f Department of Psychology, University of
Pennsylvania, Philadelphia, PA, USA
*Corresponding author. Address: Department of Clinical Neuroscience, Karolinska
Institutet, Nobels vag 9, D3, 17176 Stockholm, Sweden. Tel.: 146702130811; fax:
1468311101. E-mail address: karin.jensen@ki.se (K. B. Jensen).
Supplemental digital content is available for this article. Direct URL citations appear
in the printed text and are provided in the HTML and PDF versions of this article on
the journals Web site (www.painjournalonline.com).
PAIN 157 (2016) 12791286
2016 International Association for the Study of Pain
http://dx.doi.org/10.1097/j.pain.0000000000000517
frequently report joint activation of the thalamus, primary and
secondary somatosensory cortices (SI/SII), insula, anterior cingu-
late cortex (ACC), and prefrontal cortices.5 Despite the high
prevalence of chronic pain in the general population and the vast
number of available neuroimaging studies of chronic pain, there
have been few systematic comparisons of neuroimaging findings
from patients and healthy individuals.
Because of the complex integration of different brain regions
and varying types of pain paradigms, neuroimaging studies often
yield heterogeneous activation patterns. Many neuroimaging
studies are also statistically underpowered,38 especially in clinical
cohorts. Yet, systematic reviews may counteract this power
problem by aggregating the data. Although descriptive literature
reviews are well suited to characterize common activations
between studies based on a given variable of interest, much of the
3-dimensional (3D) spatial information that voxel-based data
consist of is lost. To counteract these problems, the meta-
analytical method of activation likelihood estimation (ALE)
for formal statistical integration of unbiased voxel-based data
from multiple studies to determine common activations across
studies and to provide a formal estimate of activation likelihood.1
In a recent study,15 the authors reported likelihood maps derived
from pain studies in healthy individuals; however, the study did
not include patients with chronic pain. Chronic pain is a common
health problem that affects more than 100 million adults in the
United States alone at any given time and leads to large economic
burdens for society.21 The aim of the present study was to use
ALE to determine, based on statistical likelihoods, core brain
regions that are activated by pain per se, irrelevant of pain
modality and origin and to compare the likelihood maps from
patients with chronic pain and healthy individuals.

June 2016
Volume 157
Number 6 www.painjournalonline.com 1279

Copyright 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited.
 1280

K.B. Jensen et al. 157 (2016) 12791286 PAIN

2. Materials and methods
2.1. Identification of publications
Only data from international peer-reviewed journals published
between January 1990 and December 2014, using functional
and spatially precise imaging methods (functional magnetic
resonance imaging and positron emission tomography), were
used for ALE analyses. We obtained these publications by
searches in the National Center for Biotechnology Informations
publications database PUBMED (http://www.ncbi.nlm.nih.gov/
pubmed/) using the keywords pain*, ache*, laser, cold, and heat
cross-paired with the keywords fMRI, PET, functional magnetic
resonance imaging, or positron emission tomography. Similar
searches were subsequently performed in the Thomson Reuters
ISI Web of Science publication database with all obtained results
cross-referenced to sort out repeated entries. In addition to these
2 databases, the reference list within each analyzed article was
searched for additional publications of interest. Data within
publications were classified as originating from either healthy or
patient populations based on the classification provided by the
authors.
2.2. Inclusion criteria
The inclusion criteria required a pain vs no-pain contrast where
brain foci were derived from whole-brain volume analyses
reported in 3D coordinates (x, y, and z) within a standardized
stereotactic space (either Montreal Neurological Institute [MNI] or
Talairach) and do not depend on region-of-interest analyses. Only
studies on human participants using either functional magnetic
resonance imaging or positron emission tomography imaging
were included. The number of subjects in a group of healthy
subjects or patients had to be larger than 5 in a given study.
Postscreening of the included studies, regarding general
methodological quality, was not performed because of 2
reasons. First, the broad inclusion strategy renders means that
the result from the ALE analyses is a better representation of the
published studies to date. Second, by adapting a broad
inclusion strategy, incidences of methodologically weak stud-
ies will have limited impact on the ALE results because the
relative contribution of each included study is smaller. Also, it is
unlikely that a significant number of studies would suffer from
such specific methodological problems that it would result in an
anatomically precise colocalization between them, something
that is needed to taint the ALE result.
2.3. Activation likelihood estimation analyses
Activation likelihood estimation analyzes the given brain coor-
dinates from published functional imaging studies and then
searches for concordance by modeling each reported foci as the
center of a 3D Gaussian probability distribution using permutation
testing. These distributions are then used to create a whole-brain
statistical map that estimates the likelihood of activation for each
individual voxel, as determined by the entire set of studies
included. Hence, the statistics reported in our ALE result tables
are quantifications of the likelihood of activation because of
a given task at a given voxel. The original ALE meta-analysis
algorithm was originally developed by Turkeltaub et al. (2002) and
continuously optimized. Detailed descriptions of the ALE
algorithm can be found elsewhere16,17,35,62,63 and this algorithm
is implemented in BrainMap (BrainMap GingerALE 2.3.2, http://
brainmap.org/ale). The most recent algorithm allows not only
automatic meta-analysis pipeline in both Talairach and MNI
space, random-effects inference, and comparison of 2 different
ALE maps but also cluster-based correction for multiple
comparisons.
During standard analysis of neuroimaging data, functional
images are spatially normalized to a stereotactic template that
might differ between studies. These anatomical templates are not
directly comparable because of minor anatomical deviances.
Therefore, to facilitate direct comparisons between publications,

Figure 1. Activation likelihood estimation (ALE) maps for noxious stimulation across all functional imaging studies between 1990 and 2014. (A) ALE results of
noxious stimulation in healthy individuals. (B) ALE results of noxious stimulation in patients with pain. In both A and B, significant ALE clusters (P , 0.05 cluster-level
corrected inference using P , 0.001 uncorrected at voxel-level as the cluster-forming threshold) are projected onto a Montreal Neurological Institute (MNI)
template provided on the GingerALE Website. The numbers in the figure correspond to the cluster numbering in the corresponding tables (letters were introduced
in case multiple foci within 1 cluster are reportedin Figure 1A: 1a 5 thalamus, 1b 5 insula, 1c 5 primary and secondary somatosensory cortices).

