Current Issue
Spring/Summer 2025, Vol. 32 No. 1
Hong Kong J. Dermatol. Venereol. (2025) 32, 20-40
Original Article
Analysis and prediction of risk factors for metastatic melanoma in different sites based on the SEER database
基於SEER數據庫不同部位轉移性黑色素瘤危險因素的分析與預測
R Chen 陳銳慶, H Yu 餘海, WK Ming 明偉傑, X Zheng 鄭炘凱, Y Lin 林鈺, J Lyu 呂軍, L Deng 鄧列華
Abstract
Objective: The aim of this study is to investigate the risk factors associated with metastasis to various sites in melanoma, construct corresponding nomograms for predicting the risk of metastasis, and provide a reference for clinical practitioners in the diagnosis and treatment of this condition. Methods: Utilising SEER database data, we analysed 11,218 metastatic melanoma patients using logistic regression to identify metastasis determinants to bone, brain, liver, and lung, illustrated with a forest plot and developed an interactive nomogram for prediction. Additionally, we investigated prognostic factors for outcomes using the same statistical methods. Results: Our results indicate that histological behaviour, sex, race, marital status, tumour site, surgery at the primary site, postoperative lymph node dissection, and radiotherapy are independent predictors of brain metastasis; histological behaviour, sex, race, marital status, surgery at the primary site, radiotherapy, chemotherapy, Breslow thickness, and ulceration are independent predictors of bone metastasis; histological behaviour, race, marital status, tumour site, surgery at the primary site, postoperative lymph node dissection, chemotherapy, Breslow thickness, and ulceration are independent predictors of liver metastasis; and histological behaviour, age, sex, race, marital status, tumour site, surgery at the primary site, postoperative lymph node dissection, radiotherapy, chemotherapy, and ulceration are independent predictors of lung metastasis. We constructed interactive nomograms based on these factors to predict the probability of metastasis at different sites. Additionally, univariate and multivariate logistic regression analyses revealed that histological behaviour, age, sex, race, marital status, income, tumour size, tumour site, surgery at the primary site, radiotherapy, chemotherapy, Breslow thickness, and ulceration are factors influencing metastatic melanoma. Conclusions: This study has successfully established a nomogram for predicting metastatic melanoma. Among the factors influencing metastasis to various sites, histological behaviour, race, marital status, primary site surgery, and postoperative lymph node dissection were identified as common determinants. This suggests that these factors play a significant role in the metastasis of melanoma. Moreover, these factors also play a crucial role in affecting the prognosis. Therefore, in the clinical diagnosis and treatment process, it is imperative to consider these factors carefully and to analyse them according to the specific circumstances of different patients.
目的：本研究旨在探討黑色素瘤不同部位轉移相關的危險因素，建立相應的列線圖，預測其轉移風險，為臨床醫生診斷和治療該疾病提供參考。方法：利用SEER數據庫數據，我們使用邏輯回歸分析了11,218例轉移性黑色素瘤患者，以確定骨、腦、肝和肺的轉移決定因素，製作森林圖，並開發了用於預測的互動式列線圖。結果：我們的研究結果表明，組織學行為、性別、種族、婚姻狀況、腫瘤部位、原發部位手術、術後淋巴結清掃和放療是腦轉移的獨立預測因素；組織學行為、性別、種族、婚姻狀況、原發部位手術、放療、化療、Breslow厚度和潰瘍是骨轉移的獨立預測因素；組織學行為、種族、婚姻狀況、腫瘤部位、原發部位手術、術後淋巴結清掃、化療、Breslow厚度和潰瘍是肝轉移的獨立預測因素；組織學行為、年齡、性別、種族、婚姻狀況、腫瘤部位、原發部位手術、術後淋巴結清掃、放療、化療和潰瘍是肺轉移的獨立預測因素。基於這些因素，我們建立了互動式列線圖來預測不同部位轉移的概率。此外，單因素和多因素邏輯回歸分析顯示，組織學行為、年齡、性別、種族、婚姻狀況、收入、腫瘤大小、腫瘤部位、原發部位手術、放療、化療、Breslow厚度和潰瘍是影響轉移性黑色素瘤的因素。結論：本研究成功建立了預測轉移性黑色素瘤的列線圖。在影響不同部位轉移的因素中，組織學行為、種族、婚姻狀況、原發部位手術和術後淋巴結清掃是共同的決定因素。這表明這些因素在黑色素瘤的轉移中起重要作用。此外，這些因素在影響預後方面也起著至關重要的作用。因此，在臨床診斷和治療過程中，必須認真考慮這些因素，並根據不同患者的具體情況進行分析。
Keywords: Bone metastasis, Brain metastasis, Liver metastasis, Lung metastasis, Metastatic melanoma, Risk factors
關鍵詞: 骨轉移、腦轉移、肝轉移、肺轉移、轉移性黑色素瘤、危險因素
Introduction
Metastatic melanoma represents a highly invasive form of skin cancer, characterised by the malignant proliferation of melanocytes and a pronounced capability for metastasis. Epidemiological data indicate an increasing incidence of metastatic melanoma globally, with a particularly higher prevalence among white populations. Despite representing a smaller fraction of skin cancer cases, metastatic melanoma accounts for a disproportionately high mortality rate.1,2
The risk of metastasis in melanoma is influenced by multiple factors including skin type (especially lighter skin), the thickness of the primary tumour, location, ulceration, and the status of the immune system. Metastasis pathways are diverse, typically initiating in nearby lymph nodes and progressing through the lymphatic or haematogenous systems to other body parts. Common metastatic sites include lymph nodes, lungs, liver, bones, and the brain, each presenting unique clinical manifestations and therapeutic challenges.3,4
Current therapeutic strategies for metastatic melanoma encompass surgical excision, chemotherapy, radiotherapy, along with targeted and immunotherapies.5 However, the primary challenges in treatment are high recurrence rates and the emergence of drug resistance. Despite significant advances in therapeutic approaches since 2011, metastasis remains a formidable challenge in melanoma management. Current treatment trends focus on targeted and immunotherapies,6 which have significantly improved patient prognosis but have not completely halted disease progression. Crucially, early diagnosis and prevention of metastasis remain key to reducing melanoma mortality rates.
Most existing research focuses on the risk factors for melanoma onset, with insufficient studies analysing the risk factors for metastasis to different sites. Therefore, this study aims to explore the risk factors for metastasis to various sites and factors affecting the prognosis of metastatic melanoma through a high-quality SEER database-based, large-scale population study, providing clinical practitioners with references for treatment.
Methods
Data source
In this investigation, data pertaining to all metastatic melanoma cases were extracted from the Surveillance, Epidemiology, and End Results (SEER) database. The selection criteria for variables included the International Classification of Diseases for Oncology, Third Edition (ICD-O-3) codes: 8720/3, 8721/3, 8742/3, 8743/3, 8744/3, with the presence of metastasis (M=1) within the TNM staging system, covering the period from 2010 to 2019. Confirmation was based on histological and cytological positivity. Data cleansing was performed by eliminating unknown variables and missing values, thereby refining the dataset required for our research objectives. This process culminated in the identification of 11,218 patients with metastatic melanoma. All pertinent information was sourced from the SEER database, negating the need for patient informed consent forms.7,8 Figure 1 outlines the methodology employed in this study.
 |
| Figure 1 Research flowchart. |
Variable classification
This study engaged in the curation of data pertaining to patients with metastatic melanoma over a decade, spanning 2010 to 2019. Leveraging the SEER database, the investigation incorporated variables inclusive of histological subtypes: malignant metastatic melanoma NOS (MMNOS), nodular metastatic melanoma (NMM), lentigo malignat metastatic melanoma (LMM), superficial spreading metastatic melanoma (SSMM), acral lentiginous metastatic melanoma (ALMM); demographic factors such as age and sex; racial categorisation; marital status; income levels; geographical distribution between urban and rural settings; tumor characteristics including size and anatomical site (head and face, trunk, limbs, among others); treatment modalities encompassing surgery of the primary site, scope of regional lymph node dissection post-surgery, radiotherapy, and chemotherapy; and clinical features such as Breslow thickness, presence of ulceration, and metastatic spread to bone, brain, liver, and lungs. Additionally, patient outcomes were classified into survival statuses. This comprehensive approach aimed at elucidating the multifaceted landscape of metastatic melanoma management and outcomes within the specified timeframe.
