Clinicopathological Features and Prognostic Factors of Primary Acral Melanomas in Caucasians

Background: Cutaneous melanomas located on the acral part of extremities (hand and foot melanoma; HFM) comprise a rare group within all melanomas in Caucasians. HFM is associated with a poor prognosis. We aimed to evaluate clinicopathological features, long-term outcomes, and prognostic factors in primary HFM in Caucasians. Methods: Medical records of all consecutive patients treated between 1997 and 2014 were revised. Patients were diagnosed with primary cutaneous melanoma at I-II clinical stage, and sentinel lymph node biopsy was conducted. The analysis was performed to define the clinicopathological factors influencing outcomes in the HFM and subungual cohort. Among 2537 consecutive patients diagnosed with primary cutaneous melanoma, 247 cases of HFM (9.7%) were found, with a median follow-up time of 7.8 years. Results: Median primary tumor Breslow thickness in subungual melanomas and HFMs was 4.0 mm and 3.3 mm, respectively, significantly higher than in the entire population (median 2.2 mm; p < 0.01). In the HFM group, 37.6% of tumors were ulcerated. Metastases to sentinel lymph node (SLN) were found in 28.3% of HFMs. The 10-year overall survival rate in the HFM group and subungual melanomas was 48.1% and 49.3%, respectively, compared to 63.0% in non-HFM melanomas. Conclusions: Our results confirm that patients with HFMs display worse overall survival compared to the entire melanoma population, with male gender and positive SLN biopsy status acting as independent negative prognostic factors.


Introduction
Cutaneous melanomas can be classified into subtypes, primarily based on histopathology and anatomic location. With regard to pathology, primary melanoma subtypes include superficial spreading melanoma (SSM), lentigo malignant melanoma (LMM), nodular melanoma (NM), acral lentiginous melanoma (ALM), desmoplastic melanoma, and amelanotic melanoma [1]. With regard to anatomy, common areas are head and neck, trunk, upper limb (without hand), lower limb (without foot), hand, foot. excisional biopsy with Breslow thickness ≥0.75 mm or presence of ulceration, feasibility for general anesthesia, sentinel lymph node (SLN) biopsy. The exclusion criteria were: incomplete medical records, metastatic disease at the moment of diagnosis, clinically palpable lymph nodes, melanoma of unknown primary. Patients were divided into non-HFM and HFM. HFMs were divided into nonsubungual lesions (non-SUM) and SUM. SUM was defined when the pigmentation was directly connected with the nail apparatus. Non-SUM was defined when the pigmentation was observed in the hand or foot but not connected with the nail apparatus. We planned to evaluate differences in clinicopathological factors and survival between non-HFM and HFM and between non-SUM and SUM. The second part of the analysis included the identification of prognostic factors for survival in all the aforementioned subgroups.
The following clinicopathological parameters were analyzed: age at the first diagnosis, sex, status of SLN biopsy, Breslow thickness (segregated into three groups with cut-off points of 1 mm and 4 mm), pathologic stage, ulceration, histopathological subtype (SSM, LMM, NM, ALM). We have chosen 5-year and 10-year OS as representative cut-offs because most of the analysis in the literature includes 5-year and 10-year OS [7,8,[11][12][13]15,19].

Statistical Methods
Pearson Chi-squared test or Fisher's exact test (if frequencies ≤ 6) was used to analyze group proportions. Mann-Whitney U test was used to evaluate differences between continuous data. OS was calculated from the date of the first diagnosis to the last follow-up (censored) or death. The Kaplan-Meier method for estimating survival functions and the Cox proportional hazards model for estimating the effects of covariates on the hazard of the occurrence of death were used. All p values < 0.05 were considered significant. Data analysis was performed using the R software/environment (R Development Core Team. R: A Language and Environment for Statistical Computing; R Foundation for Statistical Computing: Vienna, Austria, 2009) [25], version R 3.6.2, which is an open-source project that is distributed under the GNU General Public License.

