Exp Brain Res (1998) 119:205±212 Springer-Verlag 1998
RESEARCH ARTICLE
Herta Flor ´ Thomas Elbert ´ Werner Mühlnickel
Christo Pantev ´ Christian Wienbruch ´ Edward Taub
Cortical reorganization and phantom phenomena in congenital
and traumatic upper-extremity amputees
Received: 4 June 1997 / Accepted: 13 September 1997
H. Flor (
)
) ´ W. Mühlnickel
Department of Psychology, Clinical Psychology
and Behavioral Neuroscience, Humboldt-University,
Hausvogteiplatz 5±7, D-10117 Berlin, Germany
e-mail: hflor@rz.hu-berlin.de, Fax:+49-30-20377308
T. Elbert
Department of Psychology, University of Konstanz, Germany
C. Pantev ´ C. Wienbruch
Institute of Experimental Audiology, University of Münster,
Germany
E. Taub
Department of Psychology, University of Alabama at Birmingham,
USA
Abstract The relationship between phantom limb phe-
nomena and cortical reorganization was examined in five
subjects with congenital absence of an upper limb and
nine traumatic amputees. Neuromagnetic source imaging
revealed minimal reorganization of primary somatosenso-
ry cortex in the congenital amputees (M=0.69 cm, SD
0.24) and the traumatic amputees without phantom limb
pain (M=0.27 cm, SD 0.25); the amputees with phantom
limb pain showed massive cortical reorganization
(M=2.22 cm, SD 0.78). Phantom limb pain and nonpain-
ful phantom limb phenomena were absent in the congen-
ital amputees. Whereas phantom limb pain was positively
related to cortical reorganization (r=0.87), nonpainful
phantom phenomena were not significantly correlated
with cortical reorganization (r=0.34). Sensory discrimina-
tion was normal and mislocalization (referral of stimula-
tion-induced sensation to a phantom limb) was absent in
the congenital amputees. The role of peripheral and cen-
tral factors in the understanding of phantom limb pain and
phantom limb phenomena is discussed in view of these
findings.
Key words Cortical plasticity ´ Phantom limb pain ´
Traumatic amputation ´ Congenital aplasia ´
Neuromagnetic source imaging ´ Human
Introduction
The occurrence of a phantom limb, i.e., the subjective
feeling of the continued presence of a body part that is
no longer present, seems to be an almost universal conse-
quence of amputation. Phantom limbs develop in 80±
100% of traumatic amputations; in about 50±80% of those
cases they are painful (Jensen et al. 1983; Sherman et al.
1984). Whereas there is general agreement that painful
and nonpainful phantom limbs occur in traumatic ampu-
tees whose amputations took place in adult life, there is
conflicting data concerning persons whose amputation is
congenital or occurred in early childhood. Beginning with
Pick (1915), it was assumed by a number of investigators
that an image of the body develops only with experience;
congenital or early traumatic amputees were thought not
to have a well-developed body image and could therefore
not experience phantom limbs. This conception was sup-
ported in a number of studies for both congenital and
child amputees (Riese and Bruck 1950; Kolb 1954; Sim-
mel 1956, 1961; Jùrring 1971). In children, it appeared
that the presence of phantom limbs became more likely
the more advanced the age of the child was at the time
of the amputation. Whereas children who were amputated
before the age of 3 years virtually never reported phantom
limbs, the number experiencing phantom limbs increased
up to about the age of 8 years, when phantom limbs be-
came as frequent as in adult amputees (Riese and Bruck
1950; Boeri and Negri 1954; Simmel 1956).
These early results were, however, challenged when
Weinstein and his colleagues (Weinstein and Sersen
1961; Weinstein et al. 1964; Vetter and Weinstein
1967) and Poeck (1964) found that some congenital (up
to 18% in Weinstein et al. 1964) and early child amputees
did report phantom limbs. These studies were criticized
for using the testimony of children who might be more
prone to respond to demand characteristics of the situation
than adults (Skoyles 1990). Several studies assessed con-
genital amputees when they were adult; these studies re-
ported extremely low numbers of phantom limbs. For ex-
ample, Burchard (1965) carefully examined and ques-
Konstanzer Online-Publikations-System (KOPS)
URL: http://www.ub.uni-konstanz.de/kops/volltexte/2007/4139/
URN: http://nbn-resolving.de/urn:nbn:de:bsz:352-opus-41390
First publ. in: Experimental Brain Research 119 (1998), pp. 205-212
206
tioned 17 adult subjects with congenital absence of limbs;
only one of his subjects reported sensations similar to
those of a phantom limb, although that subject reported
ªimaginingº the phantom limb rather than actually ªfeel-
ingº it. Sadaah and Melzack (1994) reported the occur-
rence of phantom limbs in 4 out of 65 (7%) of the congen-
ital amputees questioned in a mail survey. However, in all
4 cases, the experience of a phantom limb occurred only
after subsequent trauma to the limb (e.g., surgery). Phan-
tom limb pain seems to be virtually absent in congenital
amputees (Simmel 1956; Jùrring 1971). Only Sohn
(1914) described a case of phantom limb pain in a con-
genital amputee. This amputee had, however, pain in
the stump radiating down to the phantom fingers. Since
this patient still had a very small rudimentary hand it is
not clear if the hand just felt inflated (and painful) due
to the severe stump pain or if there was true phantom limb
pain. In addition, one of the patients of Weinstein et al.
