Next Article in Journal
Physiopathological Features in a Three-Dimensional In Vitro Model of Hepatocellular Carcinoma: Hypoxia-Driven Oxidative Stress and ECM Remodeling
Previous Article in Journal
Benign and Malignant Tumors of the Hand: Patterns, Pathology, and Surgical Outcomes in a Large Retrospective Cohort
Previous Article in Special Issue
Topical 5% Imiquimod for the Treatment of Superficial and Nodular Periocular Basal Cell Carcinoma: A Systematic Review of Clinical Outcomes, Safety, and Treatment Strategies
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Cancers
Volume 17
Issue 18
10.3390/cancers17183080
cancers-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Development of Nomograms to Predict the Probability of Recurrence at Specific Sites in Patients with Cutaneous Melanoma

by
Eszter Anna Janka
1,2,
Imre Lőrinc Szabó
1,2,
Tünde Toka-Farkas
1,
Lilla Soltész
1,
Zita Szentkereszty-Kovács
1,
Beatrix Ványai
1,
Tünde Várvölgyi
1,
Anikó Kapitány
1,2,
Andrea Szegedi
1,2 and
Gabriella Emri
1,2,*
1
Department of Dermatology, MTA Centre of Excellence, Faculty of Medicine, University of Debrecen, 4032 Debrecen, Hungary
2
HUN-REN-UD Allergology Research Group, University of Debrecen, 4032 Debrecen, Hungary
*
Author to whom correspondence should be addressed.
Cancers 2025, 17(18), 3080; https://doi.org/10.3390/cancers17183080
Submission received: 29 August 2025 / Revised: 17 September 2025 / Accepted: 19 September 2025 / Published: 21 September 2025
(This article belongs to the Special Issue Skin Cancer: Epidemiology, Management and New Therapies)
Browse Figures
Versions Notes

Simple Summary

The risk of recurrence of cutaneous melanoma after surgical removal of the primary tumor is high and varies depending on the location of the recurrence. We analyzed clinical and pathological data from approximately 2000 patients with invasive cutaneous melanoma to identify independent prognostic factors that may be useful in predicting the pattern of recurrence. We have developed nomograms to determine the 3, 5, and 10-year probability of recurrence at specific locations. The nomogram models showed good accuracy in predicting recurrence in (1) lymph nodes, (2) skin, soft tissues (including muscle), and/or non-regional lymph nodes, (3) lung, (4) visceral sites, and (5) brain. Of note, the AJCC 8th edition pT stage and patient sex were independent risk factors for melanoma recurrence at any sites. These nomograms may help clinicians assess individualized recurrence risk and tailor follow-up, as well as make more informed decisions about the treatment of melanoma patients.

Abstract

Background: Risk assessment models are increasingly being used in oncology to improve therapeutic and follow-up decisions for individual patients. Methods: In our study, we used a university hospital registry database containing data on patients diagnosed with invasive cutaneous melanoma between 2000 and 2019 (training cohort: N = 1402; validation cohort: N = 601). Using multivariate Cox regression models, we identified clinicopathological variables that are independent risk factors for melanoma recurrence at specific sites. We then constructed nomograms to predict the probability of recurrence at 3, 5, and 10 years. Results: Age, sex, primary tumor location, histological subtype, Clark invasion level and AJCC pT category were independent prognostic factors for melanoma recurrence in regional lymph nodes. Age, sex, primary tumor location, Clark level of invasion, AJCC pT stage and regional lymph node metastasis were risk factors for skin/soft tissue (including muscle)/non-regional lymph node metastases. We found that AJCC pT category and sex were also independent prognostic factors for melanoma recurrence in the lung, visceral sites, and brain. Furthermore, the nomogram predicting recurrence in the lung and visceral sites incorporated the presence of regional lymph node and skin/soft tissue/non-regional lymph node metastases. ROC curves showed good performance of the nomograms in both the training and validation cohorts. The calibration curve showed a good fit. Conclusion: Our results support the high prognostic value of AJCC pT stage and patient sex, which remained consistent across all melanoma stages, and demonstrate the feasibility of creating nomogram models to predict recurrence risk in melanoma patients.

1. Introduction

Cutaneous melanoma of the skin has attracted the attention of the medical community due to its increasing incidence, poor prognosis, and recent advances in therapy [1,2,3]. Ways of reducing melanoma-related morbidity and mortality include improving primary prevention, early detection and primary care, appropriate follow-up methods, and effective treatment for patients at high risk of metastasis and those with metastasis [4,5,6]. Prognostic considerations are highly relevant to the proper management of patients with melanoma, with risk prediction models remaining a research focus [7,8,9,10,11,12]. Although molecular prognostic factors such as germline and somatic driver mutations, gene expression profiles, epigenetic alterations, and immunological characteristics have been extensively studied, the complex interplay between these factors poses challenges for interpreting results [2,4,5,8,13,14]. In addition, the requirement for appropriate tissue samples and molecular assays to assess these markers limits their clinical utility [2]. Clinicopathological factors are readily available and their systematic evaluation within an evidence-based framework can provide clinicians with a greater degree of confidence in formulating patient management recommendations [4,5]. In order to determine whether the prognosis for melanoma patients can be improved by taking into account clinicopathological factors beyond the American Joint Committee on Cancer’s 8th edition TNM classification, we require reliable clinical data on a sufficient number of patients, as well as an evaluation of the data using appropriate multivariable statistical models [10,11,15]. The individual and combined effects of risk factors on time to disease recurrence and death can be assessed using a Cox regression model, which provides an estimation of the relative risk of a risk factor’s presence compared to its absence while adjusting for other risk factors [16]. The present study aims to contribute data collected from a Central European melanoma patient population to research on prognostic factors in melanoma. To increase the clinical impact, nomograms incorporating multiple risk factors were constructed to provide individual risk predictions for disease recurrence [17].

