Prognostic Significance of Epidermal Growth Factor Receptor Expression in Distant Metastatic Melanoma from Primary Cutaneous Melanoma.

BACKGROUND: Epidermal growth factor receptor (EGFR) is overexpressed in many cancers. However, EGFR expression in melanoma and its role are conflicting.
OBJECTIVE: This study aimed to evaluate EGFR expression in distant metastatic melanoma and analyze its relationship with histologic and clinical characteristics and survival.
METHODS: Diagnostic tissues from 55 cases of distant metastatic melanoma was evaluated by immunohistochemistry for EGFR expression. Clinicopathologic features and survival outcomes were analyzed according to EGFR expression.
RESULTS: The positive EGFR expression in distant metastatic melanoma was significantly correlated with the absence of ulceration. The EGFR expression in distant metastatic melanoma was significantly associated with poor survival, under the conditions of male sex and primary cutaneous melanoma without ulceration or Breslow thickness ≤4.0 mm. This study bears limitations of a retrospective study in a single institution.
CONCLUSION: EGFR immunostaining had predictive values for survival outcome. The EGFR expression in distant metastatic melanoma in male, no ulcer, or Breslow thickness ≤4.0 mm appeared to be involved in disease progression.

INTRODUCTION

Epidermal growth factor receptor (EGFR) is one of the transmembrane type I receptor tyrosine kinases, which belongs to the ErbB/HER protein family1. It is a major chemoattractant for invading cancer cells and is widely known to stimulate cell proliferation, angiogenesis, differentiation, migration, survival, and adhesion by downstream complex signaling pathways2. When ligand binds to the EGFR, the receptor undergoes dimerization, which results in receptor tyrosine kinase autophosphorylation, channelling of predominant mitogenic signals, and various signaling pathway activation, including the most significant Ras/Raf/MEK/extracellular signal-regulated kinase (ERK) pathway13. The protein kinase cascade from RAF to MEK to ERK provides opportunities for feedback regulation and signal amplification4. Signal transduction participates in regulation of cell proliferation, prevents apoptosis and promotes cell invasion, initiates actin polymerization, and begins microfilament reorganization, and these processes are essential for cell migration3. The actin-rich adhesive structures secrete proteases that digest extracellular matrix (ECM) elements, making the path for cancer cells to migrate through the surrounding microenvironment5. It generates a protrusive force that allows cancer cells to form invadopodia, penetrate through ECM, and metastasize6.

Deregulation of these cascades by activated EGFR is implicated in oncogenesis, because EGFR is a critical protein for the proliferation of cells7. EGFR is often overexpressed and dysregulated in several human malignancies, including breast cancer, epithelial gastrointestinal malignancies, gliomas, non-small cell lung cancer, colorectal cancer, and head and neck squamous cell carcinoma; and its overexpression is associated with tumor progression and poor outcome prognosis8. Nowadays, EGFR is used as a target antigen for specific anticancer therapies in many malignancies9. Correspondingly, blocking the EGFR signaling pathway is well known to suppress growth and metastasis of certain types of lung cancer10.

Melanoma was one of the first cancers wherein EGFR protein expression was suggested as a probable metastatic marker11. Melanoma expresses several receptor tyrosine kinases, including EGFR family members; however, their roles have been controversial12. Many authors thought that EGFR expression could be indicative of cell maturation or of cellular proliferative capacity13. EGFR can be expressed in melanoma cells but not in normal melanocytes. The increasing EGFR protein expression turned out that it was linked with the progression of melanocytic lesions in most studies14. EGFR overexpression often occurs in the advanced stage of melanoma15. De Wit et al.16 identified significant differential EGFR expression at various stages of melanocyte tumor progression involving 61% of dysplastic nevus, 89% of primary cutaneous melanoma, and 91% of metastatic melanoma. In addition, the staining intensity was stronger in malignant lesions than in benign lesions. Furthermore, Real et al.17 found EGFR (+) only in undifferentiated melanoma cells and EGFR (−) in differentiated cell lines. However, reported results about the correlation between EGFR expression and prognosis in malignant melanoma are rather inconsistent, and there is no clear evidence about prognostic and diagnostic relevance18. Several studies reported about low EGFR expression in human melanocytes and melanoma11. Another study showed that EGFR expression did not correlate with the proliferation in melanoma cell panel. EGFR expression in the tissues also had no significant association with the proliferative state of cells17.

