Abstract
To study the territories of thin nerve fibres innervating hair follicles, we extracted single hairs from forearm skin. Scanning laser Doppler methodology was used to measure the evoked local increase of skin perfusion, the underlying assumption being that axon reflex vasodilatation would be evoked within the territory of extraction-activated thin nerve fibres. Ninety-two single hairs were extracted in 14 healthy males.
In 93 % of the cases perfusion increased transiently near the site of the extracted hair. No responses occurred when arm blood flow was occluded. In support of an underlying axon reflex mechanism the intensity of hair extraction-evoked pain correlated with the peak area of the response. In addition, after pre-extraction local anaesthesia, response components were seen in only 50 % of the cases and when they occurred they were very small.
The response had two components which could occur independently of each other. An early short-lasting component consisted of one or several separate areas with a peak total extension of 176 ± 176 mm2 (mean ± S.D.), a peak maximal intensity (in percentage of pre-extraction perfusion) of 484 ± 272 %, and a duration of 6-8 min. A later long-lasting component consisted of a single area of 51 ± 107 mm2, an intensity of 342 ± 301 % and a duration of up to approximately 60 min. Perfusion could be influenced from a single hair in an asymmetrical skin area with diameters at right angles of 23 ± 9 and 16 ± 9 mm, respectively.
We suggest that the responses were evoked by two sets of thin nerve fibres, one at a superficial level with fairly large innervation territories, and the other located more deeply close to the hair follicle and with smaller innervation territories.
The skin is innervated from several horizontal tiers of myelinated and unmyelinated nerve fibres. In many non-glabrous skin areas in humans, hairs are an important part of the sensory apparatus and the hair follicles are supplied by afferent nerve fibres from several of these tiers. Pulling or extracting a hair usually causes pain which may be short lasting and sharp, but which sometimes has a more diffuse, prolonged component. These characteristics suggest that both Aδ and C fibres may be activated. Anatomical studies have demonstrated that the epidermis contains an extensive network of unmyelinated nerve fibres (Kennedy & Wendelschafer-Crabb, 1993). Some fibres contain calcitonin gene-related polypeptide (CGRP) and/or substance P (SP) (Fundin et al. 1997), suggesting that they may be afferent axons involved in pain perception. In addition, unmyelinated fibres with immunoreactivity for different peptides have been demonstrated at different sites along the hair follicle (Fundin et al. 1997).
In contrast to such general data on the anatomy of thin fibre innervation of hairs, there is little specific information on innervation territories or functional properties of individual afferent neurones or groups of neurones at different locations on the hair follicle. The aim of the present study was to provide such information. Our methodology was based on the fact that activation of thin afferent nociceptor fibres induces vasodilatation in a local area surrounding the stimulated receptors via an axon reflex mechanism (for reviews see Holzer, 1992; Lynn, 1996). There is evidence suggesting that the area of such a response is related to the size of the innervation territory of the activated nerve fibres (Lynn et al. 1996). We hypothetised that hair extraction would also give rise to axon reflex vasodilatation which, in an analogous way, would be related to the size of nociceptive thin fibres innervating hair follicles.
METHODS
Subjects
With the permission of the human ethics committee of the University of Göteborg and with the informed consent of the subjects, 23 healthy males, aged 23-41 years (mean 30 years), participated in the study, which conformed with the Declaration of Helsinki. Hairs were extracted in 14 of the subjects and 9 participated only in control experiments. None of them suffered from any neurological or dermatological disease.
General procedure
Subjects were supine, clad in a thermal suit lined with plastic tubing, through which water of predetermined temperature could be circulated. The suit did not cover the forearms (where hairs were extracted and perfusion measurements made) and neither did it cover hands, feet and head. Subjects were warmed until fingertip temperatures were 34-35 °C before the experiment started. By then big toe temperature was usually around 30-32 °C, and the subjects had no sensation of sweating. These thermal conditions were maintained throughout the experiment (as documented by measurements of finger and toe temperatures before and after the perfusion measurements associated with each hair extraction). After five consecutive basal perfusion measurements had been made from the area to be investigated, a single hair was extracted from the dorsal side of the forearm and then, consecutive perfusion measurements were started again, until (in most cases) 18 further measurements had been made. Each perfusion measurement lasted 100 s and there was no interval between consecutive measurements. The time between the end of the last basal perfusion measurement and the hair extraction did not exceed 1 min, and after the extraction the first perfusion measurement started within 15 s. In most experimental sessions three hairs were extracted in skin areas lying approximately 8 cm apart on the dorsal side of the forearm.
After each hair extraction, subjects were instructed to rate the magnitude of the pain sensation on a scale from 0 to 3, 0 indicating no pain, 1 light, 2 moderate and 3 marked pain. They also indicated whether the pain was sharp or dull, and whether the sensation disappeared rapidly or slowly. To familiarise the subjects with the sensations evoked by hair extraction, up to 10 trial extractions were made from the contralateral forearm before the experiment, and for each extraction the subjects practised rating the sensation.