Copyright 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited.
 June 2016
Volume 157
Number 6 www.painjournalonline.com 1281

all data included in the statistical meta-analysis were transformed Table 1

into MNI space, as implemented in the GingerALE software. All Activation likelihood estimates for noxious stimulation across
coordinates were subsequently imported into the Java-based all functional imaging studies between 1990 and 2014 (A) in
version of the ALE software and analyzed. A whole-brain ALE healthy subjects and (B) in patients with chronic pain. All
map was created by modeling a Gaussian probability distribution anatomical locations are given in Montreal Neurological
centered at each reported activation coordinate. A voxel-wise Institute (MNI) coordinates.
calculation of the probability that each activation was located
Cluster no. ALE value MNI coordinates Anatomical label
within that particular voxel was then performed using a 3D (31023) x y z
Gaussian filter function with an empirically determined full-width,
half-maximum value.17 The approach of histogram permutation A. Healthy
individuals
testing was subsequently used to test the null hypothesis which
1 116.9 58 224 24 Inferior parietal lobule
states that activation foci are distributed uniformly across the 112.5 14 214 6 Thalamus
brain.16,63 To compare ALE values in healthy subjects with 105.2 40 14 2 Insula
patients, we matched a number of included contrasts to avoid 102.6 258 222 18 Postcentral gyrus
weighting the resulting ALE output toward the group with more 99.5 238 10 4 Insula
included contrasts. The ALE analysis in patients contained 40 94.4 212 212 4 Thalamus
contrasts (443 foci). As a first step, we randomly selected 40 80.1 40 218 12 Insula
contrasts (461 foci) including healthy subjects (see Supplemen- 73.2 240 218 16 Insula
59.1 254 4 4 Superior temporal gyrus
tary Table 1, available online as Supplemental Digital Content at
56.7 222 22 6 Putamen
http://links.lww.com/PAIN/A239). Then, we calculated individual
52.8 248 0 4 Precentral gyrus
ALE maps of the 2 samples, as well as a pooled ALE map, and 49.3 58 8 0 Superior temporal gyrus
performed the actual subtraction. The local maxima of 37.4 58 242 34 Inferior parietal lobule
activation clusters were labeled using a cluster-analysis step 35.8 20 10 24 Putamen
within the ALE procedure and the MNI atlas included in Mango 35.2 24 2 220 Parahippocampal gyrus
(http://ric.uthscsa.edu/mango/; version 2.5). Statistical maps 34.3 54 244 50 Inferior parietal lobule
were all corrected for multiple comparisons using P , 0.05 33.3 24 26 210 Lateral globus pallidus
cluster-level corrected inference using P , 0.001 uncorrected 33.2 226 2 214 Parahippocampal gyrus
2 107.4 4 12 38 Cingulate gyrus
at voxel-level as the cluster-forming threshold, unless explicitly
47.1 6 26 48 Superior frontal gyrus
noted. 41.1 2 36 18 Anterior cingulate gyrus
For visualization purposes, the anatomical template pro- 3 53.6 38 48 16 Middle frontal gyrus
vided on the GingerALE Website (Colin27_T1_seg_MNI.nii, 41.8 42 44 4 Middle frontal gyrus
http://brainmap.org/ale) was overlaid with the different thresh- 4 38.8 52 16 30 Middle frontal gyrus
old ALE maps using the Mango imaging software. To visualize 36.4 50 4 44 Precentral gyrus
overlaps between different pain modalities in healthy subjects, 5 46.7 28 264 230 Cerebellum
between different patients with chronic pain, or between 6 37.9 236 258 234 Cerebellum
7 37.1 234 270 222 Cerebellum
patients and healthy individuals, the different ALE maps were
B. Patients with
superimposed. chronic pain
1 30.6 40 6 2 Claustrum
23.3 54 12 0 Precentral gyrus
3. Results
19.5 42 12 210 Insula
Of the original search on the scientific search engines, 34,022 hits 17.0 38 22 24 Insula
were narrowed down using the mentioned exclusion and 16.6 34 20 2 Claustrum
inclusion criteria, rendering a total of 170 individual articles (see 15.6 28 6 0 Putamen
2 32.6 4 20 30 Cingulate Gyrus
Supplementary Tables 1 and 2 for a detailed overview, available
3 27.1 10 26 2 Thalamus
online as Supplemental Digital Content at http://links.lww.com/ 18.2 4 218 6 Thalamus
PAIN/A239). These articles were then divided into articles 15.6 24 26 6 Thalamus
exploring neural activity of pain in healthy individuals (n 5 138) 14.5 218 210 22 Medial globus pallidus
and patients (n 5 32). 14.1 16 220 4 Thalamus
13.9 24 220 6 Thalamus
13.5 212 210 6 Thalamus
3.1. Literature search resulthealthy individuals 4 23.6 56 216 10 Transverse temporal gyrus
19.7 68 220 18 Postcentral gyrus
A total of 138 functional imaging studies fulfilled all the
5 20.1 238 12 24 Insula
stipulated inclusion criteria and were included in the final 16.4 226 2 28 Putamen
analyses (see Supplementary Table 1, available online as 16.3 232 16 4 Claustrum
Supplemental Digital Content at http://links.lww.com/PAIN/ 6 19.2 262 220 18 Postcentral gyrus
A239). These 138 studies comprised a total of 183 contrasts 15.2 252 218 16 Postcentral gyrus
and 2442 activation foci that were included in ALE analyses of 7 21.6 236 266 220 Cerebellum
pain processing in healthy individuals. Most of the 2442 21.2 234 260 224 Cerebellum
included foci originated from studies using thermal pain as 8 16.9 262 2 8 Precentral gyrus
16.7 258 10 18 Inferior frontal gyrus
the means of stimulation (965 foci), followed by distension (356
9 19.9 214 6 2 Putamen
foci), electrical (314 foci), and mechanical stimulation (eg, 10 20.1 258 18 24 Inferior frontal gyrus
pressure pain; 226 foci). The most commonly stimulated body
Both analyses are based on a P , 0.05 cluster-level corrected inference using P , 0.001 uncorrected at
part was the arm (874 foci), followed by the hand (560 foci). For voxel-level as the cluster-forming threshold (also see Fig. 1).
an overview of all pain modalities and stimulated body regions, ALE, activation likelihood estimation.