Although little data exists to guide management of metastatic melanoma, the current first-line standard treatments are PD-1 blockade (nivolumab, pembrolizumab) with or without CTLA-4 blockade (ipilimumab). In addition, BRAF inhibition (vemurafenib, dabrafenib, encorafenib) combined with MEK inhibition (cobimetinib, trametinib, binimetinib) are recommended for BRAF V600-mutated metastatic melanoma19.

There is a lack of prior researches on the correlation between EGFR expression and metastatic melanoma, and the same goes for the controversial role of EGFR in melanoma. Thus, this study aims to evaluate EGFR expression in distant metastatic melanoma and analyze its relationship with histopathologic and clinical characteristics, as well as survival outcomes.

MATERIALS AND METHODS

The database of the Asan Medical Center in Korea was investigated for cutaneous melanoma patients with distant metastasis confirmed through biopsy between January 2000 and December 2017 after receiving approval from the Institutional Review Board of the Asan Medical Center (no. 2018-1248). Patients with only in-transit, satellite, or microsatellite metastases were excluded. In-transit metastases are formally classified when these lesions are located between the primary site of the tumor and the locoregional lymph nodes20.

Variables of interest

Age at diagnosis, sex, primary site of melanoma, ulceration, lymph node involvement and location of distant metastasis, follow-up results, and survival rate were collected from medical records and clinical photographs as clinical data. The researchers analyzed H&E stain sections of the primary cutaneous melanoma tissue, reviewing the Breslow thickness and Clark's levels, as well as two intervening sections for EGFR and BRAF stains as pathologic data. Overall survival (OS) data were calculated from the date of initial diagnosis to the date of death because of any cause, or the patient's last follow-up examination.

Statistical analysis

Clinicopathologic comparison between the EGFR (+) and EGFR (−) groups was performed using statistical methods of analysis. Chi-square test or Fisher's exact test was used when the dependent variable was categorical, depending on whether the number of cells whose expected frequency was less than 5 is less than 25% of the total. When the dependent variable was continuous, the independent t-test or Mann—Whitney U test was used, depending on whether normality was satisfied. Survival analysis was conducted using the Kaplan—Meier method, and the differences in survival between the subgroups were compared. All analyses were performed using Window SPSS ver. 21.0 (IBM Corp., Armonk, NY, USA). p-values of <0.05 were considered to indicate statistical significance.

RESULTS

Clinicopathologic features of patients with distant metastatic melanoma

A total of 55 cases of distant metastatic melanoma from primary cutaneous melanoma were included in this retrospective review between January 2000 and December 2017. The demographic data and clinical features of these cases are summarized in Table 1. Among the patients, 30 (54.5%) were male, and 25 (45.5%) were female. The mean age at the time of diagnosis was 60.0 years old. EGFR immunohistochemical stain of metastatic melanoma tissues was positive in 21 (38.2%) patients (Fig. 1). BRAF immunohistochemical stain was positive in 8 (14.5%) patients. The frequency of primary cutaneous melanoma location was highest in the acral area (n=24, 43.6%), followed by the extremities (n=14, 25.5%), trunk area (n=7, 12.7%), head and neck area (n=7, 12.7%), and inguinal area (n=3, 5.5%) (Table 1). This is the result in line with the previous study revealing cutaneous melanoma in Korea is most common in acral area21. Furthermore, 12 cases (21.8%) had ulcerations on the cutaneous lesion. The Breslow thickness of 37 cases (67.3%) was >4.0 mm, which indicates T4 stage in the American Joint Committee on Cancer (AJCC) melanoma staging system. In the aspect of Clark's level of invasion, 47 cases (85.5%) were level 4 or 5. A total of 35 patients had sentinel lymph node (SLN) biopsy, of which 77.1% showed positive melanoma involvement in the SLN. Moreover, 38 cases (33.3%) had metastasis in different skin areas from the primary site, and the most common location of metastasis, except for the skin, was lungs (n=26, 22.8%) (Table 1). The average follow-up time of the patients was 38.3 months. In summary, a total of 45 patients (81.8%) died during the observation period.