(1964) might have displayed a painful phantom limb
(the subject described the phantom limb as ªachingº),
but no further information is given.
More recent data suggest that children and adolescents
have a higher occurrence of phantom limbs and phantom
pain than seems to have been estimated previously (cf.,
Krane and Heller 1995; Sherman 1997). For example,
Krane and Heller (1995) report the occurrence of phantom
limbs in 100% of the 24 children and adolescents they ex-
amined, and phantom limb pain in 92%. They relied,
however, on mail surveys, which may be inadequate to
differentiate clearly between the symptoms of nonpainful
phantom limbs, painful phantom limbs, nonpainful stump
sensations, and stump pain. This possibility prompted us
to use extensive, structured interviews (see also Burchard
1965).
The reasons for the report of varying percentages of
nonpainful phantom limbs and phantom limb pain in con-
genital as compared to traumatic amputees have thus far
not been sufficiently explored. One problem with some
of the studies are the shortcomings related to the method
of determining the presence or absence of phantom phe-
nomena. In addition, both peripheral and central factors
might be different for the two groups of subjects, giving
rise in them to the experience of different types of phan-
tom phenomena. Whereas traumatic (including child) am-
putees have clearly defined deafferentation involving the
section of existing axons, congenital amputees are not ac-
tually ªamputees.º Rather their condition involves an
aplasia of the limbs such as in phocomelia (a congenital
deformity in which the limbs are extremely shortened)
and peromelia (congenital malformation). The limb often
has the appearance of having undergone an amputation
(Holmes and Borden 1974; Herrman and Opitz 1977;
Pfeiffer and Santelman 1977), but no nerves have been se-
vered as in traumatic or surgical amputation. Instead, a
lack of blood supply to the hand or malformation due to
genetic or unknown causes may lead to the congenital ab-
sence of the limb or parts of it.
So far, brain processes related to congenital aplasias
have rarely been studied, although they might elucidate
some of the differences found between congenital and
traumatic amputees. Hall et al. (1990) used transcranial
magnetic stimulation to study possible reorganization of
the motor cortex in two congenital amputees and two
traumatic amputees and four healthy controls. They found
that motor potentials on the amputated side could be
evoked with a lower threshold and from a larger area in
both congenital and one early traumatic amputee but not
in a late traumatic amputee. Unfortunately, Hall et al.
(1990) did not provide any further information on the pa-
tients they studied. In contrast, Cohen et al. (1991) report-
ed that only the traumatic amputees showed enhanced re-
activity in the musculature close to the amputation site,
but this did not occur in a congenital amputee. The find-
ings of Kew et al. (1994) are consistent with those of the
latter experiment. Kew et al. studied three traumatic and
three congenital amputees using positron emission tomog-
raphy and transcranial magnetic stimulation during paced
shoulder movements on the intact and the amputation
side. They found much larger and much more widespread
activation in the sensorimotor cortices contralateral to the
side of the amputation in the traumatic amputees than in
the congenital amputees, who showed only a slightly
but nonsignificantly higher activation of the primary so-
matosensory and primary motor (SI/MI) areas. Higher
corticospinal excitability contralateral to the missing limb
was found only in the traumatic amputees. In addition, the
deafferented SI/MI cortex of the traumatic amputees
showed some response to ipsilateral arm movement; this
activation was absent in the congenital amputees. An ad-
ditional area of activation was found in Brodmans area 5
of the posterior parietal cortex in traumatic amputees; this
was not present in the congenital amputees. Unfortunate-
ly, Kew et al. did not report on the presence of phantom
limb pain in their traumatic amputees.
