Phantom-limb pain
as
a
perceptual
correlate
of
cortical reorganization
following
arm
amputation
H.
Flor*,
T.
Elbertt,
S.
Knechti,
C.
Wienbruchi,
C.
Pantevi,
N.
Birbaumer§,
W. Larblg§
Bc
E.
Taubli
* Department
of
Psychology,
Humboldt-University,
Hausvogteiplatz 5-7, D-10117
Berlin,
Germany
t Department
of
Psychology,
University
of
Konstanz,
Postfach 5560,
D44,
D-78434 Konstanz,
Germany
t Department
of
Neurology
and Institute
of
Experimental
Audiology,
University
of
Munster,
Kardinal-von-Galen-Ring
10,
D-48129 Munster,
Germany
§ Institute
of
Medical
Psychology
and
Behavioural
Neurobiology,
University
of
Tubingen,
Gartenstr, 29, D-72074
Tubingen,
Germany
11
Department
of
Psychology,
University
of
Alabama
at
Birmingham,
Birmingham,
Alabama
35294,
USA
ALTHOUGH
phantom-limb pain is a frequent consequence
of
the
amputation
of
an extremity, little is known about its origin
l
-4.
On
the basis
of
the demonstration
of
substantial plasticity
of
the
somatosensory cortex after amputationS or somatosensory
deafferentation in adult monk
eys
6,
it has been suggested
that
cortical reorganization could account for some non-painful phan-
tom-limb phenomena in amputees and
that
cortical reorganization
has an adaptive (that is, pain-preventing) function
2
,s,7,8.
Theoretical
and empirical work on chronic back pain
9
,lo
has revealed a positive
relationship between the amount
of
cortical alteration and the mag-
nitude
of
pain, so
we
predicted
that
cortical reorganization and
phantom-limb pain should
be positively related. Using non-invasive
neuromagnetic imaging techniques to determine cortical reorganiz-
ation in humans
ll
-
13
,
we
report a very strong direct relationship
(r
=0.93) between the amount
of
cortical reorganization and the
magnitude
of
phantom limb pain (but not non-painful phantom
phenomena) experienced after arm amputation. These
data
indi-
cate
that
phantom-limb pain is related to, and may
be
a conse-
quence of, plastic changes in primary somatosensory cortex.
A brief telephone interview was used to obtain information
about
the
amount
of
phantom-limb pain in
65
upper-limb ampu-
482
tees. This information served as the sole basis for the selection
of
a representative sample
of
13
subjects with widely varying
degrees
of
phantom-limb pain. The mean age
of
the
13
subjects
was
50.1
years (s.d. = 17,2, range 27-73 yr), mean post-amputa-
tion time was 24.3 years (s.d. = 19.8, range
I to
51
yr). Twelve
men
and
one woman participated in the study, Traumatic injury
in ten cases
and
osteosarcoma in three cases
had
made the ampu-
tation necessary. Cortical reorganization was determined by
magnetic source imaging' , using the method illustrated in Fig.
1.
The
subjects underwent a comprehensive neurological
and
psychological investigation which included detailed assessments
of
phantom
pain
and
phantom
sensations, stump pain
and
stump sensations, pre-amputation pain, telescoping (the subjec-
tive experience
of
the
phantom
limb retracting towards
and
often
disappearing in the stump),
and
facial remapping (the appear-
ance
of
phantom
sensations
upon
non-painful stimulation
of
the
face with isomorphism between facial stimulation sites
and
the
location
of
phantom
sensations) (Fig. 2 legend).
A large significant positive linear relationship was found
between the
amount
of
phantom-limb pain, as measured on the
standardized pain-intensity scale,
and
the
amount
of
cortical
reorganization
(r=0.93,
P<O.OOOI;
Fig. 2); phantom-limb pain
explained almost
85'Yo
(adjusted r
2
=0.83;
F(l,11)=66.89,
P < 0.0001)
of
the variance in cortical reorganization. A similar
correlation was obtained for the visual analogue
phantom
pain
scale
(r=0.85, P<O.OOI)
and
the adjective
phantom
pain scale
(r=0,86, P<O.OOI).
The mean shift in the focus
of
cortical responsivity
to
facial
stimulation was 0.43 cm (s.d.
=0.40,
range 0.01-1.00) for the
five
pain-free subjects, whereas the mean shift
(M)
for the eight
subjects with phantom-limb pain was almost
five
times as large
(M=2.05
cm, s.d. = 1.08, range 0.52-3.86;
F(l,11)
=9.94,
P<O.OI) (Fig. 3).
