Results
Epidemiological and Clinical Data
A total of 119 granular cell tumours were studied in 114 patients. Sex ratio was 0.84, with 52 men and 62 women. The mean age (±SD) of patients at diagnosis was 43.7 years (±18; range 5–76). Eight patients (7%) had a history of multiple granular cell tumours (five men, three women). The mean age of patients with multiple granular cell tumours was 37.6 years (±14.1; range 26–67). The mean largest diameter of the tumour was 1.48 cm (±1.3; range 0.4–5 cm).
Mucosal tumours were not included. The distribution of the tumours was as follows: trunk 39%, upper limbs (except palms) 20%, head and neck 15%, genital area 12%, lower limbs (except soles) 9%, palms and soles 5%.
Evaluation of Benignity
As mentioned in Weiss and Goldblum, we considered a criterion of malignancy as present if it was prominent in the tumour, and not simply a focal change, that is, when the finding was present in more than 50% of the tumour surface.
Only eight lesions were classified as atypical, with one positive criterion (nuclear pleomorphism, cellular spindling or mitotic activity ≥2 per 10 high power fields) (figure 1). No tumour had two or more positive criteria. Therefore no lesion was classified as malignant and 111 lesions were classified as benign. Pustulo-ovoid bodies were found in 82 tumours (69%).
(Enlarge Image)
Figure 1.
Atypical features in granular cell tumours of the skin. (A) Mitotic activity; (B) spindling of the tumour cells; (C) pleomorphism (Haematoxylin, eosin, safron (HES) staining, ×400 magnification). Access the article online to view this figure in colour.
In the 25 patients with follow-up data, no local recurrence occurred in 24 patients with a median follow-up of 6 months (range: 2 to 24 months). One patient showed local recurrence 6 months after the initial resection. No metastases and no deaths were recorded during available follow-up.
Histopathological Invasive Features
Mean depth of the tumours (Breslow index) was 8.8 mm (±4.7 mm). Most of the tumours infiltrated the subcutaneous tissue (79/119; 66%) (figure 2A). The others reached the reticular dermis (3/119; 3%), or filled the reticular dermis without subcutis involvement (37/119; 31%).
(Enlarge Image)
Figure 2.
(A) Granular cell tumour of the skin infiltrating subcutaneous tissue (HES staining, ×25 magnification); (B) lymphoid aggregate at the deep edge of a granular cell tumour (HES staining, ×200 magnification). Access the article online to view this figure in colour.
The tumours showed an invasive peripheral growth pattern in 71% of cases, and were well limited in 29%.
We then examined the inflammatory infiltrate accompanying the tumour cells. It was predominantly lymphocytic, organised in intra-tumoural or peripheral aggregates. Intra-tumoural lymphoid aggregates were found in 10 tumours (8%). Peripheral lymphoid aggregates were found in 58 tumours (49%), mainly at the deep border of the tumour (figure 2B). Peripheral lymphoid aggregates were found in 65% (95% CI 43% to 83%) of the lesions on the upper limbs, 57% (95% CI 43% to 74%) on the trunk, 50% (95% CI 24% to 76%) on the lower limbs, 29% (95% CI 8% to 53%) on the head and neck area, 23% (95% CI 7% to 57%) on the genital area, and 0% (95% CI 0% to 53%) on palms and soles. The association between the presence of peripheral lymphoid aggregates and tumour localisation was statistically significant (p<0.05).
Arrector pili smooth muscles were infiltrated by tumour cells in 27 cases (23%, 95% CI 16% to 32%). This finding was confirmed with immunohistochemistry for smooth muscle actin and S100 protein (figure 3). No statistical association was found between arrector pili muscle infiltration and tumour localisation.
(Enlarge Image)
Figure 3.
Arrector pili muscle infiltration by granular cells in granular cell tumour. (A) HES staining, ×200 magnification; (B) S100 protein immunostaining, ×200 magnification; (C) smooth muscle actin immunostaining, ×200 magnification. Access the article online to view this figure in colour.