The purpose of this study was to carry out a detailed
analysis of the organization of the SI by neuromagnetic
source imaging and to determine its relationship with both
painful and nonpainful phantom phenomena in congenital
and traumatic amputees. In line with our previous results
(Flor et al. 1995), we expected to find significant phantom
limb pain and cortical reorganization in traumatic ampu-
tees, with a close association between the two, but no sig-
nificant association of nonpainful phantom limb phenom-
ena and cortical reorganization. In view of the fact that
most previous work reported an absence of phantom limb
pain in congenital amputees, we hypothesized that there
would be little or no cortical reorganization in primary so-
matosensory cortex of these individuals. In addition, we
sought to determine sensory perception and possible mis-
localization (or referral of stimulation-induced sensation
to a phantom limb) in the congenital amputees in an effort
to elucidate potential additional perceptual correlates of
cortical reorganizational changes.
207
Materials and methods
Subjects
Five congenital and nine traumatic amputees, five without and four
with phantom limb pain, participated in the study, as well as ten
healthy controls. The traumatic group comprised a subsample of a
large sample of 100 unilateral upper-extremity amputees, chosen ac-
cording to availability for the investigation in the neuromagnetom-
eter laboratory. All amputees volunteered for the project in response
to inquiry by their prosthetist or physician. Table 1 shows age, gen-
der, and time since amputation of the study sample. The congenital
amputees were younger than the traumatic amputees. Time since
amputation was not significantly different between the groups.
The mean age at the time of the amputation was 22 years in the trau-
matic amputees (ranging from 12 to 52 years). The subjects under-
went a comprehensive neurological and psychological investigation,
which included detailed assessments of phantom pain and phantom
sensations, stump pain and stump sensations, preamputation pain,
telescoping, and mislocalization (Flor et al. 1995; Knecht et al.
1996). The comparison sample of ten healthy controls was matched
with the congenital amputees for age (M=36.07 years, SD 12.05,
range 23±51 years) and gender (six women, four men).
Assessment of perceptual phenomena
The presence of referral of stimulation-induced sensation to the
phantom limb (mislocalization) was assessed by probing the surface
of the entire body with a q-tip while the patient had to indicate the
quality, location, and intensity of the primary sensation he or she ex-
perienced as well as the presence of any secondary (e.g., phantom)
sensations (Ramachandran et al. 1992). No other modalities (e.g.,
pin prick) were probed. Sensory testing included two-point discrim-
ination and electric and thermal perception, pain threshold, and pain
tolerance. These were assessed only in the congenital amputees and
the comparison group of healthy control subjects. Two-point dis-
Table 1 Demographic, perceptual, and cortical reorganizational data for congenital and traumatic amputees
Variable Congenital amputes
(n=5)
Traumatic amputes
Without phantom
limb pain (n=5)
With phantom
limb pain (n=4)
Demographic data
Age (years) Mean 31.60 50.80 58.35
SD 9.76 19.24 18.82
Gender (n) Male 1 4 4
Female 4 1 0
Time since amputation Mean 31.60 28.60 37.23
(years) SD 9.76 28.62 24.17
Age at amputation (years) Mean Prenatal 22.80 21.80
SD 17.91 7.75
Trauma to arm (n) 3 5 4
Stump length (cm) Mean 36.30 29.67 23.50
SD 14.52 15.70 8.69
Perceptual data
Stump sensation (n) Mean 0 0.80 2.50
SD 0 0.84 1.00
Phantom sensation (n) Mean 0 2.20 2.50
SD 0 2.05 2.65
Stump pain intensity VAS (0±100) Mean 6.00 9.60 41.00
SD 13.42 13.44 28.30
Phantom limb pain intensity (MPI value) Mean 0 0 3.18
SD 0 0 1.06
Two-point discrimination (mm)
Stump Mean 85.6 77.0 61.0
SD 60.2 52.4 53.5
Contralateral arm Mean 75.2 72.8 75.2
SD 38.3 49.6 38.3
Mislocalization (n) 0 1 3
Points yielding mislocalization (n) Mean 0 0.40 5.78
SD 0 0.89 5.32
Telescoping (n) 0 2 4
Physiological data
Cortical reorganization Mean 0.69 0.27 2.22
Lip (cm) SD 0.24 0.25 0.78
Toe (cm) Mean 0.72 ± ±
SD 0.31 ± ±
208
crimination was assessed using a caliper-probe opened to progres-
sively larger extensions, starting at 1 mm in three ascending series.
