Enhanced CXCL10 expression in mast cells for cutaneous neurofibroma presenting with pain and itch

Abstract

Background

Cutaneous neurofibroma (cNF) presenting with pain and itch substantially affects the quality of life. The CXCL10/CXCR3 axis, a well-known chemokine signaling pathway involved in pain and itch transmission, has recently been implicated in neurofibroma development. Our study aims to investigate the expression patterns and potential roles of the CXCL10/CXCR3 axis in pain and itch associated with cNFs.

Methods

We examined the expression of CXCL10/CXCR3 and immune cell profiles in 53 human solitary cNFs through immunohistochemical staining. The Chinese version of the Short-form McGill Pain Questionnaire and the Chinese Eppendorf Itch Questionnaire were used to assess pain and itch symptoms of cNF tumors, respectively.

Results

Elevated expression of CXCL10/CXCR3 was observed in tumoral and dermal parts of symptomatic cNFs. The percentage of mast cells expressing CXCL10, but not CXCR3, was significantly higher in symptomatic cNFs compared to asymptomatic cNFs (51.18% vs. 19.07%, respectively, p < 0.0001). The symptomatic cNFs exhibited significantly higher intraepidermal nerve fiber density compared to asymptomatic cNFs (p = 0.009).

Conclusions

Our study suggests that CXCL10, potentially mediated by mast cells, may contribute to sensory dysfunction in cNF and may be a target for treating the pain and itch symptoms associated with cNFs.

Our study suggests a model in which the CXCL10/CXCR3 pathway plays a role in inducing pain and itch in cNFs, potentially through mast cell mediation. Mast cells may increase the secretion of CXCL10, thereby contributing to pain and itch in cNF, making them a potential target for treating these symptoms. Created in BioRender. Pham, Q. (2025) https://BioRender.com/i89y356.

Introduction

Pain and itch are well-known distressing symptoms of cancer. Approximately 44.5% [1] and 13–26% [2, 3] of patients with cancer experience pain and itch, respectively. These symptoms are often reported as moderate to severe and associated with advanced disease stages and a poor quality of life [1, 4–6]. Cutaneous neurofibroma (cNF), a prevalent type of neurofibroma, frequently presents with pain and itch. cNF and its symptoms, such as pain and itch, adversely affect the quality of life, including daily activities and social interactions [7–10], resulting in diverse psychosocial challenges and affecting psychological well-being [11]. Furthermore, the effective treatment of recurrent and large or multiple cNFs remains elusive. Some indications suggest that early intervention in subclinical cNFs may be beneficial for preventing the development of refractory cases including symptomatic tumors such as painful and itchy cNFs [12].

Molecules and immune cells contributing to transduction pathways responsible for itch and pain have been identified as crucial factors in neurofibroma tumorigenesis. In particular, mast cells and macrophages are involved in cNF progression and its sensory symptoms [13–15]. A study noted an increase in the number of mast cells in actively growing and itchy cNFs, indicating an association between mast cells and itch [16]. Areas of itchy skin that appear normal can serve as indicators for cNF development, and a notable infiltration of mast cells and macrophages was found in these clinically undetectable micro-cNFs [17–19]. Furthermore, the number of nonpeptidergic C-fibers, which are involved in pain and itch transduction, increases in pre-cNF lesions [17].

The G-protein coupled receptor C-X-C motif chemokine receptor 3 (CXCR3) and its ligand, chemokine C–X–C motif ligand 10 (CXCL10), are crucial in the transmission of itch and pain as well as tumorigenesis. The CXCL10/CXCR3 axis holds a key role in the inflammation process and is expressed in several critical immune cells, such as macrophages, mast cells, T cells, and neutrophils [20–24]. These immune cells are main cellular compartments in neurofibroma microenvironment [13, 14, 16, 25, 26]. Moreover, studies have observed the direct activation of CXCL10 on cutaneous sensory nerves that trigger itch or pain through CXCR3 in certain inflammatory conditions [27–29]. Inhibiting the CXCL10/CXCR3 function reduced itch and pain sensations in vivo [22, 29–31]. Furthermore, increased CXCL10/CXCR3 expression is implicated in tumor pain, such as pancreatic cancer, breast cancer, and bone cancer pain [29, 32–34]. Recently, this axis was shown to be required for neurofibroma formation [20].

Although some progress has been made in understanding these chemokines, their precise roles and other critical factors involved in cNF tumorigenesis remain unclear. Cutaneous neurofibroma can occur either in association with the genetic neurofibromatosis type 1 (NF1) or von Recklinghausen’s disease disorder [13] or as an independent tumor [35], and it is commonly studied in the context of NF1. As a result, the specific mechanism of cNF development and its symptoms may be obscured by the NF1 background. Collectively, a considerable gap exists in the pathophysiology of cNF and its associated symptoms, leading to the absence of highly effective treatments for these symptomatic tumors. Here, we aimed to delineate the significance of CXCL10/CXCR3 in immune cells for the development of cNF and its associated symptoms. We utilised immunohistochemical (IHC) and immunofluorescent (IF) staining on non-NF1 human cNF specimens along with pain and itch questionnaires to assess characteristics of pain and itch. In addition, we suggest investigating the involvement of CXCL10/CXCR3 in neurofibroma development and associated sensory disorders as a direction for future research.