Perineural spread was a common finding, being present in 78 tumours (66%, 95% CI 56% to 74%). In 30 tumours, perineural spread was diffuse and prominent (figure 4). Intra-tumoural nerves with diameter >0.2 mm were found in 25 cases (21%). Perineural spread was found in 78% (95% CI 56% to 92%) of the lesions on upper limbs, 71% (95% CI 56% to 92%) on the head and neck area, 69% (95% CI 53% to 82%) on the trunk, 66% (95% CI 17% to 93%) on palms and soles, 50% (95% CI 24% to 76%) on the lower limbs, and 46% (95% CI 25% to 75%) on the genital area. The association between perineural spread and tumour localisation was not statistically significant.
(Enlarge Image)
Figure 4.
Perineural spread of granular cells in granular cell tumour. (A) HES staining, ×200 magnification; (B) S100 protein immunostaining, ×400 magnification. Access the article online to view this figure in colour.
Finally, vascular invasion of small or medium-sized vessels was found in 27 cases (23%, 95% CI 16% to 32%). In 17 cases, vascular invasion consisted of an infiltration of the subendothelial layers of a medium-sized vessel by tumour cells. Such mural infiltration was found in medium-sized veins and arteries with muscular wall, in the deep dermis (figure 5). In 14 cases, vascular obliteration was observed in small to medium-sized vessels, this finding being easier to detect with orcein staining (figure 6). The circular arrangement of the elastic fibres made it easy to recognise the obliterated, fully plugged, vascular structure. The blood or lymphatic nature of the fully plugged vessels could not be determined on the basis of the morphological findings. Three cases showed both vessels with subendothelial infiltration and obliterated vessels. No intravascular permeation (free-floating tumour cells in the lumen) was observed.
(Enlarge Image)
Figure 5.
(A) Infiltration of the vessel wall of a medium-sized artery in the deep dermis by granular cells (HES staining, ×100 magnification). S100 protein (B) and smooth muscle actin (D) immunostaining shows the granular cells in the media and subendothelium (×100 magnification). CD34 immunostaining (C) shows that there is no intraluminal permeation by granular cells (×100 magnification). Access the article online to view this figure in colour.
(Enlarge Image)
Figure 6.
(A) Small vessel vascular obliteration by granular cells, with orcein staining (×50 magnification). The circular arrangement of the elastic fibres made it easy to recognise the obliterated vascular structure. (B) At higher magnification in this case, there is an inconspicuous residual vascular lumen (orcein staining, ×400 magnification). (C) Subendothelial infiltration by granular cells with residual vascular lumen (orcein staining, ×200 magnification). (D) In this case, small vessel obliteration by granular cells leaves no residual lumen (orcein staining, ×200 magnification). Access the article online to view this figure in colour.
Patients with vascular invasion were significantly older than patients without vascular invasion (51.2±17.6 years vs 41.3±17.5 years; p<0.05). Vascular invasion was not statistically associated with the presence of the criteria of Fanburg-Smith. It was equally frequent in cases with single or multiple tumours. Vascular invasion occurred in 60% (95% CI 17% to 93%) of cases on palms and soles, 42% (95% CI 16% to 71%) in the genital area, 30% (95% CI 8% to 65%) on the lower limbs, 24% (95% CI 13% to 40%) on the trunk, 22% (95% CI 8% to 44%) on the upper limbs and 6% (95% CI 3% to 32%) on the head and neck. Distribution of the cases with vascular invasion was not different from cases without vascular invasion. Cases with vascular invasion tended to be larger (1.56±0.78 cm vs 1.34±0.86 cm), thicker (10±6 mm vs 8.4±4.2 mm), to have more perineural spread (78% vs 62%), and to be more associated with arrector pili muscle invasion (30% vs 19%) than cases without vascular invasion, but the difference was not statistically significant. Lymphoid aggregates, peripheral invasive growth pattern and atypical cases were equally frequent in cases with and without vascular invasion.
Among the 25 patients with follow-up data, the only tumour with local recurrence did not show vascular invasion, atypia or lymphoid aggregates, but showed prominent perineural spread. In the 24 patients with no recurrence, 19 tumours showed perineural spread, 3 were atypical, 14 showed lymphoid aggregates and 5 showed vascular invasion. No significant association was found between outcome and histopathological findings in the 25 patients.