In separate experiments on 18 subjects (9 of whom participated in hair extraction experiments) one perfusion image was obtained from forearm skin before and after vascular occlusion, respectively. The occlusion was obtained by inflating a blood pressure cuff around the upper arm to suprasystolic pressure levels. Sampling for the post-inflation image was started approximately 1 min after the inflation was completed.
Hair extraction
The hair to be extracted was marked by a minute black dot on the skin. Surrounding hairs in the area of measurement were cut close to the skin with a small pair of scissors. After the control measurements were made the hair was extracted with the aid of a small forceps, the blades of which were compressed fairly weakly to prevent the hair from being cut. This was ensured by inspecting the hair directly after extraction, making sure that a clear bulb-like expansion was visible at its base.
Measurement of perfusion
Perfusion was mapped with a laser Doppler perfusion imager (Lisca Developments, Linköping, Sweden) (Wårdell et al. 1993a) using a low-power He-Ne laser beam (wavelength 632.5 nm). By a computer-controlled optical scanner the laser light was directed step by step in a square skin area of 3 cm × 3 cm (corresponding to 40 × 40 measurement sites), with a resolution of approximately 0.8 mm, at a distance of 17 cm between the scanner head and the skin. At each of the 1600 measurement sites, the light penetrates the skin to a median sampling depth of about 0.2-0.3 mm (Jakobsson & Nilsson, 1993) and the output signal (expressed in volts) is related to tissue perfusion. Data were stored on a PC for later analysis. The black dot, marking the position of the hair to be extracted, was the position of the laser beam before the start of the measurement. In the peripheral part of the skin area to be scanned, a felt pen was used to mark three black dots with a perpendicular distance of 20 mm between the dots. To eliminate interference from ambient light, the electrical light was switched off during scanning.
Signal processing and data analysis
Scanning laser Doppler images
Minute movements during the sampling of an image were found to lead to markedly increased perfusion readings at some measurement sites. To eliminate such artefacts, the perfusion at a single measurement site which was more than twice as high as the average of the surrounding sites, was set to the median value of the surrounding sites. To reduce quantification errors due to the low resolution of the original sampling of measurement sites, each perfusion image (40 × 40 measurement sites) was resampled into an image containing 160 × 160 pixels, using linear interpolation.
Interpolated intensity images representing the total backscattered light were used for geometrical scaling and image alignment. Marked dots were localized within the intensity images and the geometrical scaling of the images was made using the known distance between dots. The area of the black dots (for geometrical scaling and marking the site of the hair) was smaller than the area of one measurement site. This, together with the interpolation procedure, had the effect that the perfusion readings at the site of the dot were only slightly lower than those at surrounding sites and therefore the error was not compensated for in the quantification procedure. To minimise errors due to movements between images, the perfusion images were aligned using an iterative Marquardt algorithm (see Bard, 1974) that maximized the correlation between each intensity image and the intensity image obtained immediately prior to the stimulus.
An image of the mean basal perfusion was calculated as the average of the five matched perfusion images obtained prior to the stimulus. Before averaging, each of the five images was filtered with a 3-term average digital filter (coefficients 0.25, 0.5, 0.25) perpendicular to the scanning direction to reduce the influence of spontaneous vasomotion. To quantify the variations between the five control images due to spontaneous vasomotion, the mean basal perfusion was subtracted from each non-filtered and matched perfusion image obtained prior to the stimulus. All subtracted perfusion values (5 × 1600) were used to calculate 3 s.d. (P99) of the variations.
To identify the perfusion response induced by hair extraction, the mean basal perfusion image was subtracted from each of the interpolated and matched perfusion images obtained after the extraction. After exclusion of image series containing obvious artefacts, an automatic algorithm was used to quantify the perfusion response. Inspection of the images revealed that the main response areas were almost always present in the first or second image after the stimulus. Therefore, the segmentation process involved identification of all areas present in these two images (we discarded increases of perfusion in later images that occurred in other areas and were not present in the first two images). The minimum criterion for detection was that the perfusion increase should occur in at least 144 neighbouring pixels (corresponding to approximately 3.5 mm2) with an intensity above P99. When a region of interest (ROI) had been identified, the same ROI was searched for and segmented in other images. The procedure involved identifying any measurement site in other images with an intensity above P99 within the appropriate area. If present, this and all neighbouring pixels satisfying the intensity criterion were included in the ROI. Thus, once the ROI was defined in the first or second image, the area criterion was omitted in subsequent images. If more than one ROI was found, the largest was labelled ROI 1, the second and third largest ROI 2 and ROI 3, respectively, and all additional regions were labelled ROI 4.