Material and methods

Patient recruitment and sample collection

We collected 53 non-NF1 tumor specimens with diagnoses of cNF at the Department of Dermatology, Taipei Medical University-Shuang Ho Hospital. This study was approved by the Research Ethics Committee of Taipei Medical University (TMU-REC No. N202103158). We included adult patients who (1) received a diagnosis of solitary cutaneous neurofibroma based on their clinical and pathological findings and (2) did not have NF1.

After obtaining written informed consent from each participant, a research associate conducted structured interviews to collect demographic data and related information, such as medical conditions, current medications, and history of pain relievers. Clinical details of each cNF, including gender, age at sampling, clinical variants (flat, sessile, globular, and pedunculated cNFs) [12], tumor distribution across five anatomical sites (head and neck, trunk, upper limb, and lower limb), and symptoms (pain and itch), were recorded. Participants also completed the Chinese Eppendorf Itch Questionnaire (CEIQ) [36, 37] and the Chinese version of the Short-form McGill Pain Questionnaire (SF-MPQ) [38, 39] to assess their itch and pain symptoms, respectively. Overall, itch and pain intensities were evaluated using a numeric rating scale (NRS) by the questionnaires. The CEIQ was divided into four categories, including intensity, affect, urge, and coping, as previously described [36]. The neuropathic itch score was calculated by the summation of scores for items in the CEIQ indicative of neuropathic itch characteristics. In particular, items 1–16, 21, 25–35, and 38 in the CEIQ were used for this calculation [36].

The formalin-fixed, paraffin-embedded tumor specimens were sectioned into 3-µm slices for the evaluation of tumor depth, CXCL10/CXCR3 expression, and immune cell profiles (mast cells, macrophages, T cells, and neutrophils). Additionally, we utilised 16 µm sections to quantify intraepidermal nerve fiber (IENF) density. These paraffin-coated slides and tumor specimens were coded to ensure blinded analysis. Tumor depth was defined as the distance between the lower border of the epidermal basement membrane and the upper border of the tumor on the S100-stained pathological image. Tumor size was determined based on the mass of the tumor after excision.

Immunohistochemistry staining

The slides were deparaffinized and then immersed in 400 mL of boiled antigen retrieval buffer (10 mM sodium citrate, 0.05% Tween 20, pH 6.0; Bioman Scientific, Taipei, Taiwan, Cat. no. D3316), followed by incubation with 3% H₂O₂ (PanReac AppliChem ITW Reagents, Darmstadt, Germany, Cat. no. 7722-84-1). To block nonspecific binding sites, 10% normal goat serum (Abbkine, Georgia, USA, Cat. no. BMS0050) was used for 1 h at room temperature. The slides were then incubated with primary antibodies in 1% normal goat serum at 4 °C overnight. Primary antibodies included rabbit anti-CXCR3 (Abcam, Cambridge, UK, Cat. no. ab288437; dilution 1:200), rabbit anti-CXCL10 (ProteinTech Group, Rosemont, IL, USA, Cat. no. 10937-1-AP, dilution 1:400), rabbit anti-PGP9.5 (Protein gene product 9.5, Abcam, Cambridge, UK, Cat. no. ab108986, dilution 1:200), mouse anti-S100 (Roche Diagnostics, Basel, Switzerland, Cat. no. 790-2914, dilution 1:1), and mouse anti-CD68 (cluster of differentiation 68, Dako, Oslo, Norway, Cat. no. M0876, dilution 1: 200). On the second day, the slides were treated with ImmPRESS HRP Horse Anti-Rabbit IgG Polymer Reagent (Vector Laboratories, CA, USA, Cat. no. MP-7801), followed by exposure to ImPACT DAB EqV working solution. Negative controls were treated with 1% normal goat serum in Tris-buffered saline instead of the primary antibody. The slides were then counterstained with hematoxylin (Cis-Biotechnology, Taiwan, Cat. no. D469) and mounted with Acrytol mounting medium (Leica, Wetzlar, Germany, Model 3801720).

Immunofluorescence staining

For the evaluation of CXCL10 and CXCR3 expression on mast cells, macrophages, T cells, and neutrophils, we conducted double immunofluorescence staining. After deparaffinization, rehydration, and antigen retrieval, the slides were blocked with 10% normal goat serum for 1 h at room temperature and then incubated at 4 °C with the primary antibodies overnight. The primary antibodies included rabbit anti-CXCR3 (Abcam, Cambridge, UK, Cat. no. ab288437; dilution 1:200), rabbit anti-CXCL10 (ProteinTech Group, Rosemont, IL, USA, Cat. no 10937-1-AP, dilution 1:400), mouse anti-tryptase (Tryp) (Abcam, Cambridge, UK, Cat. no. ab2378, dilution 1:500), mouse anti-CD68 (Abcam, Cambridge, UK, Cat. no. ab955, dilution 1:50), mouse anti-CD3 (cluster of differentiation 3, Life technologies, Thermo Fisher Scientific, Carlsbad, CA, USA, Cat. no. MHCD0300, dilution 1:150), and mouse anti-MPO (myeloperoxidase, R&D systems, Inc., Minneapolis, MN, USA, Cat. no. MAB3174, dilution 1:50). On the following day, the slides were incubated for 1 h at room temperature with the secondary antibodies, including goat anti-rabbit 488 (Alexa Fluor 488, Invitrogen, Waltham, MA, USA, Cat. no. A11034, dilution 1:200) and goat anti-mouse 594 (Alexa Fluor 594, Jackson Immuno Research Labs, West Grove, PA, USA, Cat. no. 115586146, dilution 1:150 to 1:500). A water-soluble mounting medium (Fluoromount-G, with DAPI, Invitrogen, Waltham, MA, USA, Cat. no. 00-4959-52) was used to mount the slides.

