Brain (1996), 119, 1729-1736
Paradoxical heat sensation in patients with multiple
sclerosis
Evidence for a supraspinal integration of temperature
sensation
C. Hansen,12 H. C. Hopf2 and R. D. Treede1
institute of Physiology and Pathophysiology and the
Department of Neurology, Johannes Gutenberg University,
Mainz, Germany
2
Correspondence to: Professor Dr med. Rolf-Detlef Treede,
Institute of Physiology and Pathophysiology,
Saarstrasse 21, D-55099 Mainz, Germany
Summary
Temperature thresholds were determined in 16 patients
with probable or definite multiple sclerosis, in six patients
with possible but unconfirmed multiple sclerosis and in 34
healthy subjects, using the method of limits and the thermal
sensory limen (TSL) of the MarStock technique. A significant
proportion of the patients had thresholds outside the 2.5 SD
range for normal subjects, both for warmth detection
threshold and TSL. In addition, 10 patients with probable or
definite multiple sclerosis and one patient with possible
multiple sclerosis reported a paradoxical heat sensation, i.e.
a sensation of warmth elicited by a cold stimulus. This
illusion was almost exclusively observed with the alternating
warm and cold stimuli of the TSL procedure. In contrast
to experimental nerve block or peripheral demyelinating
neuropathy, where paradoxical heat sensation has been
described by various authors, in the patients with multiple
sclerosis the demyelination sites were located in the central
nervous system. The observation that multiple sclerosis
patients had paradoxical heat sensation in addition to
threshold abnormalities supports the view that supraspinal
sites are important for the integration of temperature
sensation.
Keywords: multiple sclerosis; paradoxical heat sensation; quantitative sensory testing; thermoreception; demyelination
Abbreviations: CDT = cold detection threshold; HPC = heat, pinch, cold; PSneg = without paradoxical heat sensation;
PSpos = with paradoxical heat sensation; TSL = thermal sensory limen; WDT = warmth detection threshold
Introduction
Abnormal somatosensory sensations are common in patients
with multiple sclerosis. Cold, warmth or burning sensations
are often reported, but measurements to determine the
function of the temperature pathways are not included in
routine diagnostic procedures. Quantitative sensory testing
has been used extensively in peripheral neuropathies (Claus
et al., 1990; Yarnitsky and Ochoa, 1991; Dyck et al., 1993).
In the few studies in which these methods were applied to
multiple sclerosis patients, a variable incidence of threshold
abnormalities between 25 and 75% was found (Heijenbrok
et al., 1992; Boivie, 1994; Osterberg et al., 1994).
Thermal sensation is not a simple function of skin
temperature. Within the range of 3O-35°C, a given temperature can be perceived as cold, neutral or warm, depending
on the stimulus history; this is easily demonstrated in the
© Oxford University Press 1996
three-bowl experiment of Weber and Hering (Tritsch, 1990).
Rapid heating of the skin to temperatures around 45°C may
paradoxically be perceived as cold (Long, 1977), and cooling
the skin may be paradoxically perceived as heat (Hamalainen
et al., 1982; Greenspan et al., 1993). One way to induce the
illusion of heat during a cold stimulus is the simultaneous
application of two stimuli in close vicinity, one warm (40°C)
and the other cold (20°C) which is called the Thunberg-grill
(Green, 1977; Craig and Bushnell, 1994). Paradoxical cold
is explained by paradoxical responses of cold fibres to strong
heat stimuli (Long, 1977). Since there is no peripheral
explanation of paradoxical heat sensation, central mechanisms
must be assumed, but the central integration of cold and
warm fibre inputs is poorly understood. Recently, it has been
suggested that the central interaction of a cold-specific
1730
C. Hansen et al.
Table 1 Clinical data of the twenty-two patients studied
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Poser
criteria
Definite
Definite
Definite
Definite
Definite
Definite
Definite
Definite
Definite
Definite
Definite
Definite
Definite
Probable
Probable
Probable
Possible
Possible
Possible
Possible
Possible
Possible
Sex
M
M
M
F
F
F
M
F
M
M
M
F
F
F
F
M
F
M
F
F
F
F
Age
(years)
45
38
45
33
33
23
56
46
34
26
60
38
36
39
26
48
36
56
33
45
28
55
Disease
duration
9 years
9 years
14 years
1 year
4 years
7 years
7 years
11 years
3 weeks
6 years
17 years
15 years
16 years
2 months
4 weeks
6 months
3 weeks
3 weeks
2 weeks
2 years
4 months
3 years
MRI
PV
PV.C6
PV
PV
PV
PV,BS,CB
PV,Th7-10
n.d.
PV
PV
n.d.