In all images the response was quantified in terms of area and maximum intensity for each ROI. Since both the area and the intensity of response were greatest in the first or second image, the biggest area in either of these images was taken as the peak of the early response area. The same image was used for determination of the peak (of the maximum) perfusion intensity of the early response. To obtain the maximal perfusion intensity in a ROI, the image was smoothed with a 21 × 21 pixel filter of Hanning type with α = 0.65 (e.g. Ott et al. 1988). Following this, the maximum pixel value within each ROI was determined.
The readings of the imager did not go to zero during vascular occlusion (i.e. biological and electrical zero values differed). Therefore, to be able to calculate relative increases of perfusion after hair extraction we determined the basal perfusion level (P0), defined as the difference between the mean perfusion levels at rest and during occlusion. The mean basal perfusion from all control experiments (mean P0 = 51 arbitrary units, n = 18) was used to calculate the percentage increase of perfusion. All perfusion intensities were expressed both in arbitrary units and as a percentage of mean P0 (which was set to zero).
Following the early response peak there was a long-lasting plateau phase of increased perfusion which was quantified as the mean of the area and the maximum intensity, respectively, in images 8-12 (approximately 12-18 min) after the extraction.
Four hair extractions were exceptional in that the automatic algorithm detected no perfusion responses in the first two images after the extraction, but inspection of the measurement series revealed a clear response in later images which occurred in the same place at the site of the extracted hair. To be able to quantify these additional reactions the detection was initiated in images 8-12 and, if present there, the perfusion response was determined in other images too. No minimum size for the area was required in these cases.
RESULTS
Ninety-two hair extractions were made in 14 subjects. In 86 cases (93 %) the extraction led to a local increase of perfusion close to the relevant hair follicle. The remaining 6 extractions (7 %) evoked no response. When responses occurred, the area, strength and time course varied markedly. Most commonly (after 76 of 86 extractions, i.e. 88 %), the perfusion increased rapidly around the site of the extracted hair, and the peak area and the peak of the maximum intensity occurred in the first or second image after the extraction, i.e. within 100-200 s. After the peak there was an equally rapid decline of area and maximum intensity, and from the fourth or fifth image (around 6-8 min after the extraction) the response usually consisted of a small perfusion increase at the site of the extracted hair. The size and intensity of this central perfusion increase was relatively constant even at 30 min after the hair extraction, when imaging usually was discontinued. The typical time course of the perfusion increase is illustrated in Fig. 1.
Figure 1. Perfusion response to extraction of a single hair (location marked by cross in this and all other figures with images) from the dorsal side of the forearm.
The image in the upper left corner shows basal perfusion in this and all subsequent figures containing images. Extraction occurred immediately before the second image in the upper row in this figure and in Figs 2, 3 and 4. Three response areas were detected and enclosed by blue, red and black perimeters. The two graphs below indicate the changes with time of area and maximum intensity (in arbitrary units, AU) of increased perfusion. Blue represents the lowest and red the highest perfusion in the images (see calibration bar below images). The colours of the curves in the graphs correspond to the colours of the lines around the perimeters of the response areas. The variations of background colour (mostly yellow) in the images presumably reflect spontaneous vasomotion.
However, not all hair extractions induced both the transient and the long-lasting response components. In three cases (3 %) there was only an initial transient perfusion increase which disappeared within 13 min (Fig. 2). Conversely, in 10 cases (11 %) the initial transient increase was missing and the response consisted only of a central perfusion increase of fairly constant area and intensity (Fig. 3). These variations suggested that the perfusion response to hair extraction contained two components of different time courses; an initial, short-lasting and a relatively stable, long-lasting component.
Figure 2. Perfusion response showing an early transient but no late component.
The filled areas show the extent of the response area. Note that all traces of the response had disappeared from image 10 and onwards.
Figure 3. Perfusion response without early transient component.
The filled areas show the extent of the response. Note that the size of the response is small in the first images after extraction but increases successively and reaches a plateau in the later part of the series. The graphs below show changes of area and maximum intensity of perfusion with time.
The initial, transient perfusion increase
The area of response
In 15 cases (20 %) the initial transient perfusion increase consisted of a single response area with irregular borders. After 61 extractions (80 %) there were multiple areas of increased perfusion (Fig. 1), usually a central response area, which was much larger and of higher intensity than the others, and one or a few smaller areas, situated at some distance from the site of the extracted hair. Occasionally there was no area that was clearly larger than the others. The response areas were often asymmetrical, both in general shape and in relation to the site of the extracted hair. Exceptionally, the whole initial transient response was situated to the side of the site of the hair (Fig. 4).
Figure 4. Perfusion response with an early transient component.
The early transient component (filled area) is located at a distance from the extracted hair and the central component (open area).