Toluidine blue staining

After deparaffinization and rehydration, the slides were incubated for 5 min with toluidine blue working solution (Scytek, Logan, UT, USA, Cat. no. TQF125), which was freshly prepared by adding 500 µL of toluidine blue and 150 µL of 3% NaOH in 50 mL of distilled water. Then, the slides were washed with distilled water once for 5 min. The slides were then dehydrated and fixed with Acrytol Mounting Medium (Leica, Wetzlar, Germany, Model: 3801720).

Image analysis

The toluidine blue and IHC-stained slides were examined for the expression of mast cells and CD68⁺ macrophages in the tumoral area, as well as tumoral and dermal CXCL10/CXCR3, using the Moticeasyscan Pro scanner at 20× magnification. Image analysis was conducted using FIJI (ImageJ).

Quantification of CXCL10 and CXCR3 staining

For IHC analysis, six randomly selected regions of interest (ROIs) were extracted from both the tumoral and dermal regions on each stained slide. All selected ROIs within each respective region were of the same size. The areas of ROIs in the dermal and tumoral regions were 0.086 mm × 0.071 mm and 0.62 mm × 0.35 mm, respectively. The images were analysed using the IHC Toolbox plugin in FIJI. In brief, areas treated with 3,3’-diaminobenzidine were separated, and the channel containing exclusively 3,3’-diaminobenzidine-treated areas was converted into a binary image. For each ROI, the stained areas and fractional stained area per total area were measured.

Quantification of PGP9.5⁺ skin nerve density

The IHC-stained images were examined using a Moticeasyscan Pro scanner at 20× objective. For every 130 µm of the upper border length of the stratum granulosum layer, we counted the number of PGP9.5⁺ epidermal skin nerves using an established method [40]. The IENF density (number of IENFs/mm) was calculated from the 16 µm stained slides by dividing the average number of PGP9.5⁺ epidermal skin nerves in six randomly selected regions by the length.

Quantification of colocalization of CXCL10/CXCR3 in immune cells

The IF-stained sections were imaged using the STELLARIS 8 confocal microscope (Leica Microsystems) with a 20× objective. Six ROIs were selected for the analysis of double IF staining, each measuring 0.455 mm × 0.4 mm, using FIJI (ImageJ) software. The colocalization of two markers was calculated based on the percentage of cells double-positive for the markers out of the total number of Tryp⁺ mast cells, CD68⁺ macrophages, CD3⁺ T cells, or MPO⁺ neutrophils. These double-positive cells were counted, and the average percentage of these cells relative to the total Tryp⁺ mast cells, CD68⁺ macrophages, CD3⁺ T cells, or MPO⁺ neutrophils in the six ROIs was calculated.

To assess the overall mast cell, macrophage, T cell, and neutrophil densities, we calculated the average number of mast cells, macrophages, T cells, and neutrophils in the tumoral region based on the counts obtained from six randomly selected ROIs. Each ROI, measuring 0.62 mm × 0.35 mm for mast cells visualised by toluidine blue staining and macrophages visualised by IHC staining, and 0.455 mm × 0.4 mm for T cells and neutrophils visualised by IF staining, was included in the analysis.

Statistical analysis

We utilised GraphPad Prism version 9.5 for data analysis. Mean values accompanied by their standard errors of the mean (S.E.M) represent numerical data, while frequencies and percentages delineate categorical variables. cNF tumors were categorised into five groups: asymptomatic tumors (without pain or itch); symptomatic tumors (either pain or itch); both painful and itchy tumors (both pain and itch); painful tumors (only pain); and itchy tumors (only itch). To compare two continuous variables, an unpaired two-tailed Student’s T-test was employed. The correlation between two continuous variables was evaluated by Pearson’s correlation coefficient. A p-value below 0.05 indicates statistical significance and is denoted as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

Results

Clinical and pathological characterisation of cNFs

To investigate the clinical and pathological characteristics of cNFs and their association with pain and itch, we analysed 53 tumor specimens from 49 patients (Table 1). The average age of the patients with cNFs included in our study was 57.43 ± 1.86 years. Among the tumors, 31 were asymptomatic, 22 manifested pain or itch (symptomatic), 5 manifested both pain and itch, 10 manifested only pain, and 7 manifested only itch. There was a preponderance of painful cNFs in males (male: female ratio = 7:3, Table 1). Sessile (41.51%) and globular (37.74%) cNFs were prevalent in our collected tumors, and equally notable in asymptomatic and symptomatic tumors (Table 1). The most common location for all cNFs was the trunk (47.17%) (Table 1, Fig. 1). Head and neck regions were more frequently associated with asymptomatic tumors compared to symptomatic tumors (81.8% vs. 18.2% for frequency of asymptomatic and symptomatic tumors in head and neck region; Fig. 1f).

Table 1.

Demographic data of cutaneous neurofibromas (n = 53).