PV
PV
PV
PV
NTL
Normal
NTL
Normal
NTL
Normal
Normal
Somatosensory disturbances
SEP
PS,pos
RF
RH
RF
RF
DC
STT
LH,LF
LH,LF
RF,LF
RM,RT
RH.LH
RH,LH,RF,LF
RF,LF
RM
RH,RF
RH,LH
RH,LH,RF,LF
RF,LF
PSS
LM,LT,RT
RF,LF
RF,LF
RM,RT
RM,RT
RM.RT.LT
RT
RM,LM,RT,LT
RH.RI
+
RF,LF
RH
RF,LF
RF,LF
RF,LF
RF,LF
RF,LF
RF,LF
RF,LF
LF
LF
+
+
RT,LT
RT,LT
RH,LH,RF,LF
+
RH
RH
+
RH,RF
RF,LF
RF,LF
RT,LT
LF
Disease duration from the first symptom attributed to multiple sclerosis, even if the diagnosis was not provided at this time. Location of
lesions with MRI: PV = periventricular; BS = brainstem; CB = cerebellum or the indicated cervical (C) or thoracic (Th) spinal cord
segment; n.d. = not done; NTL = not in typical locations. Clinical signs of somatosensory disturbance were attributed to dorsal column
function (DC = hypaesthesia or numbness) or spinothalamic tract function (STT = thermhypaesthesia or hypalgesia). PSS = positive
sensory signs (paraesthesia, hyperpathia, pain). SEP = pathological somatosensory evoked potential after stimulation of the median (M)
and tibial (T) nerve, left (L) and right (R); marked SEP were pathologic. PS pos = paradoxical heat sensation occuring in the indicated
area (H = hand, F = foot).
channel and a polymodal channel, sensitive to heat, pinch,
cold), may explain paradoxical heat sensation (Craig and
Bushnell, 1994).
The aim of our investigation was to determine the incidence
of thermal sensory abnormalities in patients with CNS
demyelination. As indicated in the literature, we expected a
high incidence of threshold abnormalities. The most striking
observation, however, was the occurrence of paradoxical heat
sensations in these patients. This phenomenon was therefore
analysed in detail, and the implications for our understanding
of the central integration of temperature sensation are
discussed.
Patients and control subjects
the presence of somatosensory disturbances, i.e. symptoms,
signs or electrophysiological abnormalities. Clinical
examination showed no evidence for any other lesions of the
CNS or PNS. All patients underwent a detailed clinical
neurological examination, lumbar puncture and routine
electrophysiological examinations. All except two also had
MRI; the two patients without MRI had clinically definite
multiple sclerosis.
Thirty-four healthy volunteers (13 women, 21 men, aged
22-59 years, median 33 years) served as controls. No subject
had any sign of neurological disorder. The control values of
thermal sensory thresholds in these subjects are given in
Table 2. All patients and control subjects gave their
informed consent to the investigation procedure according to
the Declaration of Helsinki. The study was approved by the
Ethikkommission, Landesarztekammer Rheinlandpfalz.
This study was performed on 22 in-patients, 13 women and
nine men (aged 23-60 years, median 38 years, see Table 1),
who were admitted for evaluation of multiple sclerosis. Fiftysix patients with a clinically probable or definite multiple
sclerosis according to the criteria of Poser (Poser et al., 1983)
were seen during an 8 month period; 16 of them were
examined (29%). In addition, six patients with possible
multiple sclerosis, in whom the diagnosis could not be
established, were also tested. The criterion for inclusion was
Abnormalities in temperature sensibility were examined
according to the Marstock method using a thermal sensory
analyzer (TSA 2001; Medoc, Ramat, Yishai, Israel). The test
areas were in the dorsal skin of the hand and foot on both
sides. The TSA consists of a thermode including a peltier
element and two thermistors, a temperature control unit, a
Methods
Thermal sensory testing
Paradoxical heat sensation in multiple sclerosis
1731
Table 2 Control values of temperature-change thresholds from 34 healthy subjects
CDT
AT (°C)
Absolute T (°C)
WDT
AT(°C)
Absolute T (°C)
TSL
AT (-C)
Hand mean±SD
2.5 SD boundary
Foot mean±SD
2.5 SD boundary
-1.0±0.7
31.0
-2.7
29.3
-2.7±1.8
29.3
-7.2
24.8
1.4±0.6
33.4
2.9
34.9
4.4±2.2
36.4
9.9
41.9
2.5+1.1
5.3
7.5±4.0
17.5
Thresholds are expressed as deviations (AT) from the base temperature of 32°C; resulting absolute temperatures are also given.
CDT = cold detection threshold, WDT = warm detection threshold, TSL = thermal sensory limen. Mean values +2.5 SD (or -2.5 SD
in case of CDT) were taken as the boundaries for the normal range.
w*
44
w
w
w
w»
wt
w*
w
403632_
VVV V
2824.