The peak area of the total transient response was 176 ± 176 mm2 (range 8-744 mm2, n = 76). Responses in the proximal, middle or distal parts of the forearm were of a similar size. There were, however, marked variations between hairs with respect to the response to their extraction; some responses were small and seen only in a few images, others were large. Also large perfusion increases decreased rapidly but then, small fragments of the initial response could sometimes be seen 15-20 min after the extraction. In 34 cases the area even extended outside the borders of the image (see Fig. 1). Data on area and size for all responses to hair extraction in one subject are shown in Fig. 5 (upper diagram). Figure 6 displays peak areas of all early transient responses, and Table 1 summarises quantitative data on peak areas of both total responses and of ROI 1-4 from all transient responses.
Figure 5. Time course of perfusion increase.
The area and maximum intensity of perfusion increases after 9 hair extractions in subject NJ. Mean response curves indicated by large filled squares and thick line.
Figure 6. Peak areas of all early transient responses (n = 76).
The cross denotes the site of the extracted hair. All responses in the same row are from one subject (initials to the left of the row).
Table 1.
Size of the early transient perfusion increase
| Area ± S.D.(mm2) | Range(mm2) | n | |
|---|---|---|---|
| All hair extractions | 176 ± 176 | 8–744 | 76 |
| ROI 1 | 161 ± 178 | 1–744 | 76 |
| ROI 2 | 16 ± 11 | 6–49 | 61 |
| ROI 3 | 10 ± 6 | 5–37 | 30 |
| ROI 4 | 8 ± 3 | 5–13 | 12 |
In addition to the differences in peak response area between hairs, there were also interindividual differences (Fig. 6). Figure 7 summarises quantitative data on peak response areas from subjects in whom a minimum of five (range 5-10) hair extractions were made.
Figure 7. Interindividual comparison of peak areas of early transient responses from all subjects from whom 5 or more hair extractions were made.
ANOVA shows significant (P < 0. 01) differences between subjects. Number of hair extractions is shown in parentheses.
An estimate of the extent of the skin area, within which perfusion could be influenced from a single hair, was obtained by calculating the centre of gravity of the maximum total response area (i.e. including all ROIs) evoked by each hair extraction. Then, a determination was made of the length of the longest straight line that could be drawn between two points on the outer perimeter, through the centre of gravity of the total response. For all the response areas measured this distance was 23 ± 9 mm (range 4-41 mm). The corresponding distance for the longest line between two points on the outer perimeter drawn at right angles to the first one was 16 ± 9 mm (range 3-36 mm). The distance between the centre of gravity and the site of the extracted hair was 7 ± 2 mm (range 1-12 mm) (a measure of the degree of asymmetry of the response).
The intensity of the response
The intensity of the perfusion increase varied markedly both between and within the response areas. Within a response area there could be one or several small areas of high perfusion intensity (Fig. 1 and Fig. 8) but usually the greatest relative increase occurred at the site of the extracted hair. When many areas of high intensity were present the distances between them were often around 1.7-2.2 mm. The peak of the maximum intensity of the early perfusion response was 247 ± 140 arbitrary units corresponding to 484 ± 272 % of the mean basal perfusion. Figure 5 (lower diagram) illustrates the time courses of the maximum intensity for all hair extractions in one subject.
Figure 8. Examples of large early responses showing several ‘islands’ of high perfusion.
Maximum perfusion is close to the site of the extracted hair (marked by cross). Note that distances between ‘islands’ are often fairly similar (approximately 1.7-2.2 mm).
The long-lasting perfusion increase
When the initial transient perfusion increase had subsided, perfusion remained increased after 83 of 86 (97 %) hair extractions in a circumscribed area at the site of the extracted hair (Fig. 1 and Fig. 3). This area was fairly small in all subjects except one, in whom 10 hair extractions gave rise to unusually large initial transient response areas which decayed quite slowly (TK in Fig. 6). When no early transient increase occurred (Fig. 3) the long-lasting perfusion increase was often smaller (or occasionally absent) in the first images. In all cases it was still present after 30 min. To study the duration of the response 44 consecutive images were obtained after each of two hair extractions (different subjects). The area started to decrease after approximately 30 min; in one case it was gone after 60 min, whereas in the other case a small part was still present in the last image (after 66 min).
The mean area of the long-lasting perfusion increase in images 8-12 (approximately 12-18 min after hair extraction) was 51 ± 107 mm2 (range 1-604 mm2), i.e. approximately 30 % of the peak area of the maximal early transient response. The mean maximum intensity of the long-lasting perfusion increase in images 8-12 was 174 ± 154 arbitrary units corresponding to 342 ± 301 % above the intensity of the basal perfusion.