	Overall	Asymptomatic	Symptomatic	Pain + Itch	Pain	Itch
Number of tumor specimens (n, %)	53	31 (58.49%)	22 (41.51%)	5 (9.43%)	10 (18.87%)	7 (13.21%)
Sex (M:F)/ M%	29:24 (54.72%)	15:16 (48.39%)	14:8 (63.64%)	4:1 (80%)	7:3 (70%)	3:4 (42.86%)
Age (years)	57.43 ± 1.86	58.68 ± 2.12	55.68 ± 3.38	47.40 ± 6.82	58.40 ± 3.78	57.71 ± 7.9
Tumor size (cm3)	0.42 ± 0.07	0.28 ± 0.05	0.61 ± 0.13	0.51 ± 0.23	0.79 ± 0.24	0.44 ± 0.16
Tumor depth (µm)	290.5 ± 54.95	195.2 ± 15.58	424.7 ± 126.7	799.7 ± 537.4	311.2 ± 72.53	318.9 ± 78.30
Clinical variants
Flat	4 (7.54%)	1 (3.23%)	3 (13.64%)	1 (20%)	1 (10%)	1 (14.29%)
Sessile	22 (41.51%)	14 (45.16%)	8 (36.36%)	2 (40%)	4 (40%)	2 (28.56%)
Globular	20 (37.74%)	12 (38.71%)	8 (36.36%)	2 (40%)	3 (30%)	3 (42.86%)
Pedunculated	7 (13.21%)	4 (12.90%)	3 (13.64%)	0 (0%)	2 (20%)	1 (14.29%)
Distribution
Head and neck	11 (20.76%)	9 (29.03%)	2 (9.09%)	0 (0%)	2 (20%)	0 (0%)
Trunk	25 (47.17%)	12 (38.71%)	13 (59.09%)	2 (40%)	6 (60%)	5 (71.43%)
Upper limb	14 (26.42%)	9 (29.03%)	5 (22.73%)	3 (60%)	0 (0%)	2 (28.57%)
Lower limb	3 (5.66%)	1 (3.23%)	2 (9.09%)	0 (0%)	2 (20%)	0 (0%)

The demographic data for cNF tumors were recorded, including sex, age at recruitment, tumor characteristics such as clinical variants, tumor depth, tumor size, and distribution.

M male, F female.

Fig. 1. Distribution of cNFs in various body parts.

a–e Distribution of tumors was categorised by anatomic sites, including head and neck, trunk, upper limb, and lower limb in asymptomatic (n = 31), symptomatic (n = 22), both painful and itchy (n = 5), painful (n = 10), and itchy (n = 7) cNFs. f Frequency of symptomatic and asymptomatic cNFs in different anatomic areas. Created in BioRender. Pham, Q. (2025) https://BioRender.com/i89y356.

Associations of tumor size and depth with tumor pain and itch

To examine the association between common pathological markers and clinical manifestations, we compared tumor depth and size between symptomatic and asymptomatic tumors (Fig. 2a–c). cNFs relating to pain sensation were larger and deeper than asymptomatic cNFs (Fig. 2b, c). The average tumor size and depth of all cNFs were 0.42 ± 0.07 cm³ and 290.5 ± 54.95 µm, respectively. Particularly, symptomatic (424.7 ± 126.7 µm, p = 0.04), both painful and itchy (799.7 ± 537.4 µm, p = 0.005), painful (311.2 ± 72.53 µm, p = 0.02), and itchy (318.9 ± 78.30 µm, p = 0.02) tumors were located more deeply when compared with asymptomatic tumors (195.2 ± 15.58 µm) (Fig. 2b). Asymptomatic tumors were significantly smaller than painful tumors (0.28 ± 0.05 cm³ vs. 0.79 ± 0.24 cm³ for asymptomatic and painful groups, respectively, p = 0.004) (Fig. 2c).

Fig. 2. Characteristics of symptomatic and asymptomatic cNFs.

a Positive S100 staining was observed in all cNF specimens. Scale bar: 1000 µm. Tumor depth (shown in the red line in a) is the distance between the upper border of the tumor and the basement membrane of the skin epidermis layer. b, c Tumor depth and size were examined in asymptomatic (n = 31), symptomatic (n = 22), both painful and itchy (n = 5), painful (n = 10), and itchy (n = 7) tumors. Data were presented as mean ± S.E.M with error bars. The difference was evaluated by a two-sided unpaired T-test, ns: not significant, *p  <  0.05, **p < 0.01. d, e The correlation between characteristics of pain and itch, as well as tumor properties (tumor depth and tumor size), was shown. The correlation was significant when p < 0.05, Pearson’s correlation test. Asymptomatic, Asymp; Symptomatic, Symp; Both painful and itchy, Both; Painful, Pain; Itchy, Itch.

Tumor depth was correlated with specific characteristics of pain and itch. For pain characteristics, we found a positive correlation between tumor depth and several pain characteristics, such as shooting (r = 0.28, p = 0.04), stabbing (r = 0.62, p < 0.0001), sharp (r = 0.65, p < 0.0001), cramping (r = 0.64, p < 0.0001), and gnawing (r = 0.61, p < 0.0001) (Fig. 2d). Moreover, tumor depth was moderately to strongly correlated with several dimensions of itch, including intensity (r = 0.59, p < 0.0001), neuropathic itch (r = 0.51, p = 0.0001), affect (r = 0.42, p = 0.002), urge to scratch (r = 0.29, p = 0.04), coping repertoire (r = 0.55, p < 0.0001), and overall itch NRS (r = 0.41, p = 0.002) (Fig. 2e). On the contrary, we did not observe a similar correlation between tumor size and all dimensions of itch characteristics (Fig. 2e). There was a positive correlation between tumor size and some characteristics of pain, including stabbing (r = 0.42, p = 0.002), sharp (r = 0.41, p = 0.002), and overall pain NRS (r = 0.34, p = 0.01) (Fig. 2d). These results suggest a potential association between pain and/or itch symptoms and the development of cNFs with various degrees of involvement.