2. Materials and Methods

2.1. Study Population

This retrospective study included a total of 2003 patients diagnosed with invasive cutaneous melanoma between 2000 and 2019, who were followed up until 14 June 2025 or the date of death. The melanoma register data were collected from the integrated hospital information system used at the University of Debrecen (MedSolution and UDMed). The inclusion criterion was a melanoma diagnosis between 2000 and 2019. The exclusion criterion was the absence of clinicopathological and follow-up data for statistical analysis. The study was approved by the Medical Research Council Ethics Committee (certificate number IV/1711-4/2021/EKU).
The register included the following patient data: age; sex; histological subtype of primary melanoma; Breslow tumor thickness; ulceration status; localization; Clark invasion level; and primary tumor (pT) stage, according to the 8th edition of the American Joint Committee on Cancer (AJCC) TNM classification for melanoma [18]. Furthermore, we recorded the time and type of any metastases that developed from primary tumor diagnosis until the end of the follow-up period. Melanoma recurrences were identified through histopathological or imaging examinations (computed tomography (CT), magnetic resonance imaging (MRI) or positron emission tomography with 2-deoxy-2-[fluoro-18]fluoro-D-glucose integrated with computed tomography (18F-FDG PET/CT)), in accordance with routine practice.

2.2. Statistical Analysis

To validate our nomogram models, we used R statistical software to randomly divide our database into training and validation sets at a ratio of 7:3 (training cohort: N = 1402; validation cohort: N = 601). Chi-squared tests were used to analyze the categorical variables.
Recurrence-free survival (RFS) and overall survival (OS) were assessed using the Kaplan–Meier estimator, and survival probability comparisons were performed using a two-sided log-rank test. RFS was calculated by taking into account the time from diagnosis of the primary tumor to recurrence or the final follow-up. OS was calculated by taking into account the time from diagnosis of the primary tumor or first recurrence to the last follow-up or death. For occult melanoma (the AJCC 8th edition T0 category), follow-up was performed from the time of metastasis diagnosis. The median survival time (in years) is presented with a 95% confidence interval (95% CI). Univariate and multivariate hazard ratios (HR) were calculated with their respective 95% upper and lower confidence intervals using Cox regression analysis [19]. The significant independent prognostic factors for melanoma recurrence in (1) lymph nodes, (2) the skin and soft tissue (including muscle) and non-regional lymph nodes, (3) the lung, (4) non-central nervous system visceral sites and (5) the brain, as obtained during multivariate Cox regression, formed the basis for the nomograms. Accordingly, we created five nomograms to predict risk of 3-year, 5-year and 10-year recurrence at specific sites.
During the nomogram validation, the discriminatory ability of the model was evaluated based on the area under the receiver operating characteristic (ROC) curve (AUC) and the C-index. Calibration plots were created to evaluate predictive accuracy using the Bootstrap method with 1000 resamples. Decision curve analysis (DCA) was used to determine the net benefit of the model for patients.
The significance level was p < 0.05 in all cases. Statistical analyses were performed using IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA), and R software, version 4.5.1 (R Foundation for Statistical Computing, Vienna, Austria). A biostatistician, E.A. Janka, supervised the statistical analysis.

3. Results

3.1. Patient and Disease Characteristics

A total of 2003 patients diagnosed with malignant melanoma between 2000 and 2019 were included in the analysis (see Figure 1). In terms of primary tumor characteristics, the most frequent histological subtype was superficial spreading melanoma (SSM), and the most frequent location was the trunk. During follow-up, 57.3% of patients in our study population did not develop metastases. For the 1153 patients who were alive at the end of the observation period, the median follow-up time was 11.4 years from melanoma diagnosis. There were no significant differences between the training (N = 1402) and validation (N = 601) sets for any variables (Table 1).

3.2. Kaplan–Meier Survival Analysis

Kaplan–Meier survival curves were plotted to demonstrate the differences in recurrence-free (RFS) and overall survival (OS) rates among the various risk groups (Figure 2A–E). The median RFS was 0.73 years [0.44; 1.03] for regional lymph node metastasis, 2.91 years [0.66; 5.15] for skin, soft tissue (including muscle) and/or non-regional lymph node metastasis, 7.34 years [5.13; 14.26] for lung metastasis, 11.64 years [8.05; 14.59] for non-central nervous system visceral metastasis and 14.36 years [12.65; 16.07] for brain metastasis. This analysis revealed a link between RFS and the AJCC 8th edition pT stage of the primary tumor for each recurrence type. Additionally, OS probabilities differed according to AJCC pT stage and patient sex (Figure 2F,G). The site of the first recurrence influenced the probability of OS from the time of recurrence (Figure 2H).

3.3. Cox Regression Analysis and Nomogram

Cox regression models were used to evaluate prognostic factors for melanoma recurrence at specific sites. The significant results of the univariate Cox regression analysis were incorporated into the multivariate model. During the multivariate regression analysis, we identified the significant independent factors that were then included in the nomogram.