Clinicopathological Features
We found 2537 consecutive patients with primary cutaneous melanoma at clinical stage I-II, who underwent SLN biopsy between 1997 and 2014 in MSNRI. Within the cohort, 247 patients were diagnosed with primary HFM (9.74%). The others were non-HFM (90.26%). Within the HFM group, we identified 46 patients with SUM (1.8% of the entire population) and 201 with non-SUM. The results of the search and data extraction are shown in the flow diagram ( Figure 1).

HFM vs. Non-HFM Patients
Clinical and pathological features of non-HFM and HFM cohort are shown in Table 1. The mean age at the first diagnosis in the HFM group was 58.5 years (median = 60.2, range: 16-94), and most patients were diagnosed after their sixth decade of life. Patients with HFM were significantly older than patients with other anatomical sites (p < 0.05); the mean age was 51.7 years (median = 53.0, range: 14-87). Ulceration in HFMs (62.4%) was significantly more frequent than in a cohort of melanomas of other anatomical sites (44.1%, p < 0.05). SLN biopsy was performed in most of the cases, with the positive result in 28.3% of HFMs and 20.6% in other sites melanomas (p < 0.05). In the HFM group, Breslow thickness (median = 3.3 mm) was significantly higher than in the non-HFM melanoma group (median = 2.2 mm) (p < 0.05). For HFM, 53.4% were diagnosed at less than 4 mm, compared to 68.4% in a nonacral group (p < 0.05). HFM subtypes were classified into ALM (45.3%), NM (35.3%), SSM (12.9%), and LMM (6.5%). NM was the most frequent subgroup in non-HFM, accounting for 53.8%, followed by the SSM type at 38.1%, LMM at 7.9%, and ALM accounting for only 0.2% (p < 0.05). According to the 8th edition of AJCC Melanoma of the Skin Staging, 748 (34.1%) of non-HFM patients were at the pathological stage I, 721 (32.9%) were at stage II, and 723 (33%) at stage III. In the HFM group, 40 (18.7%) patients were at stage I, 64 (29.9%) patients were at stage II, and 110 (51.4%) patients were at stage III.

SUM vs. Non-SUM Patients
Patients in the SUM cohort were older, with a mean age of 62.2 years (median = 66.1, range: 36-81), compared to non-SUM patients, which showed a mean age of 57.7 years (median = 59.5, range: 16-94). However, the difference was not statistically significant (p = 0.099). In the SUM group, most of the patients were diagnosed with ALM (70.6%), followed by 23.5% of NM, and only one patient was diagnosed with SSM and LMM. In the non-SUM group, ALM was found only in 39% of patients. In the SUM cohort, 71.4% of the cases were ulcerated, which was more frequent than in non-SUM (60.2%), but not statistically significant (p = 0.242). In the SUM group, the rate of positive SLN biopsy was slightly higher (37.1%) than in non-SUM (26.2%) (p = 0.28). Breslow thickness was deeper in the SUM group (median = 4.0 mm) than in the non-SUM group (median = 3.1 mm). More clinicopathological details of the non-SUM vs. SUM group can be found in Table 2. Most HFMs (84.6%) were found on the lower limb, and there were 76 (30.8%) cases of digital melanoma. The subungual location represented 46 (18.6%) cases of digital melanoma ( Table 3). The OS was similar for cases located on fingers or toes vs. other parts of the hand or foot, as well as for cases located on hand vs. foot. According to the 8th edition of AJCC Melanoma of the Skin Staging, most patients were at the pathological stage III in both the non-SUM (n = 94, 51.3%) and SUM group (n = 16, 51.6%).
Higher Breslow thickness, presence of ulceration, positive SLN biopsy status, older age at the first diagnosis, male gender, and lower pathologic stage were all significantly associated with reduced overall survival in both non-HFM and HFM cohorts in univariate analysis. Moreover, in the non-HFM group, the pathological subtype was significantly correlated with OS. The factors associated with OS in the non-HFM and HFM cohorts are shown in Table 4. For the HFM and non-HFM subgroups, each of the aforementioned factors has also been analyzed by the Kaplan-Meier method ( Figure 2; Figure 3, respectively). The correlation between OS and all clinicopathological features was also analyzed by multivariate analysis. In the non-HFM cohort, older age (≥50 years old), male gender, thicker Breslow, presence of ulceration, and positive SLN biopsy were identified as significant prognostic factors for OS ( Figure 4). Multivariate analysis performed on the HFM cohort showed that only positive SLN biopsy status and male gender were independently correlated with worse OS ( Figure 5). Breslow thickness distribution between male and female is shown in Figure 6. Patients with HFMs had a significantly worse median OS (69.4 months) compared to non-HFM patients (95.1 months) (p < 0.05) (Figure 7). Patients with SUMs also showed a significantly worse OS when compared to non-HFM patients (median OS time = 59.3 months), but there was only a statistical tendency when the SUM group was compared to the non-SUM group (p = 0.674).