Statistical analysis
Initially, variables were characterised; categorical variables were denoted as percentages, while continuous variables were presented through means and standard deviations. The investigation into risk factors employs logistic regression analysis. Initially, univariate logistic regression identifies potential risk factors associated with melanoma metastasis to various locations. Subsequently, factors with a p-value less than 0.05 in the univariate logistic regression are incorporated into a multivariate logistic regression analysis. Independent risk factors determined in the multivariate logistic regression with a p-value less than 0.05 are utilised to construct an interactive nomogram model for predicting metastasis to different locations, facilitating clinical application.9 Statistical analysis for this study was conducted using SPSS Statistics software (version 27.0, Chicago, IL, USA) and R software (version 4.0.1), with statistical significance set at p<0.05.
Results
Basic information of the patient
In this investigation, the melanoma tissue subtypes encompassed the commonly encountered clinical variants, including NMM, LMM, SSMM, ALMM, and MMNOS, with MMNOS accounting for the majority (74.2%) of cases. The average age of the cohort was 64.8 years (±15.0). Demographically, the patient population was predominantly male (65.3%) and white (92.7%), with an even distribution across the upper three income brackets. Over 80% of the cases were urban residents (85.1%), with the trunk being the primary site of lesion in 62.9% of instances. Approximately half of the patients underwent surgical intervention at the primary site (41.7%), while a minority received radiotherapy (22.3%) and chemotherapy (15.4%). Ulceration was present in 11.5% of the cases. Metastasis was most frequently observed in the lungs (29.5%), followed by the brain, liver, and bones. The median survival duration for the entire cohort was 17 months. Table 1 delineates the foundational characteristics of the patient demographic.
| Table 1 Fundamental characteristics of patients with metastatic melanoma | ||||
|---|---|---|---|---|
| Alive (N=4760) | Death (N=6458) | Total (N=11218) | ||
| Hist_behav | NMM | 285 (6.0%) | 532 (8.2%) | 817 (7.3%) |
| LMM | 287 (6.0%) | 117 (1.8%) | 404 (3.6%) | |
| SSMM | 1309 (27.5%) | 297 (4.6%) | 1606 (14.3%) | |
| ALMM | 28 (0.6%) | 34 (0.5%) | 62 (0.6%) | |
| MMNOS | 2851 (59.9%) | 5478 (84.8%) | 8329 (74.2%) | |
| Age (years) | Mean (SD) | 60.5 (15.0) | 68.1 (14.1) | 64.8 (15.0) |
| Sex | Male | 2852 (59.9%) | 4476 (69.3%) | 7328 (65.3%) |
| Female | 1908 (40.1%) | 1982 (30.7%) | 3890 (34.7%) | |
| Race | White | 4181 (87.8%) | 6223 (96.4%) | 10404 (92.7%) |
| Black | 29 (0.6%) | 94 (1.5%) | 123 (1.1%) | |
| Other | 550 (11.6%) | 141 (2.2%) | 691 (6.2%) | |
| Marital | Married | 1982 (41.6%) | 3248 (50.3%) | 5230 (46.6%) |
| Single | 592 (12.4%) | 1145 (17.7%) | 1737 (15.5%) | |
| Other | 2186 (45.9%) | 2065 (32.0%) | 4251 (37.9%) | |
| Income | Low | 1084 (22.8%) | 1617 (25.0%) | 2701 (24.1%) |
| Mediate | 2173 (45.7%) | 2997 (46.4%) | 5170 (46.1%) | |
| High | 1503 (31.6%) | 1844 (28.6%) | 3347 (29.8%) | |
| Distribution | Urban | 4011 (84.3%) | 5536 (85.7%) | 9547 (85.1%) |
| Rural | 749 (15.7%) | 922 (14.3%) | 1671 (14.9%) | |
| Size (mm) | Mean (SD) | 5.42 (29.0) | 8.24 (44.0) | 7.04 (38.4) |
| Site | HF | 764 (16.1%) | 891 (13.8%) | 1655 (14.8%) |
| Truck | 2535 (53.3%) | 4520 (70.0%) | 7055 (62.9%) | |
| Limbs | 1450 (30.5%) | 1011 (15.7%) | 2461 (21.9%) | |
| Other | 11 (0.2%) | 36 (0.6%) | 47 (0.4%) | |
| Surg_Prim_Site | No | 2067 (43.4%) | 4476 (69.3%) | 6543 (58.3%) |
| Yes | 2693 (56.6%) | 1982 (30.7%) | 4675 (41.7%) | |
| Reg_LN_Sur | No | 4120 (86.6%) | 5567 (86.2%) | 9687 (86.4%) |
| Yes | 640 (13.4%) | 891 (13.8%) | 1531 (13.6%) | |
| Radiotherapy | No | 4189 (88.0%) | 4529 (70.1%) | 8718 (77.7%) |
| Yes | 571 (12.0%) | 1929 (29.9%) | 2500 (22.3%) | |
| Chemotherapy | No | 4375 (91.9%) | 5120 (79.3%) | 9495 (84.6%) |
| Yes | 385 (8.1%) | 1338 (20.7%) | 1723 (15.4%) | |
| Breslow_thickness (mm) | Mean (SD) | 0.840 (1.86) | 1.03 (2.36) | 0.951 (2.16) |
| Ulceration | No | 4367 (91.7%) | 5565 (86.2%) | 9932 (88.5%) |
| Yes | 393 (8.3%) | 893 (13.8%) | 1286 (11.5%) | |
| Mets_bone | No | 4475 (94.0%) | 5197 (80.5%) | 9672 (86.2%) |
| Yes | 285 (6.0%) | 1261 (19.5%) | 1546 (13.8%) | |
| Mets_brain | No | 4307 (90.5%) | 4517 (69.9%) | 8824 (78.7%) |
| Yes | 453 (9.5%) | 1941 (30.1%) | 2394 (21.3%) | |
| Mets_liver | No | 4481 (94.1%) | 4994 (77.3%) | 9475 (84.5%) |
| Yes | 279 (5.9%) | 1464 (22.7%) | 1743 (15.5%) | |
| Mets_lung | No | 3995 (83.9%) | 3909 (60.5%) | 7904 (70.5%) |
| Yes | 765 (16.1%) | 2549 (39.5%) | 3314 (29.5%) | |
| Hist_behav: Histological behaviour; NMM: Nodular metastatic melanoma; LMM: Lentigo maligna metastatic melanoma; SSMM: Superficial spreading metastatic melanoma; ALMM: Acral lentiginous metastatic melanoma, malignant; MMNOS: Malignant metastatic melanoma, NOS; HF: Head and Face; Surg_Prim_Site: surgery of the primary site; Reg_LN_Sur: scope of regional lymph node dissection post-surgery; Mets_bone: Metastases bone; Mets_brain: Metastases brain; Mets_liver: Metastases liver; Mets_lung: Metastases lung. | ||||
Analysis and prediction of risk factors for bone metastasis
Utilising univariate and multivariate logistic regression analyses, this study elucidates the determinants of bone metastasis in melanoma patients. Our findings illuminate histological behave, sex, race, marital status, primary site surgery, radiotherapy, chemotherapy, Breslow thickness, and ulceration as independent prognostic indicators for bone metastasis (Table 2). Particularly, Acral Lentiginous Melanoma (ALMM) and Malignant Melanoma Not Otherwise Specified (MMNOS) exhibit a heightened susceptibility to bone metastasis. Comparative analysis reveals a diminished likelihood of bone metastasis in females versus males. Among ethnic groups, Caucasians bear a significantly elevated risk of bone metastasis. Marital status analysis indicates that married individuals have a predisposition towards bone metastasis. Surprisingly, both radiotherapy and chemotherapy emerged as risk factors for bone metastasis, contrary to their therapeutic intentions. Conversely, surgical intervention at the primary tumour site markedly reduces the risk of bone metastasis. Breslow thickness and ulceration are identified as risk factors, underscoring their clinical relevance in metastasis prediction. Factors such as income, geographic distribution, tumour size, and tumor site exhibited minimal impact on bone metastasis occurrence. Forest plots (Figure 2) provide a visual representation of risk factors including ALMM, MMNOS, radiotherapy, chemotherapy, Breslow thickness, and ulceration, alongside protective factors such as female gender, unmarried status, non-Caucasian ethnicity, and primary site surgery. Leveraging these independent factors, an interactive nomogram (Figure 3) was developed to estimate the probability of bone metastasis, tailored to individual patient profiles, thus facilitating personalised risk assessment.
