Professional Documents
Culture Documents
www.EuropeanJournalPain.com
Abstract
We have compiled a comprehensive QST protocol as part of the German Research Network on Neuropathic Pain (DFNS) using well
established tests for nearly all aspects of somatosensation. This protocol encompasses thermal as well as mechanical testing procedures.
Our rationale was to test for patterns of sensory loss (small and large nerve fiber functions) or gain (hyperalgesia, allodynia, hyper-
pathia), and to assess both cutaneous and deep pain sensitivity. The practicality of the QST protocol was tested in 18 healthy subjects,
21–58 years, half of them female. All subjects were tested bilaterally over face, hand and foot. We determined thermal detection and pain
thresholds including a test for the presence of paradoxical heat sensations, mechanical detection thresholds to von Frey filaments and a
64-Hz tuning fork, mechanical pain thresholds to pinprick stimuli and blunt pressure, stimulus–response-functions for pinprick and
dynamic mechanical allodynia (pain to light touch), and pain summation (wind-up ratio) using repetitive pinprick stimulation.
The full protocol took 27 ± 2.3 min per test area. The majority of QST parameters were normally distributed only after loga-
rithmic transformation (secondary normalization) except for the frequency of paradoxical heat sensations, cold and heat pain
thresholds, and for vibration detection thresholds. Thresholds were usually lowest over face, followed by hand, and then foot. Only
thermal pain thresholds, wind-up ratio and vibration detection thresholds were not significantly dependent on the body region.
There was no significant right-to-left difference for any of the QST parameters; left-to-right correlation coefficients ranged between
0.78 and 0.97, thus explaining between 61% and 94% of the variance. This study has shown that a complete somatosensory profile of
one affected area and one unaffected control area, which will be necessary to characterize patients with a variety of diseases, can be
obtained within 1 h. Case examples of selected patients illustrate the value of z-transformed QST data for an easy survey of indi-
vidual symptom profiles.
! 2005 European Federation of Chapters of the International Association for the Study of Pain. Published by Elsevier Ltd. All
rights reserved.
Keywords: QST; Sensory profile; Allodynia; Hyperalgesia; Hypoesthesia; Hypoalgesia; Paradoxical heat sensation; Data transformation; Z-scores
Abbreviations: ALL, Dynamic mechanical allodynia; CDT, Cold detection threshold; CPT, Cold pain threshold; DFNS, Deutscher Forschun-
gsverbund Neuropathischer Schmerz; HPT, Heat pain threshold; MDT, Mechanical detection threshold; MPS, Mechanical pain sensitivity; MPT,
Mechanical pain threshold; PHS, Paradoxical heat sensation; PPT, Pressure pain threshold; TSL, Thermal sensory limen; VDT, Vibration detection
threshold; WDT, Warm detection threshold; WUR, Wind-up ratio.
*
Corresponding author. Tel.: +49 6131 3925715; fax: +49 6131 3925902.
E-mail address: treede@uni-mainz.de (R.-D. Treede).
1090-3801/$32 ! 2005 European Federation of Chapters of the International Association for the Study of Pain. Published by Elsevier Ltd. All rights
reserved.
doi:10.1016/j.ejpain.2005.02.003
78 R. Rolke et al. / European Journal of Pain 10 (2006) 77–88
Fig. 1. QST-a battery of sensory tests: figure of methods. The standardized QST protocol assesses 13 parameters in seven test procedures (A–G). All
procedures are presented including a time frame for testing over one area. A. Thermal testing comprises detection and pain thresholds for cold, warm,
or hot stimuli (C- and A-delta fiber mediated): cold detection threshold (CDT); warm detection threshold (WDT); number of paradoxical heat
sensations (PHS) during the thermal sensory limen procedure (TSL) for alternating warm and cold stimuli; cold pain threshold (CPT); heat pain
threshold (HPT). B. mechanical detection threshold (MDT) tests for A-beta fiber function using von Frey-filaments. C. Mechanical pain threshold
(MPT) tests for A-delta fiber mediated hyper- or hypoalgesia to pinprick stimuli. D. Stimulus–response-functions: mechanical pain sensitivity (MPS)
for pinprick stimuli, and dynamic mechanical allodynia (ALL) assess A-delta mediated sensitivity to sharp stimuli (pinprick), and also A-beta fiber
mediated pain sensitivity to stroking light touch (CW = cotton wisp; QT = cotton wool tip; BR = brush). E. Wind-up ratio (WUR) compares the
numerical ratings within five trains of a single pinprick stimulus (a) with a series (b) of 10 repetitive pinprick stimuli to calculate WUR as the ratio:
b/a. F. Vibration detection threshold (VDT) tests for A-beta fiber function using a Rydel–Seiffer 64 Hz tuning fork. G. Pressure pain threshold (PPT)
is the only test for deep pain sensitivity, most probably mediated by muscle C- and A-delta fibers. For details regarding the testing procedures also see
Section 2.
Since it was our aim to use simple hand-held de- 2.2. Thermal detection, thermal pain thresholds and
vices whenever available, we did not use an automated paradoxical heat sensations
vibrameter or pressure algometer, which are useful in
specialized sensory laboratory tests. For thermal The tests for thermal sensation were performed using
testing, no simple devices are available yet (Cruccu a TSA 2001-II (MEDOC, Israel) thermal sensory testing
et al., 2004). device (Fruhstorfer et al., 1976; Yarnitsky et al., 1995).