Table 1

Clinicopathologic features of patients with metastatic melanoma

Clinicopathological feature		Number (%)	Mean (SD)
Sex
	Male	30 (54.5)
	Female	25 (45.5)
Age at time of diagnosis (yr)			60.0 (11.7)
EGFR
	Negative	34 (61.8)
	Positive	21 (38.2)
BRAF
	Negative	47 (85.5)
	Positive	8 (14.5)
Location of primary melanoma
	Acral	24 (43.6)
	Extremities	14 (25.5)
	Head and neck	7 (12.7)
	Trunk	7 (12.7)
	Inguinal area	3 (5.5)
Ulceration
	Absent	43 (78.2)
	Present	12 (21.8)
Breslow thickness (mm)
	≤4.0	18 (32.7)
	>4.0	37 (67.3)
Clark level
	3	8 (14.5)
	4	26 (47.3)
	5	21 (38.2)
Sentinel lymph node involvement
	Negative	8 (22.9)
	Positive	27 (77.1)
Location of metastasis
	Skin	38 (33.3)
	Lung	26 (22.8)
	Brain	14 (12.3)
	Liver	11 (9.6)
	Bone	11 (9.6)
	Digestive system	9 (7.9)
	Spine	5 (4.4)
Follow-up duration (mo)			38.3 (31.1)
Mortality
	Dead	45 (81.8)
	Alive	10 (18.2)

SD: standard deviation, EGFR: epidermal growth factor receptor.

Fig. 1

Epidermal growth factor receptor (EGFR) immunostaining in melanoma sections. (A, B) EGFR protein is strongly expressed on the melanoma section of the labia minora of a 63-year-old female patient. (C,D) EGFR protein is not expressed on the melanoma section of the left 2nd toe of a 61-year-old female patient. Original magnification: (A) ×100, (B) ×200, (C) ×100, (D) ×200.

Association between EGFR expression and clinicopathologic variables

The clinicopathologic variables of distant metastatic melanoma were stratified depending on the expression of EGFR in tumor tissues to assess their associations (Table 2). There were significant correlations between EGFR expression and absence of ulceration (p=0.019). Among 21 patients with EGFR (+) metastatic melanoma, only one patient had a skin lesion with ulceration. The BRAF (+) rate was higher in the EGFR (+) group compared with the EGFR (−) group, but there was no statistically significant difference (p=0.236). The proportion of patients with Breslow thickness >4.0 mm of melanoma was lower in the EGFR (+) group than in the EGFR (−) group; however, no statistical significance was found (p=0.064). Any other significant association with EGFR expression was not found in the other factors, including SLN involvement. A comparison between EGFR expression and metastatic location was also not statistically meaningful, although the ratio of brain metastasis in thee EGFR (+) group was slightly higher than that in the EGFR (−) group (p=0.091) (Table 3).

Table 2

Epidermal growth factor receptor (EGFR) expression and clinicopathologic parameters

Variable		EGFR		p-value
		Negative (n=34)	Positive (n=21)
Sex				0.389†
	Female	17	8
	Male	17	13
Age (yr)		59.3±12.0	61.1±11.3	0.583∥
BRAF				0.236§
	Negative	31	16
	Positive	3	5
Location of primary melanoma				0.554†
	Acral	17	7
	Extremities	9	5
	Head and neck	4	3
	Trunk	3	4
	Inguinal area	1	2
Ulceration				0.019*§
	Absent	23	20
	Present	11	1
Breslow thickness (mm)		6.79±3.17	6.00±2.92	0.064†
	≤4.0	8	10	0.291‡
	>4.0	26	11
Clark level				0.243†
	3	3	5
	4	16	10
	5	15	6
Sentinel lymph node involvement				1.000§
	Negative	6	2
	Positive	20	7
Follow-up duration (mo)		40.3±30.7	35.0±27.1	0.549∥
Mortality				0.287§
	Dead	26	19
	Alive	8	2

Values are presented as number only or mean±standard deviation. Statistically significant (*p<0.05). †Chi square test; ‡Mann—Whitney U test; §Fisher's exact test; ∥Independent t-test.