Copyright 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited.
 1282

K.B. Jensen et al. 157 (2016) 12791286 PAIN

Figure 2. Representation of significant ALE maps. A conjunction of all foci from both healthy subjects and patients is shown. The activation likelihood maps are
projected onto a template provided on the GingerALE Website. The numbers in the figure correspond to the cluster numbering in the corresponding table. All
anatomical locations are given in Montreal Neurological Institute coordinates (MNI).

see Supplementary Table 2 (available online as Supplemental reveals an overlap across the 4 most common pain modalities in
Digital Content at http://links.lww.com/PAIN/A239). the right insula, bilateral SII, and right ACC.

3.2. Literature search resultpatients 3.4. Significant activation likelihood estimation

resultspatients with chronic pain
Thirty-two studies, reporting a total of 40 contrasts and 443
activation foci, fulfilled the inclusion criteria and were included in The ALE analysis in patients with chronic pain, including 443 foci,
the final ALE analysis of patients with chronic pain (see revealed 10 different clusters that were likely activated during
Supplementary Table 3, available online as Supplemental Digital
Content at http://links.lww.com/PAIN/A239).
A total of 13 different categories of patients with pain were
Table 2
included in the 32 studies. Most foci were reported from patients
with fibromyalgia (83 foci), followed by irritable bowel syndrome Activation likelihood estimates for noxious stimulation across
(76 foci), and headache (76 foci). The most common methods of all functional imaging studies between 1990 and 2014. All
experimental stimulation in patients with pain were pressure pain anatomical locations are given in Montreal Neurological
(167 foci), distension (116 foci), and thermal pain (80 foci). The Institute coordinates (MNI).
most commonly stimulated body parts were the hand (144 foci), Cluster no. ALE value MNI coordinates Anatomical
face or head (95 foci), and rectum (76 foci). For representation of the (31023) x y z label
different patient categories, pain modalities, and stimulated body Conjunction
parts, see Supplementary Table 4 (available online as Supplemental healthy subjects
Digital Content at http://links.lww.com/PAIN/A239). and patients
with chronic pain
1 30.6 40 6 2 Claustrum
3.3. Significant activation likelihood estimation 23.3 54 12 0 Precentral gyrus
resultshealthy individuals 19.5 42 12 210 Insula
17.0 38 22 24 Insula
The ALE analysis for all 2442 foci revealed 7 clusters that had
16.6 34 20 2 Claustrum
a significant likelihood of activation during pain, namely, bilateral 15.6 28 6 0 Putamen
thalamus, bilateral insula, left SI and SII, right ACC, right prefrontal 2 32.6 4 20 30 Cingulate gyrus
cortices (middle-frontal gyrus), and cerebellum (Fig. 1A, Table 1; 3 27.1 10 26 2 Thalamus
Supplementary Fig. 1 for complete brain coverage, available 18.2 4 218 6 Thalamus
online as Supplemental Digital Content at http://links.lww.com/ 15.6 24 26 6 Thalamus
PAIN/A239). 14.1 16 220 4 Thalamus
To assess commonly reported brain regions independent of pain 13.8 218 210 0 Medial globus pallidus
modality, we first calculated the individual ALE maps for the 4 most 13.5 212 210 6 Thalamus
common methods of pain stimulation (thermal, distension, 4 23.6 56 216 10 Transverse temporal gyrus
18.2 68 220 18 Postcentral gyrus
electrical, and mechanical stimulation). We then calculated areas
5 20.1 238 12 24 Insula
of overlap between these individual ALE maps to represent areas
16.3 232 16 4 Claustrum
commonly activated by pain stimulation. The commonly activated 6 19.2 262 220 18 Postcentral gyrus
brain regions in healthy individuals were further explored by 15.2 252 218 16 Postcentral gyrus
a common subtraction method, ie, every voxel in the brain that did 7 16.9 262 2 8 Precentral gyrus
have a significant ALE value in all the individual analyses was kept; 8 19.8 236 268 220 Cerebellum
all others were given a value of zero, and thus removed. Although 9 13.9 24 220 6 Cerebellum
this is not a statistical approach stricto sensu, it has the potential to 10 14.3 234 260 230 Cerebellum
elucidate core brain regions that process pain through its common 11 14.7 212 2 4 Putamen
presence in all conditions. For a visual representation of the overlap 12 13.5 224 2 212 Putamen
13 12.5 216 4 6 Putamen
between the 4 different modalities (conjunction analysis), see
14 12.6 254 10 12 Inferior frontal gyrus
Figure 2 for discussed areas and Supplementary Figure 2 for
Displayed are the results of a conjunction analysis of healthy subjects and patients with chronic pain (see also
complete brain coverage (available online as Supplemental Digital Fig. 2).
Content at http://links.lww.com/PAIN/A239). The overlap image ALE, activation likelihood estimation.