80 R. Rolke et al. / European Journal of Pain 10 (2006) 77–88
Cold detection threshold (CDT) and warm detection Stimulus–response-functions for dynamic mechani-
threshold (WDT) were measured first. The number of cal allodynia (ALL) were determined using a set of
paradoxical heat sensations (PHS) was determined dur- three light tactile stimulators (Baumgärtner et al.,
ing the thermal sensory limen procedure (TSL, the dif- 2002; LaMotte et al., 1991): a cotton wisp exerting a
ference limen for alternating cold and warm stimuli), force of "3 mN, a cotton wool tip fixed to an elastic
followed by cold pain threshold (CPT), and heat pain strip exerting a force of "100 mN, and a standardized
threshold (HPT). The mean threshold temperature of brush (Somedic, Sweden) exerting a force of "200–400
three consecutive measurements was calculated. All mN. The three tactile stimuli were applied five times
thresholds were obtained with ramped stimuli (1 "C/s) each with a single stroke of approximately 1–2 cm in
that were terminated when the subject pressed a button. length over the skin. They were intermingled with the
Cut-off temperatures were 0 and 50 "C. The baseline pinprick stimuli in balanced order and subjects were
temperature was 32 "C (center of neutral range) and asked to give a rating on the same scale as for pinprick
the contact area of the thermode was 7.84 cm2. During stimuli.
the experiment, the subjects were not able to watch the
computer screen. All thermal tests were first demon- 2.6. Wind-up ratio – the perceptual correlate of temporal
strated over an area that was not tested later during pain summation for repetitive pinprick stimuli
the QST session.
In this test of temporal summation, the perceived
2.3. Mechanical detection threshold for modified magnitude of a single pinprick stimulus was compared
von Frey filaments with that of a train of 10 pinprick stimuli of the same
force repeated at a 1/s rate (128 mN, when tested over
Modified von Frey filaments (MDT) was measured face, and 256 mN, when tested over hand and foot).
with a standardized set of modified von Frey hairs The train of pinprick stimuli was given within a small
(Optihair2-Set, Marstock Nervtest, Germany) that exert area of 1 cm2 and the subject was asked to give a pain
forces between 0.25 and 512 mN (Fruhstorfer et al., rating representing the pain at the end of the train using
2001; Von Frey, 1896; Weinstein, 1968). The contact a numerical rating scale. In contrast to the more sophis-
area of the von Frey hairs with the skin was of uniform ticated technique of VAS-ratings at a 1/s rate (Magerl
size and shape (rounded tip, 0.5 mm in diameter) to et al., 1998) this method (modified from Sieweke et al.,
avoid sharp edges that would facilitate nociceptor acti- 1999) is likely more appropriate for clinical routine
vation. The final threshold was the geometric mean of assessment. Single pinprick stimuli were alternated with
five series of ascending and descending stimulus intensi- a train of 10 stimuli until both were done five times at
ties (Baumgärtner et al., 2002). five different skin sites within the same body region.
The mean pain rating of trains divided by the mean pain
2.4. Mechanical pain threshold for pinprick stimuli rating to single stimuli was calculated as wind-up ratio
(WUR).
Mechanical pain threshold (MPT) was measured
using a set of seven custom-made weighted pinprick 2.7. Vibration detection threshold
stimulators (flat contact area of 0.2 mm diameter) that
exert forces between 8 and 512 mN (Baumgärtner Vibration detection threshold (VDT) test was per-
et al., 2002; Chan et al., 1992; Magerl et al., 1998). formed with a Rydel–Seiffer tuning fork (64 Hz, 8/8
Again using the ‘‘method of limits’’, the final threshold scale) that was placed over a bony prominence (cheek,
was the geometric mean of five series of ascending and processus styloideus ulnae, malleolus medialis). Vibra-
descending stimulus intensities. tion threshold was determined with three series of
descending stimulus intensities (Fagius and Wahren,
2.5. Stimulus–response-functions: mechanical pain 1981; Goldberg and Lindblom, 1979; Rydel and Seiffer,
sensitivity for pinprick stimuli and dynamic mechanical 1903).
allodynia for stroking light touch
2.8. Pressure pain threshold
Mechanical pain sensitivity (MPS) was tested using
the same weighted pinprick stimuli as for MPT. To ob- The final test in the protocol was performed with a
tain a stimulus–response-function, these seven pinprick pressure gauge device (FDN200, Wagner Instruments,
stimuli were applied in a balanced order, five times each, USA) with a probe area of 1 cm2 (probe diameter of
and the subject was asked to give a pain rating for each 1.1 cm) that exerts pressure up to 20 kg/cm2/"200
stimulus on a 0–100 numerical rating scale ("0! indicating N/cm2/"2000 kPa (Fischer, 1987; Kilo et al., 1994; Ko-
‘‘no pain’’, and "100! indicating ‘‘most intense pain sek et al., 1999; Rolke et al., 2005). The pressure pain
imaginable’’). threshold (PPT) is determined with three series of
R. Rolke et al. / European Journal of Pain 10 (2006) 77–88 81
ascending stimulus intensities, each applied as a slowly are familiar with the concept of threshold elevations,
increasing ramp of 50 kPa/s. we decided to use this procedure, because the general