30
1
60
90
1—
120
150
time (s)
180
210
240
Fig. 1 Time course of the examination in one patient with
paradoxical heat sensation after stimulation of the left foot. For
the cold- and warm-detection threshold, four stimuli each were
given. Pressing the switch returned the temperature to baseline
(32°C) and the quality of the sensation was described by the
patient (c = cold, w = warm, wT = warmer compared with the
previous stimuli). The TSL procedure (from 120 s on) consisted
of alternating heat and cold stimuli. Pressing the switch by the
patient altered the sign of the temperature change. Cooling
following the first heat stimulus initially led to a neutral
temperature perception (not shown). Upon further cooling, the
patient perceived the stimulus as heat instead of cold. This
paradoxical heat sensation was reproducible throughout the TSL
procedure.
patient feedback unit and a computer. The peltier element
cools and heats the contact plate while its temperature is
measured by two thermistors in the thermode. The patient
feedback unit (a switch) is connected with the computer and
the patient's signals terminate the heating or cooling process
(Yarnitsky et ai, 1995). The area of the thermode was 46X29
mm2 (13.3 cm2). The base temperature was set to 32°C,
which is in the neutral range, and the rate of temperature
change was 1 °C s~' for all test runs. The thermode temperature
was limited to 0-50°C to prevent damage to the skin. The
testing sequence started with determination of the cold
detection threshold (CDT) followed by warmth detection
threshold (WDT) and TSL.
The cold and warmth detection thresholds were measured
by the method of limits. The subject was asked to press the
button immediately when he or she noticed the temperature
change (see Fig. 1). After each stimulus, the temperature
automatically returned to baseline (with 2°C s"1 for cooling
and 4°C s ' for heating). The detection threshold was
calculated as the mean value of four stimuli. For the TSL
procedure, alternating warm and cold stimuli were given and
the patient pressed a button when recognizing either warmth
or cold. Pressing the button reversed the temperature change.
Eight stimuli were given (four each) and the mean of the
temperature differences was taken as the thermal sensory
limen (Fruhstorfer et al., 1976). After each stimulus, the
patient had to indicate the quality of the perception, i.e. cold,
warm, hot, burning or any other sensation.
The method of limits is a reaction-time dependent test.
This leads to an over-estimation of the peripheral threshold,
because the temperature continues to change during the time
between the receptor activation and the subject's reaction
(Yarnitsky and Ochoa, 1990a). The time is longest with
testing the warmth threshold at the foot, i.e. the thresholds
are artificially high due to the long conduction distance and
the slow conduction velocity of the C-fibres. The advantage
of the method of limits as compared with all other methods
is the short time needed for the testing (Claus et al., 1990)
which reduces the influence of inattention.
Healthy subjects with paradoxical heat
sensation
Four control subjects showed paradoxical heat sensations in
the TSL procedure. In a second session, they were tested for
reproducibility and rate dependence of paradoxical sensation.
One subject was excluded, because his ratings of 'warmth'
and 'cold' were erratic throughout the second session; he
knew before the session that his temperature perception was
thought to be abnormal. With the remaining three subjects,
clinical sensibility testing of both feet was carefully reassessed
to exclude any changes that had occurred since the previous
examination. Also the sensation evoked by a menthol in
ethanol solution which activates cold-receptors was
documented (Hensel and Zotterman, 1951). The cold and
warmth detection threshold and TSL procedure (base
temperature 32°C, temperature change 1°C s"1) were
repeatedly determined on both feet to test the reproducibility.
Then, the TSL procedure was performed at three different
1732
C. Hansen et al.
Table 3 Incidence of threshold abnormalities and paradoxical heat sensations in control subjects and in patients with
possible, probable or definite multiple sclerosis (MS)
Control subjects
Threshold abnormalities
CDT: subjects
affected areas
WDT: subjects
affected areas
TSL: subjects
affected areas
Paradoxical heat sensations
PSpoS: subjects
affected areas
In particular,
TSL test induced warmth
TSL test induced burning
CDT test induced warmth
Patients with
possible MS
probable or definite MS
2/34
2 feet, 2 hands
3/34
1 foot, 2 hands
3/34
2 feet, 1 hand
1/6
2 feet, 2 hands
2/6
3 hands
4/6*
6 hands
6/16
4 feet, 2 hands
9/16*
6 feet, 8 hands
6/16
6 feet, 6 hands
4/34
4 feet
1/6
1 foot
10/16**
12 feet, 2 hands
1 foot
10 feet, : hand
1 foot, 1 hand
2 feet
4 feet
The affected areas were hand and/or foot; several subjects had more than one affected site. Warmth and burning were the descriptions
given by the subject during a cold stimulus. The Yates-corrected %2 test was used to check differences between patients and control
subjects. *P < 0.05; **P < 0.001.
rates of temperature change (0.1, 0.5 and 2°C s '), with
eight stimuli per test series applied. The order of the three
rates of temperature change was counterbalanced across
subjects. All subjects were first re-examined on the site
where paradoxical heat sensation was found in the
first session.