The nature of the perfusion responses
The effect of occlusion
In five subjects a hair was extracted while a cuff around the upper arm was inflated to suprasystolic levels ( = circulatory arrest). Three images were obtained before and two after the extraction. In no case could an early transient response be identified. In five other subjects circulatory arrest was applied after the disappearance of the early transient response but when a long-lasting perfusion increase still was present. Two images were obtained during and two after release of occlusion. In all cases the long-lasting response area disappeared during occlusion but returned again after deflation of the cuff.
Effects of local anaesthesia
In 11 experiments nerves around the hair follicle and in the neighbouring skin were anaesthetized with 1.0-1.2 ml of mepivacain (Carbocain 1 %, Astra, Södertälje, Sweden) which was injected from below, at the site of the hair to be extracted. To achieve this, a hypodermic needle was inserted 2-3 cm from the hair, moved subcutaneously towards the site of the hair and then directed up towards the skin surface, where the injection was made. One hour later imaging was started and the hair extracted. The results are shown in Table 2 and can be summarised as follows. One hair extraction evoked no perfusion response, four extractions induced an early transient response only, five extractions evoked a late sustained response only, and in one case both response components were seen. All responses were small (9-32 mm2 for the early and 2-31 mm2 for the late responses) and two of the early transient responses were in the peripheral part of the image, to the side of the hair. Three hair extractions evoked no sensation, but in eight cases there was a vague sensation that something happened, but no pain component.
Table 2.
Effects of hair extractions after local anaesthesia
| Response type |
|||
|---|---|---|---|
| Hair ID | Early(mm2) | Long lasting(mm2) | Comment |
| CO1 | 0 | 31 | |
| SS1 | 11 | 3 | |
| SS2 | 9 | 0 | |
| MT1 | 0 | 4 | |
| MT2 | 32 | 0 | Area in peripheral part of image |
| TT1 | 0 | 6 | |
| TT2 | 15 | 0 | Area in peripheral part of image |
| AL1 | 0 | 2 | |
| AL2 | 0 | 0 | |
| PP1 | 0 | 4 | |
| PP2 | 9 | 0 | |
Relationship between perception and perfusion responses
There was a weak positive correlation (r = 0.23, P < 0.05, n = 86) between the degree of pain evoked by the hair extraction and the peak area of the early transient (but not the late long-lasting) response. The size of the perfusion response was not related to the character of the pain, i.e. sharp, diffuse, short-lasting and long-lasting sensations were associated with similar perfusion responses.
Of the six hair extractions which evoked no perfusion response, two gave rise to pain of grade 2, three to pain of grade 1 and one to no pain. All sensations disappeared rapidly.
DISCUSSION
The present study confirmed our hypothesis that extraction of a single hair from forearm skin induces a local perfusion increase around, or close to, the extracted hair. The time course of the response comprised both an early transient and a weaker sustained component and since these sometimes occurred independently of each other, their sources are unlikely to be identical.
Underlying mechanisms
Since no hair extraction-induced perfusion increases occurred during vascular occlusion, we conclude that both the early transient and the late response components were dependent on ongoing perfusion. Consequently, they were not due to bleeding induced by the trauma of the extraction.
After local anaesthesia the hair extractions evoked no pain, i.e. many neurones innervating the hair follicle were anaesthetised. Concomitantly, the incidence of both components of the perfusion increase was reduced to about 50 %, and when responses did occur, the areas were small. These findings suggest that both the early and the late components of the response were evoked by action potentials in nerve fibres activated by the extraction of the hair. The reason why all vascular responses did not disappear may be incomplete anaesthesia. Since it was difficult to determine the exact location of the needle tip, and since it was crucial to avoid irritating or damaging the skin with the tip, the anaesthetic solution may have been deposited too deeply. If the spread of fluid to the dermis was incomplete, occasional small nerve branches in the upper part of the dermis or epidermis may still have been able to propagate impulses in the distal segment. This would agree with the fact that when perfusion responses did occur after anaesthesia, they were small or, in the case of the two largest early transient responses, located at a distance from the site of hair.
Action potentials in the terminal branches of some unmyelinated and thin myelinated cutaneous fibres are known to cause liberation of neuropeptides which evoke (axon reflex) vasodilatation in an area around the activated terminals (for reviews see Holzer, 1992; Lynn, 1996). We suggest that both components of the perfusion increase induced by hair extraction were axon reflex effects. Duration and intensity of axon reflex vasodilatation have been found to increase in parallel with increasing number of impulses (Magerl et al. 1987; Hornyak et al. 1990) and, depending on the stimulus conditions, durations ranging from 1 to 40 min have been reported in human skin (Blumberg & Wallin, 1987; Wårdell et al. 1993b). In the present study results were different. Our early transient response component was more intense but had a much shorter duration than the late component. A possible explanation would be that the initial short-lasting response component was due to fairly weak activation of one set of superficial neurones, and the sustained response component to strong activation of a different set of neurones at a deeper location. This would agree with the fact that the deeper a vessel is located, the weaker its relative contribution to the perfusion recorded by the laser Doppler imager (Wårdell et al. 1993a; Nilsson & Nilsson, 1998).