readership may find it conceptually easier to think in
2.9. Data evaluation terms of gain or loss of sensory function.
After this Z-transformation it is straightforward to
All data were analyzed for their distribution proper- compare a single patient with the group mean of healthy
ties. We calculated skewness, kurtosis, Kolmogorov– controls, since the 95% confidence interval (CI) of a
Smirnov!s d for raw data and log-transformed data. standard normal distribution is defined as follows:
The product of the geometric mean of skewness and 95% CI ¼ Meancontrols ' 1:96SDcontrols :
kurtosis combined and the geometric mean of Kolmogo-
rov–Smirnov!s d (for continuous test of normality of dis-
tribution) was calculated as a compound measure of
goodness of normality. Log-transformation was consid- 3. Results
ered to be superior, when the ratio for raw data to log-
transformed data exceeded a factor of 3. It was possible to obtain all QST data in all 18
For pain ratings to pinprick and light touch a small healthy subjects at all sites tested. The testing procedures
constant (+0.1) was added prior to log-transformation were easily feasible with a mean duration of 27.0 ± 2.3
to avoid a loss of zero rating values (Bartlett, 1947; min for the full QST protocol tested over one test area.
Magerl et al., 1998). All statistical calculations were per- Thus, assessing six sites in healthy subjects took about
formed by using the Statistica software package, release 3 h. In patients, assessment of two sites (one affected,
6.0 for Windows (StatSoft Inc., USA). Differences be- one normal area) is expected to be finished within 1 h.
tween areas (face, hand, foot), right and left sides of
the body were compared using a two-way analysis of 3.1. QST report form
variance for repeated measures (ANOVA). Post hoc
comparisons were calculated using LSD-post hoc tests QST data were entered into an EXCEL-spreadsheet
(LSD = least significant difference). Log-data of thresh- (Microsoft, USA) automatically generating thresholds
olds were retransformed to linear values representing the and average ratings, and numbers of observed symp-
original unit of each test. toms (in the case of PHS). These data are summarized
in a single sheet QST report form which eases compari-
2.10. Z-transformation of QST data to create profiles son of the test and control areas (Fig. 2).
of sensory changes
3.2. The majority of QST parameters are lognormally
To compare a patient!s QST data profile with distributed
control data independent of the different units of mea-
surement across QST parameters, the patient data were Some QST parameters were not normally distributed,
Z-transformed for each single parameter by using the but normal distribution was achieved by logarithmic
following expression (Glass and Stanley, 1970; Gauss, transformation (secondary normal distribution). A typi-
1863): cal example is shown in Fig. 3 (warm detection thresh-
olds in the hand dorsum). Table 1 comprises skewness,
Z-score ¼ ðXsingle patient % Meancontrols Þ=SDcontrols :
kurtosis and Kolmogorov–Smirnov!s d as markers to test
This procedure results in a QST profile where all param- for normality of distribution in raw and log-transformed
eters are presented as standard normal distributions data. Based on a weighted comparison of distribution
(zero mean, unit variance). For clarity of data presenta- parameters we recommend to execute log-transforma-
tion we adjusted the algebraic sign of Z-score values for tion in the following QST parameters: CDT, WDT,
each parameter so that it reflects the patient!s sensitivity TSL, MDT, MPT, MPS, ALL, WUR, and PPT.
for this parameter. Z-values above ‘‘0’’ indicate a gain of
function when the patient is more sensitive to the tested 3.3. QST procedures show highly significant differences
stimuli compared with controls, while Z-scores below across test areas
‘‘0’’ indicate a loss of function referring to a lower sen-
sitivity of the patient. Thus, elevations of threshold There were highly significant differences between test
(CDT, WDT, TSL, HPT, CPT, MDT, MPT, VDT, areas for most QST parameters (Table 2), except for
PPT) resulted in negative Z-scores, whereas increases cold pain threshold (CPT), heat pain threshold (HPT),
in ratings (MPS, ALL, WUR) resulted in positive wind-up ratio (WUR), and vibration detection threshold
Z-scores. Paradoxical heat sensations (PHS) were inter- (VDT). Differences between test areas could not be as-
preted as a loss of thermodiscriminative function result- sessed for paradoxical heat sensations (PHS) and dy-
ing in negative Z-scores. Even though QST specialists namic mechanical allodynia (ALL) because there was
82 R. Rolke et al. / European Journal of Pain 10 (2006) 77–88
Fig. 2. QST report form. The QST report form allows easy comparisons of parameters over control vs. test site supplemented by the corresponding
Z-score compared to healthy age and gender matched subjects. These comparisons might be between different test areas, e.g., hand vs. foot in the case
of a patient with symmetrical neuropathy. Most important will be direct comparisons of right and left side of the body, e.g., in a patient with an
unilateral chronic pain syndrome over distal limb.
no significant occurrance of these parameters in healthy any significant interactions with other factors (Table 2),
subjects. Mean value differences across areas are illus- we compared data for left and right body sides to assess
trated in Table 1. Thresholds were usually lowest over intra-individual variability of QST testing. The correla-
face, followed by hand and foot with the exceptions of tion coefficients were highly significant (r = 0.78–0.97,
vibration detection threshold presenting the highest sen- all p < 0.001), and their squared values indicate that
sitivity in the hand (Table 1), and stimulus–response- systematic interindividual differences accounted for
function for pinprick presenting highest sensitivity over between 61% and 94% of the total variance of the
face, followed by foot, then hand. No differences were QST parameters. These findings suggest lower variabil-
found between body sides (Table 2), and there were no ity of QST parameters within subjects than across
interactions between test area and body side. These find- subjects.
ings confirm that each area of the body needs its own set
of QST reference data. Due to the narrow age range and 3.5. QST profiles of Z-transformed data in selected
small number of healthy subjects in the present study we patients
did not address age or gender differences.