Discussion
This study represents a large single-institution study of HFM in Caucasians with very long followup time. In our study, the mean age of HFM patients was 58.5 years, in line with the literature reporting a mean age for HFM and ALM varying from 55.3 to 69 years [5,9,21,[26][27][28][29][30]. At the time of diagnosis,

Discussion
This study represents a large single-institution study of HFM in Caucasians with very long follow-up time. In our study, the mean age of HFM patients was 58.5 years, in line with the literature reporting a mean age for HFM and ALM varying from 55.3 to 69 years [5,9,21,[26][27][28][29][30]. At the time of diagnosis, patients with SUM were older than patients with melanoma located in other sites of the body, with a mean age of 62.2 years. The mean age of the other patients was 57.7 years. In our study, the median age in the non-HFM cohort was 52.5 years old, which is consistent with data in the literature [31]. There were slightly more women in the HFM (59.5%) and SUM groups (58.7%), which is consistent with the previous literature [11,30,32]. We have also shown that the most common histological subtype in HFM is ALM, which stays in concordance with studies of other authors such as Durbec et al., Albreski et al., Jung et al., Nunes et al. [2,9,28,30]. Many HFMs were ulcerated (62.4%), which is consistent with the retrospective literature in this MM group [2,30,33]. In the HFM group, we found a higher median Breslow thickness (3.3 mm) than reported by Bello (2.1 mm) [5] but thinner than in the Nunes cohort (5.0 mm) [30]. In Durbec et al.'s review of HFM, melanoma was more frequently located on the foot than on the hand, with hand/foot melanoma distribution ranging from 1/13 to 1/4, which is similar to our results in which foot melanoma constituted 84.6% of all HFM [28].
In our group, ALM represented 5.6% of all MMs, most of which (95%) were found on acral locations, while in the large United States population-based study of Bradford et al., ALM frequency was calculated to be about 1.5% of all MMs in the non-Hispanic white skin population [21]. The most prevalent subtype of SUM in our series was ALM, which confirms the previous results [25]. In our study, SUM comprised 18.6% of acral melanomas, while in other studies, the percentage was slightly higher. Indeed, Jung et al. reported a frequency of 28.9% for SUM in Korean patients [28]. The mean Breslow thickness was 5.03 mm, which confirmed that patients with subungual melanoma display thick lesions [11,24,32].
We confirmed a reduced survival in the HFM group vs. patients with MMs in other sites, as observed in previous studies [5,30]. The five-year survival rate in HFM patients was 59.3%, in line with the literature reporting a range of 59-63% [7,8]. In a study performed on an English cohort, Banfield et al. reported a five-year OS of 51% [11]. However, in the investigations conducted by Nunes et al. and Keith et al., the five-year OS was higher, at 61% and 59%, respectively [8,11,32]. On the other side, the 10-year OS was only 48.1%. The 5-year and 10-year OS in the SUM cohort was 62.1% and 49.3%, respectively, which is slightly higher than in HFM patients.
Depending on the study, different prognostic factors were described as related to survival rates in HFM. The prognostic factors in melanoma previously established in the literature such as age at diagnosis, tumor thickness, sex, presence of ulceration, pathologic stage, and SLN biopsy status were also significantly associated with OS in all melanomas in both non-HFM and HFM groups when analyzed by univariate analysis in our study [33]. Our multivariate analysis in the HFM group found only the positive SLN biopsy status and male gender as independent significant prognostic factors for survival. We observed just a tendency to worse survival in older age, ALM subtype, presence of ulceration, and thick Breslow. Some studies