| Table 2 Univariate and multivariate analyses of melanoma bone metastas | |||||
|---|---|---|---|---|---|
| Dependent: Mets_bone | No (N=9672) | Yes (N=1546) | OR (univariable) | OR (multivariable) | |
| Hist_behav | NMM | 712 (7.4%) | 105 (6.8%) | ||
| LMM | 397 (4.1%) | 7 (0.5%) | 0.12 (0.06-0.26, p<0.001) | 0.32 (0.14-0.72, p=0.005) | |
| SSMM | 1565 (16.2%) | 41 (2.7%) | 0.18 (0.12-0.26, p<0.001) | 0.49 (0.33-0.74, p<0.001) | |
| ALMM | 52 (0.5%) | 10 (0.6%) | 1.30 (0.64-2.64, p=0.462) | 2.19 (1.03-4.64, p=0.041) | |
| MMNOS | 6946 (71.8%) | 1383 (89.5%) | 1.35 (1.09-1.67, p=0.006) | 1.73 (1.33-2.25, p<0.001 | |
| Age (years) | Mean (SD) | 64.9±15.1 | 64.3±14.2 | 1.00 (0.99-1.00, p=0.146) | 1.00 (0.99-1.00, p=0.288 |
| Sex | Male | 6223 (64.3%) | 1105 (71.5%) | ||
| Female | 3449 (35.7%) | 441 (28.5%) | 0.72 (0.64-0.81, p<0.001) | 0.87 (0.77-0.99, p=0.038) | |
| Race | White | 8914 (92.2%) | 1490 (96.4%) | ||
| Black | 104 (1.1%) | 19 (1.2%) | 1.09 (0.67-1.79, p=0.723) | 1.03 (0.62-1.72, p=0.898) | |
| Other | 654 (6.8%) | 37 (2.4%) | 0.34 (0.24-0.47, p<0.001) | 0.64 (0.45-0.91, p=0.013 | |
| Marital | Married | 4342 (44.9%) | 888 (57.4%) | ||
| Single | 1462 (15.1%) | 275 (17.8%) | 0.92 (0.79-1.07, p=0.267) | 0.88 (0.75-1.03, p=0.107) | |
| Other | 3868 (40%) | 383 (24.8%) | 0.48 (0.43-0.55, p<0.001) | 0.79 (0.69-0.91, p=0.001) | |
| Income | Low | 2372 (24.5%) | 329 (21.3%) | ||
| Mediate | 4481 (46.3%) | 689 (44.6%) | 1.11 (0.96-1.28, p=0.150) | 1.05 (0.89-1.23, p=0.557) | |
| High | 2819 (29.1%) | 528 (34.2%) | 1.35 (1.16-1.57, p<0.001) | 1.26 (1.06-1.50, p=0.091) | |
| Distribution | Urban | 8188 (84.7%) | 1359 (87.9%) | ||
| Rural | 1484 (15.3%) | 187 (12.1%) | 0.76 (0.65-0.89, p<0.001) | 0.93 (0.77-1.13, p=0.462) | |
| Size (mm) | Mean (SD) | 6.5±35.4 | 10.2±53.0 | 1.00 (1.00-1.00, p=0.001) | 1.00 (1.00-1.00, p=0.078) |
| Site | HF | 1467 (15.2%) | 188 (12.2%) | ||
| Truck | 5921 (61.2%) | 1134 (73.4%) | 1.49 (1.27-1.76, p<0.001) | 0.92 (0.76-1.11, p=0.389) | |
| Limbs | 2246 (23.2%) | 215 (13.9%) | 0.75 (0.61-0.92, p=0.006) | 0.83 (0.67-1.04, p=0.106) | |
| other | 38 (0.4%) | 9 (0.6%) | 1.85 (0.88-3.88, p=0.105) | 1.31 (0.60-2.85, p=0.501) | |
| Surg_Prim_Site | No | 5399 (55.8%) | 1144 (74%) | ||
| Yes | 4273 (44.2%) | 402 (26%) | 0.44 (0.39-0.50, p<0.001) | 0.50 (0.43-0.59, p<0.001) | |
| Reg_LN_Sur | No | 8369 (86.5%) | 1318 (85.3%) | ||
| Yes | 1303 (13.5%) | 228 (14.7%) | 1.11 (0.95-1.29, p=0.175) | 0.96 (0.81-1.13, p=0.613) | |
| Radiotherapy | No | 7866 (81.3%) | 852 (55.1%) | ||
| Yes | 1806 (18.7%) | 694 (44.9%) | 3.55 (3.17-3.97, p<0.001) | 2.37 (2.10-2.67, p<0.001) | |
| Chemotherapy | No | 8417 (87%) | 1078 (69.7%) | ||
| Yes | 1255 (13%) | 468 (30.3%) | 2.91 (2.57-3.29, p<0.001) | 1.83 (1.60-2.09, p<0.001) | |
| Breslow_thickness (mm) | Mean (SD) | 0.9±2.0 | 1.3±2.7 | 1.07 (1.05-1.09, p<0.001) | 1.15 (1.11-1.18, p<0.001 |
| Ulceration | No | 8629 (89.2%) | 1303 (84.3%) | ||
| Yes | 1043 (10.8%) | 243 (15.7%) | 1.54 (1.33-1.79, p<0.001) | 1.40 (1.12-1.75, p=0.003 | |
| Hist_behav: Histological behaviour; NMM: Nodular metastatic melanoma; LMM: Lentigo maligna metastatic melanoma; SSMM: Superficial spreading metastatic melanoma; ALMM: Acral lentiginous metastatic melanoma, malignant; MMNOS: Malignant metastatic melanoma, NOS; HF: Head and Face; Surg_Prim_Site: surgery of the primary site; Reg_LN_Sur: scope of regional lymph node dissection post-surgery | |||||
 |
| Figure 2 Forest plot of bone metastasis in melanoma. |
 |
| Figure 3 Interactive nomogram for predicting the risk of bone metastasis. Hist_behav: Histological behaviour; Surg_Prim_Site :surgery of primary site. |
Analysis and prediction of risk factors for brain metastasis
Utilising both univariate and multivariate logistic regression analyses, this study elucidated the determinants of cerebral metastasis in melanoma. The findings identified histological behave, sex, race, marital status, tumour location, primary site surgery, postoperative lymph node dissection, and radiation therapy as independent predictors of brain metastasis (Table 3). Compared to NMM, LMM and SSMM exhibited a reduced propensity for cerebral metastasis. Females, as opposed to males, were found to have a lower likelihood of developing cerebral metastases. Non-white ethnicities faced a significantly heightened risk of cerebral metastasis compared to their counterparts. Individuals who were unmarried showed a greater tendency towards cerebral metastasis. Tumours located on the trunk were associated with an increased risk of cerebral metastasis. Patients who underwent surgery at the primary tumour site followed by lymph node dissection exhibited significantly reduced risks of cerebral metastasis. Radiation therapy emerged as a risk factor for cerebral metastasis. Factors such as income, regional distribution, and tumor size were not significantly associated with cerebral metastasis. Based on the forest plot (Figure 4), it can be visually discerned that NMM, being unmarried, tumours located on the trunk, and undergoing radiation therapy are risk factors for brain metastasis, whereas being female, non-Caucasian, and undergoing surgery at the primary site followed by lymph node dissection serve as protective factors against cerebral metastasis. Leveraging these independent predictors, an interactive nomogram (Figure 5) was developed to calculate the probability of cerebral metastasis for individual patients, offering a tailored risk assessment based on their specific circumstances.