Table 3

Epidermal growth factor receptor (EGFR) expression and location of metastasis

Variable		EGFR		p-value
		Negative (%) (n=34)	Positive (%) (n=21)
Skin				0.768†
	No	11 (32.4)	6 (28.6)
	Yes	23 (67.6)	15 (71.4)
Lung				0.088†
	No	21 (61.8)	8 (38.1)
	Yes	13 (38.2)	13 (61.9)
Brain				0.091†
	No	28 (82.4)	13 (61.9)
	Yes	6 (17.6)	8 (38.1)
Liver				0.731‡
	No	28 (82.4)	16 (76.2)
	Yes	6 (17.6)	5 (23.8)
Bone				1.000‡
	No	27 (79.4)	17 (81.0)
	Yes	7 (20.6)	4 (19.0)
Digestive system				0.719‡
	No	29 (85.3)	17 (81.0)
	Yes	5 (14.7)	4 (19.0)
Spine				1.000‡
	No	31 (91.2)	19 (90.5)
	Yes	3 (8.8)	2 (9.5)

Values are presented as number (%). †Chi square test; ‡Fisher's exact test.

Survival outcomes in distant metastatic melanoma patients

The EGFR (+) group showed a shorter average follow-up duration than the EGFR (−) group as shown in Table 2, but it was not statistically significant (p=0.549). The mortality rate observed in the EGFR (+) group was higher compared with the EGFR (−) group, but there was no statistical significance either (p=0.287). The researchers also performed multivariate analysis with EGFR expression, sex, age, ulceration, Breslow thickness, SLN involvement, and BRAF expression as independent variables to consider all variables affecting follow-up duration. However, no statistically significant variable can be obtained from the multiple regression analysis.

The median OS of all the patients with metastatic melanoma obtained using Kaplan—Meier survival analysis was 31.2 months (95% confidence interval [CI], 18.3∼44.1 months). The median OS of the EGFR (+) group was 22.2 (95% CI, 18.3∼36.1) months, and it was shorter than that of the EGFR (−) group, whose median OS was 37.0 months (95% CI, 31.8∼42.2). However, the researchers could not find a significant difference between the cumulative survival rates of both groups (p=0.076).

The OS data in this research are graphed in Fig. 2. The blue lines represent the EGFR (−) group, and the green lines represent the EGFR (+) group. At first, EGFR expression had no meaningful predictive value for the OS in patients with metastatic melanoma (Fig. 2A; p=0.076). When analyzed by subgroups, EGFR expression on metastatic melanoma in case of male (Fig. 2B; n=30, p=0.029), cutaneous melanoma without ulceration (Fig. 2C; n=43, p=0.043), or Breslow thickness ≤4.0 mm (Fig. 2D; n=18, p=0.031) was significantly associated with poor prognosis based on the survival data. In other subgroups, including female, cutaneous melanoma with ulcer, or Breslow thickness >4.0 mm, no significant difference were found concerning the survival outcomes depending on whether EGFR expression was present or not.

Fig. 2

Kaplan—Meier survival analysis in metastatic melanoma for comparison according to epidermal growth factor receptor (EGFR) expression. (A) Overall survival of all cases of metastatic melanoma showed no difference between the EGFR positive group and EGFR negative group. Overall survival outcomes in metastatic melanoma between the EGFR positive group and EGFR negative group were significantly different under the condition of (B) male, (C) without ulcer, or (D) Breslow thickness ≤4.0 mm. *Statistically significance (p<0.05).

DISCUSSION

In a previous study, EGFR gene amplification has been found in primary cutaneous melanoma, and it was associated with poor prognosis22. Rákosy et al.11 also studied with 81 samples of cutaneous melanoma about EGFR gene copy number index. They found that tumors with a higher gene copy number formed metastasis within 5 years after diagnosis, and high-level gene amplification that is usually the whole gain of chromosome 7 was only observed in metastatic tumors. Generally, melanoma patients with highly amplified or extra EGFR gene copies showed poor outcomes with higher invasion capacity. EGFR gene copy number alternation was associated with elevated EGFR mRNA expression. However there was no strong correlation between gene copy number alteration and EGFR protein expression level. In this regard, the researchers conducted a study to investigate the relationship between EGFR protein expression and clinical data of patients with metastatic melanoma from primary cutaneous melanoma, which has the poorest prognosis and rare treatment options.

When comparing our research results about EGFR expression and clinicopathologic parameters with a previous study23, EGFR protein expression was not less frequently observed in metastatic melanoma patients with a positive SNL (Table 2). Nevertheless, there was no statistically significant correlation between SNL involvement and EGFR expression.