3.3.1. Cox Regression Analyses of Melanoma Recurrence in Regional Lymph Nodes

Multivariate Cox regression analysis showed no significant difference in the occurrence of regional lymph node metastasis between melanoma cases diagnosed under the age of 40 and those diagnosed over the age of 50 (Table 2). However, the incidence of lymph node metastasis was significantly higher in cases diagnosed in the 40–49 age group than in cases diagnosed under the age of 40 (40–49 years vs. <40 years: hazard ratio (HR) 1.54 [1.11; 2.12]). Men were found to be at a higher risk of lymph node metastasis than women (male vs. female: HR 1.34 [1.13; 1.58]). Lymph node metastasis occurred with similar frequency in primary melanomas of the head and neck, upper extremities, and trunk, but more frequently in primary melanomas of the lower extremities or when the primary tumor site could not be identified (lower extremities vs. head and neck: HR 1.63 [1.23; 2.15], occult vs. head-neck: HR 4.86 [3.02; 6.72]). The incidence of lymph node metastasis was significantly lower in the SSM histological subtype than in the other subtypes, except for LMM (NM vs. SSM: HR 1.79 [1.38; 2.32]). Clark IV and V invasion levels were associated with a significantly higher probability of lymph node metastasis than Clark I-III invasion levels (Clark level V vs. I-III: HR 1.60 [1.20; 2.13]; Clark level IV vs. I-III: HR 1.39 [1.11; 1.74]). The probability of lymph node metastasis was closely correlated with the AJCC T category (T4b vs. T1a: HR 10.66 [7.13; 15.93]; T3b-T4a vs. T1a: HR 8.26 [5.57; 12.23]; T2b-T3a vs. T1a: HR 5.92 [4.02; 8.71]; T1b-T2a vs. T1a: HR 3.39 [2.39; 4.81]). Figure 3 shows a nomogram that integrates the independent variables to predict the probability of regional lymph node metastasis at 3, 5 and 10 years after melanoma diagnosis.

3.3.2. Cox Regression Analyses of Melanoma Recurrence in the Skin, Soft Tissue (Including Muscle), and/or Non-Regional Lymph Nodes

The likelihood of skin, soft tissue (including muscle) and/or non-regional lymph node metastases (equivalent to stage M1a in the 8th edition of the AJCC) was significantly higher in melanoma patients diagnosed at over 70 years of age than in those diagnosed at under 40 years of age (≥70 years vs. <40 years: HR 1.75 [1.23; 2.48]) (Table 3). For other age groups, there was no significant difference in the probability of this type of metastasis between melanoma diagnosed under the age of 40 and melanoma diagnosed at 40 years of age or older. Men were more likely to experience such metastases than women. These metastases occurred with similar frequency in primary melanomas of the head and neck region, extremities, and trunk, but more frequently in occult primary tumors. Histological subtype did not influence the incidence of this type of metastasis. Clark V invasion levels were associated with a significantly higher likelihood of these metastases than Clark I–III invasion levels. The odds of these metastases showed a strong statistical correlation with the AJCC T category. The probability of distant metastasis in the AJCC M1a category was significantly higher in cases of regional lymph node metastasis (yes vs. no: HR5.70 [4.48; 7.25]). Figure 4 shows a nomogram for predicting the probability of recurrence within three, five and ten years after melanoma diagnosis at sites classified as AJCC M1a stage.

3.3.3. Cox Regression Analyses of Melanoma Recurrence in the Lung, Non-Central Nervous System Visceral Sites and the Brain

Using multivariate Cox regression models, we found that the likelihood of lung, visceral or brain metastasis did not differ according to age at primary melanoma diagnosis, primary tumor location, histological subtype or Clark invasion level (Table 4, Table 5 and Table 6). However, the probability of all types of distant metastasis showed a strong statistical correlation with the AJCC T category (lung metastasis: T4b vs. T1a: HR 5.53 [3.29; 9.30]; T3b-T4a vs. T1a: HR 2.93 [1.73; 4.94]; T2b-T3a vs. T1a: HR 2.49 [1.50; 4.13]; T1b-T2a vs. T1a: HR 3.39 1.94 [1.24; 3.02]; T0 vs. T1a: HR 2.32 [1.03; 4.00]; visceral metastasis: T4b vs. T1a: HR 2.46 [1.51; 4.00]; T3b-T4a vs. T1a: HR 1.93 [1.19; 3.12]; T2b-T3a vs. T1a: HR 1.69 [1.05; 2.69]; T0 vs. T1a: HR 2.86 [1.44; 5.66]; brain metastasis: T4b vs. T1a: HR 5.00 [2.73; 9.18]; T3b-T4a vs. T1a: HR 2.82 [1.54; 5.19]; T2b-T3a vs. T1a: HR 2.32 [1.28; 4.20]; T0 vs. T1a: HR 3.29 [1.33; 8.13]) and was significantly higher in men than in women (lung metastasis: male vs. female: HR 1.30 [1.03; 1.63]; visceral metastasis: male vs. female: HR 1.29 [1.04; 1.61]; brain metastasis: male vs. female: HR 1.49 [1.13; 1.96]). The likelihood of lung metastasis was notably higher in cases involving regional lymph node metastasis (yes vs. no: HR 2.95 [2.18; 4.00]), skin and soft tissue (including muscle) and/or non-regional lymph node metastasis (yes vs. no: HR 2.63 [2.02; 3.43]). Visceral metastasis was significantly more likely in cases of regional lymph node metastasis (yes vs. no: HR 2.62 [1.96; 3.51]), skin and soft tissue (including muscle) and/or non-regional lymph node metastasis (yes vs. no: HR 2.31 [1.79; 2.79]) and lung metastasis (yes vs. no: HR 4.00 [3.18; 5.04]). Likewise, the likelihood of brain metastasis was higher in cases involving skin, soft tissue (including muscle) and/or non-regional lymph node metastases (yes vs. no: HR 2.11 [1.53; 2.92]), lung metastases (yes vs. no: HR 2.45 [1.79; 3.36]) and visceral metastases (yes vs. no: HR 2.10 [1.53; 2.89]). Nomograms were created to combine the independent variables and predict the likelihood of lung, visceral and brain metastases occurring within three, five and ten years (Figure 5, Figure 6 and Figure 7).

