Spatial summation of thermal stimuli assessed by a standardized, randomized, single-blinded technique

1 Introduction

Quantitative sensory testing (QST) of thermal perception also known as Quantitative Thermal Testing (QTT) is a neuro-physiological method used in clinical and experimental evaluation of small nerve fibre function [1]. QTT is a “classical” psychophysical method examining the relationship between a graded thermal stimulus, and the perceived, subjective response, i.e., warmth, cool or heat pain. Assessment of warmth and heat pain thresholds reflects unmyelinated C-fibre function, while cool detection threshold is correlated to myelinated Aδ-fibre function [2, 3].

Spatial summation denotes either a decrease in numerical threshold-value accompanying increased stimulation areas, or an increase in perceived stimulation intensity for increased stimulation areas with constant, stimulation intensity [4]. Spatial summation has been observed for both non-noxious and noxious thermal [5, 6, 7], and, mechanical stimulation [8].

Advantageous use of QTT in clinical testing and in experimental research requires the method to be standardized [9, 10]. A number of studies have contributed to standardization by evaluating the appropriate site of testing, effect of skin temperature, method for threshold determination, the rate of temperature change, baseline temperature of the thermode, and differences related to gender of test subjects [7, 11, 12, 13, 14]. The Peripheral Neuropathy Association more than two decades ago recommended specific validation procedures for thermo-electronic units [9, 10], but these are rarely reported in QTT studies. The German Research Network on Neuropathic Pain (DFNS) has conducted numerous studies, propagating for improvement and development of standardized protocols for QST [15, 16]. The DFNS has implemented multicenter-studies procuring normative databases for different patient phenotypes, with established age- and gender-matched values [15]. Though studies in recent years show more general consensus on the use of QST for the identification of subgroups of patients with different underlying pain mechanisms, prediction of therapeutic outcomes and quantification of therapeutic interventions in pain therapy [17], there still seems to be a lack of studies with standardized procedures focusing on thermal stimulation areas, which may facilitate interpretation of data across studies using differently sized thermodes. In the present study we therefore evaluated spatial summation of thermal thresholds with a standardized technique using a randomized procedure blinded to the test subjects. The aims of this study thus were, first, to examine spatial summation of QTT by three contact thermodes with different thermal stimulation areas, and, second, to evaluate if these differences are of a relevant magnitude affecting clinical and experimental acumen.

2 Material and methods

2.2 Methods

2.2.1 Randomization procedure

The study used a randomized, single-blind design. The number of subjects required to demonstrate a difference of 20% of the mean in the compared samples with a variation observed from a previous study [18] with the large thermode (heat pain threshold: mean [SD] = 48.0°C [1.5 °C]) was calculated for double-sided α = 0.05 (type I error of 5%) and β = 0.20 (type II error of 20%, i.e., a power of 80%). The estimated number of subjects required was 16.

QTT for warmth, cool and for heat pain were assessed by three thermodes with identical exterior size, but with different thermally active areas: 3.0 cm² (1.2 × 2.5 cm²), 6.25 cm² (2.5 × 2.5 cm²) and 12.5 cm² (5.0 × 2.5 cm²; Fig. 1). Thus six different testing sequences of thermodes were possible (small-medium-large, medium-large-small etc.). In order to get an even distribution of testing sequences, allowing en bloc randomization, three complete sets of testing sequences were made. Paper slips with each of the 18 sequences indicated were each placed in an unmarked closed envelope and the envelopes were shuffled. Prior to each test an envelope for each subject was drawn by the investigator. The subjects were throughout the study kept blinded to thermode size and the results of the threshold assessments. At the end of the session each subject was asked to indicate the testing sequence of the thermodes.

Fig. 1

The thermodes with identical exterior size (contact surface: 3.8 cm × 9.6 cm) but with differing thermally active areas: 3.0 cm², 6.25 cm² and 12.5 cm².

2.2.2 Testing procedure

Test sites were the medial part of the lower leg and the volar part of the distal arm. Subjects were instructed to shave the test- sites on the legs at least one day prior to testing. Perception thresholds were determined in a 2–3 h session. During the first hour the procedure was explained, the test-sites were outlined and a training session with the large thermode (12.5 cm²) was performed until the subjects seemed comfortable with the pro-cedure [19]. Perception thresholds were determined in a specific crossover sequence: leg on dominant side, non-dominant arm, leg on non-dominant side and dominant arm. All tests took place in a quiet room with an ambient temperature of 22 °C. Subjects were seated in a comfortable reclined armchair with their legs supported.