Statistical analysis
Threshold differences between groups were tested using the
t test for independent samples. The y} test, including Yates
correction, was employed to test for different incidences.
Results
Paradoxical heat sensation
The quality of temperature sensation during all test procedures
was specified by the patients. In the example given in Fig.
1, the cold stimulus during determination of the CDT was
recognized as 'cold' and all stimuli of the WDT procedure
were perceived as 'warm'. A drift in temperature threshold
due to adaptation of the receptors is evident from the
increasing amplitude. During the TSL, alternating warm and
cold stimuli were applied. The first stimulus was warm and
was recognized as such. Subsequent cooling led to a neutral
sensation, but further cooling below 32°C elicited a warm
rather than a cold sensation. This paradoxical sensation of
warmth, heat or burning is termed 'paradoxical heat
sensation'. Paradoxical heat sensation disappeared when the
skin was heated after the next reversal of temperature change;
further temperature rise evoked a heat sensation again. All
heat sensations that appeared during the following stimuli
were separated by a period of neutral temperature perception,
probably coinciding with the crossing of the 32°C base
temperature (not shown in the figure).
During the TSL procedure, consisting of alternating warm
and cold stimuli, the warm stimulus was recognized as
'warm' or 'hot' by all of the patients. The cold stimulus was
correctly identified as 'cold' by 11 patients but as 'warm'
(paradoxical heat sensation) by 11 patients, i.e. by 50% of
the patients {see Table 3). The latter were 10 of the 16
patients with probable or definite multiple sclerosis (incidence
63%, P < 0.001 versus control) and one of six patients with
possible multiple sclerosis (incidence 17%, not significantly
different from controls). Paradoxical heat sensation was also
elicited in the lower extremity of four of the 34 control
subjects who were neurologically healthy (incidence 12%).
During CDT, only two of the 22 patients indicated a 'warm'
or 'burning' sensation with the cold stimuli. The following
warm stimuli (WDT) were correctly described as 'warm' or
'hot' by all 22 patients.
Paradoxical heat sensation occurred in the lower limbs
more frequently (13 feet in 11 patients) than in the upper
limbs (two hands in 11 patients) and nearly always during
the TSL procedure (see Table 3). Four of the 11 patients had
two affected areas which were either both feet or one hand
and the ipsilateral foot. In seven patients, only one of the
four tested areas was affected (in six cases the foot and in
one the hand, see Table 1). The patients with shorter disease
duration (< 1 year) had a lower incidence of paradoxical heat
sensation (three out of 10) than those with longer (2-17
years) disease duration (eight out of 12); these differences,
however, were not significant (x2 test).
Paradoxical heat sensation in multiple sclerosis
Threshold abnormalities
Ten of the patients with probable or definite multiple sclerosis
had abnormal thresholds (see Table 3), most often the WDT
(nine out of 16), which was significantly different from the
control subjects (P < 0.05). The incidence of abnormalities
in CDT and TSL (six out of 16 for each) was not significantly
elevated versus the control group. In contrast to paradoxical
heat sensation, threshold abnormalities were almost equally
frequent in hands and feet. Four out of six patients with
possible multiple sclerosis showed abnormal thresholds, all
with the TSL procedure and mainly when testing the hand
(P < 0.05 versus the control group). Differences between
the two patient groups were not significant.
The incidence of paradoxical heat sensation was
significantly raised only in the group of patients with a
probable or a definite multiple sclerosis. Since paradoxical
sensations predominantly occurred during testing the foot,
we analysed the lower limb thresholds in more detail. Cold
detection threshold, WDT and TSL results were compared
between three groups: there were 11 patients with paradoxical
heat sensation (PSpos, affecting a total of 13 feet, i.e. one or
two areas in each patient), 11 patients (22 feet) without
paradoxical heat sensation (PSneg) and 34 controls (68 feet).
The incidence of abnormal thresholds was significantly higher
in PSpos than in controls for the WDT (four out of 13 versus
one out of 68, P < 0.005, %2 t e s t ) a n d f o r t n e TSL (three
out of 13 versus two out of 68, P < 0.005) but not for the
CDT (one out of 13 patients of the PSpos group versus two
out of 68 controls). Thus, paradoxical heat sensation occurred
independent of cold threshold abnormalities. The low
incidence of threshold abnormalities in PSp<,s patients may
be due to the wide range of normal values in thermal
sensory testing.