None
of
the measures for stump pain, non-
painful stump sensations
or
non-painful
phantom
sensations
showed a significant relationship to
amount
of
cortical reorgani-
zation. The presence, frequency
and
intensity
of
telescoping,
or
the length
of
the telescoped
phantom,
as well as reports
of
facial
remapping (which was present in 4
of
the
13
subjects) were
unrelated
to
cortical reorganization,
None
of
the
five
patients
without
phantom
pain reported the experience
of
pre-amputa-
tion pain. However,
half
of
the eight patients with
phantom
pain
had
experienced pain before the
amputation
(three due to
traumatic injury,
and
one due
to
osteosarcoma); the
other
half
First publ. in: Nature 375 (1995), No. 6531, pp. 482-484
Konstanzer Online-Publikations-System (KOPS)
URL: http://www.ub.uni-konstanz.de/kops/volltexte/2008/6371/
URN: http://nbn-resolving.de/urn:nbn:de:bsz:352-opus-63710
FIG.
1 Illustration
of
the
method used
to
determine
the
amount
of
cortical reorganization. The centres of the magnetic responses
to
stimu-
lation
on
the face and digits are superimposed onto a magnetic reson-
ance image of an individual subject.
METHODS.
In
all subjects
at
least the following four sites were stimula-
ted
by
using light superficial pressure applied via a pneumatic stimula-
tor: (1) the first and (2)
the
fifth digits of the (intact) hand; (3) the chin
below
the
left corner
of
the lip (first five subjects) or the lower lip near
the left corner
of
the mouth (eight subjects) and (4) the chin below the
right corner
of
the
lips (first five subjects) or the lower lip near
the
right
corner
of
the mouth (eight SUbjects). The recording from the lip was
found
to
yield a much better cortical response than the recording from
the chin and was therefore preferred in the latter series
of
recordings.
During these four conditions, the sensor array was positioned over the
hemisphere contralateral
to
the
site
of
stimulation. At each site
1,000
stimuli were delivered
at
an average stimulation rate
of
0.5
Hz
(interval
between stimulus onsets:
500
±
50
ms). The sequence
of
sites
at
which
stimuli were presented was varied according
to
a fixed irregular order
across subjects. After each train
of
1,000
stimuli, the subjects indicated
the primary and secondary (if any) location
of
the perceived stimula-
tions, as well as its quality and intensity. Using a
BTi
neuromagneto-
meter, magnetic fields were recorded from
37
locations over a circular
concave area
(14.4
cm diameter) above the parietotemporal cortex
contralateral
to
the site of
the
stimulation. Recordings were carried
out
in
a magnetically shielded room. Subjects lay
in
a lateral position with
their
whole body supported
by
vacuum cushions. The
MEG
was sampled
at
a rate
of
520.5
Hz.
The evoked magnetic responses from each stimu-
lation site were averaged (from
-100
to
+250
ms) and digitally filtered
with a bandpass
of
0.01
to
100
Hz.
To
exclude artefacts, a response
was omitted from the average
if
its range exceeded 2
pT
in
any of
the
MEG
channels. For each magnetic field distribution, a single equivalent
current dipole
(ECD)
model (best-fitting local sphere) was fitted within
the latency range from
35
to
75
ms. From the points with a goodness-
of-fit larger than
0.95
and a confidence volume smaller than
300
mm
3
,
the region with
the
maximal field power (measured as root-mean-square
across channels) was selected.
To
illustrate
the
measure
of
reorganiza-
tion used, these
ECD
locations were mapped onto the cortical surface
of
area 3b, which was reconstructed from a magnetic resonance image
scan as described
2D
Distances between locations along the curved
surface
of
the somatosensory map can be represented
by
the distance
in
the
three cortical dimensions (black rods). The arrows connecting the
cortical representation of
the
lip and (1) the midpoint between digits 1
and 5
on
the intact side (lower arrow) and (2)
the
midpoint between
the mirror images
of
the first and fifth digit and the lip on
the
amputated
side (upper arrow), represent the cortical-distance measure used. The
mirror images of digits 1 and 5 were obtained
by
projecting the centres
of magnetic activity
on
the intact side across the midsagittal plane
onto the opposite hemisphere (representing the amputation side of the
body).
To
obtain an estimate of the extent
of
the reorganization
that
had occurred, a comparison was made
of
the
difference in the two
distance measures (that
is,
the mean coronal
shift
in
the
dipole
of
the
amputation-side face area relative
to
the
fingers).
had
never experienced
pre-amputation
pain (Yates-corrected
X",
NS).