Copyright 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited.
 June 2016
Volume 157
Number 6
pain: the bilateral thalamus, insula, SI, SII, right-cingulate gyrus,
and left cerebellum (Fig. 1B, Table 1).
3.5. Significant activation likelihood estimation
resultsoverlap between healthy individuals and patients
with chronic pain
A conjunction analysis of common ALE results for healthy
individuals and patients with chronic pain revealed a considerable
pain-processing overlap in 14 clusters, eg, in the insula,
cingulum, thalamus, and cerebellum (Fig. 2, Table 2).
3.6. Significant activation likelihood estimation
resultsdifferences between healthy individuals and
patients with chronic pain
Because of the large amount of literature suggesting that chronic
pain is associated with functional brain aberrations, we in-
vestigated the differences in activation likelihood between
patients and healthy individuals. Statistical subtraction was used
to elucidate the differences in ALE results between healthy
individuals and patients with chronic pain (P , 0.05 uncorrected
at voxel-level). Ten different clusters had a significantly higher
likelihood of activation in healthy individuals than in patients,
including the cingulate gyrus, thalamus, insula, middle-frontal
gyrus, and the cerebellum (Fig. 3, Table 3). The reverse contrast,
ie, assessing where patients had a higher likelihood of activity
than healthy individuals, revealed 4 different clusters reaching
statistical significance for higher likelihood of activation: the
cerebellum, inferior-frontal gyrus, precentral gyrus, and the
cingulate gyrus (Fig. 3, Table 3; for complete brain coverage,
see Supplementary Fig. 3, available online as Supplemental
Digital Content at http://links.lww.com/PAIN/A239).
4. Discussion
Here, a quantitative assessment of brain regions involved in
human processing of pain, and a unique comparison of brain
activations in patients with pain and healthy individuals, was
performed using ALE. The insula, ACC, SII, and thalamus
were commonly activated by pain, irrelevant of pain modality
Figure 3. Results of the subtraction analysis of significant ALE maps. The cluster maps show the contrast where healthy subjects have higher ALE values than
patients in green color-coding. The contrast where patients have higher ALE values than healthy subjects is shown in purple color-coding. The activation likelihood
maps are projected onto a Montreal Neurological Institute (MNI) template provided on the GingerALE Website. The numbers in the figure correspond to the cluster
numbering in the corresponding table.
Copyright 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited.
www.painjournalonline.com 1283
or stimulated body part, both in patients and healthy
individuals. The high predominance of insular and ACC
activations in our analyses furthers their role as key regions
for the experience of pain. Our results are also made
available as a spatial brain mask for free download and use
in future neuroimaging analyses, eg, when analyses are
confined to the core pain network or when independently
and functionally defined pain areas for region of interest
analyses are desired.
4.1. Activation likelihood during pain processing
In line with recent data suggesting that the insula plays
a fundamental role in human pain,11,58 we report robust likelihood
of activations of the entire insula during pain, with peaks in the
anterior insula (AI). Although the dorsal-posterior insula has been
described as the cortical representation of incoming nociceptive
signals,11 the AI has been associated with the integration of
emotional and interoceptive states.10,46 For example, exposure to
noxious stimuli may induce the activation of both the posterior and
anterior insula, whereas the subjective evaluation of these stimuli is
represented in the AI.31 Based on the significant likelihood of
activating the insula across pain modalities, with peaks in the AI, we
suggest that the AI is an essential component for the subjective
experience of pain. Previous ALE meta-analyses also found insular
involvement in pain processing,15,36,43 with peaks in anterior,
middle, and the posterior insula. Yet, is there coherence between
experimental studies suggesting that the posterior insula plays
a fundamental role in pain and ALE analyses suggesting that the AI
is key for pain processing? It has been suggested that the posterior
insula is a nonspecific way station for sensory input and coding of
stimulus intensity, rather than a specific pain center,13 as opposed
to the evaluative role of the AI. Hence, the 2 regions reflect different
aspects of pain processing. Neuroanatomical studies support the
notion that the right AI is part of an afferent path for interoceptive
representations of pain and homeostatic drive.9 The AI is thereby
thought to provide meta-representations of the state of the body, or
the feeling self, combined with motivational drive for bodily
protection.9 Involvement of the AI in cognitive meta-
representations of pain has also been found in studies of empathy
for pain where the AI is activated both by ones own pain and by
 1284