On the average, however, P S ^ patients needed
significantly larger temperature changes for detection
thresholds than control subjects (Fig. 2). For CDT, the skin
was cooled by a mean of 5.4°C to a final temperature of
26.6°C, which is twice as much as in the control group
(P < 0.05). The WDT needed a mean temperature increase
of 8.5°C to a final temperature of 40.5°C which is also twice
the control value (P < 0.0001). The differences were most
striking in the TSL procedure (PSpos, 17.0°C; controls, 7.5°C;
P < 0.0001). Thus, quantitatively the WDT and TSL were
most affected in the PSpoS patients, and these parameters
were the only ones which were significantly greater than
those in PSneg patients. In the PSneg group, threshold changes
were small and differences from the control group were not
significant, except in the TSL. This suggests that paradoxical
heat sensations occur preferentially in those multiple sclerosis
patients who have a relatively high WDT and consequently
high TSL values. The average peak temperature during TSL
was 42.9±2.7°C in P S ^ patients, 39.0±2.5°C in PSneg
patients and 37.9± 2.6°C in controls. The differences between
all three groups were significant (P < 0.005; t test).
Since the TSL procedure consisted of four repetitions,
1733
paradoxical heat sensation could occur up to four times per
subject and area, documented as 0/4-4/4. In the 11 PSpos
patients, the frequency was 2.7/4± 1.3 (mean±SD) of the
cold stimuli given during TSL. No significant correlation
was found between the frequency of paradoxical heat
sensation and TSL (r = 0.30), WDT (j = 0.34), or CDT
(r = -0.15).
Rate dependence of paradoxical heat sensation
in control subjects
In areas showing paradoxical heat sensation in the first test,
the sensation also occurred at re-examination. In addition, one
other area was symptomatic during the second examination.
Paradoxical heat sensation was never reported at 0.1 °C s~'.
Increasing the rate of cooling from 0.5 to 2CC s"1 led to
an increase in the overall incidence of paradoxical heat
sensation (seven with 0.5°C s"1, 10 with 1°C s"1, 12 with
2°C s"1, see Table 4). The cold-sensibility, tested clinically
and with ethanol- or menthol-solution applied to the skin,
revealed no difference between the areas with or without
paradoxical heat sensation. All subjects felt appropriately
'cool' or 'cold'.
Discussion
Cooling the skin normally evokes a sensation of cold. We
observed that patients with multiple sclerosis often perceived
warmth instead of cold when their skin was cooled; this
phenomenon is called 'paradoxical heat sensation'.
Paradoxical heat sensation was found particularly during
alternating warm and cold stimuli, though single cold stimuli
could be recognized correctly. Paradoxical heat sensations
have not yet been described in patients with lesions of the
central nervous system.
Paradoxical heat sensations are known to occur in
peripheral nerve disorders such as uremic polyneuropathy
(Yosipovitch et al., 1995). They can also be induced in
healthy subjects, when the myelinated afferents are blocked
by pressure or ischaemia (Fruhstorfer, 1984; Wahren et al.,
1989; Yarnitsky and Ochoa, 1990/?). The loss of cold sensation
upon cooling of the skin during a preferential A-fibre block
is attributed to conduction block of myelinated cold fibres.
However, it is more difficult to explain how skin cooling can
paradoxically evoke the impression of heating. It has been
hypothesized that this paradoxical heat sensation is mediated
by unmyelinated nociceptive afferents that are spared by the
block (Fruhstorfer, 1984). Two types of projection neurons
have been identified in lamina I of the cat's spinal cord
(Craig and Kniffki, 1985) that may provide the central basis
for paradoxical heat sensations (Fig. 3): 'cold' cells receive
A8-fibre cold specific input and their activation should lead
to a sensation of cold. Cells which respond to heat, pinch
and cold (HPC cells) receive AS- and C-fibre polymodal
input, and their C-fibre input must include cold-sensitive
1734
C. Hansen et al.
n.s.
THERMAL SENSORY
LIMEN (°C)
WARM DETECTION
THRESHOLD (°C)
COLD DETECTION
THRESHOLD (°C)
n.s.
n.s.
20
10
- 10
10
-5
contr PS neg PSpo S
contr
contr
Fig. 2 Mean cold and warm detection thresholds expressed as differences from the baseline of 32°C
(AT) and thermal sensory limen of the feet of 34 control subjects (contr, left), 11 patients without
(PSneg, middle) and 13 affected feet of 11 patients with paradoxical heat sensation (PSpos, right).
*P < 0.05; **P < 0.01; ***P < 0.0001 in the t test; n.s. = not significant.