CXCL10/CXCR3 is upregulated in cNF with pain and itch

Given the common roles of the CXCL10/CXCR3 axis in neurofibroma formation and the pathogenesis of pain and itch, we next examined the association of CXCL10/CXCR3 expression in different parts of the skin with tumor pain and itch (Fig. 3a–c). We observed an increase in the expression of CXCL10 in both the tumoral and dermal regions of symptomatic cNFs (18.74% vs. 13.99%, for tumoral parts in symptomatic and asymptomatic groups, respectively, p = 0.0005; 18.02% vs. 12.67%, for dermal parts in symptomatic and asymptomatic groups, respectively, p = 0.001), both painful and itchy cNFs (19.01% vs. 13.99%, for tumoral parts in both painful and itchy groups; and asymptomatic groups, respectively, p = 0.04; 14.05% vs. 12.67%, for dermal parts in both painful and itchy groups; and asymptomatic groups, respectively, p = 0.56), painful cNFs (17.99% vs. 13.99%, for tumoral parts in painful and asymptomatic groups, respectively, p = 0.02; 18.94% vs. 12.67%, for dermal parts in painful and asymptomatic groups, respectively, p = 0.006), and itchy cNFs (19.63% vs. 13.99%, for tumoral parts in itchy and asymptomatic groups, respectively, p = 0.01; 19.54% vs. 12.67%, for dermal parts in itchy and asymptomatic groups, respectively, p = 0.003) compared with asymptomatic ones (Fig. 3b).

Fig. 3. Expression of CXCL10/CXCR3 in cNFs.

a Immunohistochemistry staining visualised CXCL10/CXCR3 expression in symptomatic and asymptomatic cNFs, with the dermal and tumoral parts delineated by a red dashed line. Scale bar: 600 µm (all tumor), 60 µm (dermal part), and 30 µm (tumoral part). b, c The proportion of areas positive for CXCL10 or CXCR3 in symptomatic and asymptomatic cNFs in dermal and tumoral regions (n = 53). Data were presented as mean ± S.E.M with error bars. The difference was evaluated by a two-sided unpaired T-test, ns: not significant, *p  <  0.05, **p < 0.01, ***p < 0.001. d, e Specific pain and itch characteristics were associated with increased expression of CXCL10/CXCR3 (n = 53). The correlation was significant when p < 0.05, Pearson’s correlation test. Asymptomatic, Asymp; Symptomatic, Symp; Both painful and itchy, Both; Painful, Pain; Itchy, Itch.

The proportion of CXCR3⁺ areas in the dermal region was markedly higher in symptomatic cNFs (7.83% vs. 4.71%, for symptomatic and asymptomatic groups, respectively, p = 0.008), and painful cNFs (7.78% vs. 4.71%, for painful and asymptomatic groups, respectively, p = 0.04). We also observed a trend for elevated dermal CXCR3⁺ signals in both painful and itchy cNFs (8.7% vs. 4.71%, for both painful and itchy groups; and asymptomatic groups, respectively, p = 0.07), and itchy cNFs (7.28% vs. 4.71%, for itchy and asymptomatic groups, respectively, p = 0.16) than in asymptomatic tumors. Similarly, tumoral CXCR3 expression was higher in the symptomatic cNFs (0.89% vs. 0.52%, for symptomatic and asymptomatic groups, respectively, p = 0.02), both painful and itchy cNFs (0.96% vs. 0.52%, for both painful and itchy groups; and asymptomatic groups, respectively, p = 0.1), painful cNFs (0.78% vs. 0.52%, for painful and asymptomatic groups, respectively, p = 0.08), and itchy cNFs (1.01% vs. 0.52%, for itchy and asymptomatic groups, respectively, p = 0.005) than in the asymptomatic cNFs (Fig. 3c).

CXCL10/CXCR3 is associated with specific characteristics of pain and itch

Detailed characterisation of pain and itch using the SF-MPQ and CEIQ indicated that individual characteristics correlated differently with CXCL10 and CXCR3 expression in different parts of cNF. For pain characteristics, there were higher degrees of correlation with CXCR3 than with CXCL10 expression (Fig. 3d). CXCR3 expression in dermal and tumoral parts was moderately to strongly correlated with specific pain characteristics, particularly throbbing (r = 0.36, p = 0.008, and r = 0.63, p < 0.0001, for dermal and tumoral parts, respectively), gnawing (r = 0.39, p = 0.004, and r = 0.51, p < 0.0001, for dermal and tumoral parts, respectively), and tender (r = 0.29, p = 0.03, for dermal parts) (Fig. 3d). However, only the tender pain exhibited a moderate correlation (r = 0.32, p = 0.02) with dermal CXCL10 expression (Fig. 3d).

CXCL10 and CXCR3 expressions were both associated with specific itch characteristics. Tumoral CXCL10 expression was positively associated with overall itch NRS (r = 0.27, p = 0.048) and itch characteristics of painful (r = 0.32, p = 0.02), only itch feeling (r = 0.27, p = 0.048), and localised itch (r = 0.29, p = 0.04) (Fig. 3e). Meanwhile, dermal CXCL10 expression was correlated with itch characteristic as decreased when scratching (r = 0.27, p = 0.049), and the dermal CXCR3 expression was correlated with itch characteristics of itch at rest (r = 0.3, p = 0.03), satisfaction when scratching (r = 0.3, p = 0.03), ecstasy when scratching (r = 0.3, p = 0.03), and decreased when scratching (r = 0.3, p = 0.03) (Fig. 3e). We did not observe a similar correlation between tumoral CXCR3 with all itch characteristics.