2.2.3 Assessment of thresholds

Thermal stimuli were delivered by computer controlled contact thermodes operating by the Peltier principle (Modular Sensory Analyzer, Somedic AB, Sweden). Baseline temperature was adjusted to 32°C [7, 12], and the ramp rate for both heating and cooling was set to ±1 °C/s and the cut-off limits were 50°C and 25 ° C, respectively. For warmth detection threshold (WDT) and cool detection threshold (CDT) the subject was instructed to press a button immediately when a change in temperature was perceived. For heat-pain thresholds (HPT) the subject was instructed to press the button at the first sensation of pain or discomfort. Thermal thresholds were determined as the median value of three consecutive assessments randomly separated by an interval of 4–6 s. Median values were chosen since assessments exceeding the cut-off limits for WDT and HPT, and CDT were assigned a value of 51 °C, respectively 24 °C.

2.2.4 Statistics

Statistical analyses were carried out using MedCalc (12.3.0.0, Mariakerke, Belgium) and SAS 9.1.3 (SAS Institute Inc., Cary, NC, USA) software. Data sets were initially assessed for normality by the Kolmogorov–Smirnov test and inspection of residual plots.

A mixed model with random effect for subject and fixed-effects for the variables, site (arm/leg), thermode area (ln thermode area) and side (dominant/non-dominant), was used for each of the outcomes: WDT, CDT and HPT. Non-significant (P > 0.05) fac-tors, beginning with interactions, were excluded until all included factors attained significance. HPT-data included cut-off values (all values exceeding 50 °C were given the value of 51 °C) potentially violating the assumption of a normal data distribution. Therefore a Tobit regression model for right censored data [20],^[1] and including a random effect for subject was tested and compared to the mixed effect model.

Simple pairwise comparisons of all thresholds were by Wilcoxon signed-rank test since HPT-data were non-parametrically distributed. A double-sided significance value of 0.01 was chosen in order to reduce the probability of inflicting a type I error due to multiple comparisons. Data are presented as mean (95% confidence interval [CI]) or median (non-parametric CI), as appropriate.

3 Results

Data from two individuals, #15 and #16, were excluded due to a computer error.

3.2 Pairwise comparisons

Highly significant inverse relationships between stimulation areas and thermal thresholds performed at the arms, were confirmed for all thermode sizes (P < 0.001; Fig. 2A–C), except for CDT (P =0.04; medium vs. large thermode). Similar findings of spatial summation were observed at the legs for all thermode sizes (P <0.001; Fig. 2A-C), in regard to WDT, CDT and HPT. The statistical association was weaker for CDT (P = 0.002; small vs. medium thermode) and HPT (P <0.009; medium vs. large thermode).

Fig. 2

The panels illustrate (A) warmth detection threshold (WDT), (B) cool detection threshold (CDT) and (C) heat pain threshold (HPT) for the three thermodes (small = 3.0cm², medium = 6.3cm², large = 12.5cm²) for arms and legs. By convention WDT- and CDT-values represent temperatures relative to base-line (32°C), while HPT-values represent absolute temperature values. Bar in box represents median value, box limits are 25th and 75th percentiles and whiskers are 2.5th and 97.5th percentiles. Outliers (circles) are located 1½ box height from the 2.5th percentile and extreme outliers (star) are located more than 2 box heights from the 2.5th percentile. *P <0.01, **P <0.001, ***P <0.0001.

The WDT and CDT were significantly lower for the arms compared to the legs for all thermode sizes (P < 0.0036; Fig. 2A and B). No statistical differences in HPT were seen for the small (P = 0.10), medium (P = 0.15) or large (P = 0.71) thermode.

In 17 of 18 comparisons regarding size of thermal heating area (small, medium, large), thermal thresholds (WDT, CDT, HPT) and stimulation sites (arm, leg) we observed a significant spatial summation for QTT (Fig. 3).

Fig. 3

Thermal thresholds (°C) from the arm (mean values from both arms) as a function of log thermode size (square cm) showing individual patient’s thermal trajectories. Please, observe the dual y-axes: for heat pain thresholds (HPT) left y-axis, for warmth detection thresholds (WDT) and cool detection thresholds (CDT) the right y-axis. By convention HPT-values represent absolute temperature values, and, WDT- and CDT-values represent temperatures relative to base-line (32 °C).

4 Discussion

The present study showed significant spatial summation for thermal thresholds using calibrated, uniform contact thermodes with thermal active areas in the range of 3.0–12.5 cm². The medial lower leg and volar forearm were chosen for test-sites since these are large enough for the thermodes to adjust properly to the skin surface. These sites have been used in other studies of thermal thresholds, giving valid results [11, 12]. In a study of determination paradigm for thermal perception thresholds, the method of limits by separate determinations, which is the method used in the present study, has been recommended since it is less timeconsuming, and has a good reproducibility [13, 21]. A baseline of 32°C has previously been shown to give reproducible results [12, 22].