Possible anatomical correlates
The early transient response
The area of an axon reflex vasodilatation probably reflects the size of the innervation zone of the terminal branches of the activated afferent neurones (see Lynn et al. 1996). However, since a cutaneous axon reflex vasodilatation has been found to extend 1-2 mm into anaesthetised skin (Wårdell et al. 1993b), the vascular response area is somewhat larger than the size of the territory of the activated nerve fibres. Our mean peak area of 176 mm2 for the total early transient response corresponds to a circle with a diameter of approximately 15 mm. Since this calculation did not take into account the fact that the total response was often composed of several response areas at some distance from each other, we also estimated the extent of the skin area within which perfusion could be influenced from a single hair. The result was an asymmetrical area with mean perpendicular diameters being approximately 23 and 16 mm. For several reasons both calculations are uncertain and may underestimate the true area (e.g. the response area often extended outside the image), but the numbers nevertheless suggest that a set of thin nerve fibres connected with a hair may have large arborisations into neighbouring skin areas.
Whether the response was due to activation of more than one nerve fibre is unclear. The multiple response areas may correspond to different branches of a single neurone or, alternatively, to activation of several neurones. In addition, the marked differences in response between hairs may be due to differences either in the number of activated nerve fibres or in the size of the innervation zones of individual fibres. Furthermore, the interindividual differences in response size suggest that there may be systematic differences between subjects, either in thin fibre innervation of hairy skin and/or in the spread of released neuropeptides in the tissue.
Hairs are supplied by both myelinated and unmyelinated afferent nerve fibres from several horizontal tiers. CGRP- and SP-containing axons deriving from the most superficial tier have been demonstrated in the epidermis, approximately 100 μm below the skin surface (Fundin et al. 1997) and may be the anatomical correlate with the early transient axon reflex responses seen in the present study. Recent evidence from the pig (Lynn et al. 1996) and from humans (Magerl & Treede, 1996; Schmelz et al. 2000) suggest that the unmyelinated afferent fibres giving rise to cutaneous axon reflex vasodilatation are heat sensitive and mechano-insensitive, rather than polymodal C fibres. Aδ fibres have been shown to evoke cutaneous axon reflex vasodilatation in the rat (Jänig & Lisney, 1989) but whether such fibres contribute in humans is unclear.
The long-lasting response
Compared to the early transient response component, the size of the long-lasting component was smaller with an area of 51 mm2, corresponding to a circle with a diameter of 8 mm. It was always located around or close to the site of the hair. A possible anatomical correlate would be unmyelinated axons encircling the hair follicle neck or associating with the piloneural complexes (Fundin et al. 1997). The piloneural complex is situated approximately 0.3-0.4 mm below the skin surface and although this is below the median operating range of the scanning laser Doppler equipment (0.2-0.3 mm) (Jacobsson & Nilsson, 1993) increases of blood flow in larger vessels at this depth will still be able to contribute to the perfusion signal (Nilsson & Nilsson, 1998).
Relationship between perception and perfusion responses
We found a correlation between pain intensity and peak area of the early transient perfusion response. In principle, this agrees with the idea that the two main fibre types associated with axon reflex vasodilatation, unmyelinated and thin myelinated fibres, are known to contribute to pain perception. However, the correlation was very weak and in addition, five of the six hair extractions that gave no perfusion response did evoke pain. Why the coupling between pain and vascular response was so weak is unclear. A possible explanation is that some of the nerve fibres that are important for pain perception, such as polymodal C fibres, give rise to little axon reflex vasodilatation (see Lynn et al. 1996; Magerl & Treede, 1996; Schmelz et al. 2000). A second alternative (which, however, has no support in the literature) would be that fibre types other than those associated with pain perception evoke axon reflex vasodilatation. The explanation may also be related to the fact that arm hairs have a life cycle of 6-8 months (Saitoh et al. 1970) comprising both an anagen phase when the hair grows and maturates and a telogen phase when it degenerates and falls off. Our hairs were probably extracted at different phases of the life cycle and since the coupling between hair, nerves and blood vessels changes during the cycle (Botchkarev et al. 1997), this may be a confounding factor.
Functional significance
Cutaneous axon reflex vasodilatation is usually regarded as part of a nocifencor response, i.e. an inflammatory reaction counteracting damage to the skin (see Lembeck, 1983) and the perfusion increase after hair extraction may fall in the same category. The duration of the late component of our response was remarkably long which, as mentioned above, may indicate a strong neural activation with associated peptide release (e.g. of CGRP, which induces a very long-lasting vasodilatation; see Holzer, 1992, for references) at the site of the empty hair follicle. This may also initiate processes that facilitate the outgrowth of a new hair. Whether the (presumed) hair extraction-evoked high local concentration of neuropeptides also has more distant effects is unclear.