In terms of metastasis, when EGFR is expressed in melanoma, there was a tendency to show metastasis at a higher rate in the lungs (1.62 times, p=0.088) and the brain (2.16 times, p=0.091) compared with the event when EGFR expression was negative, although it was not statistically significant (Table 3). In the 8th edition of the AJCC melanoma staging system, M1a is defined as distant metastasis to skin, soft tissues, and/or non-regional lymph nodes. M1b is also defined as distant metastasis to the lungs with or without M1a sites. If there is a metastasis to the central nervous system, it is unconditionally defined as M1d in order for the poor prognosis24. From this point of view, the tendency that the proportion of lung metastasis corresponding to M1b and brain metastasis corresponding to M1d may be higher in the EGFR (+) group, suggesting that there may be a possibility of affecting patient survival in metastatic melanoma. When comparing the mortality among the metastatic melanoma patients in our study, it was higher in the the EGFR (+) group, but there was no statistical significance either (Table 2). The median OS was 31.2 months (95% CI, 18.3∼44.1) in 55 metastatic melanoma patients. To further analyze survival outcome in more detail, there was no significant difference in OS according to EGFR expression (p=0.076) in the Kaplan—Meier survival analysis. It was not consistent with the results of the past study about 44 nodular melanoma patients, which documented that EGFR overexpression showed a correlation with shorter OS18. Our results showed also slightly shorter median OS in the EGFR (+) group.

When analyzing metastatic melanoma patients per subgroup, the OS of the EGFR (+) group was found to be statistically significantly shorter under the condition of male (p=0.029), without ulceration of primary cutaneous melanoma lesion (p=0.043), or Breslow thickness ≤4.0 mm of primary cutaneous melanoma lesion (p=0.031; Fig. 2). On another note, statistically significant results were not obtained in subgroups of female, with ulceration, or Breslow thickness >4.0 mm.

There was no statistically significant difference in the sex ratio (p=0.389) of metastatic melanoma patients according to EGFR expression (Table 2). In this regard, the EGFR expression of male metastatic melanoma patients may be considered as a factor influencing prognosis. Moreover, there was one study on sex differences and EGFR expression in melanoma patients, which reported the higher rate of deletion and low level EGFR gene amplification in male patients11.

Generally, ulceration is known as the third most powerful predictor of survival outcome in melanoma25. When analyzing ulceration in primary cutaneous melanoma with metastasis and EGFR expression, ulceration is rarely accompanied in the EGFR (+) group compared with the EGFR (−) group (p=0.019; Table 2). This is in conflict with the findings of previous researches that the presence of ulceration in melanoma has a higher EGFR gene copy number index, and the high EGFR expression level of nodular melanoma significantly presented ulceration more often1118 Furthermore, metastatic melanoma without ulceration showed significant poorer prognosis when accompanied by EGFR expression in our data (p=0.043; Fig. 2C). This might be due to the rapid exacerbation of metastasis when EGFR is expressed even before primary lesion ulceration is observed by the naked eye, encouraging patients to seek medical attention.

Although there was no statistical significance, the proportion of Breslow thickness ≤4.0 mm was higher in the EGFR (+) group (p=0.064; Table 2). Based on the AJCC 8th edition, if the Breslow thickness is >4.0 mm, it is the T4 stage. This results are consistent with the negative correlation between membrane EGFR expression and melanoma thickness in the prior study18. However, there was also a study that reported that the EGFR gene copy number index was higher when the Breslow thickness was >4.0 mm11. In our study, when the Breslow thickness was ≤4.0 mm, the prognosis of the EGFR (+) group was found to be worse (p=0.031; Fig. 2D). It can be inferred that, as in ulceration, melanoma metastasizes rapidly due to EGFR expression and develops the systemic spreading function early before the primary cutaneous lesion becomes severe.

Looking at the data of the patients, no significant relationship was observed between EGFR and BRAF expression findings in metastatic melanoma (p=0.236, Table 2). The ratio of EGFR positivity (38.2%) was higher than that of BRAF positivity (14.5%) in the samples of metastatic melanoma tissue. The difference in survival rate depending on EGFR expression did not show any statistical significance regardless of BRAF mutation in the Kaplan—Meier survival analysis.

In conclusion, EGFR expression in distant metastatic melanoma had predictive values for survival outcomes under the certain conditions. EGFR expression in distant metastatic melanoma in male, without ulceration, or Breslow thickness ≤4.0 mm of the primary cutaneous melanoma lesion may be involved in the progression of the disease.