Our threshold data are in general agreement with normative data previously reported (Table 2). WDT and CDT were significantly lower for the volar forearm than for the lower leg, corroborating data from several other studies [7, 11, 23, 24]. Furthermore, there were no differences in HPT between the two sites, which are in agreement with a previous study [11]. However, in comparison with another study [12] a remarkably large difference in WDT, between the forearm and lower leg assessed by the small thermode, was observed (Table 2). Interestingly a recent study observed a difference between body sides with a lower threshold for the left side for HPT (Table 2) [1], a finding that is not corroborated in a very large study [15]. There is no obvious explanation, but since CDT-data are comparable between the studies [1, 15] and since the general variability in our measurements with the small thermode is of the same magnitude as that found in other studies (Table 2), it is not likely that our findings represent a random error.

In a large multicentre study (n = 180) by DFNS [15] presenting a standardized protocol, two different sizes of thermodes were used to obtain reference values: 9.0 cm² and 12.5 cm². The authors stated, that “The small difference in thermode size would at most lead to a 0.5 ° C difference in threshold” citing a previously published normative study [5]. From this cited study it appears that the difference in HPT between the thermal areas 9.0 cm² and 12.5 cm², is in the order of 0.9 °C. However, no information on WDT and CDT is presented in the study.

The regression data presented in Table 1 indicate that differences in threshold-values, across the 9.0 cm² and 12.5 cm² thermodes, assessed at the leg, for WDT, CDT, and HPT, are 1.1 °C (0.9–1.3 °C), 0.6 °C (0.5–0.8°C) and 0.7 °C (0.6–0.9°C), respectively. The relative differences in WDT, CDT and HPT, comparing the 12.5 cm² thermode with the 9 cm² thermode, are 20%, 28% and 6%,^[2] respectively. These data indicate that considerable deviations may occur if absolute values, across these thermode sizes, are used uncritically: a procedure that may lead to inaccurate clinical and scientific conclusions.

Data in the present study were evaluated for normality and median values were reported. Data distributions in previous studies were not reported and only mean values were given [12].

Furthermore, in the present study subjects were blinded to size of stimulation area. QTT is a psychophysical assessment method and it is reasonable to assume that awareness of the size of stimulating area may affect accuracy of assessments. Blinding has not routinely been used in studies of spatial summation. The number of subjects estimating none, one or all of the thermodes correctly did in the present study not differ from a random distribution. The likelihood of guessing none, one or all of the thermodes correctly is 1/3, 1/2 and 1/6, respectively. The numbers in the present study were 3/14, 7/14 and 4/14 (P = 0.2). This difficulty in assessing magnitude of stimulation area by the test subjects has previously been recognized [6]. This study was blinded using several circular thermodes, the number of correctly reported activated thermodes occurred for 20-31% of the stimuli which was close to chance performance.

The importance of performing QTT with reliable, standardized equipment is demonstrated in two studies arriving at conflicting results with the use of different methods. One study regulated stimulation areas by the number of thermodes and by changing the heating aperture with application of 0.3 mm insulating plastic material on the thermode [5]. The authors concluded that the observed spatial summation of perceived heat pain intensity could be attributed to a reduction in HPT. Thus spatial summation of pain above heat pain threshold was eliminated. Another study [6] investigated the spatial summation of heat pain within and between dermatomes. In contrast the conclusion was that summation does exist also for supra-threshold heat stimuli. This study used five sep-arate circular thermodes of the same size and spatial summation was assessed by activating a variable number of thermodes. The thermodes were separated by 5 cm to 9 cm, while the former study used separation by 2.5 cm. It is likely that a sheet of only 0.3 mm is not fully reliable as insulation of the heating surface, and therefore the actual areas of testing could be inaccurate. Due to the very different methods and equipment used in the two studies, the results are largely incomparable.

5 Conclusions

In conclusion, this standardized, randomized and single-blinded study of quantitative testing of thermal thresholds by standardized contact thermodes with stimulation areas of 3.0 cm², 6.3 cm² and 12.5 cm² confirmed spatial summation for warmth detection threshold, cool detection threshold and heat pain threshold. Thus, the area of the investigating thermode is an important variable in the assessment of thermal thresholds, and standardization is highly recommended in order to facilitate interpretation of research data and normative data.

6 Implications

Data from the present study are valuable in calibration procedures, allowing estimation of thermal thresholds with differing thermode sizes, enabling important comparisons across studies to be made. The data also indicate that significant deviations, leading to erroneous clinical and scientific conclusions, may occur, if absolute values, across these thermode sizes, are used uncritically.

Highlights

• Thermal thresholds were assessed by three contact thermodes (3.0, 6.3 and 12.5 cm²).

• A significant relationship between thermode size and thermal thresholds was demonstrated.

• Spatial summation was confirmed in a randomized, single-blind study design.

• The study demonstrates that data obtained with a 9 cm² and 12.5 cm2 cannot be used interchangeably.

• Data from the present study enable estimation of thermal thresholds with differing thermode size.