| Table 3 Univariate and multivariate analyses of melanoma brain metastasis | |||||
|---|---|---|---|---|---|
| Dependent: Mets_brain | No (N=8824) | Yes (N=2394) | OR (univariable) | OR (multivariable) | |
| Hist_behav | NMM | 685 (7.8%) | 132 (5.5%) | ||
| LMM | 396 (4.5%) | 8 (0.3%) | 0.10 (0.05-0.22, p<0.001) | 0.19 (0.09-0.40, p<0.001) | |
| SSMM | 1568 (17.8%) | 38 (1.6%) | 0.13 (0.09-0.18, p<0.001) | 0.20 (0.13-0.30, p<0.001) | |
| ALMM | 58 (0.7%) | 4 (0.2%) | 0.36 (0.13-1.00, p=0.051) | 0.53 (0.17-1.70, p=0.288) | |
| MMNOS | 6117 (69.3%) | 2212 (92.4%) | 1.88 (1.55-2.28, p<0.001) | 0.88 (0.67-1.15, p=0.332) | |
| Age (years) | Mean (SD) | 65.2±15.2 | 63.5±13.9 | 0.99 (0.99-1.00, p<0.001) | 1.00 (0.99-1.00, p=0.039) |
| Sex | Male | 5606 (63.5%) | 1722 (71.9%) | ||
| Female | 3218 (36.5%) | 672 (28.1%) | 0.68 (0.62-0.75, p<0.001) | 0.86 (0.76-0.97, p=0.016) | |
| Race | White | 8066 (91.4%) | 2338 (97.7%) | ||
| Black | 112 (1.3%) | 11 (0.5%) | 0.34 (0.18-0.63, p<0.001) | 0.25 (0.13-0.51, p<0.001) | |
| Other | 646 (7.3%) | 45 (1.9%) | 0.24 (0.18-0.33, p<0.001) | 0.56 (0.39-0.80, p=0.002) | |
| Marital | Married | 3895 (44.1%) | 1335 (55.8%) | ||
| Single | 1248 (14.1%) | 489 (20.4%) | 1.14 (1.01-1.29, p=0.031) | 1.38 (1.18-1.61, p<0.001) | |
| Other | 3681 (41.7%) | 570 (23.8%) | 0.45 (0.41-0.50, p<0.001) | 0.88 (0.77-1.01, p=0.066) | |
| Income | Low | 2157 (24.4%) | 544 (22.7%) | ||
| Mediate | 4060 (46%) | 1110 (46.4%) | 1.08 (0.97-1.22, p=0.169) | 1.01 (0.86-1.18, p=0.923) | |
| High | 2607 (29.5%) | 740 (30.9%) | 1.13 (0.99-1.27, p=.063) | 1.02 (0.86-1.21, p=0.844) | |
| Distribution | Urban | 7460 (84.5%) | 2087 (87.2%) | ||
| Rural | 1364 (15.5%) | 307 (12.8%) | 0.80 (0.70-0.92, p=0.001) | 0.97 (0.81-1.17, p=0.772) | |
| Size (mm) | Mean (SD) | 7.1±36.8 | 6.7±43.8 | 1.00 (1.00-1.00, p=0.666) | 1.00 (1.00-1.00, p=0.405) |
| Site | HF | 1462 (16.6%) | 193 (8.1%) | ||
| Truck | 5085 (57.6%) | 1970 (82.3%) | 2.93 (2.50-3.44, p<0.001) | 1.63 (1.33-1.98, p<0.001) | |
| Limbs | 2233 (25.3%) | 228 (9.5%) | 0.77 (0.63-0.95, p=0.013) | 1.00 (0.79-1.27, p=0.992) | |
| Other | 44 (0.5%) | 3 (0.1%) | 0.52 (0.16-1.68, p=0.272) | 0.41 (0.11-1.49, p=0.175) | |
| Surg_Prim_Site | No | 4614 (52.3%) | 1929 (80.6%) | ||
| Yes | 4210 (47.7%) | 465 (19.4%) | 0.26 (0.24-0.29, p<0.001) | 0.59 (0.50-0.69, p<0.001) | |
| Reg_LN_Sur | No | 7524 (85.3%) | 2163 (90.4%) | ||
| Yes | 1300 (14.7%) | 231 (9.6%) | 0.62 (0.53-0.72, p<0.001) | 0.51 (0.43-0.62, p<0.001) | |
| Radiotherapy | No | 7918 (89.7%) | 800 (33.4%) | ||
| Yes | 906 (10.3%) | 1594 (66.6%) | 17.41 (15.61-19.42, p<0.001) | 13.08 (11.64-14.71, p<0.001) | |
| Chemotherapy | No | 7732 (87.6%) | 1763 (73.6%) | ||
| Yes | 1092 (12.4%) | 631 (26.4%) | 2.53 (2.27-2.83, p<0.001) | 1.10 (0.96-1.27, p=0.180) | |
| Breslow_thickness (mm) | Mean (SD) | 1.0±2.2 | 0.7±2.0 | 0.93 (0.90-0.95, p<0.001) | 1.02 (0.99-1.06, p=0.230) |
| Ulceration | No | 7765 (88%) | 2167 (90.5%) | ||
| Yes | 1059 (12%) | 227 (9.5%) | 0.77 (0.66-0.89, p<0.001) | 1.13 (0.88-1.44, p=0.345) | |
| Hist_behav: Histological behaviour; NMM: Nodular metastatic melanoma; LMM: Lentigo maligna metastatic melanoma; SSMM: Superficial spreading metastatic melanoma; ALMM: Acral lentiginous metastatic melanoma, malignant; MMNOS: Malignant metastatic melanoma, NOS; HF: Head and Face; Surg_Prim_Site : surgery of the primary site; Reg_LN_Sur: scope of regional lymph node dissection post-surgery | |||||
 |
| Figure 4 Forest plot of brain metastasis in melanoma. |
 |
| Figure 5 Interactive nomogram for predicting the risk of brain metastasis. Hist_behav: Histological behaviour; Surg_Prim_Site: surgery of primary site; Reg_LN_Sur: Scope regional lymph nodes removed after surgery. |
Analysis and prediction of risk factors for liver metastasis
Employing both univariate and multivariate logistic regression analyses, this study elucidates the determinants of hepatic metastasis in melanoma. The findings underscore histological behave, race, marital status, tumour location, primary site surgery, postoperative lymph node dissection, chemotherapy, Breslow thickness, and ulceration as independent predictors of liver metastasis (Table 4). Notably, NMM types exhibited a heightened propensity for liver metastasis compared to others. In comparison to white individuals, those of non-white ethnicity were found to have a reduced likelihood of developing hepatic metastases. Unmarried individuals demonstrated a lower probability of liver metastasis. The risk of hepatic metastasis was significantly increased in tumours located on the trunk. Individuals undergoing surgery at the primary tumour site and postoperative lymph node dissection showed a markedly reduced risk of liver metastasis. Chemotherapy, Breslow thickness, and the presence of ulceration within the tumor were identified as risk factors for hepatic metastasis. Factors such as sex, income, geographic distribution, and tumour size were deemed to have negligible impact on the incidence of liver metastasis. Forest plots (Figure 6) visually highlight NMM, tumours located on the trunk, chemotherapy, Breslow thickness, and ulceration as risk factors for hepatic metastasis, whereas non-white race, being unmarried, and undergoing primary site surgery and postoperative lymph node dissection serve as protective factors against bone metastasis. Based on these independent factors, an interactive nomogram (Figure 7) was developed, enabling the calculation of individual patients' risk probabilities for liver metastasis according to their specific characteristics.