The
overall correlation between
pre-amputation
pain
and
phantom
pain was thus only
moderate
and
not
significant.
There was a trend towards a significant relationship between the
amount
of
cortical reorganization and pre-amputation pain (I' =
0.51,
F(I,
11) = 3.91, P = 0.07) which was entirely due to the posi-
tive
though
nonsignificant relationship
of
phantom-limb pain
and
pre-amputation
pain (partial correlation
of
reorganization
and
pre-amputation
pain with phantom-limb pain removed:
I'
=
-0.132).
Age
of
patients
at
time
of
testing, age
at
time
of
amput-
ation,
and
time elapsed since the
amputation
were unrelated to
both
phantom-limb pain
and
cortical reorganization (all P
values NS). A final stepwise regression analysis, with all pain,
sensation, demographic and clinical variables entered to predict
the
amount
of
cortical reorganization, yielded a single predictor:
amount
of
phantom-limb
pain experienced by the subjects.
The results
of
this study reveal a strong positive relationship
between the
amount
of
cortical reorganization occurring after
unilateral upper-extremity
amputation
and the intensity
of
phan-
tom pain. The present findings suggest the intriguing possibility
that
the shift
of
the cortical
map
following
amputation
might
be a potential neurophysiological basis
of
phantom-limb
pain.
The result contradicts
current
influential theoretical analyses
of
phantom-limb pain
2
.
5
,
in which an inverse relationship between
painful phantom-limb
phenomena
and cortical reorganization
was postulated,
but
is
consistent with Melzack's' assumption
of
a
'neuromatrix'
underlying phantom-limb pain. The involvement
of
cortical reorganization in
phantom-limb
pain
is
not
inconsist-
ent with the extensive evidence
that
peripheral mechanisms also
contribute to
phantom
pain'·2. The remodelling
of
the functional
architecture
of
the cortex in response
to
nervous-system damage
could serve as an adaptive compensatory mechanism by restor-
5
4
5
5
c
4
4,5
'0;;
35
4
E
0..
3
()
E
3
5
~
~
2
5
3
e.n
2
2
5
0
'"
)5
~
.c
0-
1
2
-;;
],5
.~
0
5
1
1:
0
0
0
5
U
0
FIG.
2 Amount
of
cortical reorganization (in cm) for each subject plotted
against intensity of phantom-limb pain, as measured with the Multi-
dimensional Pain Inventory.
METHODS.
Phantom-limb and
stump
pain were assessed
by
three
methods: (1) the pain intensity scale
of
the West Haven-Yale Multidi-
mensional Pain
Inventorl
'
,22,
a reliable and valid measure of the
amount
of pain experienced, administered separately for phantom and
stump pain; (2) a phantom-and-stump phenomena interview
that
included the pain experience scale (Schmerzempfindungsskala
23
) con-
sisting
of
24
pain adjectives derived from the
McGi11
pain
questionnaire
24
. Each
of
these
24
descriptors was scored
on
a 4-point
scale according
to
the extent
to
which it accurately described
the
sub-
jects'
pain experience. The scale was administered separately for
phantom-limb pain,
stump
pain, and pre-amputation pain; (3) a
10-cm
visual analogue scale with the endpoints 'no pain' and 'unbearable
pain'. Non-painful phantom phenomena (such as telescoping) and non-
painful stump sensations (such as itching, pressure) were measured
in
the phantom-and-stump phenomena interview. The presence
of
facial
remapping was assessed
by
probing the surface
of
the entire body with
a cotton-wool bud while
the
patient indicated the presence of phantom
sensations
8
,1'.
483
FIG.
3 A representative subject with intense phantom-limb pain (black
symbols with white surround) and one subject who experienced no
phantom-limb pain (white symbols with black surround) were selected
from the patient sample. Squares represent the
ECD
locations
of
the
digits, circies represent the location
of
the lip projected onto a coronal
section of
the
brain. Hemisphere contralateral
to
intact side (left) and
amputation side (right). A pronounced hemispheric asymmetry in the
location
of
the lips can be observed in the phantom-pain patient.
ing activity in a zone deprived
of
its afferent input
and
could
provide a basis for recovery
of
function.
In
patients with
phan-
tom-limb pain, this mechanism would
appear
to have become
maladaptive, possibly related to a lasting hyperexcitability
of
nociceptive pathways induced
by
previous pain
14,
imbalance
of
nociceptive
and
non-nociceptive inputs after deafferentation
l5
,
or
similar mechanisms.