A Medoc thermal stimulator was used to determine thermal thresh-
olds, which were measured using three ascending series. Electric
stimulation thresholds were also assessed in three ascending series
using a Tönnies electric stimulator, which delivered unipolar im-
pulses of 100 ms duration. The stimuli were applied to the thenar
eminence of the intact hand (both hands in the healthy controls),
6 cm proximal to the amputation line on the amputated arm, a ho-
mologous site on the intact arm, and both corners of the mouth.
Phantom limb and stump pain were assessed by three methods: (a)
the pain intensity scale of the West Haven-Yale Multidimensional
Pain Inventory (MPI; Kerns et al. 1985; Flor et al. 1990), a reliable
and valid measure of the amount of pain experienced, which was ad-
ministered separately for phantom and stump pain; (b) a phantom
and stump phenomena interview, including the Pain Experience
Scale (Schmerzempfindungsskala; Geissner 1997),which consists
of 24 pain adjectives derived from the McGill pain questionnaire.
These 24 descriptors were scored on a 4-point scale according to
the extent to which they accurately described the subjects pain ex-
perience. The scale was administered separately for phantom limb
pain, stump pain, and preamputation pain. (c) A 10-cm visual analog
scale with the endpoints ªno painº and ªunbearable pain.º Nonpain-
ful phantom phenomena (e.g., telescoping) and nonpainful stump
sensations (e.g., itch, pressure) were measured in the phantom and
stump phenomena interview (Flor et al. 1995).
Assessment of cortical reorganization
Cortical reorganization was determined by neuromagnetic source
imaging. In all subjects (except for one congenital amputee for
whom only subjective data could be obtained), at least the follow-
ing four sites were stimulated by using light superficial pressure
applied via a pneumatic stimulator: (1) the first and (2) fifth digits
of the (intact) hand, the lower lip near the (3) left and (4) right
corners of the mouth. In two congenital amputees, the first toe
was stimulated bilaterally in order to determine the full extent
of the reorganization along the sensory homunculus. Using a
BTi neuromagnetometer, 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 stimula-
tion. Recordings were carried out in a magnetically shielded
room. Subjects lay in a lateral position with their whole body sup-
ported by vacuum cushions. At each stimulation site, 1000 stimuli
were delivered at an average 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 1000 stimuli, the sub-
jects indicated the primary and secondary (if any) location of
the perceived stimulation as well as its quality and intensity.
The magnetoencephalogram (MEG) was sampled at a rate of
520.5 Hz. The evoked magnetic responses from each stimulation
site were averaged (from ±100 to +250 ms) and digitally filtered
with a bandpass of 0.01±100 Hz. In order to exclude artifacts, a
response was omitted from the mean 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 20 to 70 ms.
From the points with a goodness of fit larger then 0.95 and a con-
fidence volume smaller then 300 mm
3
, the region with the maxi-
mal field power (measured as root-mean-square across channels)
was selected. To illustrate the measure of reorganization used,
these ECD locations were mapped onto the cortical surface, which
was reconstructed from a magnetic resonance image using the
procedure described by Lütkenhöner et al. (1995). The mirror im-
ages of digits 1 and 5 were obtained by projecting the centers of
magnetic activity on the intact side across the midsagittal plane
onto the hemisphere contralateral to the amputation side. To ob-
tain an estimate of the extent of the reorganization that had oc-
curred, a comparison was made of the distances in the coronal
projection. The mean difference in the two distance measures
(i.e., mean coronal shift in the dipole of the amputation side face
area relative to the mean of the fingers) was used as the measure
of cortical reorganization (Flor et al. 1995).
Fig. 1 Locations of the representation of the first (D1, circle) and
fifth (D5, square) digits and both sides of the the mouth (triangles)
in the three groups of patients superimposed on a coronal section of
a magnetic resonance (MR) image of a healthy control. The cortical
representation of the amputated side is depicted on the left side
(right hemisphere) of each MR image; the cortical representation
of the intact side is shown on the right side (left hemisphere). The
dipole locations of the digits and the mouth, which can be located
in the depths of sulci or within cortex on the MR image, were pro-
jected onto the cortical surface based on the algorithm described by
Dale and Sereno (1993). It involves modeling the surface of a two-
dimensional projection of the coronal section of the cortex as a one-
dimensional line; a deep dipole location is projected onto this line at
the point that is closest to the dipole (i.e., a line from the dipole per-
pendicular to the closest representation of the cortical surface). Note
that the cortical representation of the mouth on the amputation side
has shifted toward the hand region in the traumatic amputees with
phantom limb pain but not in the congenital amputees or amputees
without phantom limb pain
209
Results
Cortical reorganization
The most extensive reorganization was observed in the
traumatic amputees with phantom limb pain (M=2.22 cm,
SD 0.78), whereas the congenital amputees (M=0.69 cm,
SD 0.24) and the traumatic amputees without phantom
limb pain displayed almost no reorganization (M=
0.27 cm, SD 0.25). These group differences were signifi-
cant (F
2,10
=20.10, P<0.001; see Fig. 1). Individual post
hoc tests revealed significant differences between the
pain-free amputees and the amputees with phantom pain
(P<0.001) and the congenital amputees and the amputees
with phantom limb pain (P<0.01), but not between the
congenital and pain-free amputees. In the two congenital
amputees where the location of the toe was assessed, the
dipole of the cortical representation of the toe on the am-
putation side was very similar to that of the intact side (eu-
clidean distance: M=0.71 cm, SD 0.32). Phantom limb
pain was significantly positively correlated with cortical
reorganization (r=0.87, N=13, P<0.001), but not with
nonpainful phantom sensations (r=0.34, N=13, n.s.) or
any other amputation-related variables such as time since
amputation or stump length.