| Table 4 Univariate and multivariate analyses of melanoma liver metastasis | |||||
|---|---|---|---|---|---|
| Dependent: Mets_liver | No (N=9475) | Yes (N=1743) | OR (univariable) | OR (multivariable) | |
| Hist_behav | NMM | 687 (7.3%) | 130 (7.5%) | ||
| LMM | 400 (4.2%) | 4 (0.2%) | 0.05 (0.02-0.14, p<0.001) | 0.12 (0.04-0.32, p<0.001) | |
| SSMM | 1579 (16.7%) | 27 (1.5%) | 0.09 (0.06-0.14, p<0.001) | 0.20 (0.13-0.31, p<0.001) | |
| ALMM | 56 (0.6%) | 6 (0.3%) | 0.57 (0.24-1.34, p=0.196) | 0.99 (0.41-2.42, p=0.985) | |
| MMNOS | 6753 (71.3%) | 1576 (90.4%) | 1.23 (1.01-1.50, p=0.035) | 1.21 (0.94-1.54, p=0.133) | |
| Age (years) | Mean (SD) | 64.7±15.2 | 65.5±13.6 | 1.00 (1.00-1.01, p=0.058) | 1.00 (1.00-1.01, p=0.175 |
| Sex | Male | 6101 (64.4%) | 1227 (70.4%) | ||
| Female | 3374 (35.6%) | 516 (29.6%) | 0.76 (0.68-0.85, p<0.001) | 0.92 (0.81-1.03, p=0.144) | |
| Race | White | 8727 (92.1%) | 1677 (96.2%) | ||
| Black | 97 (1%) | 26 (1.5%) | 1.39 (0.90-2.16, p=0.135) | 1.27 (0.81-1.98, p=0.301) | |
| Other | 651 (6.9%) | 40 (2.3%) | 0.32 (0.23-0.44, p<0.001) | 0.63 (0.44-0.88, p=0.007) | |
| Marital | Married | 4265 (45%) | 965 (55.4%) | ||
| Single | 1396 (14.7%) | 341 (19.6%) | 1.08 (0.94-1.24, p=0.275) | 1.07 (0.92-1.24, p=0.385) | |
| Other | 3814 (40.3%) | 437 (25.1%) | 0.51 (0.45-0.57, p<.001) | 0.80 (0.70-0.91, p<0.001) | |
| Income | Low | 2329 (24.6%) | 372 (21.3%) | ||
| Mediate | 4347 (45.9%) | 823 (47.2%) | 1.19 (1.04-1.35, p=0.012) | 1.10 (0.94-1.28, p=0.222) | |
| High | 2799 (29.5%) | 548 (31.4%) | 1.23 (1.06-1.41, p=0.005) | 1.12 (0.95-1.32, p=0.180) | |
| Distribution | Urban | 8019 (84.6%) | 1528 (87.7%) | ||
| Rural | 1456 (15.4%) | 215 (12.3%) | 0.77 (0.66-0.90, p=0.001) | 0.95 (0.80-1.14, p=0.611) | |
| Size (mm) | Mean (SD) | 6.7±35.7 | 9.0 50.3 | 1.00 (1.00-1.00, p=0.029) | 1.00 (1.00-1.00, p=0.171) |
| Site | HF | 1479 (15.6%) | 176 (10.1%) | ||
| Truck | 5687 (60%) | 1368 (78.5%) | 2.02 (1.71-2.39, p<0.001) | 1.30 (1.08-1.57, p=0.005) | |
| Limbs | 2272 (24%) | 189 (10.8%) | 0.70 (0.56-0.87, p=0.001) | 0.78 (0.62-0.98, p=0.033) | |
| Other | 37 (0.4%) | 10 (0.6%) | 2.27 (1.11-4.65, p=0.025) | 1.44 (0.69-3.02, p=0.337) | |
| Surg_Prim_Site | No | 5181 (54.7%) | 1362 (78.1%) | ||
| Yes | 4294 (45.3%) | 381 (21.9%) | 0.34 (0.30-0.38, p<0.001) | 0.41 (0.35-0.48, p<0.001) | |
| Reg_LN_Sur | No | 8161 (86.1%) | 1526 (87.6%) | ||
| Yes | 1314 (13.9%) | 217 (12.4%) | 0.88 (0.76-1.03, p=0.113) | 0.83 (0.71-0.99, p=0.034) | |
| Radiotherapy | No | 7514 (79.3%) | 1204 (69.1%) | ||
| Yes | 1961 (20.7%) | 539 (30.9%) | 1.72 (1.53-1.92, p<0.001) | 1.03 (0.91-1.16, p=0.631) | |
| Chemotherapy | No | 8221 (86.8%) | 1274 (73.1%) | ||
| Yes | 1254 (13.2%) | 469 (26.9%) | 2.41 (2.14-2.72, p<0.001) | 1.75 (1.54-1.99, p<0.001) | |
| Breslow_thickness (mm) | Mean (SD) | 0.9±2.1 | 1.1±2.5 | 1.04 (1.01-1.06, p=0.002) | 1.12 (1.08-1.15, p<0.001) |
| Ulceration | No | 8434 (89%) | 1498 (85.9%) | ||
| Yes | 1041 (11%) | 245 (14.1%) | 1.33 (1.14-1.54, p<0.001) | 1.52 (1.22-1.89, p<0.001) | |
| Hist_behav: Histological behaviour; NMM: Nodular metastatic melanoma; LMM: Lentigo maligna metastatic melanoma; SSMM: Superficial spreading metastatic melanoma; ALMM: Acral lentiginous metastatic melanoma, malignant; MMNOS: Malignant metastatic melanoma, NOS; HF: Head and Face; Surg_Prim_Site: surgery of the primary site; Reg_LN_Sur: scope of regional lymph node dissection post-surgery | |||||
 |
| Figure 6 Forest plot of liver metastasis in melanoma. |
 |
| Figure 7 Interactive nomogram for predicting the risk of Liver metastasis. Hist_behav: Histological behaviour; Surg_Prim_Site: surgery of primary site; Reg_LN_Sur: Scope regional lymph nodes removed after surgery. |
Analysis and prediction of risk factors for lung metastasis
Utilising both univariate and multivariate logistic regression analyses, this study elucidates the determinants of melanoma metastasis to the lungs. The investigation identifies histological behaviour, age, sex, race, marital status, tumour location, surgical intervention at the primary site, postoperative lymph node dissection, chemotherapy, Breslow thickness, and ulceration as independent factors influencing lung metastasis (Table 5). Notably, NMM types exhibit a heightened predisposition towards lung metastasis. The analysis further reveals an augmented risk of lung metastasis with advancing age, whereas females demonstrate a reduced risk compared to their male counterparts. Compared to Caucasians, individuals of non-Caucasian ethnicity have a lower probability of lung metastasis. Unmarried individuals show a decreased likelihood of lung metastasis. Tumours located on the trunk are associated with an increased risk of lung metastasis. Surgical removal of the primary tumour and subsequent lymph node dissection significantly mitigate the risk of lung metastasis. Radiation therapy, chemotherapy, Breslow thickness, and the presence of ulceration within the tumour emerge as risk factors for lung metastasis. Conversely, income, geographical distribution, and tumour size exert negligible influence on the incidence of lung metastasis. Forest plots (Figure 8) visually underscore NMM, age, trunk location of the tumour, radiation therapy, chemotherapy, Breslow thickness, and ulceration as risk factors for lung metastasis. In contrast, female gender, non-Caucasian ethnicity, being unmarried, undergoing surgery at the primary site, and postoperative lymph node dissection serve as protective factors against lung metastasis. Building on these independent factors, we developed an interactive nomogram (Figure 9), which, based on individual patient characteristics, facilitates the estimation of lung metastasis risk probabilities.
| Table 5 Univariate and multivariate analyses of melanoma lung metastasis | |||||
|---|---|---|---|---|---|
| Dependent: Mets_lung | No (N=7904) | Yes (N=3314) | OR (univariable) | OR (multivariable) | |
| Hist_behav | NMM | 580 (7.3%) | 237 (7.2%) | ||
| LMM | 384 (4.9%) | 20 (0.6%) | 0.13 (0.08-0.20, p<0.001) | 0.26 (0.16-0.42, p<0.001) | |
| SSMM | 1541 (19.5%) | 65 (2%) | 0.10 (0.08-0.14, p<0.001) | 0.22 (0.16-0.30, p<0.001) | |
| ALMM | 49 (0.6%) | 13 (0.4%) | 0.65 (0.35-1.22, p=0.179) | 1.19 (0.62-2.31, p=0.603) | |
| MMNOS | 5350 (67.7%) | 2979 (89.9%) | 1.36 (1.16-1.60, p<0.001) | 1.10 (0.90-1.34, p=0.354) | |
| Age (years) | Mean (SD) | 64.2±15.5 | 66.2±13.6 | 1.01 (1.01-1.01, p<0.001) | 1.01 (1.01-1.02, p<0.001) |
| Sex | Male | 4961 (62.8%) | 2367 (71.4%) | ||
| Female | 2943 (37.2%) | 947 (28.6%) | 0.67 (0.62-0.74, p<0.001) | 0.87 (0.79-0.96, p=0.004) | |
| Race | White | 7196 (91%) | 3208 (96.8%) | ||
| Black | 87 (1.1%) | 36 (1.1%) | 0.93 (0.63-1.37, p=0.709) | 0.87 (0.58-1.31, p=0.510) | |
| Other | 621 (7.9%) | 70 (2.1%) | 0.25 (0.20-0.32, p<0.001) | 0.55 (0.42-0.73, p<0.001) | |
| Marital | Married | 3324 (42.1%) | 1906 (57.5%) | ||
| Single | 1129 (14.3%) | 608 (18.3%) | 0.94 (0.84-1.05, p=0.279) | 1.03 (0.91-1.16, p=0.692) | |
| Other | 3451 (43.7%) | 800 (24.1%) | 0.40 (0.37-0.44, p<0.001) | 0.66 (0.60-0.74, p<0.001) | |
| Income | Low | 1972 (24.9%) | 729 (22%) | ||
| Mediate | 3622 (45.8%) | 1548 (46.7%) | 1.16 (1.04-1.28, p=0.006) | 1.04 (0.92-1.18, p=0.526) | |
| High | 2310 (29.2%) | 1037 (31.3%) | 1.21 (1.09-1.36, p<0.001) | 1.06 (0.93-1.22, p=0.386) | |
| Distribution | Urban | 6640 (84%) | 2907 (87.7%) | ||
| Rural | 1264 (16%) | 407 (12.3%) | 0.74 (0.65-0.83, p<0.001) | 0.88 (0.76-1.03, p=0.104) | |
| Size (mm) | Mean (SD) | 6.5±36.7 | 8.5±42.1 | 1.00 (1.00-1.00, p=0.015) | 1.00 (1.00-1.00, p=0.036) |
| Site | HF | 1310 (16.6%) | 345 (10.4%) | ||
| Truck | 4469 (56.5%) | 2586 (78%) | 2.20 (1.93-2.50, p<0.001) | 1.44 (1.24-1.67, p<0.001) | |
| Limbs | 2091 (26.5%) | 370 (11.2%) | 0.67 (0.57-0.79, p<0.001) | 0.81 (0.68-0.97, p=0.022) | |