Additionally, elevated CXCR3 expression was associated with large tumors. In the dermal region, CXCR3 expression showed a mild correlation with tumor size (r = 0.28, p = 0.046; Supplementary Fig. S1c), and in the tumoral region, this correlation was moderate (r = 0.4, p = 0.003; Supplementary Fig. S1d).

Decreased density of mast cells in symptomatic cNFs

Next, we looked into the possible sources of CXCL10/CXCR3 by evaluating the overall densities of mast cells, macrophages, T cells, and neutrophils which are implicated in the development of cNFs and their associated sensory symptoms [13–15]. To our surprise, the density of mast cells was significantly lower in symptomatic tumors (93.19 ± 5.85 cells/mm² vs. 126.1 ± 9.76 cells/mm², for symptomatic and asymptomatic tumors, p = 0.01), and both painful and itchy tumors (72.88 ± 17.79 cells/mm² vs. 126.1 ± 9.76 cells/mm², for both painful and itchy tumors; and asymptomatic tumors, p = 0.04); and marginally lower in painful tumors (92.10 ± 4.31 cells/mm² vs. 126.1 ± 9.76 cells/mm², for painful and asymptomatic tumors, p = 0.06), and itchy tumors (109.3 ± 9.67 cells/mm² vs. 126.1 ± 9.76 cells/mm², for itchy and asymptomatic tumors, p = 0.43) when compared with asymptomatic tumors (Fig. 4b). No significant difference in the density of CD68⁺ macrophages (Fig. 4c; 197.8 ± 17.53 cells/mm² vs. 174.7 ± 13.14 cells/mm², for symptomatic and asymptomatic tumors, p = 0.29), of CD3⁺ T cells (Supplementary Fig. S2a, b; 56.78 ± 4 cells/mm² vs. 58.11 ± 4.3 cells/mm², for symptomatic and asymptomatic tumors, p = 0.82), and of MPO⁺ neutrophils (Supplementary Fig. S2a, c; 59.94 ± 10.58 cells/mm² vs. 60.05 ± 4.67 cells/mm², for symptomatic and asymptomatic tumors, p = 0.99) was observed between symptomatic and asymptomatic tumors.

Fig. 4. Frequency of mast cells and macrophages in cNFs.

a Mast cells were visualised and analysed by toluidine blue (top). Scale bar: 60 µm and 30 µm (inset). Macrophages were presented in cNF tumors by CD68 staining (bottom). Scale bar: 50 µm and 10 µm (inset). b, c Number of mast cells in asymptomatic (n = 31), symptomatic (n = 22), both painful and itchy (n = 5), painful (n = 10), and itchy (n = 7) tumors and CD68⁺ macrophages in asymptomatic (n = 30), symptomatic (n = 22), both painful and itchy (n = 5), painful (n = 10), and itchy (n = 7) tumors, per mm² was calculated and compared between groups. Data were presented as mean ± S.E.M with error bars. The difference was evaluated by a two-sided unpaired T-test, ns: not significant, *p  <  0.05. Asymptomatic, Asymp; Symptomatic, Symp; Both painful and itchy, Both; Painful, Pain; Itchy, Itch.

Increased CXCL10 expression in mast cells for symptomatic cNFs

We examined CXCL10/CXCR3 expression in immune cells to investigate their potential association with symptomatic cNFs. We evaluated the degree of colocalization of CXCL10/CXCR3 with markers of mast cells, macrophages, T cells, and neutrophils separately. The percentage and density of CXCL10-expressing mast cells, as visualised by tryptase, were significantly higher in symptomatic tumors than in asymptomatic tumors. Specifically, CXCL10⁺ Tryp⁺ mast cells constituted 51.18% in symptomatic tumors, 35.55% in both painful and itchy tumors, 54.62% in painful tumors, and 57.42% in itchy tumors compared with 19.07% in asymptomatic tumors (p < 0.0001, p = 0.007, p < 0.0001, and p < 0.0001, respectively; Fig. 5c). Additionally, the absolute number of mast cells expressing CXCL10 was increased in symptomatic cNFs (36.42 ± 3.64 cells/mm² vs. 18.13 ± 2.07 cells/mm², for symptomatic and asymptomatic groups, respectively, p < 0.0001), painful cNFs (40.48 ± 5.45 cells/mm² vs. 18.13 ± 2.07 cells/mm², for painful and asymptomatic groups, respectively, p < 0.0001), and itchy cNFs (38.46 ± 6.04 cells/mm² vs. 18.13 ± 2.07 cells/mm², for itchy and asymptomatic groups, respectively, p = 0.0003). However, a significant difference in the proportion of mast cells expressing CXCR3 was not found among these groups: symptomatic cNFs (66.14% vs. 63.42%, for symptomatic and asymptomatic groups, respectively, p = 0.57), painful cNFs (71.14% vs. 63.42%, for painful and asymptomatic groups, respectively, p = 0.16), and itchy cNFs (72.45% vs. 63.42%, for itchy and asymptomatic groups, respectively, p = 0.11) (Fig. 5d).