Table 4 Reproducibility of paradoxical heat sensation
during TSL tests in control subjects and its dependence on
the rate of temperature change
Foot
Right
Left
40°C ^
Examination I Examination II
PCs" 1
20'
l°Cs-'
2°C s"1
0.5°C s"1
3/4
3/4
4/4
4/4
2/4
+
COLD
HPC
O.rCs" 1
spino-thalamic
Right
Left
2/4
4/4
4/4
1/4
Right
Left
2/4
1/4
2/4
4/4
Three control subjects with paradoxical heat sensation (PS^) were
tested in a second session. The test area was the foot, the TSL
procedure was performed with one cooling/heating velocity
(1°C s~') in the first examination; in the second four rates were
used as indicated; ratios indicate the incidence of PSpoS during the
TSL test; each dash (-) indicates absence of PSpos during the
sequence of eight stimuli.
nociceptors which show thresholds around normal room
temperature (~20°C). The most sensitive C-nociceptors in
rats and monkeys are, indeed, activated by very mild cold
stimuli (LaMotte and Thalhammer, 1982; Kajander et al.,
1994; Simone et al., 1994). In this model, cooling the skin
during preferential A-fibre blockade activates only the HPC
channel, leading to a sensation of heat or pain. Under normal
conditions, cold stimuli activate both the cold and the HPC
channel. A central inhibition of the HPC channel by the cold
channel has been postulated (Craig and Bushnell, 1994), as
otherwise cold stimuli would always be perceived as hot.
This inhibitory interaction must occur at the thalamo-cortical
thalamo-cortical
Fig. 3 Model of central interaction of cold and warm stimuli. A
stimulus of 20°C activates spinal cold-specific cells via A5-cold
fibres and the polymodal HPC-cells (responsive to heat, pinch,
cold) via polymodal C-fibres. A 40°C stimulus inhibits cold-cells
(either by peripheral suppression of cold-receptors or by central
inhibition of cold-cells via warm fibres) and is just below
threshold for activation of the HPC-cells in the cat. In humans
and monkeys, however, this temperature is enough to cause
lowgrade input via polymodal C-fibres (dashed line). The cold
and the HPC channels terminate in different regions in the
thalamus; in the thalamo-cortical loop an inhibitory connection
from cold to HPC exists. Drawn according to Craig and Bushnell
(1994).
level because both types of neurons in lamina I are output
neurons that project to the thalamus. The thalamic termination
areas have been suggested as the posterior part of the ventral
medial nucleus (VMpo) for cold (projecting further to the
insula) and the ventral caudal part of the medial dorsal
nucleus (MDvc) for HPC cells (projecting to the anterior
cingulate, Craig, 1994).
This model suggests a supraspinal decoding of pain and
temperature sensation. Thus, the model predicts that this
integration can be interfered with by CNS lesions. Our
finding of paradoxical heat sensation in 63% of the patients
with probable or definite multiple sclerosis verified this
prediction. If temperature-related sensations were decoded at
Paradoxical heat sensation in multiple sclerosis
the spinal level, supraspinal lesions would only attenuate
temperature sensation, and could not cause paradoxical
sensations. Because there was no evidence of peripheral
nerve lesions in our patients with multiple sclerosis, the
increased incidence of paradoxical heat sensation must have
been related to impairment of the pain and temperature
channels within the CNS. Although paradoxical heat sensation
was not restricted to those few patients with clinical or MRI
evidence of myelitis, these methods did not definitely exclude
spinal demyelinating lesions in the other patients. Therefore,
a common lesion site could not be identified in this
heterogenous group of patients.
If the lesions in our multiple sclerosis patients simply
blocked the cold channel at a central level, the cold
detection threshold (CDT) change should be similar to
peripheral cold fibre block. Ischaemic or pressure block of
peripheral A-fibres lowers CDT by 3.5°C and increases the
WDT by 1.5°C (Fruhstorfer, 1984; Wahren et al., 1989). In
our patients the average CDT was 2.7°C lower and the
average WDT 4.1°C higher than in healthy controls. This
substantial increase in WDT is in contrast to the peripheral
nerve block studies. It could be explained by a central block
of the cold channel, if the sensation of heat was mediated
by a decreased activity of the cold channel (see footnote 14
in Craig and Bushnell, 1994). Such a convergence of warm
and cold fibres on spinal cold cells would be in line with the
very small number of spinal neurons that are activated by
non-noxious heating.
According to the model, peripheral or central blockade of
the cold channel would cause all cold stimuli to be perceived
as heat, as observed with A-fibre blockade. However, we
found that cold stimuli of the CDT series were correctly
perceived. Also specific chemical activation of cold fibres
by menthol (Hensel and Zotterman, 1951) was correctly
perceived as cold by all subjects tested. In most cases, only
the alternating cold and warm stimuli during TSL induced
the paradoxical heat sensation. An interaction of cold and
warm stimuli thus seems to be crucial. For this interaction,
two possibilities exist: either both stimuli are applied at the
same time in two areas that are very close together, or both
are applied in the same area of the skin one following the
other. Both combinations have been described in the literature.