3.4. Nomogram Validation

The nomograms were evaluated and validated using multiple metrics. First of all, the C-index of the training cohort and validation cohort in case of regional lymph node metastasis nomogram was 0.81 [95% CI 0.78; 0.84] and 0.83 [95% CI 0.79; 0.87], respectively. In the case of distant skin, soft tissue (including muscle) and/or non-regional lymph node metastases, the C-index was 0.84 [0.81; 0.87] in the training cohort and 0.84 [0.80; 0.88] in the validation cohort, while in the case of lung metastasis, it was 0.84 [0.81; 0.87] and 0.83 [0.79; 0.88], for visceral metastasis 0.86 [0.84; 0.90] and 0.85 [0.81; 0.90], and for brain metastasis 0.85 [0.82; 0.89] and 0.82 [0.76; 0.89], respectively.
In the training cohort, the AUC of predicted nomogram for 3-, 5- and 10-year recurrence probability of lymph node metastasis were 0.864 [0.846; 0.882], 0.878 [0.861; 0.895], and 0.887 [0.870; 0.904], while in the validation cohort, the AUC for 3-, 5- and 10-year recurrence probability were 0.881 [0.847; 0.915], 0.877 [0.843; 0.911] and 0.891 [0.858; 0.924] (Figure S1A,B). In case of distant skin, soft tissue (including muscle) and/or non-regional lymph node metastases nomogram the AUC for 3-, 5- and 10-year recurrence probability were 0.898 [0.876; 0.920], 0.921 [0.901; 0.941] and 0.934 [0.916; 0.952] in the training set, and 0.883 [0.846; 0.920], 0.888 [0.852; 0.924] and 0.900 [0.865; 0.935] in the validation set (Figure S1C,D). In case of lung metastases nomogram, the AUC for 3-, 5- and 10-year recurrence probability were 0.884 [0.855; 0.913], 0.890 [0.862; 0.918] and 0.931 [0.908; 0.954] in the training set, and 0.883 [0.840; 0.926], 0.870 [0.826; 0.914] and 0.883 [0.840; 0.926] in the validation set (Figure S1E,F). In case of visceral metastases nomogram, the AUC for 3-, 5- and 10-year recurrence probability were 0.901 [0.876; 0.926], 0.910 [0.886 0.934] and 0.942 [0.923; 0.961] in the training set, and 0.862 [0.817; 0.907], 0.879 [0.836; 0.922] and 0.917 [0.881; 0.953] in the validation set (Figure S1G,H). In the case of brain metastases nomogram the AUC for 3-, 5- and 10-year recurrence probability were 0.881 [0.847; 0.915], 0.899 [0.867; 0.931] and 0.939 [0.914; 0.964] in the training set, and 0.855 [0.798; 0.912], 0.868 [0.813; 0.923] and 0.885 [0.798; 0.912] in the validation set (Figure S1I,J). The C-index and ROC curves indicated that the models had good discriminatory power.
Calibration plots, which were used to predict the probability of recurrence in the training and validation cohorts, showed good agreement for all types of melanoma recurrence (Figure S2A,J). DCA demonstrated the significant net benefit of the predictive model for all nomogram types in both cohorts (Figure S3A,J).