show that only a higher stage at diagnosis is the predictor for a shorter survival rate [34]. Multivariate analysis in Slingluff et al.'s study on stage I acral melanomas showed that race and ulceration were significant as independent factors while in Nunes et al.'s study, Breslow thickness and ulceration were independent factors for HFM [29,30]. Differences between our cohort compared to other groups may arise from a different ethnical or genetic profile of MMs, as a result of, for example, a difference in exposure to UVR. Because of a limited number of patients and survival rates, we did not perform an analysis of prognostic factors for survival in SUM. In literature, the main prognostic factors affecting survival were Breslow thickness and ulceration. Moreover, Clark level, patient's race, amputation level, bone invasion, stage at presentation, aneuploidy fraction, and S-phase fraction have been identified as prognostic factors in some studies [8,17,18,32].
Our analysis shows that HFMs are associated with a worse prognosis than non-HFMs. However, there was no significant difference in overall survival between non-SUM and SUM cohorts. Also, the Korean study shows that there is no difference in survival rates between volar and subungual lesions or between weight-bearing and non-weight-bearing sites of the soles [28]. Inversely, Barnes et al. reported that subungual locations of foot melanomas were associated with lower survival rates (10-year OS was 17%). However, the number of cases was small, and the difference was not significant compared to the remaining part of the foot [35].
At the same time, Zebary et al. have shown that factors such as age at diagnosis, tumor thickness, Clark's level of invasion, anatomical site (hand versus foot), and ulceration were significantly associated with OS in ALMs. However, the status of the SLN biopsy was not evaluated [27]. The prognostic value of SLN biopsy status has been confirmed by Ito et al. in ALMs and further supported by Parvi et al., who reported that positive SLN biopsy, increasing age, increasing thickness, and ulceration were significantly associated with worse OS in ALM patients [36,37]. Nunes et al. reported Breslow thickness, ulceration, and SLN biopsy status as prognostic factors for OS for Brazilian patients [38].
The study has limitations. First, because of its retrospective design, additional clinical information, such as trauma history or exposure to UV, was not available.
Moreover, treatment options have changed over the past several decades that may affect patients' survival. Nevertheless, our study represents one of the largest analyses focused on acral melanomas in Caucasians treated in one institution with long-term follow-up with additional data on prognostic factors in this rare melanoma subtype.

Conclusions
We showed that HFM possesses specific epidemiological features that differ from melanoma in other anatomical sites. We also showed that HFM occurs later in life than melanomas of other anatomic sites. We confirmed a reduced survival in the HFM cohort vs. patients with MMs in other sites, with two independent significant prognostic factors for survival: positive SLN biopsy and male gender. However, there was no significant difference in overall survival between non-SUM and SUM patients. In our study, we identified prognostic factors for OS in HFM and SUM patients, some of them previously described in the literature. We believe that the specific epidemiological features of HFM may be clinically significant to determine treatment options for metastatic acral and non-acral cutaneous melanoma patients. Generally, further investigation is needed to identify molecular factors associated with HFM and SUM. Funding: This work has been supported by Maria-Sklodowska Curie National Research Institute of Oncology statutory funding.

Conflicts of Interest:
The authors declare no conflict of interests.