To illustrate the potential use of QST profiles, Fig. 4
3.4. Intra- and interindividual variability of QST data presents QST profiles of three selected patients. Some of
the Z-values were beyond the 95% confidence interval
Since there were no significant differences in thresh- (grey zone) of healthy subjects showing three different
olds between the right and left sides of the body, nor patterns of sensory changes.
R. Rolke et al. / European Journal of Pain 10 (2006) 77–88 83
4. Discussion
Table 1
Distribution parameters: the majority of QST parameters are normally distributed only after logarithmic transformation (secondary normalization)
Parameter Mean ± SD Mean ± SD Skewness Kurtosis K–S 0 d Weighted Recommended data
(raw data) (log data) (raw/log) (raw/log) (raw/log) ratio transformation
(raw/log)
CDT (DT, "C) Face %0.67 ± 0.33 %0.213 ± 0.173 %2.28/1.01 6.51/1.04 0.25/0.19
Hand %0.91 ± 0.44 %0.081 ± 0.179 %1.67/0.81 2.30/0.28 0.20/0.12 6.92 Log
Foot %1.71 ± 1.47 0.187 ± 0.280 %0.13/%0.08 3.45/0.12 0.17/0.12
WDT (DT, "C) Face 1.05 ± 0.49 %0.019 ± 0.186 1.21/0.47 0.77/%0.55 0.18/0.13
Hand 1.87 ± 0.82 0.237 ± 0.173 1.63/0.23 3.66/0.75 0.14/0.08 5.90 Log
Foot 4.57 ± 2.30 0.615 ± 0.195 1.50/0.46 2.38/%0.34 0.17/0.11
TSL (DT, "C) Face 1.37 ± 0.84 0.056 ± 0.293 1.24/%1.11 0.88/3.90 0.24/0.16
Hand 2.81 ± 1.36 0.403 ± 0.200 1.55/%0.06 3.30/0.61 0.22/0.13 4.04 Log
Foot 6.80 ± 2.71 0.802 ± 0.165 1.01/0.21 0.83/%0.61 0.13/0.07
PHSa (x/3) Face 0±0 %1.000 ± 0
Hand 0±0 %1.000 ± 0 None
Foot 0.11 ± 0.40 %0.905 ± 0.321
CPT ("C) Face 10.36 ± 10.35 0.410 ± 1.021 0.53/%0.54 %1.21/%1.56 0.18/0.24
Hand 7.73 ± 7.82 0.436 ± 0.865 1.00/%0.82 0.11/%0.87 0.16/0.21 0.77 None
Foot 5.96 ± 7.74 0.202 ± 0.910 1.62/%0.35 1.77/%1.54 0.22/0.21
HPT ("C) Face 44.96 ± 3.31 1.652 ± 0.033 %0.77/%0.94 0.10/0.39 0.14/0.15
Hand 45.39 ± 3.60 1.656 ± 0.036 %0.63/%0.81 %0.09/0.34 0.10/0.11 1.21 None
Foot 45.80 ± 2.61 1.660 ± 0.025 %0.68/%0.79 %0.24/%0.002 0.14/0.15
MDT (mN) Face 0.21 ± 0.05 %0.682 ± 0.093 1.33/1.04 0.82/0.03 0.27/0.28
Hand 1.93 ± 2.08 0.124 ± 0.366 3.01/0.36 11.46/0.07 0.25/0.10 3.78 Log
Foot 3.52 ± 3.46 0.367 ± 0.413 1.93/%0.11 3.55/%0.29 0.18/0.09
MPT (mN) Face 55.7 ± 58.6 1.537 ± 0.456 2.17/%0.23 5.22/%0.49 0.22/0.10
Hand 129.3 ± 95.5 1.971 ± 0.394 0.99/%0.71 0.49/0.13 0.15/0.10 7.44 Log
Foot 88.2 ± 74.4 1.764 ± 0.452 1.28/%0.66 1.48/%0.13 0.16/0.12
MPS (rating) Face 1.79 ± 2.18 %0.039 ± 0.519 1.74/0.24 2.17/%0.96 0.23/0.11
Hand 0.65 ± 0.79 %0.409 ± 0.425 2.59/0.51 7.95/%0.45 0.28/0.11 7.76 Log
Foot 0.94 ± 1.11 %0.292 ± 0.471 1.49/0.61 0.94/%1.05 0.27/0.17
ALLa (rating) Face 0±0 %1.000 ± 0
Hand 0.001 ± 0.006 %0.995 ± 0.023 Logb
Foot 0.001 ± 0.003 %0.998 ± 0.013
WUR (ratio) Face 3.11 ± 2.10 0.419 ± 0.247 2.06/0.54 5.37/%0.31 0.18/0.12
Hand 2.67 ± 1.94 0.338 ± 0.268 1.45/0.74 1.01/%0.59 0.29/0.18 3.31 Log
Foot 3.20 ± 2.14 0.420 ± 0.271 1.05/0.44 %0.21/%1.14 0.20/0.13
VDT (x/8) Face 7.20 ± 0.75 %0.266 ± 0.500 %0.63/%0.43 %0.77/%1.30 0.21/0.20
Hand 7.66 ± 0.43 %0.564 ± 0.445 %1.30/0.24 1.35/%1.62 0.26/0.30 1.96 None
Foot 7.25 ± 0.86 %0.319 ± 0.517 %1.79/%0.23 4.09/%1.34 0.19/0.21
PPT (kPa) Face 212 ± 55.7 2.313 ± 0.113 0.62/0.01 %0.11/%0.20 0.11/0.08
Hand 512 ± 191.6 2.683 ± 0.152 1.13/0.37 1.24/%0.50 0.15/0.09 4.69 Log
Foot 572 ± 199.8 2.732 ± 0.154 0.46/%0.03 %0.71/%1.13 0.15/0.15
DT, difference from baseline temperature 32 "C; K–S 0 d, Kolmogorov–Smirnov!s d.