4. Discussion

Cutaneous melanoma can be characterized by an increasing incidence associated with an aging population and high rate of UV exposure, a high cure rate when diagnosed and completely excised at an early stage, and a poor prognosis in advanced cases [1,2,3]. Patterns concerning the probability, timing and site of disease recurrence allow us to categorize patients according to their risk of relapse and their need for follow-up care and treatment [4,5,6,20]. The most powerful prognostic variables are represented in the American Joint Committee on Cancer (AJCC) TNM classification (8th edition) [21]. However, existing prognostic heterogeneity supports continued research for independent, readily available prognostic factors that could improve clinical decision-making. Identifying prognostic factors that are independent of the AJCC T, N and M categories and developing risk prediction models are highly relevant to improving morbidity and mortality outcomes in melanoma patients [22,23]. Furthermore, identifying significant prognostic factors across different databases (geographical regions) can inform molecular research. Recently, several risk prediction models have been published that demonstrate good predictive performance for survival in patients with non-metastatic [11,24] or metastatic [12,25,26,27] melanoma; elderly patients [28,29]; and patients with specified histopathological types [30,31,32] or localizations [10,15,33,34,35,36,37] of melanoma. In most studies, the Surveillance, Epidemiology, and End Results (SEER) database was used to obtain patient demographic and clinical information. The most common independent prognostic factors were age, AJCC stage, primary site, sex, histological subtype and marital status. Selected clinical outcomes included overall survival, cancer-specific survival and, in some studies, recurrence-free survival [9,11,38] and early death [12,27,28]. Notably, the published nomogram models showed that Breslow thickness, ulceration of the primary tumor, histological subtype and primary site were independent predictors of survival in patients with metastatic melanoma [25,26,27]. Our previous studies also found that primary tumor characteristics maintained their prognostic power as independent variables for survival in metastatic melanoma patients treated with immune checkpoint inhibitors or BRAF and MEK inhibitor combinations [39,40].
For the present study, we extracted demographic, clinicopathological and outcome data from a university hospital-based melanoma registry in order to identify independent prognostic factors that are readily available and associated with disease recurrence. We employed univariate and multivariate Cox regression models and then constructed a nomogram model to predict selected patient outcomes. This study is novel in that we aimed to identify independent prognostic factors associated with melanoma recurrence at specific sites, and to construct nomograms to predict recurrence-free survival. We believe that these nomograms could be useful in providing a more personalized prognosis assessment and in supporting clinical decisions regarding adjuvant treatment and follow-up.
We found that the risk of melanoma recurrence in regional lymph nodes can be more accurately assessed by considering additional clinical parameters alongside the thickness and ulceration of the primary tumor. Furthermore, our results suggest that the site and histological subtype of primary melanomas, as well as the Clark invasion level, provide valuable insights into melanoma biology that remain largely unexplored. Although risk prediction models developed to predict sentinel lymph node status using clinicopathological features have been shown to reduce unnecessary sentinel lymph node biopsies (SLNB), there is concern that clinical nomograms may not be accurate enough to provide clear benefits to patients with early-stage melanoma [7,41,42]. The clinicopathological and gene expression profile (CP-GEP) model may be more effective in identifying low-risk melanoma patients who can safely avoid SLNB [8]. This model incorporates two clinicopathological factors: age and Breslow thickness. Notably, the CP-GEP model considers an age of less than 40 years to be an indication for SLNB. However, our study showed that the 40–49 age group was at a higher risk of regional lymph node metastasis than those diagnosed under 40 years of age (although occult and clinically detected lymph node metastasis cases were not considered separately). It is also worth mentioning that integrating artificial intelligence platforms into risk prediction models has recently gained attention in predicting SLNB positivity [43].
As expected, we found that patients’ overall survival estimates depended on the site of metastases [44,45]. Metastasis in regional lymph nodes increased the risk of metastasis at a distant site, which in turn increased the likelihood of metastasis developing in additional sites. It should be noted that the outcomes of patients in this study reflect the era before adjuvant treatment with immune checkpoint inhibitors and BRAF and MEK inhibitor combinations. As previously mentioned, there is evidence that primary tumor characteristics are independent prognostic factors for overall survival in patients with advanced melanoma [25,26,27,39,40]. In this study, we found that the AJCC 8th edition T classification was an independent predictor of the risk of melanoma recurrence at distant sites. Furthermore, while the other features of the primary tumor did not predict the risk of distant metastases, except for the prognostic significance of Clark invasion level and the occult nature of primary melanoma in the occurrence of distant skin, soft tissue (including muscle) and/or non-regional lymph node metastases, we should bear in mind that these characteristics had a prognostic effect on the risk of regional lymph node metastases. This was also an independent predictor of lung and non-central nervous system visceral metastases.
In terms of age, patients over 70 were found to have a relatively lower risk of regional lymph node metastasis than those under 40. However, they exhibited a significantly higher risk of distant skin, soft tissue (including muscle) and/or non-regional lymph node metastases. Interestingly, age-related alterations in lymphatic integrity may explain this finding [46]. Finally, we found that patient sex was an independent prognostic factor for melanoma recurrence in all cases. The prognostic value of sex for survival in melanoma patients has been evaluated in multiple studies, with some reporting it to be an independent prognostic variable and others not [47,48]. However, a recent study employing a machine learning model found that incorporating patient age and sex into the TNM prognostic system was more accurate at predicting cancer-specific survival in patients with cutaneous melanoma than TNM staging alone [22]. A worse prognosis for male patients than female patients with melanoma may be due to differences in hormonal status, as well as sex-related genetic and epigenetic factors that can influence tumor growth and the anti-tumor immune response [49].
This study’s strength lies in its use of multivariate logistic regression analysis to identify independent prognostic factors for disease recurrence in melanoma patients at specified sites. Nomograms were constructed and validated by integrating these variables. A limitation of our study is that it was retrospective and single-center; further prospective validation on independent melanoma cohorts is required. As the definitions of mitotic index and microscopic satellites varied in the histological reports of primary tumors, these prognostic factors could not be included in the nomogram. We were unable to distinguish between occult and clinically detected lymph node metastases because SLNB was not consistently performed on all patients with melanoma measuring 1 mm or above in thickness. Data on interferon therapy, radiotherapy and chemotherapy were excluded from the statistical analysis due to heterogeneity in the indications and therapeutic regimens applied. Furthermore, melanoma-specific survival could not be evaluated as an outcome measure.