Limitations of the study
The perfusion response to hair extraction is influenced both by neural and vascular factors, which may be difficult to separate from each other. For example, resting skin blood flow displays irregular spontaneous variations which may be evoked by sympathetic nerve traffic or by non-neural mechanisms. In the present study, one of our criteria for a significant perfusion response to hair extraction was that the increase of perfusion should exceed 3 s.d. of such spontaneous variations occurring in the five control images. The criterion is conservative and may have led to an underestimation of the true extension of the initial response area (and if so, overestimation of the incidence of multiple response areas). In support of this possibility the response area was sometimes surrounded by a rim in which perfusion was ‘non-significantly’ higher than in the periphery of the image. On the other hand, a less conservative criterion would probably have given larger variability between successive images. For example, we noted that during long series of images one or several images often displayed (localised or more general) changes of perfusion, which were greater than those seen during the control sequence (see Fig. 1). The origin of such changes is unclear but reflex effects (e.g. due to sudden sounds from the corridor, emotional reactions to sudden thoughts, and periods of drowsiness) are possible alternatives.
Since the initial transient perfusion response often had a maximum in the first image it cannot be excluded that the true shape and extent of the response was distorted by the long duration of the scanning procedure (e.g. if a response had started to decrease before the scanning of an image was completed). Such effects would be more pronounced the shorter the duration of the initial response and could, for example, exaggerate the apparent asymmetry of a response area in relation to the location of the extracted hair.
There is evidence that sympathetic vasoconstriction and vasodilatation due to axon reflex effects are competitive in the control of skin blood flow (Hornyak et al. 1990; Habler et al. 1997). An ongoing vasoconstrictor drive would then be expected to reduce the size and strength of the axon reflex dilatations and lead to an underestimation of the size of the innervation territories. Therefore, we warmed the subjects until they had a fingertip temperature of 34-35 °C. In such a comfortably warm thermal state there is little skin vasoconstrictor activity (Bini et al. 1980), which should ensure that the axon reflex vasodilatation was more or less unopposed by sympathetically mediated vasoconstriction.
The cutaneous vascular anatomy may also influence the perfusion response. In human abdominal skin, laser Doppler flowmetry has detected ascending arterioles situated 1.5-1.7 mm apart (Braverman et al. 1990). Since our perfusion imager has a resolution of 0.8 mm, it seems likely that such vessels contribute to the occurrence of the multiple peaks of perfusion seen within the perfusion responses (Fig. 8). In addition, since each perfusion measurement lasts only a fraction of a second, the different measurement points were obtained during different parts of the cardiac cycle, i.e. the perfusion data will include a random variability due to the variations of volume flow during the cardiac cycle.
Acknowledgments
We thank Tomas Karlsson for technical assistance. This work was supported by Swedish Medical Research Council Grant no. 3546.
References
- Bard Y. Nonlinear Parameter Estimation. New York: Academic Press; 1974. The Marquardt Method. [Google Scholar]
- Bini G, Hagbarth K-E, Hynninen P, Wallin B G. Thermoregulatory and rhythm-generating mechanisms governing the sudomotor and vasoconstrictor outflow in human cutaneous nerves. Journal of Physiology. 1980;306:537–552. doi: 10.1113/jphysiol.1980.sp013413. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Blumberg H, Wallin B G. Direct evidence of neurally mediated vasodilatation in hairy skin of the human foot. Journal of Physiology. 1987;382:105–121. doi: 10.1113/jphysiol.1987.sp016358. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Botchkarev V A, Eichmüller S, Johansson O, Paus R. Hair cycle-dependent plasticity of skin and hair follicle innervation in normal murine skin. Journal of Comparative Neurology. 1997;386:376–395. doi: 10.1002/(sici)1096-9861(19970929)386:3<379::aid-cne4>3.0.co;2-z. [DOI] [PubMed] [Google Scholar]