a
Paradoxical heat sensation (PHS) and allodynia (ALL) did not significantly occur in healthy subjects.
b
Recommendation on data transformation for allodynia (pain to light touch) was derived from patient studies and from studies of experimentally
induced hyperalgesia (Baumgärtner et al., 2002; Magerl et al., 2001).
sample size. These differences may not always represent other exception from the rule, since MPT was lower over
true regional differences, since stimuli, which were ap- foot than hand, likely because the hands usually are
plied as increasing ramps of temperature or pressure more exposed to environmental influences than feet,
(CDT, WDT, TSL, CPT, HPT, PPT) always included e.g., ultraviolet radiation and exhibit a significant thick-
reaction time artefacts, following the rule that the longer ening of the epidermis resulting in a higher mechanical
the distance to the brain, the larger the reaction time resistance to stimuli causing shear stress by very local-
artefact (see, e.g., Tillman et al., 1995; Yarnitsky and ized strong indentations (Holbrook and Odland, 1974;
Ochoa, 1991). In contrast, the higher sensitivity of the Plewig and Marples, 1970).
hand to vibratory stimuli documented in previous stud-
ies (e.g., Goldberg and Lindblom, 1979) just missed 4.2. Z-score QST profiles for easy data analysis and
significance in our data (p = 0.08) and may even be presentation
underestimated due to a reaction time artefact, since it
was measured as a disappearance threshold. The pain The abundance of tested parameters in the QST
thresholds for pinprick stimuli (MPT) constituted an- protocol of the DFNS indicates the need for an easily
R. Rolke et al. / European Journal of Pain 10 (2006) 77–88 85
Table 2
ANOVA and correlation analysis: QST data vary significantly over different areas with a lack of side differences
Factor Test area (1) Body side (2) 1 · 2 Interaction Side-to-side correlation
Parameter F p F P F p r p r2
CDT 31.1 <0.001 0.38 n.s. 0.67 n.s. 0.78 <0.001 0.61
WDT 89.4 <0.001 0.16 n.s. 2.47 n.s. 0.79 <0.001 0.62
TSL 64.3 <0.001 1.86 n.s. 1.46 n.s. 0.88 <0.001 0.77
PHS No significant occurrence of PHS in healthy subjects
CPT 2.80 n.s. 0.52 n.s. 0.08 n.s. 0.89 <0.001 0.79
HPT 0.97 n.s. 0.45 n.s. 0.26 n.s. 0.80 <0.001 0.64
MDT 77.5 <0.001 1.21 n.s. 0.82 n.s. 0.91 <0.001 0.83
MPT 11.2 <0.001 0.69 n.s. 1.56 n.s. 0.89 <0.001 0.79
MPS 5.92 <0.01 3.14 n.s. 0.56 n.s. 0.95 <0.001 0.90
ALL No significant occurrence of ALL in healthy subjects
WUR 2.76 n.s. 3.15 n.s. 1.16 n.s. 0.90 <0.001 0.81
VDT 3.21 n.s. 2.63 n.s. 0.30 n.s. 0.86 <0.001 0.74
PPT 160.0 <0.001 1.03 n.s. 1.65 n.s. 0.97 <0.001 0.94
F- and p-values as derived from 2-way ANOVA for repeated measurements (for list of abbreviations, see Section 2). For some parameters (PHS and
ALL) data exhibited close-to-zero variance and thus the near singular data matrix could not be inverted, i.e., ANOVA could not be calculated,
n.s. = not significant.
applicable standard presentation. At the same time, a tended those findings to mechanical and thermal detec-
standard presentation has to account for the fact that tion thresholds (cf. Haanpää et al., 1999; Weinstein,
different parameters come in different units of measure- 1968), but not thermal pain thresholds. Logarithmic
ment and possible data ranges differ vastly across vari- transformation of the latter are inadequate, since the
ables (e.g., 0–3 for PHS vs. 0.0–31.9 "C for CPT). temperature scale is arbitrary and there is no natural
Moreover, a clear definition of hyper- and hypophe- zero in the stimulus dimension. In contrast, the wind-
nomena is essential, if QST is to gain wider acceptance. up ratio (Price et al., 1994; Vierck et al., 1997) like all
Grouping under the heading of ‘‘abnormal finding’’ as ratios follows a geometrical distribution, which is ade-
often done in the existing literature is of little value quately accounted for by logarithmic transformation
and may potentially obscure a view on mechanisms of (cf. Magerl et al., 1998).
a pathology (Hansson et al., 2001). All of these require- In the resulting Z-score QST profile the differences
ments are fulfilled by the Z-transformed QST profiles as between different areas like hands or feet become irrele-
presented for three selected patients in Fig. 4. To enable vant due to site-specific normalization. Therefore, this
the reader to interpret the meaning of a deviation from type of data presentation will allow at-a-glance identifi-
normality, we adjusted the signs of the Z-score values in cation of symptom patterns, e.g., to identify local vs.
such a way that they specify uniformly whether a change bilateral or generalized alterations of the somatosensory
represents gain or loss of sensitivity. For example, a system. Localized changes can easily be judged com-
drop in pain threshold and an increase in suprathreshold pared to an unaffected control area, while generalized
pain ratings both indicate a gain in pain sensitivity with changes can only be identified using absolute reference
positive Z-scores (e.g., in MPT and MPS in patient C). values.