5. Conclusions

Stratifying cutaneous melanoma patients according to prognostic biomarkers could improve disease outcomes. Our study suggests that risk prediction models can be successfully constructed using clinicopathological variables, potentially increasing the accuracy of prognostic considerations. However, the created nomograms require further validation before they can be used in daily clinical practice. Moreover, there is an urgent need to identify molecular biomarkers that can be easily measured using standardized assays and incorporated into a multivariate prognostic model.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/cancers17183080/s1, Figure S1. 3-, 5- and 10-year ROC of regional lymph node metastasis nomogram in training cohort (A) and in validation cohort (B), 3-, 5- and 10-year ROC of skin, soft tissue (including muscle) and/or non-regional lymph node metastasis nomogram in training cohort (C) and in validation cohort (D), 3-, 5- and 10-year ROC of lung metastasis nomogram in training cohort (E) and in validation cohort (F), 3-, 5- and 10-year ROC of visceral metastasis nomogram in training cohort (G) and in validation cohort (H), 3-, 5- and 10-year ROC of brain metastasis nomogram in training cohort (I) and in validation cohort (J). Figure S2. 3-, 5- and 10-year calibration plots of regional lymph node metastasis nomogram in training cohort (A) and in validation cohort (B), 3-, 5- and 10-year calibration plots of skin, soft tissue (including muscle) and/or non-regional lymph node metastasis nomogram in training cohort (C) and in validation cohort (D), 3-, 5- and 10-year calibration plots of lung metastasis nomogram in training cohort (E) and in validation cohort (F), 3-, 5- and 10-year calibration plots of visceral metastasis nomogram in training cohort (G) and in validation cohort (H), 3-, 5- and 10-year calibration plots of brain metastasis nomogram in training cohort (I) and in validation cohort (J). Figure S3. 3-, 5- and 10-year DCA of regional lymph node metastasis nomogram in training cohort (A) and in validation cohort (B), 3-, 5- and 10-year DCA of skin, soft tissue (including muscle) and/or non-regional lymph node metastasis nomogram in training cohort (C) and in validation cohort (D), 3-, 5- and 10-year DCA of lung metastasis nomogram in training cohort (E) and in validation cohort (F), 3-, 5- and 10-year DCA of visceral metastasis nomogram in training cohort (G) and in validation cohort (H), 3-, 5- and 10-year DCA of brain metastasis nomogram in training cohort (I) and in validation cohort (J).

Author Contributions

Writing—original draft: E.A.J. and G.E.; data curation: E.A.J., I.L.S., B.V., T.T.-F., A.K., L.S., Z.S.-K. and T.V.; formal analysis: E.A.J.; methodology: E.A.J. and G.E.; statistical analysis: E.A.J.; visualization: E.A.J.; investigation: T.V., I.L.S., L.S., Z.S.-K. and B.V.; supervision: G.E. and A.S.; conceptualization: A.S. and G.E.; writing—review and editing: I.L.S., L.S., G.E., E.A.J., A.K. and A.S. All authors have read and agreed to the published version of the manuscript.

Funding

Supported by the EKÖP-24-4-II. University Research Scholarship Program of the Ministry for Culture and Innovation from the Source of the National Research, Development and Innovation Fund. This research was funded by the National Research Development and Innovation Fund (grant number K-142348).

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Regional Ethics Committee (Certificate number: IV/1711-4/2021/EKU; Approved date: 16 April 2021).

Informed Consent Statement

Informed consent was obtained from all patients involved in the study.