Simultaneous application of alternating strips of cold
(20°C) and warm (40°C) silver bars ('Thunberg grill') to the
palm of the hand of healthy subjects increased the impression
of heat, whereas the sensation of cold diminished; pain was
felt by nearly all subjects (Craig and Bushnell, 1994). Isolated
cold and warm stimuli applied to the same area were
recognized as such without any painful component. A similar
interaction between thermal stimuli to adjacent finger tips
was called 'synthetic heat' (Green, 1977). The sensation of
heat in the Thunberg grill experiment was explained as a
central disinhibition phenomenon (Craig and Bushnell, 1994).
According to the model mentioned above, simultaneous warm
and cold stimuli increase the spinal HPC-cell activity by
summation effects of cold and heat activation, and decrease
1735
cold-cell activity due to inhibition by warmth. The lower
cold-cell activity results in a higher activity in the supraspinal
HPC-channel because of the lack of inhibition (Fig. 3).
Alternating warm and cold stimuli of the TSL procedure
may create a temporal pattern of primary afferent input
equivalent to the Thunberg grill spatial pattern. Notably, the
Thunberg grill illusion works best when heating precedes
cooling by 5 s (footnote 7 in Craig and Bushnell, 1994). If
an interaction of the cold-induced and heat-induced afferent
input was the clue to paradoxical heat sensations, they should
be facilitated with faster rates of temperature changes. This
prediction was verified in the healthy subjects with
paradoxical heat sensation. With faster rates of temperature
change paradoxical heat sensation occurred more often than
with the slow changes. In healthy subjects, the temporal
pattern appears to be less efficient than the spatial pattern,
since paradoxical heat sensations were observed in only four
out of 34 subjects, whereas the Thunberg grill illusion is
supposedly present in every subject.
Paradoxical heat sensations occur more frequently if the
skin has previously been heated (Hamalainen et al., 1982).
Without preheating, cooling caused paradoxical sensations in
~10% of healthy subjects; painful preheating to ~50°C
increased the incidence to 30-40%. These authors
hypothesized that polymodal nociceptors are sensitized to
cooling stimuli by a preceding heat stimulus. Peripheral
sensitization of polymodal nociceptors can easily be induced
by a strong heat stimulus. In our data, there was no painful
heat stimulus before the TSL procedure, and the peak
temperatures reached during TSL testing in the patients with
paradoxical heat sensation (42.9±2.7°C) are not sufficient to
sensitize nociceptors. These temperatures, however, may
elicit a few action potentials in C-fibre nociceptors (Tillman
et al., 1995). Thus, weak nociceptor activation may have
preceded the cold stimuli during the TSL procedure in our
patients. By causing a vasodilatation (flare) that spreads
several centimetres, such a nociceptor activation can create
a spatial temperature pattern: a cold stimulus applied after a
heat stimulus provides cold receptor input from a small area
within the skin area that is warmed due to the vasodilatation.
In summary, patients with central demyelination showed
a high incidence of paradoxical heat sensation when cooling
the skin (63% in the selected sample studied). Extrapolating
our results to the whole population with probable or definite
multiple sclerosis (Poser criteria), one of five patients (18%)
is expected to show this phenomenon which would make this
a frequent finding in multiple sclerosis. However, paradoxical
heat sensation currently has no diagnostic significance per se,
since it also occurs in healthy subjects. From the physiological
point of view, the observation that CNS lesions facilitate
paradoxical heat sensation, supports the model that temperature sensation is integrated at the thalamo-cortical rather
than at the spinal level.
Acknowledgements
We wish to thank Dr W. Magerl for help with the statistics
and Dr A. D. Craig for critical reading of the manuscript.
1736
C. Hansen et al.
References
Boivie J. Sensory abnormalities in patients with central nervous
system lesions as shown by quantitative sensory tests. In: Boivie J,
Hansson P, Lindblom U, editors. Touch, temperature, and pain in
health and disease: mechanisms and assessments. Progress in pain
research and management, Vol. 3. Seattle: IASP Press, 1994: 179-91.
Claus D, Hilz MJ, Neundbrfer B. Thermal discrimination thresholds:
a comparison of different methods. Acta Neural Scand 1990; 81:
Craig AD. Spinal and supraspinal processing of specific pain and
temperature. In: Boivie J, Hansson P, Lindblom U, editors. Touch,
temperature, and pain in health and disease: mechanisms and
assessments. Progress in pain research and management, Vol. 3.