This way, we have chosen to present QST data accord-
ing to the general concept of ‘‘loss or gain of sensory 4.3. Strengths and limitations of quantitative sensory
function’’, which has a longstanding tradition through- testing
out the neurological sciences. We suggest to use Z-trans-
formed QST data (i.e., data presentation as values from Quantitative sensory tests are psychophysical in
a standard normal distribution of a reference database) nature, with an objective physical stimulus but a subjec-
in order to judge the significance of sensory changes in a tive report from a patient or control subject as the
single patient with reference to healthy controls. response. In contrast to electrophysiological, imaging
The Z-transformation has to be done separately for and biopsy techniques, QST requires cooperation from
each QST parameter and for each area tested. Most of the subject. The size of the effects of malingering and
the QST parameters required a logarithmic data trans- other non-organic factors on QST findings is currently
formation to conform to a normal (Gaussian) distribu- unresolved (Shy et al., 2003), eliminating its use in
tion. For mechanical pain thresholds to blunt and medico-legal matters. On the other hand, QST can as-
pricking stimuli, and for pain ratings this transforma- sess both large and small fiber function as illustrated
tion had previously been identified as adequate (e.g., in patients A and B, whereas standard electrophysiol-
Magerl et al., 1998; Rolke et al., 2005). We have now ex- ogy is limited to large fibers (Cruccu et al., 2004).
86 R. Rolke et al. / European Journal of Pain 10 (2006) 77–88
establish a population-based reference database, which is Dworkin RH, Backonja M, Rowbotham MC, Allen RR, Argoff CR,
currently developed by a nation-wide multi-center study Bennett GJ, et al. Advances in neuropathic pain: diagnosis,
mechanisms, and treatment recommendations. Arch Neurol
(DFNS – German Research Network on Neuropathic 2003;60:1524–34.
Pain). Such a database will provide data on the age Dyck PJ, O!Brain PC, Kosanke JL, Gillen DA, Karnes JL. A 4, 2, and
and gender-dependence of all QST parameters. Inter- 1 stepping algorithm for quick and accurate estimation of
observer and test–retest reliability have to be established cutaneous sensation threshold. Neurology 1993a;43:1508–12.
to confirm the adequacy of this QST protocol. Once val- Dyck PJ, Zimmerman I, Gillen DA, Johnson D, Karnes JL, O!Brien
PC. Cool, warm, and heat-pain detection thresholds – testing
idated with known neural mechanisms and against exter- methods and inferences about anatomic distribution of receptors.
nal standards, this protocol is expected to help in the Neurology 1993b;43:1500–8.
mechanism-based diagnosis of neuropathic pain. Fagius J, Wahren LK. Variability of sensory threshold determination
in clinical use. J Neurol Sci 1981;51:11–27.
Fields HL, Rowbotham M, Baron R. Postherpetic neuralgia: irritable
nociceptors and deafferentation. Neurobiol Dis 1998;5:209–27.
Acknowledgements Fischer AA. Pressure algometry over normal muscles. Standard values,
validity, and reproducibility of pressure thresholds. Pain
This work was supported by BMBF Grant 1987;30:115–26.
01EM0107 (German Research Network on Neuropathic Fitzek S, Baumgärtner U, Fitzek C, Magerl W, Urban P, Thömke F,
Pain, DFNS), and MAIFOR Grants of the Johannes et al. Mechanisms and predictors of chronic facial pain in lateral
medullary infarction. Ann Neurol 2001;49:493–500.
Gutenberg-University, Mainz. We are indebted to the Fruhstorfer H, Lindblom U, Schmidt WG. Method for quantitative
subjects who participated in the study for their consent estimation of thermal thresholds in patients. J Neurol Neurosurg
and co-operation. We thank G. Günther and G. Schatt Psychiatry 1976;39:1071–5.
for excellent technical support. Fruhstorfer H, Gross W, Selbmann O. Technical note: von Frey hairs:
new materials for a new design. Eur J Pain 2001;5:341–2.
Gauss CF. Wahrscheinlichkeitsrechnung und Geometrie. In: Diete-
rich, editor. Werke: Gauss; 1863.
References Gray L, Stevens JC, Marks LE. Thermal stimulus thresholds: sources
of variability. Physiol Behav 1982;29:355–60.
Arendt-Nielsen L. Induction and assessment of experimental pain Gescheider GA, Bolanowski SJ, Hardick KR. The frequency selectiv-
from human skin, muscle and viscera. In: Jensen TS, Turner JA, ity of information-processing channels in the tactile sensory system.
Wiesenfeld-Hallin Z, editors. Proceedings of the 8th world congress Somatosens Mot Res 2001;18:191–201.
on pain, progress in pain research and management. Seattle: IASP Glass GV, Stanley JC, editors. Statistical methods in education and
Press; 1997. p. 393–425. psychology; 1970. p. 534.