- Braverman I M, Keh A, Goldminz D. Correlation of laser doppler wave pattern with underlying microvascular anatomy. Journal of Investigative Dermatology. 1990;95:283–286. doi: 10.1111/1523-1747.ep12484917. [DOI] [PubMed] [Google Scholar]
- Fundin B T, Arvidsson J, Aldskogius H, Johansson O, Rice S N, Rice F L. A comprehensive immunofluorescence and lecitin binding analysis of intervibrissal fur innervation in the mystacial pad of the rat. Journal of Comparative Neurology. 1997;385:185–206. doi: 10.1002/(sici)1096-9861(19970825)385:2<185::aid-cne2>3.0.co;2-y. [DOI] [PubMed] [Google Scholar]
- Habler H J, Wasner G, Jänig W. Interaction of sympathetic vasoconstriction and antidromic vasodilatation in the control of skin blood flow. Experimental Brain Research. 1997;113:402–410. doi: 10.1007/pl00005594. [DOI] [PubMed] [Google Scholar]
- Holzer P. Peptidergic sensory neurons in the control of vascular functions: mechanisms and significance in the cutaneous and splanchnic vascular beds. Review of Physiology, Biochemistry and Pharmacology. 1992;121:49–146. doi: 10.1007/BFb0033194. [DOI] [PubMed] [Google Scholar]
- Hornyak M E, Naver H K, Rydenhag B, Wallin B G. Sympathetic activity influences the vascular axon reflex in the skin. Acta Physiologica Scandinavica. 1990;139:77–84. doi: 10.1111/j.1748-1716.1990.tb08899.x. [DOI] [PubMed] [Google Scholar]
- Jacobsson A, Nilsson G E. Prediction of sampling depth and photon pathlength in laser Doppler flowmetry. Medical and Biological Engineering and Computing. 1993;31:301–307. doi: 10.1007/BF02458050. [DOI] [PubMed] [Google Scholar]
- Jänig W, Lisney S J W. Small diameter myelinated afferents produce vasodilatation but not plasma extravasation in rat skin. Journal of Physiology. 1989;415:477–486. doi: 10.1113/jphysiol.1989.sp017732. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kennedy W R, Wendelschafer-Crabb G. The innervation of the human epidermis. Journal of the Neurological Sciences. 1993;115:184–190. doi: 10.1016/0022-510x(93)90223-l. [DOI] [PubMed] [Google Scholar]
- Lembeck F. Sir Thomas Lewis's nocifensor system, histamine and substance-P-containing primary afferent nerves. Trends in Neurosciences. 1983;6:106–108. [Google Scholar]
- Lynn B. Neurogenic inflammation caused by cutaneous polymodal receptors. Progress in Brain Research. 1996;113:361–368. doi: 10.1016/s0079-6123(08)61098-5. [DOI] [PubMed] [Google Scholar]
- Lynn B, Schütterle S, Pireau F K. The vasodilator component of neurogenic inflammation is caused by a special subclass of heat-sensitive nociceptors in the skin of the pig. Journal of Physiology. 1996;494:587–593. doi: 10.1113/jphysiol.1996.sp021516. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Magerl W, Szolcsányi J, Westerman R A, Handwerker H O. Laser Doppler measurements of skin vasodilatation elicited by percutaneous electrical stimulation of nociceptors in humans. Neuroscience Letters. 1987;82:349–354. doi: 10.1016/0304-3940(87)90281-3. [DOI] [PubMed] [Google Scholar]
- Magerl W, Treede R-D. Heat-evoked vasodilatation in human hairy skin: axon reflexes due to low-level activity of nociceptive afferents. Journal of Physiology. 1996;497:837–848. doi: 10.1113/jphysiol.1996.sp021814. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Nilsson H, Nilsson G E. Proceedings of Optical Diagnostics of Biological Fluids. Vol. 3252. The International Society for Optical Engineering; 1998. Monte Carlo simulations of the light interaction with blood vessels in human skin in the red wavelength region. III, SPIE series. [Google Scholar]
- Ott R J, Flower M A, Babich J W, Marsden P K. The physics of radioisotope imaging. In: Webb S, editor. The Physics of Medical Imaging. Bristol and Philadelphia: Adam Hilger; 1988. p. 197. [Google Scholar]
- Saitoh M, Uzuka M, Sakamoto M. Human hair cycle. Journal of Investigative Dermatology. 1970;54:65–81. doi: 10.1111/1523-1747.ep12551679. [DOI] [PubMed] [Google Scholar]
- Schmelz M, Michael K, Weidner C, Schmidt R, Torebjörk H E, Handwerker H O. Which nerve fibres mediate the axon reflex flare in human skin. NeuroReport. 2000;11:645–648. doi: 10.1097/00001756-200002280-00041. [DOI] [PubMed] [Google Scholar]
- Wårdell K, Jakobsson A, Nilsson G E. Laser Doppler perfusion imaging by dynamic light scattering. IEEE Transactions of Biomedical Engineering. 1993a;40:309–316. doi: 10.1109/10.222322. [DOI] [PubMed] [Google Scholar]
- Wårdell K, Naver H K, Nilsson G E, Wallin B G. The cutaneous vascular axon reflex in humans characterized by laser Doppler perfusion imaging. Journal of Physiology. 1993b;460:185–199. doi: 10.1113/jphysiol.1993.sp019466. [DOI] [PMC free article] [PubMed] [Google Scholar]