Seattle: IASP Press, 1994:421-37.
Kajander KC, Allen BJ, Allard BL, Courteix C, Simone DA.
Responses of rat C-fiber nociceptors to noxious cold [abstract]. Soc
Neurosci Abstr 1994; 20: 1571
LaMotte RH, Thalhammer JG. Response properties of high-threshold
cutaneous cold receptors in the primate. Brain Res 1982; 244:
279-87.
Long RR. Sensitivity of cutaneous cold fibers to noxious heat:
paradoxical cold discharge. J Neurophysiol 1977; 40: 489-502.
Osterberg A, Boivie J, Holmgren H, Thuomas K-A, Johansson I.
The clinical characteristics and sensory abnormalities of patients
with central pain caused by multiple sclerosis. In: Gebhart GF,
Hammond DL, Jensen TS, editors. Proceedings of the 7th World
Congress on Pain. Seattle: IASP Press, 1994: 789-96.
Craig AD, Bushnell MC. The thermal grill illusion: unmasking the
burn of cold pain. Science 1994; 265: 252-5.
Poser CM, Paty DW, Scheinberg L, McDonald WI, Davis FA, Ebers
GC, et al., New diagnostic criteria for multiple sclerosis: guidelines
for research protocols. Ann Neurol 1983; 13: 227-31.
Craig AD, Kniffki K-D. Spinothalamic lumbosacral lamina I cells
responsive to skin and muscle stimulation in the cat. J Physiol
(Lond) 1985; 365: 197-221.
Simone DA, Gilchrist HD, Kajander KC. Responses of primate
cutaneous nociceptors evoked by noxious cold [abstract]. Soc
Neurosci Abstr 1994; 20: 1379
Dyck PJ, Zimmerman I, Gillen DA, Johnson D, Karnes JL, O'Brien
PC. Cool, warm, and heat-pain detection thresholds - testing methods
and inferences about anatomic distribution of receptors. Neurology
1993; 43: 1500-8.
Tillman D-B, Treede R-D, Meyer RA, Campbell JN. Response of
C fibre nociceptors in the anaesthetized monkey to heat stimuli:
estimates of receptor depth and threshold. J Physiol (Lond) 1995;
485: 753-65.
Fruhstorfer H. Thermal sensibility changes during ischemic nerve
block. Pain 1984; 20: 355-361.
Tritsch MF. Temperature sensation: the '3-bowls experiment'
revisited. Naturwissenschaften 1990; 77: 288-9.
Fruhstorfer H, Lindblom U, Schmidt WG. Method for quantitative
estimation of thermal thresholds in patients. J Neurol Neurosurg
Psychiatry 1976; 39: 1071-5.
Wahren LK, Torebjbrk E, Jorum E. Central suppression of coldinduced C fibre pain by myelinated fibre input. Pain 1989; 38: 313-9.
Green BG. Localization of thermal sensation: an illusion and
synthetic heat. Perception and Psychophysics 1977; 22: 331-7.
Greenspan JD, Taylor DJ, McGillis SLB. Body site variation of
cool perception thresholds, with observations on paradoxical heat.
Somatosens Mot Res 1993; 10: 467-74.
Hamalainen H, Vartiainen M, Karvanen L, Jarvilehto T. Paradoxical
heat sensations during moderate cooling of the skin. Brain Res
1982; 251: 77-81.
Yarnitsky D, Ochoa JL. Studies of heat pain sensation in man:
perception thresholds, rate of stimulus rise and reaction time. Pain
1990a; 40: 85-91.
Yarnitsky D, Ochoa JL. Release of cold-induced burning pain by
block of cold-specific afferent input. Brain 1990b; 113: 893-902.
Yarnitsky D, Ochoa JL. Warm and cold specific somatosensory
systems. Brain 1991; 114: 1819-26.
Yarnitsky D, Sprecher E, Zaslansky R, Hemli JA. Heat pain
thresholds: normative data and repeatability. Pain 1995; 60: 329-32.
Heijenbrok MW, Anema JR, Faes TJC, Bertelsmann FW, Heimans
JJ, Polman CH. Quantitative measurement of vibratory sense and
temperature sense in patients with multiple sclerosis. Electromyogr
Clin Neurophysiol 1992; 32: 385-8.
Yosipovitch G, Yarnitsky D, Mermelstein V, Sprecher E, Reiss J,
Witenberg C, et al. Paradoxical heat sensation in uremic
polyneuropathy. Muscle Nerve 1995; 18: 768-71.
Hensel H, Zotterman Y. The effect of menthol on the
thermoreceptors. Acta Physiol Scand 1951; 24: 27-34.
Received December 15, 1995. Revised April 29, 1996.
Accepted June 14, 1996