K.B. Jensen et al. 157 (2016) 12791286 PAIN

Table 3 as the AI.10 The AI and ACC are often jointly activated, consistent
Activation likelihood estimates for noxious stimulation across with the idea that they represent complementary limbic sensory
all functional imaging studies between 1990 and 2014. All and motor regions that work together.10 More specifically, the AI
anatomical locations are given in Montreal Neurological might be the site for pain awareness because of its afferent
Institute (MNI) coordinates. representation of bodily interoception and the ACC, the site for the
initiation of behavioral response. The ACC has been frequently
Cluster no. ALE MNI coordinates Anatomical label
value associated with the emotional-motivational aspect of pain, as
x y z
supported by lesion and imaging studies.19,20,54 For example,
Healthy subjects vs selective changes in pain unpleasantness, but not pain intensity,
chronic pain patients have been associated with ACC modulation.50
1 3719.0 1 7 35 Cingulate gyrus
In line with analyses by Duerden and Albanese,15 we found that
3540.1 28 12 36 Cingulate gyrus
3352.8 8 8 38 Cingulate gyrus
the AI, ACC, and thalamus were consistently activated across
2635.6 12 0 40 Cingulate gyrus pain modalities, suggesting a consistent core network of brain
2427.6 12 12 50 Medial frontal gyrus regions involved in pain processing.
2144.4 6 26 46 Paracentral lobule
2 3719.0 18 213 14 Thalamus
3540.1 23 211 7 Lateral globus pallidus 4.2. Comparison between pain processing in health
3352.8 24 210 26 Lateral globus pallidus and disease
2604.5 41 13 11 Insula As cumulating evidence suggests that chronic pain is man-
2186.2 42 18 2 Insula ifested in the central nervous system,22,44,57 a number of
3 3238.9 241 14 8 Insula
studies have investigated the cerebral changes associated with
3090.2 240 8 6 Insula
2549.1 226 4 4 Putamen
chronic pain2,55,56; including structural,4,14,18,32,52 func-
2373.9 226 22 24 Claustrum tional,7,24,45 and neurochemical brain alterations.23,29,65 Here,
2263.6 224 26 10 Putamen a formal comparison between the ALE results from a large
1863.4 250 8 26 Insula number of patients and healthy individuals demonstrates that
4 3540.1 55 230 30 Inferior parietal lobule patients are less likely to activate the AI and thalamus. As
3352.8 53 232 26 Inferior parietal lobule previously discussed, the AI is involved in the subjective
2911.2 42 222 14 Insula experience and interoceptive representation of pain.9 As
2878.2 44 224 10 Transverse temporal gyrus patients had a less likelihood to activate the AI, it may reflect
2820.2 38 216 18 Insula a disrupted pain interoceptive function, which in turn could
2106.1 52 230 14 Superior temporal gyrus
explain the absence of pain inhibitory activations in patients
5 2549.1 40 52 20 Middle frontal gyrus
2270.1 44 48 6 Middle frontal gyrus
during evoked pain.26,28
2033.5 30 44 20 Middle frontal gyrus Patients less likelihood to activate key nociceptive regions
6 3035.7 232 212 14 Claustrum may seem counterintuitive because patients have persistent
1789.1 226 220 4 Lentiform nucleus pain. One methodological explanation could be that patients
7 2706.5 262 230 28 Inferior parietal lobule relatively lower likelihood to activate, eg, the AI and cingulate
2467.7 256 230 30 Inferior parietal lobule gyrus, indicates a ceiling effect, as patients have constant
1827.7 258 218 32 Postcentral gyrus noxious input to nociceptive brain regions and are thus not able
8 3155.9 30 272 230 Cerebellum to display additional activation during added experimental
3035.7 24 273 229 Cerebellum pain.36 Furthermore, recent data suggest that the transition
9 3035.7 51 14 29 Inferior frontal gyrus
from acute to chronic pain entails a shift from lateral (sensory-
10 2341.6 8 40 18 Anterior cingulate gyrus
1926.8 0 40 22 Cingulate gyrus
discriminative) to medial (affective-motivational) neural activity.3
1872.0 10 44 12 Anterior cingulate gyrus Our results may thus support the hypothesis that chronic pain
Chronic pain patients involves decreased sensory and enhanced emotional processes
vs healthy subjects because ALE differences were also found in the thalamus, a region
1 3155.9 236 258 222 Cerebellum with a significantly less likelihood of pain activation in patients. For
2988.9 240 262 222 Cerebellum several different categories of patients with pain, there are reports
2 3035.7 261 10 19 Inferior frontal gyrus of thalamic structural changes,14,53,59 attenuated thalamic activity
2947.8 256 12 18 Inferior frontal gyrus during rest,34,41 and attenuated pain-evoked activations.26,33 Our
3 2536.4 61 15 6 Precentral gyrus results thereby support the notion that pain pathophysiology may
4 2512.1 6 16 24 Cingulate gyrus
involve thalamocortical dysrhythmia.6,30,37 According to the
5 2101.5 58 216 4 Superior temporal gyrus
dysrhythmia theory, thalamocortical disruption may lead to
Displayed are the results of a subtraction analysis between healthy subjects and patients with chronic pain
(P , 0.05 uncorrected at voxel-level, also see Fig. 3). disturbances of sensation, motor performance, cognition, and
ALE, activation likelihood estimation. ultimately, to disabling chronic disorders, such as chronic pain. In
a longitudinal neuroimaging study,27 where fibromyalgia patients
were scanned before and after treatment with cognitive behavioral
watching pain in others.47,60 Interestingly, Ochsner et al.47 found therapy, there was increased connectivity between the thalamus
that the right AI was more engaged during the processing of ones and lateral prefrontal cortex compared to waitlist controls. Because
own pain than watching others pain.47 Combined with the central chronic pain is associated with disrupted thalamocortical connec-
role of the AI suggested here, the AI could be essential for tivity, the increase in thalamocortical connectivity may reflect
attributing pain to ones own body. a normalization of pain pathophysiology.
Conclusive ALE findings support the importance of ACC We found that patients had a less likelihood to activate parts
activation for pain processing. The ACC is a limbic brain structure of the middle-frontal gyrus, possibly reflecting patients
that shares afferent projections from the same spinothalamic tract decreased activation of the brains pain inhibitory