Fig. 5. Expression of CXCL10/CXCR3 on mast cells in the tumoral region.

a, b Double immunofluorescence staining was performed to evaluate the expression of CXCL10/CXCR3 (green) on Tryp⁺ mast cells (red) in asymptomatic (n = 29), symptomatic (n = 22), both painful and itchy (n = 5), painful (n = 10), and itchy (n = 7) cNFs. Scale bar: 30 µm. c, d Proportion and number of mast cells expressing CXCL10 or CXCR3 in symptomatic and asymptomatic cNFs. Data were presented as mean ± S.E.M with error bars. The difference was evaluated by a two-sided unpaired T-test, ns: not significant, *p  <  0.05, **p  <  0.01, ***p < 0.001, ****p < 0.0001. Asymptomatic, Asymp; Symptomatic, Symp; Both painful and itchy, Both; Painful, Pain; Itchy, Itch.

In other immune cells, we observed no difference in the percentage of CXCL10-expressing macrophages, as identified by CD68⁺ cells, between asymptomatic and symptomatic cNFs (25.83% vs. 27.43%, p = 0.72; Supplementary Fig. S3a, b). Furthermore, the percentage of CXCR3-expressing macrophages did not significantly differ between asymptomatic and symptomatic tumors (15.45% vs. 15.16%, p = 0.93; Supplementary Fig. S3a, c). Similarly, in asymptomatic and symptomatic tumors, the percentage of CXCL10- and CXCR3-expressing T cells, identified as CD3⁺ cells, showed no significant difference (36.47% vs. 38.85%, for CXCL10-expressing T cells in asymptomatic and symptomatic tumors, p = 0.6; and 22.76% vs. 21.91%, for CXCR3-expressing T cells in asymptomatic and symptomatic tumors, p = 0.83) (Supplementary Fig. S3d–f). Additionally, the proportion of MPO⁺ neutrophils expressing CXCL10 and CXCR3 was not remarkably different between asymptomatic and symptomatic cNFs (25.57% vs. 31.35%, for CXCL10-expressing neutrophils in asymptomatic and symptomatic tumors, p = 0.26; and 25.58% vs. 26.15%, for CXCR3-expressing neutrophils in asymptomatic and symptomatic tumors, p = 0.92) (Supplementary Fig. S3g–i). These results suggested that increased expression of CXCL10 by the mast cells may be associated with itch and pain in cNFs.

Increased intraepidermal nerve fiber density in symptomatic cNFs

To investigate the association between tumor-related pain and itch and the intraepidermal nerve fiber (IENF), we measured the density of IENF in cNFs. All IENFs were visualised using antibody targeting PGP9.5, a widely used pan-axonal marker [40]. The density of IENF was significantly higher in symptomatic tumors (26.98 ± 1.14 nerves/mm vs. 22.54 ± 1.11 nerves/mm, for symptomatic and asymptomatic tumors, respectively, p = 0.009), both painful and itchy tumors (30.51 ± 2.79 nerves/mm vs. 22.54 ± 1.11 nerves/mm, for both painful and itchy tumors; and asymptomatic tumors, respectively, p = 0.01), painful tumors (28.21 ± 1.34 nerves/mm vs. 22.54 ± 1.11 nerves/mm, for painful and asymptomatic tumors, respectively, p = 0.01), but not significantly different in itchy tumors (22.71 ± 1.39 nerves/mm vs. 22.54 ± 1.11 nerves/mm, for itchy and asymptomatic tumors, respectively, p = 0.94) when compared with asymptomatic tumors (Fig. 6a, b). Additionally, although not statistically significant, there was a trend for increased nerve branching in symptomatic cNFs than in asymptomatic cNFs (Fig. 6c).

Fig. 6. Intraepidermal nerve fiber (IENF) density in symptomatic and asymptomatic cNFs.

a PGP9.5 staining revealed different IENF densities (IENFs/mm) in asymptomatic (n = 31), symptomatic (n = 22), both painful and itchy (n = 5), painful (n = 10), and itchy (n = 7) cNFs. Scale bar: 100 µm. b Difference in IENF density between asymptomatic and symptomatic cNFs (n = 53). c The frequency of branching in symptomatic tumors (n = 22, 4.25 ± 0.85 branches/mm), both painful and itchy tumors (n = 5, 5.64 ± 2.58 branches/mm), painful tumors (n = 10, 4.1 ± 1.08 branches/mm), and itchy tumors (n = 7, 3.48 ± 1.36 branches/mm), and asymptomatic tumors (n = 31, 3.6 ± 0.66 branches/mm) was shown. Data were presented as mean ± S.E.M with error bars. The difference was evaluated by a two-sided unpaired T-test, ns: not significant, *p  <  0.05, **p  <  0.01. Asymptomatic, Asymp; Symptomatic, Symp; Both painful and itchy, Both; Painful, Pain; Itchy, Itch.

Discussion

Cutaneous neurofibromas and their associated sensory symptoms substantially affect the quality of life [8]. Here, we analysed 53 human cNF tumor specimens. Our study investigated the characteristics and key molecules underlying pain and itch in cNFs. We evaluated specific clinical and pathological features in symptomatic tumors, highlighting the potential roles of these factors and proposing a model for the activity of the CXCL10/CXCR3 axis in cNF presenting with pain and itch, as summarised in Graphical abstract.

Our findings identified distinct characteristics of symptomatic cNFs, including their clinical phenotypes, distribution, size, and depth. The cervical region exhibited a higher propensity to the development of asymptomatic tumors than that of symptomatic tumors (Fig. 1f). Such propensity was observed in the mouse model of neurofibromatosis 1, suggesting some shared features in human and mouse neurofibroma [41]. We observed an association between larger or deeper cNFs and an increased prevalence of tumors relating to pain (Fig. 2). Additionally, tumor depth was found to be positively correlated with all major categories of itching and some pain characteristics (Fig. 2d, e). Tumor size exhibited a moderate positive correlation with dermal and tumoral CXCR3 expression (Supplementary Fig. S1c, d). Clinically, tumor size and CXCR3 expression in dermal and tumoral regions exhibited a certain of moderate to strong association with some painful sensations (Figs. 2d and 3d). These observations are consistent with a common role of the CXCL10/CXCR3 pathway in neurofibroma development as well as pain and itch signaling [20, 42].