Perceptual phenomena
None of the congenital amputees had any painful or non-
painful phantom phenomena, although three of the con-
genital amputees had experienced trauma to the stump
(surgery, painful inflammation, stump pain). Stump pain
was significantly more severe in the traumatic amputees
with phantom limb pain than in the congenital group,
but not significantly different in the two traumatic groups
(F
2,11
=4.66, P<0.05; t-test, P<0.05; see Table 1). Phan-
tom sensations and painful stump sensations were only
present in the two traumatic amputee groups, with no sig-
nificant differences between them. Mislocalization of sen-
sation to the missing limb was nonexistent in the congen-
ital amputees, but present in one of five traumatic ampu-
tees without phantom limb pain and three of four traumat-
ic amputees with phantom limb pain (overall c
2
(2)=7.25,
P<0.05).
When the number of points that elicited mislocaliza-
tion were compared, the congenital amputees reported
none, the traumatic amputees without pain 0.40, and the
traumatic amputees with pain 5.75 on the average (Krus-
kal-Wallis=6.87, P<0.05). Telescoping was not reported
by any of the congenital amputees, but was present in
all four traumatic amputees with pain and in two of five
traumatic amputees without pain. Two-point discrimina-
tion at the stump was not significantly better in any of
the amputee groups than at the homologous site on the in-
tact arm (see Table 1). Thermal and electrical perception,
pain thresholds, and pain tolerance values are presented in
Table 2 Thermal and electrical perception threshold, pain threshold, and pain tolerance levels for congenital amputees
and a healthy control group
Congenital amputees (n=5) Healthy controls (n=10)
Thermal Intact
hand
Stump Intact
arm
Intact
side lip
Ampu-
tation
side lip
Dominant
hand
Non-
dominant
hand
Dominant
hand
Non-
dominant
hand
Dominant
hand
Non-
dominant
hand
Perception threshold (C)
Mean 33.48 32.96 33.51 32.75 32.74 34.15 33.87 34.68 35.20 33.14 33.13
SD 0.33 0.27 0.59 0.42 0.35 2.45 1.80 1.01 0.64 1.24 0.95
Pain threshold (C)
Mean 41.84 39.53 38.63 39.17 38.33 40.12 41.15 40.90 42.02 37.11 37.30
SD 5.32 3.62 3.28 3.77 3.59 3.73 3.90 4.64 3.36 3.06 2.74
Pain tolerance (C)
Mean 47.70 43.75 43.00 42.43 41.83 45.59 45.73 46.70 48.12 42.54 42.54
SD 3.82 3.26 3.81 2.71 3.64 3.10 3.79 2.92 1.97 3.98 3.98
Electrical Stimulation
Perception threshold (mA)
Mean 2.75 2.94 2.01 2.10 2.90 4.79 4.72 3.26 3.30 3.15 3.34
SD 1.48 2.61 1.68 1.28 1.99 1.81 1.78 0.79 0.87 0.63 0.55
Pain threshold (mA)
Mean 9.53 9.71 7.21 9.13 9.46 10.78 10.85 11.60 11.60 7.59 7.97
SD 8.50 8.48 5.70 8.24 8.14 3.63 4.26 2.07 1.82 2.34 2.63
Pain tolerance (mA)
Mean 11.85 11.98 10.79 10.81 10.78 15.46 14.86 17.22 17.82 12.91 12.67
SD 8.25 8.42 7.81 7.30 7.42 4.20 5.18 3.29 3.51 4.90 5.15