Data Availability Statement

The original contributions presented in the study are included in the article/Supplementary Materials; further inquiries can be directed to the corresponding author/s.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Kim, H.J.; Kim, Y.H. Molecular Frontiers in Melanoma: Pathogenesis, Diagnosis, and Therapeutic Advances. Int. J. Mol. Sci. 2024, 25, 2984. [Google Scholar] [CrossRef]
  2. Waseh, S.; Lee, J.B. Advances in melanoma: Epidemiology, diagnosis, and prognosis. Front. Med. 2023, 10, 1268479. [Google Scholar] [CrossRef] [PubMed]
  3. Arnold, M.; Singh, D.; Laversanne, M.; Vignat, J.; Vaccarella, S.; Meheus, F.; Cust, A.E.; de Vries, E.; Whiteman, D.C.; Bray, F. Global Burden of Cutaneous Melanoma in 2020 and Projections to 2040. JAMA Dermatol. 2022, 158, 495–503. [Google Scholar] [CrossRef] [PubMed]
  4. Garbe, C.; Amaral, T.; Peris, K.; Hauschild, A.; Arenberger, P.; Basset-Seguin, N.; Bastholt, L.; Bataille, V.; Brochez, L.; Del Marmol, V.; et al. European consensus-based interdisciplinary guideline for melanoma. Part 1: Diagnostics—Update 2024. Eur. J. Cancer 2025, 215, 115152. [Google Scholar] [CrossRef]
  5. Amaral, T.; Ottaviano, M.; Arance, A.; Blank, C.; Chiarion-Sileni, V.; Donia, M.; Dummer, R.; Garbe, C.; Gershenwald, J.E.; Gogas, H.; et al. Cutaneous melanoma: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up. Ann. Oncol. 2025, 36, 10–30. [Google Scholar] [CrossRef]
  6. Pathak, S.; Puckett, Y. Cutaneous Malignant Melanoma: Guideline-Based Management and Interprofessional Collaboration; StatPearls Publishing: Treasure Island, FL, USA, 2025. [Google Scholar]
  7. Mesgarzadeh, S.; Amamoo, R.S.; Ameneni, G.; Gong, A.H.; Ayoade, O.O.; Stratton, D.B.; Latour, E.; Yu, W.; Curiel-Lewandrowski, C.; Abraham, I.; et al. Validation of risk prediction models for sentinel lymph node metastasis in melanoma in a high UV index region. JAAD Int. 2025, 21, 61–69. [Google Scholar] [CrossRef] [PubMed]
  8. Weitemeyer, M.B.; Helvind, N.M.; Klausen, S.; Clasen-Linde, E.; Schmidt, G.; Chakera, A.H.; Holmich, L.R. Validation of a Clinicopathologic and Gene Expression Model for Predicting Sentinel Node Metastasis in Melanoma: A Multicenter Danish Cohort Study. J. Surg. Oncol. 2025. [Google Scholar] [CrossRef]
  9. Verver, D.; Rekkas, A.; Garbe, C.; van Klaveren, D.; van Akkooi, A.C.J.; Rutkowski, P.; Powell, B.; Robert, C.; Testori, A.; van Leeuwen, B.L.; et al. The EORTC-DeCOG nomogram adequately predicts outcomes of patients with sentinel node-positive melanoma without the need for completion lymph node dissection. Eur. J. Cancer 2020, 134, 9–18. [Google Scholar] [CrossRef]
  10. Shi, Y.; Mo, R.; Chen, Y.; Ma, Z.; Wen, B.; Tan, Q. Establishment and Validation of Prognostic Nomograms for Nonmetastatic Melanoma of the Limbs-A SEER-Based Study. J. Investig. Surg. 2024, 37, 2401125. [Google Scholar] [CrossRef]
  11. Stassen, R.C.; Maas, C.; van der Veldt, A.A.M.; Lo, S.N.; Saw, R.P.M.; Varey, A.H.R.; Scolyer, R.A.; Long, G.V.; Thompson, J.F.; Rutkowski, P.; et al. Development and validation of a novel model to predict recurrence-free survival and melanoma-specific survival after sentinel lymph node biopsy in patients with melanoma: An international, retrospective, multicentre analysis. Lancet Oncol. 2024, 25, 509–517. [Google Scholar] [CrossRef]
  12. Guo, Q.; Lv, X.; Zheng, Z.; Lan, G. How to predict early mortality in patients with metastatic melanoma. Sci. Rep. 2025, 15, 25395. [Google Scholar] [CrossRef]
  13. Tímár, J.; Ladányi, A. Molecular Pathology of Skin Melanoma: Epidemiology, Differential Diagnostics, Prognosis and Therapy Prediction. Int. J. Mol. Sci. 2022, 23, 5384. [Google Scholar] [CrossRef] [PubMed]
  14. Szabó, I.L.; Emri, G.; Ladányi, A.; Tímár, J. Clinical Applications of the Molecular Landscape of Melanoma: Integration of Research into Diagnostic and Therapeutic Strategies. Cancers 2025, 17, 1422. [Google Scholar] [CrossRef]
  15. Gao, M.; Wu, B.; Bai, X. Establishment and validation of a nomogram model for predicting the specific mortality risk of melanoma in upper limbs based on the SEER database. Sci. Rep. 2024, 14, 9623. [Google Scholar] [CrossRef]
  16. Abd ElHafeez, S.; D’Arrigo, G.; Leonardis, D.; Fusaro, M.; Tripepi, G.; Roumeliotis, S. Methods to Analyze Time-to-Event Data: The Cox Regression Analysis. Oxidative Med. Cell. Longev. 2021, 2021, 1302811. [Google Scholar] [CrossRef]
  17. Balachandran, V.P.; Gonen, M.; Smith, J.J.; DeMatteo, R.P. Nomograms in oncology: More than meets the eye. Lancet Oncol. 2015, 16, e173–e180. [Google Scholar] [CrossRef]
  18. Gershenwald, J.E.; Scolyer, R.A.; Hess, K.R.; Sondak, V.K.; Long, G.V.; Ross, M.I.; Lazar, A.J.; Faries, M.B.; Kirkwood, J.M.; McArthur, G.A.; et al. Melanoma staging: Evidence-based changes in the American Joint Committee on Cancer eighth edition cancer staging manual. CA Cancer J. Clin. 2017, 67, 472–492. [Google Scholar] [CrossRef] [PubMed]
  19. Christensen, E. Multivariate survival analysis using Cox’s regression model. Hepatology 1987, 7, 1346–1358. [Google Scholar] [CrossRef] [PubMed]
  20. Papageorgiou, C.; Apalla, Z.; Manoli, S.M.; Lallas, K.; Vakirlis, E.; Lallas, A. Melanoma: Staging and Follow-Up. Dermatol. Pract. Concept. 2021, 11 (Suppl. 1), e2021162S. [Google Scholar] [CrossRef]