Our study revealed an association of CXCL10 and CXCR3 signaling with pain and itch symptoms in cNFs. We noted elevated levels of CXCL10 and CXCR3 expression in the dermal and tumoral regions of symptomatic cNFs compared with asymptomatic cNFs (Fig. 3a–c). The CXCL10/CXCR3 signaling pathway is among the critical pathways associated with pain and itch transmission [27, 28, 42]. CXCL10 contributes to itch and pain symptoms in neuroinflammatory conditions by directly stimulating sensory neurons expressing the CXCR3 receptor or indirectly recruiting immune cells, such as neutrophils and mast cells [21, 27, 28]. In our study, mast cells and macrophages, which have been proposed to contribute to itch or pain in cNFs [13–15], exhibited no significant increase in the number in symptomatic cNFs (Fig. 4).

Mast cells are commonly observed in cNF as one of its distinct pathological features, yet their roles in tumorigenesis and associated symptoms remain debated. Our study suggested that mast cells with derived CXCL10 are associated with pain and itch in cNFs. Previous studies have observed mast cell accumulation in itchy, actively growing cNFs [16], while other studies indicated a bystander role of mast cells in neurofibroma [41]. We noted a significantly higher proportion and absolute counts of mast cells expressing CXCL10 in symptomatic cNFs than in asymptomatic cNFs (Fig. 5c). Intriguingly, there was no difference in the proportion of mast cells expressing CXCR3 between symptomatic and asymptomatic cNFs (Fig. 5d), and a lower mast cell density in symptomatic cNFs was observed (Fig. 4b). Since CXCR3 is the only known functional receptor for CXCL10 [43], it is likely secreted by mast cells in our cNFs rather than being bound to their receptor from an external source. Moreover, macrophages, T cells, and neutrophils are abundant in the immune cell profiles of neurofibroma [25, 44], but we did not find a difference in CXCL10 or CXCR3 expression in these cells between asymptomatic and symptomatic cNFs (Fig. S3). These observations imply that a potential involvement of mast cells in the pain and itch associated with CXCL10/CXCR3 upregulation in cNFs. Thus, specifically targeting CXCL10/CXCR3 activity could enhance the effectiveness of anti-mast cell therapies in the treatment of cNF.

The involvement of CXCL10/CXCR3 in symptomatic cNFs may be linked to their effect on IENF, which is associated with dysfunction in pain and itch perception in neuropathic disorders [45–47] and malignant diseases [48]. A recent study observed skin hyperinnervation in the early stage of cutaneous T cell lymphomas, which was associated with itch establishment and mast cell accumulation [48]. Additionally, the CXCL10/CXCR3 axis and mast cells can interact with skin sensory neurons in pain and itch transmission [27, 28, 49]. Consistent with these findings, we observed a significant increase in IENF density in symptomatic cNFs compared with asymptomatic cNFs (Fig. 6b).

This study has some limitations. The limited number of cNF samples, particularly those associated with both pain and itch, restricted our ability to fully characterise symptomatic cNFs. For instance, we observed a tendency for increased tumor size (Fig. 2c) and dermal CXCR3 expression (Fig. 3c) in both painful and itchy groups (denoted as “both”) compared with the asymptomatic groups. However, the differences were not statistically significant, which might be due to the small sample size. With further recruitment of cNF patients in the future, we may better understand the differences in the groups with small sample sizes. Furthermore, the challenges in simultaneously visualising IENF and CXCR3 to demonstrate their co-localisation in the epidermis limit our investigation of their correlation. Additional research is required to identify particular cellular pathways implicated in the activity of CXCL10/CXCR3 throughout the development and manifestation of symptoms in cNF.

In summary, we propose a model in which CXCL10/CXCR3 contributes to the induction of pain and itch in symptomatic cNFs, potentially mediated by mast cells. Specifically, mast cells may enhance CXCL10 secretion, thereby activating sensory neurons expressing CXCR3 to induce pain and itch. Further exploration of CXCL10/CXCR3 signaling and its pathways in the cNF could reveal effective treatments for patients, particularly in mitigating pain and itch symptoms.

Author contributions

HJW and TTQP designed the study and performed experiments, analysed the data, wrote the manuscript, and prepared the figures. CPL, WRL, KHL, TSY, YHJS, YHL and HJW supervised the study, discussed, and contributed to the manuscript. HJW, WRL, CLC, YHS, YWC, DL, HCW, and BJC recruited the patients and processed the human specimens. All authors reviewed and approved the final revised version.

Data availability

All data generated in this study are accessible from the corresponding author upon reasonable inquiry.

Competing interests

The authors declared no competing interests.

Ethics approval and consent to participate

This study was approved by the Research Ethics Committee of Taipei Medical University (TMU-REC No. N202103158). Patients were recruited at the Department of Dermatology, Taipei Medical University-Shuang Ho Hospital. The recruitment of patients and collection of tumor specimens adhered to the principles of the Declaration of Helsinki. All participants provided written informed consent.

Supplementary information

The online version contains supplementary material available at 10.1038/s41416-025-02956-z.