| Other | 34 (0.4%) | 13 (0.4%) | 1.45 (0.76-2.78, p=0.261) | 1.02 (0.52-2.02, p=0.945) | |
| Surg_Prim_Site | No | 4017 (50.8%) | 2526 (76.2%) | ||
| Yes | 3887 (49.2%) | 788 (23.8%) | 0.32 (0.29-0.35, p<0.001) | 0.50 (0.44-0.57, p<0.001) | |
| Reg_LN_Sur | No | 6774 (85.7%) | 2913 (87.9%) | ||
| Yes | 1130 (14.3%) | 401 (12.1%) | 0.83 (0.73-0.93, p=0.002) | 0.76 (0.66-0.87, p<0.001) | |
| Radiotherapy | No | 6622 (83.8%) | 2096 (63.2%) | ||
| Yes | 1282 (16.2%) | 1218 (36.8%) | 3.00 (2.74-3.29, p<0.001) | 1.84 (1.66-2.03, p<0.001) | |
| Chemotherapy | No | 7032 (89%) | 2463 (74.3%) | ||
| Yes | 872 (11%) | 851 (25.7%) | 2.79 (2.51-3.09, p<0.001) | 1.84 (1.64-2.06, p<0.001) | |
| Breslow_thickness (mm) | Mean (SD) | 0.9±2.1 | 1.0±2.3 | 1.00 (0.98-1.02, p=0.895) | 1.05 (1.02-1.08, p<0.001) |
| Ulceration | No | 7061 (89.3%) | 2871 (86.6%) | ||
| Yes | 843 (10.7%) | 443 (13.4%) | 1.29 (1.14-1.46, p<0.001) | 1.76 (1.47-2.10, p<0.001) | |
| Hist_behav: Histological behaviour; NMM: Nodular metastatic melanoma; LMM: Lentigo maligna metastatic melanoma; SSMM: Superficial spreading metastatic melanoma; ALMM: Acral lentiginous metastatic melanoma, malignant; MMNOS: Malignant metastatic melanoma, NOS; HF: Head and Face; Surg_Prim_Site: surgery of the primary site; Reg_LN_Sur: scope of regional lymph node dissection post-surgery | |||||
 |
| Figure 8 Forest plot of lung metastasis in melanoma. |
 |
| Figure 9 Interactive nomogram for predicting the risk of Lung metastasis. Hist_behav: Histological behaviour; Surg_Prim_Site: surgery of primary site; Reg_LN_Sur: Scope regional lymph nodes removed after surgery. |
Prognosis analysis of metastatic melanoma
Based on both univariate and multivariate logistic regression analyses of factors influencing the prognosis of metastatic melanoma (Table 6), our findings reveal that histological behaviour, age, sex, race, marital status, income, tumour size, tumour site, surgery at the primary site, radiotherapy, chemotherapy, Breslow thickness, and the presence of ulceration serve as independent prognostic factors for metastatic melanoma. Compared to NMM, patients with LMM and SSMM exhibit superior prognoses. Prognosis deteriorates with increasing age and Breslow thickness. Females experience more favourable outcomes than males. Relative to Caucasians, African Americans have a poorer prognosis, while patients of other ethnicities fare better. Unmarried patients exhibit worse prognoses compared to their married counterparts. A higher income is associated with a better prognosis. Surprisingly, tumours located on the head, face, or limbs have better outcomes compared to other sites. Patients undergoing surgery at the primary site followed by lymph node dissection demonstrate significantly improved prognoses. Conversely, the administration of radiotherapy, chemotherapy, and the presence of tumour ulceration are associated with a marked decline in prognosis. The size of the tumor and its regional distribution do not significantly impact prognosis.
| Table 6 Univariate and multivariate analysis of prognostic factors in metastatic melanoma | |||||
|---|---|---|---|---|---|
| Dependent: Status | Alive (N=4760) | death (N=6458) | OR (univariable) | OR (multivariable) | |
| Hist_behav | NMM | 285 (6%) | 532 (8.2%) | ||
| LMM | 287 (6%) | 117 (1.8%) | 0.22 (0.17-0.28, p<0.001) | 0.34 (0.25-0.46, p<0.001) | |
| SSMM | 1309 (27.5%) | 297 (4.6%) | 0.12 (0.10-0.15, p<0.001) | 0.27 (0.21-0.33, p<0.001) | |
| ALMM | 28 (0.6%) | 34 (0.5%) | 0.65 (0.39-1.09, p=0.105) | 1.14 (0.64-2.04, p=0.658) | |
| MMNOS | 2851 (59.9%) | 5478 (84.8%) | 1.03 (0.89-1.20, p=0.707) | 1.01 (0.83-1.22, p=0.949) | |
| Age (years) | Mean (SD) | 60.5±15.0 | 68.1±14.1 | 1.04 (1.03-1.04, p<0.001) | 1.04 (1.04-1.05, p<0.001) |
| Sex | Male | 2852 (59.9%) | 4476 (69.3%) | ||
| Female | 1908 (40.1%) | 1982 (30.7%) | 0.66 (0.61-0.72, p<0.001) | 0.85 (0.77-0.93, p<0.001) | |
| Race | White | 4181 (87.8%) | 6223 (96.4%) | ||
| Black | 29 (0.6%) | 94 (1.5%) | 2.18 (1.43-3.31, p<0.001) | 1.84 (1.18-2.88, p=0.008) | |
| Other | 550 (11.6%) | 141 (2.2%) | 0.17 (0.14-0.21, p<0.001) | 0.35 (0.28-0.44, p<0.001) | |
| Marital | Married | 1982 (41.6%) | 3248 (50.3%) | ||
| Single | 592 (12.4%) | 1145 (17.7%) | 1.18 (1.05-1.32, p=0.004) | 1.78 (1.56-2.03, p<0.001) | |
| Other | 2186 (45.9%) | 2065 (32%) | 0.58 (0.53-0.63, p<0.001) | 1.05 (0.95-1.17, p=0.328) | |
| Income | Low | 1084 (22.8%) | 1617 (25%) | ||
| Mediate | 2173 (45.7%) | 2997 (46.4%) | 0.92 (0.84-1.02, p=0.105) | 0.80 (0.71-0.91, p<0.001) | |
| High | 1503 (31.6%) | 1844 (28.6%) | 0.82 (0.74-0.91, p<0.001) | 0.66 (0.58-0.75, p<0.001) | |
| Distribution | Urban | 4011 (84.3%) | 5536 (85.7%) | ||
| Rural | 749 (15.7%) | 922 (14.3%) | 0.89 (0.80-0.99, p=0.032) | 0.91 (0.79-1.04, p=0.165) | |
| Size (mm) | Mean (SD) | 5.4±29.0 | 8.2±44.0 | 1.00 (1.00-1.00, p<0.001) | 1.00 (1.00-1.00, p=0.024) |
| Site | HF | 764 (16.1%) | 891 (13.8%) | ||
| Truck | 2535 (53.3%) | 4520 (70%) | 1.53 (1.37-1.70, p<0.001) | 1.09 (0.95-1.24, p=0.238) | |
| Limbs | 1450 (30.5%) | 1011 (15.7%) | 0.60 (0.53-0.68, p<0.001) | 0.80 (0.69-0.93, p=0.003) | |
| Other | 11 (0.2%) | 36 (0.6%) | 2.81 (1.42-5.55, p=0.003) | 1.52 (0.74-3.13, p=0.255) | |
| Surg_Prim_Site | No | 2067 (43.4%) | 4476 (69.3%) | ||
| Yes | 2693 (56.6%) | 1982 (30.7%) | 0.34 (0.31-0.37, p<0.001) | 0.47 (0.42-0.52, p<0.001) | |
| Reg_LN_Sur | No | 4120 (86.6%) | 5567 (86.2%) | ||
| Yes | 640 (13.4%) | 891 (13.8%) | 1.03 (0.92-1.15, p=0.592) | 0.95 (0.83-1.08, p=0.421) | |
| Radiotherapy | No | 4189 (88%) | 4529 (70.1%) | ||
| Yes | 571 (12%) | 1929 (29.9%) | 3.12 (2.82-3.46, p<0.001) | 2.07 (1.85-2.32, p<0.001) | |
| Chemotherapy | No | 4375 (91.9%) | 5120 (79.3%) | ||
| Yes | 385 (8.1%) | 1338 (20.7%) | 2.97 (2.63-3.35, p<0.001) | 2.26 (1.97-2.59, p<0.001) | |
| Breslow_thickness (mm) | Mean (SD) | 0.8±1.9 | 1.0±2.4 | 1.04 (1.02-1.06, p<0.001) | 1.05 (1.02-1.08, p<0.001) |
| Ulceration | No | 4367 (91.7%) | 5565 (86.2%) | ||
| Yes | 393 (8.3%) | 893 (13.8%) | 1.78 (1.57-2.02, p<0.001) | 1.76 (1.47-2.10, p<0.001) | |
| Hist_behav: Histological behaviour; NMM: Nodular metastatic melanoma; LMM: Lentigo maligna metastatic melanoma; SSMM: Superficial spreading metastatic melanoma; ALMM: Acral lentiginous metastatic melanoma, malignant; MMNOS: Malignant metastatic melanoma, NOS; HF: Head and Face ; Surg_Prim_Site : surgery of the primary site; Reg_LN_Sur: scope of regional lymph node dissection post-surgery | |||||
Discussion
The incidence of melanoma has been escalating across many countries at an annual rate of 3-7%, predominantly affecting the population aged 60 and above. This demographic trend has consequently led to an increase in cases of metastatic melanoma.10 Despite recent breakthroughs in the treatment of metastatic melanoma, mortality rates remain high,11 posing significant challenges to melanoma management and exerting substantial pressure on public health systems.12
Our comprehensive analysis underscores the intricate interplay among histological types, demographic factors (such as sex, race, and marital status), and therapeutic interventions (including surgery, lymph node dissection, radiotherapy, and chemotherapy) in the metastatic behaviour of melanoma. These variables significantly influence the likelihood of metastasis to the brain, bone, liver, and lungs, highlighting the imperative for personalized management strategies in melanoma patients. Histological types, sex, race, and marital status emerge as independent predictors for recurrences in most metastatic sites (brain, bone, and liver). This suggests a complex interplay between biological, demographic, and potential socio-economic factors during the metastatic progression of melanoma. For instance, the significance of marital status may reflect its role in early detection and adherence to treatment protocols.