Arezzo JC, Bolton CF, Boulton A, Casey KL, Dyck PJ, Cornblath D, Goldberg JM, Lindblom U. Standardised method of determining
et al. Quantitative sensory testing: a consensus report from the vibratory perception thresholds for diagnosis and screening in
peripheral neuropathy association. Neurology 1993;43:1050–2. neurological investigation. J Neurol Neurosurg Psychiatry
Baron R, Saguer M. Mechanical allodynia in postherpetic neuralgia: 1979;42:793–803.
evidence for central mechanisms depending on nociceptive c-fiber Gracely RH. Pain measurement. Acta Anaesthesiol Scand
degeneration. Neurology 1995;45:S63–5. 1999;43:897–908.
Bartlett MS. The use of transformations. Biometrics 1947;3:39–52. Gracely RH, Grant MA, Giesecke T. Evoked pain measures in
Baumgärtner U, Magerl W, Klein T, Hopf HC, Treede R-D. fibromyalgia. Best Pract Res Clin Rheumatol 2003;17:593–609.
Neurogenic hyperalgesia versus painful hypoalgesia: two distinct Granot M, Sprecher E, Yarnitsky D. Psychophysics of phasic and
mechanisms of neuropathic pain. Pain 2002;96:141–51. tonic heat pain stimuli by quantitative sensory testing in healthy
Bendtsen L, Jensen R, Olesen J. Qualitatively altered nociception in subjects. Eur J Pain 2003;7:139–43.
chronic myofascial pain. Pain 1996;65:259–64. Greenspan JD. Quantitative assessment of neuropathic pain. Curr
Bolanowski Jr SJ, Gescheider GA, Verrillo RT, Checkosky CM. Four Pain Headache Rep 2001;5:107–13.
channels mediate the mechanical aspects of touch. J Acoust Soc Greenspan JD, McGillis SLB. Thresholds for the perception of
Am 1988;84:1680–94. pressure, sharpness, and mechanically evoked cutaneous pain:
Bove GM, Robichaud DR, Grigg P. Three-dimensional load analysis effects of laterality and repeated testing. Somatosens Motor Res
of indentation stimulators. J Neurosci Meth 2003;123:23–30. 1994;11:311–7.
Brennum J, Kjeldsen M, Jensen K, Jensen TS. Measurements of Gustorff B, Samal D, Magerl W, Sycha T, Rolke R, Treede R-D.
human pressure–pain thresholds on fingers and toes. Pain Primary and secondary hyperalgesia in the human UV-B sunburn
1989;38:211–7. model. Neuropathic pain – an international congress of NeuPSIG,
Brune K, Handwerker HO, editors. Hyperalgesia: molecular mecha- Book of abstracts; 2004. p. 64.
nisms and clinical implications; 2004. Haanpää ML, Laippala PA, Nurmikko TJ. Thermal and tactile
Chan AW, MacFarlane IA, Bowsher D, Campbell JA. Weighted perception thresholds in acute herpes zoster. Eur J Pain
needle pinprick sensory thresholds: a simple test of sensory 1999;3:375–86.
function in diabetic peripheral neuropathy. J Neurol Neurosurg Hansen C, Hopf HC, Treede R-D. Paradoxical heat sensation
Psychiatry 1992;55:56–9. in patients with multiple sclerosis. Evidence for a supraspi-
Chesterton LS, Barlas P, Foster NE, Baxter GD, Wright CC. Gender nal integration of temperature sensation. Brain 1996;119:
differences in pressure pain threshold in healthy humans. Pain 1729–1736.
2003;101:259–66. Hansson P, Lacerenza M, Marchettini P. Aspects of clinical and
Cruccu G, Anand P, Attal N, Garcia-Larrea L, Haanpää M, Jorum E, experimental neuropathic pain: the clinical perspective. In: Hans-
et al. EFNS guidelines on neuropathic pain assessment. Eur J son P, Fields HL, Hill RG, Marchettini P, editors. Neuropathic
Neurol 2004;11:153–62. pain: pathophysiology and treatment; 2001. p. 1–18.
88 R. Rolke et al. / European Journal of Pain 10 (2006) 77–88
Hansson P, Lindblom U. Quantitative evaluation of sensory distur- Rydel A, Seiffer W. Untersuchungen über das Vibrationsgefühl oder
bances accompanying focal or referred nociceptive pain. In: die sogenannte, ‘‘Knochensensibilität’’ (Pallästhesie). Archiv für
Vecchiet L, Albe-Fessard D, Lindblom U, Giamberardino MA, Psychiatrie und Nervenkrankheiten 1903;37:488–536.
editors. New trends in referred pain and hyperalgesia; 1993. p. 251– Sieweke N, Birklein F, Riedl B, Neundörfer B, Handwerker HO.
8. Patterns of hyperalgesia in complex regional pain syndrome. Pain
Hilz MJ, Axelrod FB, Hermann K, Haertl U, Duetsch M, Neundörfer 1999;80:171–7.
B. Normative values of vibratory perception in 530 children, Shy ME, Frohman EM, So YT, Arezzo JC, Cornblath DR,
juveniles and adults aged 3–79 years. J Neurol Sci 1998;159:219–25. Giuliani MJ, Kincaid JC, Ochoa JL, Parry GJ, Weimer LH.