Copyright 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited.
 June 2016
Volume 157
Number 6
network.26,28 The prefrontal cortex is a key region for pain
inhibition, eg, during reappraisal of pain and placebo
analgesia,48,64 suggesting that the attenuated activation of
the frontal gyrus in patients may represent decreased pain
inhibition. Yet, our results were inconclusive, as patients
displayed an increased likelihood of activating some parts of
the prefrontal cortex, whereas the same structure had
a decreased likelihood of activation on the right side. Because
the laterality and specificity of the different prefrontal
subregions during noxious stimulation is not fully understood,
further studies of the role of the prefrontal cortex in pain
chronification are needed.
4.3. Future outlook and limitations
Most neuroimaging paradigms in human subjects do not allow for
the distinction between nociception and pain because the
experimental stimulation of peripheral nociceptors is inherently
coupled to the subjective experience of pain. Therefore, our study
will not be able to shed light on the possible segregation between
nociception and pain. Furthermore, it is important to note that
pain activations share a considerable overlap with other cognitive
and emotional processes.42 Even if we strive to describe the
minimum denominators of pain, specific patterns of brain activity
are merely correlated to the subjective experience of pain, and not
pain per se.51
The results reported within this meta-analysis are based
entirely on results reported in previous publications. This means
that although the analyses performed are random-effect analyses
in a statistical sense, the ALE values reported are referring to the
likelihood that these voxels are activated in any given neuro-
imaging study on pain processing. Hence, our results should be
interpreted and used within balanced limits of previous probabil-
ities and reverse inferences in neuroimaging studies.
In this study, we only included data originating from publica-
tions reporting their results using 3D coordinates within a stan-
dardized stereotactic space. Although studies assessing links
between brain areas and pain processing in patients with
localized lesions are very informative and, some argue, with
a stronger causality, they fall outside the scope of this statistical
meta-analysis. Future attempts should be made to quantify lesion
extensions in coordinate space and merge the information with
that from coordinate-based inference studies to render a more
inclusive meta-analysis.
5. Conclusions
The results from the present study suggest that (1) insular and
ACC activation are central for pain perception and (2)
functional differences in pain processing between patients
with chronic pain and healthy individuals may explain behav-
ioral differences and further our understanding of pain
pathology, especially when pain is viewed as a homeostatic
function or as a transition from lateral (sensory) to medial
(emotional) processing of pain.
Conflict of interest statement
The authors have no conflicts of interest to declare.
This material is based on work supported by a grant from the
Knut and Alice Wallenberg Foundation (KAW 2012.0141)
awarded to J. N. Lundstrom. Johnson & Johnson Inc provided
partial salary support during the data collection phase. C.
Regenbogen is supported by a postdoctoral fellowship of the
Copyright 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited.
www.painjournalonline.com 1285
DAAD (German Academic Exchange Service). J. Frasnelli is
supported by grants from the Research Center of Sacre-Coeur
Hospital, Montreal, the University of Quebec in Trois-Rivieres, the
Natural Sciences and Engineering Research Council (Canada),
and the Fonds de Recherche du QuebecSante. The funders
had no role in study design, data collection and analysis, decision
to publish, or preparation of the manuscript.
Acknowledgements
The authors thank Dr Marco Loggia for valuable comments on
a previous version of this manuscript as well as Kajsa Forsberg and
Anna Sjoholm Norling for assisting in data collection and extraction.
Appendix A. Supplemental Digital Content
Supplemental Digital Content associated with this article can be
found online at http://links.lww.com/PAIN/A239.
Article history:
Received 6 October 2015
Received in revised form 3 January 2016
Accepted 25 January 2016
Available online 11 February 2016
References
[1] Albrecht J, Kopietz R, Frasnelli J, Wiesmann M, Hummel T, Lundstrom
JN. The neuronal correlates of intranasal trigeminal function-an ALE
meta-analysis of human functional brain imaging data. Brain Res Rev
2010;62:18396.
[2] Apkarian AV, Baliki MN, Geha PY. Towards a theory of chronic pain. Prog
Neurobiol 2009;87:8197.
[3] Apkarian AV, Bushnell MC, Treede RD, Zubieta JK. Human brain
mechanisms of pain perception and regulation in health and disease. Eur
J Pain 2005;9:46384.
[4] Burgmer M, Gaubitz M, Konrad C, Wrenger M, Hilgart S, Heuft G,
Pfleiderer B. Decreased gray matter volumes in the cingulo-frontal cortex
and the amygdala in patients with fibromyalgia. Psychosom Med 2009;
71:56673.
[5] Bushnell MC, Apkarian V. Representation of pain in the brain. In:
McMahon SB, Koltzenburg M, editor. Textbook of pain, Vol. 5. Edinburgh:
Elsevier Churchill Livingstone, 2006. p. 10724.
[6] Cauda F, Sacco K, DAgata F, Duca S, Cocito D, Geminiani G, Migliorati
F, Isoardo G. Low-frequency BOLD fluctuations demonstrate altered
thalamocortical connectivity in diabetic neuropathic pain. BMC Neurosci
2009;26:138.
[7] Cauda F, Sacco K, Duca S, Cocito D, DAgata F, Geminiani G, Canavero
S. Altered resting state in diabetic neuropathic pain. PLoS One 2009;4:
4542.
[8] Craig AD. How do you feel? Interoception: the sense of the physiological
condition of the body. Nat Rev Neurosci 2002;3:65566.
[9] Craig AD. A new view of pain as a homeostatic emotion. Trends Neurosc
2003;26:3037.
[10] Craig AD. How do you feelnow? The anterior insula and human
awareness. Nat Rev Neurosci 2009;10:5970.
[11] Craig AD. Topographically organized projection to posterior insular cortex
from the posterior portion of the ventral medial nucleus in the long-tailed
macaque monkey. J Comp Neurol 2014;522:3663.
[12] Davis KD. Neuroimaging of pain: what does it tell us? Curr Opin Support
Palliat Care 2011;5:11621.
[13] Davis KD, Bushnell MC, Ianetti GD, Lawrence K, Coghill RC. Evidence
against pain specificity in the dorsal posterior insula. F1000Res 2015;4:
16.
[14] Davis KD, Pope G, Chen J, Kwan CL, Crawley AP, Diamant NE. Cortical
thinning in IBS: implications for homeostatic, attention, and pain
processing. Neurology 2008;70:1534.
[15] Duerden EG, Albanese MC. Localization of pain-related brain activation:
a meta-analysis of neuroimaging data. Hum Brain Mapp 2013;34:
10949.
[16] Eickhoff SB, Bzdok D, Laird AR, Kurth F, Fox PT. Activation likelihood
estimation meta-analysis revisited. Neuroimage 2012;59:234961.
 1286
K.B. Jensen et al. 157 (2016) 12791286
[17] Eickhoff SB, Laird AR, Grefkes C, Wang LE, Zilles K, Fox PT. Coordinate-
based activation likelihood estimation meta-analysis of neuroimaging
data: a random-effects approach based on empirical estimates of spatial
uncertainty. Hum Brain Mapp 2009;30:290726.
[18] Flor H. Cortical reorganisation and chronic pain: implications for
rehabilitation. J Rehabil Med 2003;41:6672.
[19] Foltz EL, White LE. The role of rostral cingulumotomy in pain relief. Int J
Neurol 1968;6:35373.
[20] Foltz EL, White LE Jr. Pain relief by frontal cingulumotomy. J Neurosurg
1962;19:89100.
[21] Gaskin DJ, Richard P. The economic costs of pain in the United States.
J Pain 2012;13:71524.
[22] Giesecke T, Gracely RH, Grant M, Nachemson A, Petzke F, Williams D,
Clauw DJ. Evidence of augmented central pain processing in idiopathic
chronic low back pain. Arthritis Rheum 2004;50:61323.
[23] Harris RE, Clauw DJ, Scott DJ, McLean SA, Gracely RH, Zubieta JK.
Decreased central m-opioid receptor availability in fibromyalgia.
J Neurosci 2007;27:100006.
[24] Hsieh JC, Belfrage M, Stone-Elander S, Hansson P, Ingvar M. Central
representation of chronic ongoing neuropathic pain studied by positron
emission tomography. PAIN 1995;63:22536.
[25] Ingvar M. Pain and functional imaging. Philos Trans R Soc B Biol Sci 1999;
354:134758.
[26] Jensen KB, Kosek E, Petzke F, Carville S, Fransson P, Marcus H, Williams
SCR, Choy E, Giesecke T, Mainguy Y, Gracely R, Ingvar M. Evidence of
dysfunctional pain inhibition in Fibromyalgia reflected in rACC during
provoked pain. PAIN 2009;144:95100.
[27] Jensen KB, Kosek E, Wicksell R, Kemani M, Olsson G, Merle JV, Kadetoff
D, Ingvar M. Cognitive behavioral therapy increases pain-evoked
activation of the prefrontal cortex in patients with fibromyalgia. PAIN
2012;153:1495503.
[28] Jensen KB, Loitoile R, Kosek E, Petzke F, Carville S, Fransson P, Marcus
H, Williams SC, Choy E, Mainguy Y, Vitton O, Gracely RH, Gollub R,
Ingvar M, Kong J. Patients with fibromyalgia display less functional
connectivity in the brains pain inhibitory network. Mol Pain 2012;8:32.
[29] Jones AK, Cunningham VJ, Ha-Kawa S, Fujiwara T, Luthra SK, Silva S,
Derbyshire S, Jones T. Changes in central opioid receptor binding in
relation to inflammation and pain in patients with rheumatoid arthritis. Br J
Rheumatol 1994;33:90916.
[30] Jones EG. Thalamocortical dysrhythmia and chronic pain. PAIN 2010;
150:45.
[31] Kong J, White NS, Kwong KK, Vangel MG, Rosman IS, Gracely RH,
Gollub RL. Using fMRI to dissociate sensory encoding from cognitive
evaluation of heat pain intensity. Hum Brain Mapp 2006;27:71521.
[32] Kuchinad A, Schweinhardt P, Seminowicz DA, Wood PB, Chizh BA,
Bushnell MC. Accelerated brain gray matter loss in fibromyalgia patients:
premature aging of the brain? J Neurosci 2007;27:40047.
[33] Kwan CL, Diamant NE, Pope G, Mikula K, Mikulis DJ, Davis KD. Abnormal
forebrain activity in functional bowel disorder patients with chronic pain.
Neurology 2005;65:126877.
[34] Kwiatek R, Barnden L, Tedman R, Jarrett R, Chew L, Rowe C, Pile K.
Regional cerebral blood flow in fibromyalgia: single-photon-emission
computed tomography evidence of reduction in the pontine tegmentum
and thalami. Arthritis Rheum 2000;43:282333.
[35] Laird AR, Fox PM, Price CJ, Glahn DC, Uecker AM, Lancaster JL,
Turkeltaub PE, Kochunov P, Fox PT. ALE meta-analysis: controlling the
false discovery rate and performing statistical contrasts. Hum Brain Mapp
2005;25:15564.