The findings of this investigation underscore the role of histological types as independent prognostic factors across various metastatic sites, highlighting the critical influence of melanoma's biological diversity on its metastatic potential. This discovery reinforces the notion that histopathological analysis remains a cornerstone in the assessment of melanoma metastasis and prognosis, potentially guiding the selection of therapeutic approaches. Our comprehensive results demonstrate that the likelihood of metastasis to disparate sites for LMM and SSMM is lower compared to ALMM and NMM, aligning with prior research.13,14 However, the mechanisms underlying these differences remain unelucidated, necessitating further investigation.
Our investigation underscores the pivotal role of primary site surgery and postoperative lymph node dissection in modulating the risk of metastasis. The beneficial impact of surgical intervention and lymph node clearance on metastatic risk highlights the significance of conventional therapeutic approaches in the management of melanoma. It emphasizes the importance of early detection and treatment for metastasis, aligning with findings from existing studies.15 Similarly, surgical treatment at the primary site is markedly advantageous for the prognosis of metastatic melanoma. However, postoperative lymph node dissection has not demonstrated a positive effect on prognosis. Two randomised trials have indicated that complete lymph node dissection (CLND) does not enhance survival rates for patients with a positive sentinel node biopsy, thus, CLND is no longer considered mandatory.16,17 Therefore, while surgery and postoperative lymph node clearance are crucial for the metastasis of melanoma, the decision to proceed with lymph node dissection for prognosis should be tailored to the individual patient's circumstances.
In an integrative overview, the influence of age on the metastasis of melanoma to various anatomical sites appears to be minimal, in contrast to its significant impact on prognosis. It is widely acknowledged that with advancing age, there is a gradual decline in bodily functions and the performance of various organs,18 leading to unfavourable outcomes under the onslaught of cancer. However, age does not serve as a meaningful determinant in the metastatic spread of tumours. Additionally, demographic factors such as sex, race, marital status, and socioeconomic status may influence the risk of metastasis through potential biological differences, access to care, and the availability of social support systems. Addressing disparities in healthcare opportunities and bolstering support mechanisms may be crucial in improving the prognosis for patients with melanoma.
Sex, marital status, and ethnicity exert significant impacts across all metastatic sites, with females and non-Caucasians exhibiting lower risk of metastasis and more favourable prognoses. This disparity may be attributed to variations in estrogen levels within females and their heightened attentiveness towards their bodily health,18 though the precise mechanisms warrant further exploration. Conversely, Caucasians present higher incidence rates but often benefit from greater access to healthcare resources, subsequently mitigating the likelihood of metastases. Prognostically, compared to Caucasians, African Americans face worse outcomes, whereas individuals of other ethnicities fare better. Among different metastatic locations, married individuals demonstrate increased incidence rates in bone, liver, and lung metastases, whereas single status emerges as a risk factor specifically for brain metastases. Moreover, singlehood is identified as a prognostic risk factor in metastatic melanoma. Studies across various cancers indicate that unmarried patients are more susceptible to cancer metastasis and face a significantly elevated risk of cancer-induced mortality, potentially due to poorer adherence to treatment protocols and reduced social support available to them.19-21
The therapeutic efficacies of radiotherapy and chemotherapy manifest distinct significance across various metastatic sites; however, this study elucidates that both modalities predominantly act as risk factors for melanoma metastasis. Such discrepancy underscores the imperative to tailor treatment methodologies based on a comprehensive risk profile assessment of metastatic potential. Predominantly, radiotherapy and chemotherapy are administered to patients with advanced-stage or high-risk melanoma22 - a demographic inherently associated with an elevated risk of metastasis, suggesting that the correlation between these treatments and increased metastasis may not directly stem from the therapeutic interventions themselves. Instead, this relationship could be attributed to baseline characteristics of the patients or poor responses to treatment among certain subtypes. Additionally, the adverse effects associated with radiotherapy and chemotherapy, such as subcutaneous fibrosis, pain, neurotoxicity, and cytotoxicity, may exacerbate the risk of metastasis.23-25 Further research is necessitated to elucidate the precise mechanisms underlying this association, potentially encompassing more in-depth studies on biomarkers, clinical feature analysis, and extended follow-up investigations. In clinical practice, for patients contemplating radiotherapy or chemotherapy, a nuanced discussion regarding the potential risks and benefits, tailored to the individual's circumstances, is advisable.
The biological attributes of metastatic melanoma, notably Breslow thickness and ulceration, play a pivotal role in the dissemination process. The parameters of Breslow thickness and ulceration serve as risk factors in the metastasis to the liver, lungs, and bones, yet they do not act as independent determinants in brain metastasis. A greater Breslow thickness indicates a deeper tumour invasion, thereby elevating the propensity for metastasis.26,27 Similarly, tumours with concurrent ulceration are more prone to infiltrate surrounding lymph nodes, leading to further dissemination. Current research underscores that Breslow thickness and ulceration are paramount in influencing prognosis and metastatic potential, with each factor exerting a mutual influence.28 Moreover, the impact of ulceration varies among patients with melanoma of different thicknesses. Additionally, studies have revealed that the likelihood of tumour metastasis to various sites varies with mutation status; BRAF mutations are associated with initial lymph node metastasis and positivity in sentinel lymph nodes. Mutations in BRAF and NRAS are linked to metastasis in the central nervous system and liver, with NRAS mutations also related to lung metastasis.23 Hence, the co-presence of these factors in patients does not unequivocally designate them as risk factors for metastasis or adverse prognosis.
Investigations into the influence of income, geographical distribution, and tumour size on the metastasis of melanoma revealed that these factors do not hold statistical significance; however, their impact on prognosis is profoundly notable. Notably, income exerts a significant influence on the prognosis, with higher income levels correlating with better outcomes for metastatic melanoma. It is evident that individuals with higher income levels have access to superior healthcare services, facilitating earlier detection and treatment, thereby naturally reducing mortality and metastasis rates.29 Conversely, individuals in rural areas are less likely to undergo recommended surgical interventions, resulting in notably poorer prognoses.30
The capability to predict metastatic risk based on a synthesis of factors enables stratification of patients into distinct risk categories, thereby facilitating the implementation of more personalised follow-up strategies and intervention plans. Understanding the specific factors contributing to the risk of metastasis to various organs allows for the more targeted development of therapeutic approaches, potentially improving outcomes. Insight into these factors enhances the advisory on patient prognosis. Nomograms constructed from these independent factors offer clinicians practical tools for predicting the likelihood of metastasis to specific sites, aiding in making informed decisions regarding monitoring and intervention strategies. This personalised approach, through early detection and treatment, may reduce the risk of distant metastasis, thereby enhancing the prognosis for patients with metastatic tumors, and augmenting survival rates and quality of life for patients with melanoma.