Holbrook KA, Odland GF. Regional differences in the thickness (cell Therapeutics and technology assessment subcommittee of the
layers) of the human stratum corneum: an ultrastructural analysis. American academy of neurology. Quantitative sensory testing:
J Invest Dermatol 1974;62:415–22. report of the therapeutics and technology assessment subcom-
Jääskeläinen SK, Teerijoki-Oksa T, Virtanen A, Tenovuo O, Forssell mittee of the American academy of neurology. Neurology
H. Sensory regeneration following intraoperatively verified trigem- 2003;60:898–904.
inal nerve injury. Neurology 2004;62:1951–7. Susser E, Sprecher E, Yarnitsky D. Paradoxical heat sensation in
Jensen K, Andersen HO, Olesen J, Lindblom U. Pressure-pain healthy subjects: peripherally conducted by A delta or C fibres?.
threshold in human temporal region. Evaluation of a new Brain 1999;122:239–46.
algometer. Pain 1986;25:313–23. Talbot WH, Darian-Smith I, Kornhuber HH, Mountcastle VB. The
Jensen TS, Baron R. Translation of symptoms and signs into sense of flutter-vibration: comparison of the human capacity with
mechanisms in neuropathic pain. Pain 2003;102:1–8. response patterns of mechanoreceptive afferents from the monkey
Kilo S, Schmelz M, Koltzenburg M, Handwerker HO. Different hand. J Neurophysiol 1968;31:301–34.
patterns of hyperalgesia induced by experimental inflammation in Tillman DB, Treede R-D, Meyer RA, Campbell JN. Response of C
human skin. Brain 1994;117:385–96. fibre nociceptors in the anaesthetized monkey to heat stimuli:
Koltzenburg M, Lundberg LER, Torebjörk HE. Dynamic and static correlation with pain threshold in humans. J Physiol
components of mechanical hyperalgesia in human hairy skin. Pain 1995;485:767–74.
1992;51:207–19. Tölle TR, Baron R. Neuropathic pain. Basic principles for successful
Kosek E, Ekholm J, Hansson P. Pressure pain thresholds in different therapy. MMW Fortschr Med 2002;144:41–4.
tissues in one body region. The influence of skin sensitivity in Treede R-D, Lorenz J, Baumgartner U. Clinical usefulness of laser-
pressure algometry. Scan J Rehabil Med 1999;31:89–93. evoked potentials. Neurophysiol Clin 2003;33:303–14.
LaMotte RH, Shain CN, Simone DA, Tsai EFP. Neurogenic Treede R-D, Rolke R, Andrews K, Magerl W. Pain elicited by blunt
hyperalgesia: psychophysical studies of underlying mechanisms. J pressure: neurobiological basis and clinical relevance. Pain
Neurophysiol 1991;66:190–211. 2002;98:235–40.
Lang S, Klein T, Magerl W, Treede R-D. Modality-specific sensory Vierck CJ, Cannon RL, Fry G, Maixner W, Whitsel BL. Character-
changes parallel the induction of human nociceptive long-term istics of temporal summation of second pain sensations elicited by
potentiation (LTP). Soc Neurosci Abstracts 2004 Program No. brief contact of glabrous skin by a preheated thermode. J
644.17. Neurophysiol 1997;78:992–1002.
Lindblom U, Verillo RT. Sensory funtions in chronic neuralgia. J Von Frey M. Untersuchung über die Sinnesfunctionen der menschli-
Neurol Neurosurg Psychiatry 1979;42:422–35. chen Haut. Erste Abhandlung: Druckempfindung und Schmerz.
Magerl W, Fuchs PN, Meyer RA, Treede RD. Roles of capsaicin- Abhandlungen der mathematisch-physischen Klasse der Königli-
insensitive nociceptors in cutaneous pain and secondary hyperal- chen Sächsischen Gesellschaft der Wissenschaften; 1896. p. 23.
gesia. Brain 2001;124:1754–64. [Translated in: Handwerker HO, Brune K, editors. Classical
Magerl W, Wilk SH, Treede R-D. Secondary hyperalgesia and German contributions to pain research; 1987. p. 69–131].
perceptual wind-up following intradermal injection of capsaicin Weinstein S. Intensive and extensive aspects of tactile sensitivity as a
in humans. Pain 1998;74:257–68. function of body part, sex, and laterality. In: Kenshalo D,
Ochoa JL, Yarnitsky D. Mechanical hyperalgesias in neuropathic pain Springfield R, Thomas C, editors. The skin senses; 1968. p. 195–
patients: dynamic and static subtypes. Ann Neurol 1993;33:465–72. 222.
Plewig P, Marples BM. Regional differences of cell size in the human Woolf CJ, Bennett GJ, Doherty M, Dubner R, Kidd B, Koltzenburg
stratum corneum. Part I. J Invest Dermatol 1970;54:13–8. M, et al. Towards a mechanism-based classification of pain?. Pain
Price DD, Hu JW, Dubner R, Gracely RH. Peripheral suppression of 1998;77:227–9.
first pain and central summation of second pain evoked by noxious Yarnitsky D, Ochoa JL. Warm and cold specific somatosensory
heat pulses. Pain 1977;3:57–68. systems. Psychophysical thresholds, reaction times and peripheral
Price DD, Mao J, Frenk H, Mayer DJ. The N-methyl-D-aspartate conduction velocities. Brain 1991;114:1819–26.
receptor antagonist dextromethorphan selectively reduces temporal Yarnitsky D, Sprecher E. Thermal testing: normative data and
summation of second pain in man. Pain 1994;59:165–74. repeatability for various test algorithms. J Neurol Sci
Rolke R, Andrews Campbell K, Magerl W, Treede R-D. Deep pain 1994;125:39–45.
thresholds in the distal limbs of healthy human subjects. Eur J Pain Yarnitsky D, Sprecher E, Zaslansky R, Hemli JA. Heat pain
2005;9:39–48. thresholds: normative data and repeatability. Pain 1995;60:329–32.