[36] Lanz S, Seifert F, Maihofner C. Brain activity associated with pain,
hyperalgesia and allodynia: an ALE meta-analysis. J Neural Transm
(Vienna) 2011;118:113954.
[37] Llinas RR, Ribary U, Jeanmonod D, Kronberg E, Mitra PP.
Thalamocortical dysrhythmia: a neurological and neuropsychiatric
syndrome characterized by magnetoencephalography. Proc Natl Acad
Sci U S A 1999;96:152227.
[38] Logothetis NK. What we can do and what we cannot do with fMRI. Nature
2008;453:86978.
[39] Mazzola L, Isnard J, Peyron R, Guenot M, Mauguiere F. Somatotopic
organization of pain responses to direct electrical stimulation of the
human insular cortex. PAIN 2009;146:99104.
[40] Melzack R. From the gate to the neuromatrix. PAIN 1999(suppl 6):
S1216.
[41] Mountz J, Bradley L, Modell J, Alexander R, Triana-Alexander M, Aaron L,
Stewart K, Alarcon G, Mountz J. Fibromyalgia in women. Abnormalities of
Copyright 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited.
PAIN
regional cerebral blood flow in the thalamus and the caudate nucleus are
associated with low pain threshold levels. Arthritis Rheum 1995;38:92638.
[42] Mouraux A, Diukova A, Lee MC, Wise RG, Iannetti GD. A multisensory
investigation of the functional significance of the pain matrix.
Neuroimage 2011;54:223749.
[43] Mutschler I, Ball T, Wankerl J, Strigo IA. Pain and emotion in the insular
cortex: evidence for functional reorganization in major depression.
Neurosci Lett 2012;520:2049.
[44] Naliboff BD, Derbyshire SWG, Munakata J, Berman S, Mandelkern M,
Chang L, Mayer EA. Cerebral activation in patients with irritable bowel
syndrome and control subjects during rectosigmoid stimulation.
Psychosom Med 2001;63:36575.
[45] Napadow V, LaCount L, Park K, AsSeine S, Clauw D, Harris RE. Intrinsic
brain connectivity in fibromyalgia is associated with chronic pain intensity.
Arthiritis Rheum 2010;62:254555.
[46] Nieuwenhuys R. The insular cortex: a review. Prog Brain Res 2012;195:
12363.
[47] Ochsner KN, Zaki J, Hanelin J, Ludlow DH, Knierim K, Ramachandran T,
Glover GH, Mackey SC. Your pain or mine? Common and distinct neural
systems supporting the perception of pain in self and other. Soc Cogn
Affect Neurosci 2008;3:14460.
[48] Petrovic P, Kalso E, Petersson KM, Andersson J, Fransson P, Ingvar M. A
prefrontal non-opioid mechanism in placebo analgesia. PAIN 2010;150:
5965.
[49] Peyron R, Laurent B, Garca-Larrea L. Functional imaging of brain
responses to pain. A review and meta-analysis (2000). Neurophysiol Clin
2000;30:26388.
[50] Rainville P, Duncan GH, Price DD, Carrier B, Bushnell MC. Pain affect
encoded in human anterior cingulate but not somatosensory cortex.
Science 1997;277:96871.
[51] Robinson ME, Staud R, Price DD. Pain measurement and brain activity:
will neuroimages replace pain ratings? J Pain 2013;14:3237.
[52] Ruscheweyh R, Deppe M, Lohmann H, Stehling C, Floel A, Ringelstein
EB, Knecht S. Pain is associated with regional grey matter reduction in the
general population. PAIN 2011;152:90411.
[53] Schmidt-Wilcke T, Luerding R, Weigand T, Jurgens T, Schuierer G,
Leinisch E, Bogdahn U. Striatal grey matter increase in patients suffering
from fibromyalgiaa voxel-based morphometry study. PAIN 2007;132:
10916.
[54] Schnitzler A, Ploner M. Neurophysiology and functional neuroanatomy of
pain perception. J Clin Neurophysiol 2000;17:592603.
[55] Schweinhardt P, Bushnell MC. Pain imaging in health and diseasehow
far have we come? J Clin Invest 2010;120:378897.
[56] Schweinhardt P, Kalk N, Wartolowska K, Chessell I, Wordsworth P,
Tracey I. Investigation into the neural correlates of emotional
augmentation of clinical pain. Neuroimage 2008;40:75966.
[57] Schweinhardt P, Kuchinad A, Pukall CF, Bushnell MC. Increased gray
matter density in young women with chronic vulvar pain. PAIN 2008;140:
41119.
[58] Segerdahl AR, Mezue M, Okell TW, Farrar JT, Tracey I. The dorsal
posterior insula subserves a fundamental role in human pain. Nat
Neurosci 2015;18:499500.
[59] Seminowicz DA, Labus JS, Bueller JA, Tillisch K, Naliboff BD, Bushnell
MC, Mayer EA. Regional gray matter density changes in brains of
patients with irritable bowel syndrome. Gastroenterology 2010;139:
4857.e42.
[60] Singer T, Seymour B, ODoherty J, Kaube H, Dolan RJ, Frith CD. Empathy
for pain involves the affective but not sensory components of pain.
Science 2004;303:115762.
[61] Tracey I, Mantyh PW. The cerebral signature for pain perception and its
modulation. Neuron 2007;55:37791.
[62] Turkeltaub PE, Eden GF, Jones KM, Zeffiro TA. Meta-analysis of the
functional neuroanatomy of single-word reading: method and validation.
Neuroimage 2002;16(3 pt 1):76580.
[63] Turkeltaub PE, Eickhoff SB, Laird AR, Fox M, Wiener M, Fox P. Minimizing
within-experiment and within-group effects in activation likelihood
estimation meta-analyses. Hum Brain Mapp 2012;33:113.
[64] Wager TD, Rilling JK, Smith EE, Sokolik A, Casey KL, Davidson RJ,
Kosslyn SM, Rose RM, Cohen JD. Placebo-induced changes in
fMRI in the anticipation and experience of pain. Science 2004;303:
11627.
[65] Wood PB, Schweinhardt P, Jaeger E, Dagher A, Hakyemez H, Rabiner
EA, Bushnell MC, Chizh BA. Fibromyalgia patients show an abnormal
dopamine response to pain. Eur J Neurosci 2007;25:357682.