Despite the identification of these influencing factors, the underlying mechanisms remain elusive. Further investigation into the molecular mechanisms behind the factors associated with the observed effects could unveil novel therapeutic targets. Additionally, prospective studies are required to validate the predictive accuracy of the nomograms developed in this study.
Conclusion
This study has successfully established a predictive nomogram for metastatic melanoma. Among the factors influencing metastasis at various sites, histological type, ethnicity, marital status, primary site surgery, and postoperative lymph node dissection were identified as common determinants. These findings underscore the significant impact of these factors on melanoma metastasis. Additionally, these elements also play a crucial role in prognosis. Therefore, in clinical diagnosis and treatment processes, it is imperative to consider these factors carefully and conduct analyses tailored to the specific circumstances of individual patients.
Authors' contributions
All authors had full access to all of the data in the study. Doc. Ruiqing Chen, Hai Yu, Jun Lyu and Liehua Deng take responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design: Ruiqing Chen, Hai Yu. Acquisition, analysis, or interpretation of data: Ruiqing Chen, Hai Yu. Drafting of the manuscript: Ruiqing Chen. Critical revision of the manuscript for important intellectual content: All authors. Statistical analysis: Ruiqing Chen, Hai Yu and Wai-kit Ming. Administrative, technical, or material support: Xinkai Zheng and Linyu. Supervision: Jun Lyu and Liehua Deng. All authors contributed to writing of the manuscript and approved the final version.
Conflicts of Interest
The authors declare no conflict of interest. The funders had no role in the design of the study.
Ethics approval
The NCI SEER study is retrospective in nature, and the ethics committee waived consent due to the study's anonymized data and guarantee of patient privacy.
Data availability statement
Publicly available datasets were analysed in this study. This data can be found at: https://seer.cancer.gov.
Acknowledgments
We thank all SEER database staff and scientists. We are also very grateful to China Medical Education Association for their support.
Funding
This research was funded by Key Scientific Problems and Medical Technical Problems Research Project of China Medical Education Association (grant numbers 2022KTZ009, 2024KTZ014) and Guangdong Provincial Key Laboratory of Traditional Chinese Medicine Informatization (grant number 2021B1212040007), and Jiangmen Science and Technology Bureau (project number 2024YL10007).
References
1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2020. CA Cancer J Clin 2020;70:7-30.
2. Lehrer EJ, Stoltzfus KC, Jones BM, Jones BM, Gusani NJ, Walter V, et al. Trends in Diagnosis and Treatment of Metastatic Cancer in the United States. Am J Clin Oncol 2021;44:572-9.
3. Cagney DN, Martin AM, Catalano PJ, Redig AJ, Lin NU, Lee EQ, et al. Incidence and prognosis of patients with brain metastases at diagnosis of systemic malignancy: a population-based study. Neuro Oncol 2017;19:1511-21.
4. Abdel-Rahman O. Clinical correlates and prognostic value of different metastatic sites in patients with malignant melanoma of the skin: a SEER database analysis. J Dermatolog Treat 2018;29:176-81.
5. Maverakis E, Cornelius LA, Bowen GM, Phan T, Pate EB, Fitzmaurice S, et al. Metastatic melanoma - a review of current and future treatment options. Acta Derm Venereol2015;95:516-24.
6. Napolitano S, Brancaccio G, Argenziano G, Martinelli E, Morgillo F, CiardielloF, et al. It is finally time for adjuvant therapy in melanoma. Cancer Treat Rev 2018;69:101-11.
7. Che W, Li Y, Tsang C, Wang YJ, Chen Z, Wang XU, et al. How to use the Surveillance, Epidemiology, and End Results (SEER) data: research design and methodology. Military Medical Research 2023;10:50.
8. Xu Y, Zheng X, Li Y, Ye X, Cheng H, Wang H, et al. Exploring patient medication adherence and data mining methods in clinical big data: A contemporary review. J Evid Based Med. 2023;16:342-75.
9. Wu WT, Li YJ, Feng AZ, Li L, Huang T, Xu AD, et al. Data mining in clinical big data: the frequently used databases, steps, and methodological models. Mil Med Res 2021:8:44.
10. Whiteman DC, Green AC, Olsen CM. The Growing Burden of Invasive Melanoma: Projections of Incidence Rates and Numbers of New Cases in Six Susceptible Populations through 2031. J Invest Dermatol 2016;136:1161-71.
11. Ghazawi FM, Cyr J, Darwich R, Le M, Rahne E, Moreau L, et al. Cutaneous malignant melanoma incidence and mortality trends in Canada: A comprehensive population-based study. J Am Acad Dermatol 2019;80: 448-59.
12. Elliott TM, Whiteman DC, Olsen CM, Gordon LG. Estimated Healthcare Costs of Melanoma in Australia Over 3 Years Post-Diagnosis. Appl Health Econ Health Policy 2017;15:805-16.
13. Lattanzi M, Lee Y, Simpson D, Morran U, Darvishian F, Kim RH, et al. Primary Melanoma Histologic Subtype: Impact on Survival and Response to Therapy. J Natl Cancer Inst 2019;111:180-8.
14. Lowe L. Metastatic melanoma and rare melanoma variants: a review. Pathology 2023;55:236-44.
15. Enomoto LM, Levine EA, Shen P, Votanopoulos KI. Role of Surgery for Metastatic Melanoma. Surg Clin North Am 2020;100:127-39.
16. Testori A, Blankenstein SA, van Akkooi A. Surgery for Metastatic Melanoma: an Evolving Concept. Curr Oncol Rep 2019;21:98.
17. Leiter U, Stadler R, Mauch C, Hohenberger W, Brockmeyer N, Berking C, et al. Complete lymph node dissection versus no dissection in patients with sentinel lymph node biopsy positive melanoma (DeCOG-SLT): a multicentre, randomised, phase 3 trial. Lancet Oncol 2016;17:757-67.
18. Enninga E, Moser JC, Weaver AL, Markovic SN, Brewer JD, Leontovich AA, et al. Survival of cutaneous melanoma based on sex, age, and stage in the United States, 1992-2011. Cancer Med. 2017;6:2203-12.
19. Ding W, Ruan G, Lin Y, Zhu J, Tu C, Li Z. Dynamic changes in marital status and survival in women with breast cancer: a population-based study. Sci Rep 2021;11:5421.
20. Buja A, Lago L, Lago S, Vinelli A, Zanardo C, Baldo V. Marital status and stage of cancer at diagnosis: A systematic review. Eur J Cancer Care (Engl) 2018;27(1).
21. Xing LX, Zhang J, Shen H, Tang XL, He L, Wu JZ, et al. Association of marital status with stage and survival in patients with mycosis fungoides: A population-based study. Cancer Med 2021;10:7320-9.
22. Howard MD. Melanoma Radiological Surveillance: A Review of Current Evidence and Clinical Challenges. Yale J Biol Med 2020;93:207-13.
23. Adler NR, Wolfe R, Kelly JW, Haydon A, McArthur GA, McLean CA, et al. Tumour mutation status and sites of metastasis in patients with cutaneous melanoma. Br J Cancer 2017;117:1026-35.
24. Wada-Ohno M, Ito T, Furue M. Adjuvant Therapy for Melanoma. Curr Treat Options Oncol 2019;20:63.
25. Jilaveanu LB, Aziz SA, Kluger HM. Chemotherapy and biologic therapies for melanoma: do they work. Clin Dermatol 2009;27:614-25.
26. Longvert C, Saiag P. [Melanoma update]. Rev Med Interne 201;40:178-83.
27. Bunnell AM, Nedrud SM, Fernandes RP. Classification and Staging of Melanoma in the Head and Neck. Oral Maxillofac Surg Clin North Am 2022;34:221-34.
28. Wei X, Wu D, Chen Y, Li H, Zhang R, Yao H, et al. Prognostic value of ulceration varies across Breslow thicknesses and clinical stages in acral melanoma: a retrospective study. Br J Dermatol 2022;86=:977-87.
29. Joshi P, Ogobuiro I, Hernandez A, De La Cruz Ku G, Noe DG, Feun L, et al. Racial and ethnic minorities in lower income brackets are associated with late-stage diagnosis of non-ocular melanoma. Cancer Epidemiol 2023;87:102489.
30. Hernandez AE, Benck KN, Huerta CT, Ogobuiro I, De La Cruz Ku G, Möller MG. Rural Melanoma Patients Have Less Surgery and Higher Melanoma-Specific Mortality. Am Surg 2024;90:510-7.