The
reorganization observed in some
subjects with
phantom-limb
pain (exceeding 3 cm in two sub-
jects)
probably
encompasses the full extent
of
the cortical area
3b representing the
hand
and
lower arm area before
amputation.
It
is possible
that
reorganization may have occurred in additional
areas
of
SI
as well as SIC which would
not
have been assessed
by
our
methods.
It
has been suggested
5
.
16
that
the reorganization
could be due to synaptic remodelling within the arborization
zones
of
thalamocortical neurons projecting to areas
of
somato-
sensory cortex
that
had retained their normal sources
of
input
and
which were adjacent to the affected cortical areas
that
had
lost their afferents (cortical face
or
head areas). This mechanism,
however,
can
account
for only a small fraction
of
the full topo-
graphic extent
of
the massive reorganization in the present study.
To
what
extent chronically altered afferent inputs produced by
sprouting
17
, increase in synaptic strength
IX
or
unmasking
of
intracortical connections
l9
contribute to the reorganization
observed here needs to be investigated further. In addition, deter-
mination
of
behavioural
methods
or
pharmacological agents
affecting cortical reorganization might eventually be effective
in
relieving
phantom-limb
pain. D
Received
14
February; accepted
29
March
1995.
1. Jensen.
T.
S.
& Rasmussen,
P.
in
Textbook
of
Pain (eds Wall,
P.
& Melzack,
R.
A.)
651-
666
(Churchill-Livingstone, Edinburgh,
1994).
2. Katz,
J.
Can.
J.
Psychiat.
37,
282-291
(1992).
3.
Melzack,
R.
A.
Can. Psychol.
30,
1-16
(1989).
4.
Sherman,
R.
A.,
Arena,
J.
C.,
Sherman,
C.
J.
& Ernst,
J.
C.
Biof
Self-Regul.
14,
267-280
(1989).
5.
Merzenich,
M.
et. al.
J.
comp. Neurol.
224,591--605
(1984).
6.
Pons,
T.
et
al. Science
252,
1857-1860
(1991).
7.
Ramachandran,
V.,
Rogers-Ramachandran,
D.
& Stewart,
M.
Science
258,
1159-1160
(1992).
8. Ramachandran, V
..
Stewart,
M.
& Rogers-Ramachandran,
D.
Neuroreport
3,
583-586
(1992).
9.
Flor,
H.
& Birbaumer,
N.
Am. Pain Soc.
1.
3,
118-127
(1994).
10. Fiar,
H.
et
al.
in
Recent Advances in
Biomagnetism
(eds Deecke. L., Baumgartner,
C"
Stromk,
G.
& Williamson,
S.
J.)
(Elsevier, Amsterdam:
In
the press).
11. Elbert,
T.
et al.
Neuroreport
5,
2593-2597
(1994).
12.
Yang.
T.
et
al.
Neuroreport
5,
701-704
(1994).
13.
Yang.
T.
et
al.
Nature
368,
592-593
(1994).
484
14. Willis,
W.
D.
Jr
in
Proc. 7th World Congr. on Pain (eds Gebhart,
G.
F.,
Hammond.
D.
L.
&
Jensen,
T.
5.)
301-324
(IASP,
Seattle,
1994).
15. Pain and Central Nervous System Disease (ed. Casey,
K.
L.) (Raven, New York
1991).
16.
Lund,
J.
T.,
Sun,
G.
D.
& Lamarre.
Y.
Science
265,
546-548
(1994).
17.
Danan-Smlth,
C.
& Gilbert,
C.
D.
Nature
368,
737-740
(1994).
18.
Gilbert.
C.
D.
Curr. Opin. Neurobiol.
3,
100-103
(1993).
19.
Calford,
M.
B.
& Tweedale,
R.
Proc.
R.
Soc.
B.
243,
269-275
(1991).
20. Lutkenhbner,
B.
et
al. Brain Topography (in
the
press).
21. Kerns,
R.
D.,
Turk,
D.
C.
& Rudy,
T.
E.
Pain
23,
345-356
(1985).
22. Flor,
H.,
Rudy,
T.
E.,
Birbaumer,
N.,
Streit,
B.
& Schugens,
M. M.
Der Schmerz
4,
82-87
(1990).
23. Geissner,
E.
Die Schmerzempfindungsskala (Beltz, Weinhelm,
in
the
press).
24.
Melzack,
R.
A.
Pain
1,
277-299
(1975).
ACKNOWLEDGEMENTS. This work was supported
by
grants from the Deutsche
Forschungsgemeinschaft.