  21. Hynes, M.C.; Nguyen, P.; Groome, P.A.; Asai, Y.; Mavor, M.E.; Baetz, T.D.; Hanna, T.P. A population-based validation study of the 8th edition UICC/AJCC TNM staging system for cutaneous melanoma. BMC Cancer 2022, 22, 720. [Google Scholar] [CrossRef]
  22. Yang, C.Q.; Wang, H.; Liu, Z.; Hueman, M.T.; Bhaskaran, A.; Henson, D.E.; Sheng, L.; Chen, D. Integrating additional factors into the TNM staging for cutaneous melanoma by machine learning. PLoS ONE 2021, 16, e0257949. [Google Scholar] [CrossRef] [PubMed]
  23. Acar, C.; Yuksel, H.C.; Sahin, G.; Acar, F.P.; Gunenc, D.; Karaca, B. Prognostic Utility of the CALLY Index in Metastatic Melanoma: Building a Nomogram for Patients on Anti-PD-1 Therapy. Clin. Transl. Oncol. 2025, 27, 3770–3780. [Google Scholar] [CrossRef] [PubMed]
  24. He, X.Y.; Gao, Y.F.; Huang, Y.P.; Zou, X.W.; Wang, D.; Su, H. Development of a prognostic prediction model for patients with cutaneous malignant melanoma: A study based on the SEER database. Eur. Rev. Med. Pharmacol. Sci. 2022, 26, 9437–9446. [Google Scholar] [CrossRef] [PubMed]
  25. Xi, Q.; Lu, X.; Zhang, J.; Wang, D.; Sun, Y.; Chen, H. A practical nomogram and risk stratification system predicting the cancer-specific survival for patients aged >50 with advanced melanoma. Front. Oncol. 2023, 13, 1166877. [Google Scholar] [CrossRef]
  26. Li, W.; Xiao, Y.; Xu, X.; Zhang, Y. A Novel Nomogram and Risk Classification System Predicting the Cancer-Specific Mortality of Patients with Initially Diagnosed Metastatic Cutaneous Melanoma. Ann. Surg. Oncol. 2021, 28, 3490–3500. [Google Scholar] [CrossRef]
  27. Li, S.; Yin, C.; Yang, X.; Lu, Y.; Wang, C.; Liu, B. Risk factors and predictive models for early death in patients with advanced melanoma: A population-based study. Medicine 2023, 102, e35380. [Google Scholar] [CrossRef]
  28. Lei, Y.; Wang, S.; Chen, J.; Liu, L.; Huang, L.; Wu, X.; Xu, H.; Yang, Y. A novel tool for predicting the risk of cancer-specific early death in older patients with primary malignant melanoma of skin: A population-based analysis. Front. Oncol. 2024, 14, 1387014. [Google Scholar] [CrossRef]
  29. Yan, J.; Wang, H.; Lu, X.; Li, F. Development and validation of a nomogram for elderly patients with ulcerative melanoma. Melanoma Res. 2024, 34, 207–214. [Google Scholar] [CrossRef]
  30. Ji, Q.; Tang, J.; Li, S.; Chen, J. Prognostic model for predicting overall and cancer-specific survival among patients with superficial spreading melanoma: A SEER based study. Medicine 2022, 101, e32521. [Google Scholar] [CrossRef]
  31. Yin, T.; Zhao, Y.; Yang, Y.; Xu, H.; Zheng, D.; Lyu, J.; Fu, G. Nomogram for Predicting Overall Survival in Acral Lentiginous Melanoma: A Population-based Study. Int. J. Gen. Med. 2021, 14, 9841–9851. [Google Scholar] [CrossRef]
  32. Xu, C.; Yu, X.; Ding, Z.; Fang, C.; Gao, M.; Liu, W.; Liu, X.; Yin, C.; Gu, R.; Liu, L.; et al. Artificial intelligence-assisted metastasis and prognosis model for patients with nodular melanoma. PLoS ONE 2024, 19, e0305468. [Google Scholar] [CrossRef] [PubMed]
  33. Ma, C.; Hao, L.; Luo, J.; He, B.; Liao, S.; Wang, F.; Fu, H.; Zhang, S. Development and validation of a nomogram for predicting overall survival in patients with lower extremity melanoma: Based on the Surveillance, Epidemiology, and End Results (SEER) database. Arch. Dermatol. Res. 2025, 317, 653. [Google Scholar] [CrossRef]
  34. Zhang, S.; Tian, S.; Qin, X.; Mou, H.; He, B.; Wang, Y.; Xue, B.; Lin, W. Development of a prognostic model for patients with nodular melanoma of the lower extremities: A study based on the SEER database. Arch. Dermatol. Res. 2024, 316, 563. [Google Scholar] [CrossRef]
  35. Ma, C.; Qin, R.; Cao, Y.; Dai, Y.; Hua, M.; Wang, L.; Cao, L.; Fan, L.; Li, K. Nomogram Predicts Prognostic Factors for Head and Neck Cutaneous Melanoma: A Population-Based Analysis. World Neurosurg. 2024, 187, e839–e851. [Google Scholar] [CrossRef]
  36. Fan, F.; Gou, J.; Li, S.; Ji, Q. A Prognostic Model for Cancer-Specific Survival Among Patients with Nodular Melanoma in Head and Neck. Laryngoscope 2024, 134, 3611–3619. [Google Scholar] [CrossRef]
  37. Huang, J.N.; Yu, H.; Wan, Y.; Ming, W.K.; Situ, F.; Zhu, L.; Jiang, Y.; Wu, U.T.; Huang, W.E.; Chen, W.; et al. A prognostic nomogram for the cancer-specific survival of white patients with invasive melanoma at BANS sites based on the Surveillance, Epidemiology, and End Results database. Front. Med. 2023, 10, 1167742. [Google Scholar] [CrossRef] [PubMed]
  38. Li, A.T.; Law, J.; Ch’ng, S.; Scolyer, R.A.; Thompson, J.F.; Lo, S.N.; Varey, A.H.R. External validation in an Australian population of the EORTC-DeCOG nomogram predicting recurrence, distant metastasis and overall mortality in melanoma patients with positive sentinel lymph nodes. Eur. J. Cancer 2023, 189, 112901. [Google Scholar] [CrossRef]
  39. Janka, E.A.; Szabó, I.L.; Kollár, S.; Toka-Farkas, T.; Ványai, B.; Várvölgyi, T.; Kapitány, A.; Shabu, H.; Szegedi, A.; Emri, G. Prognostic Role of Clinicopathological Characteristics and Serum Markers in Metastatic Melanoma Patients Treated with BRAF and MEK Inhibitors. Cancers 2024, 16, 2981. [Google Scholar] [CrossRef]
  40. Janka, E.A.; Ványai, B.; Szabó, I.L.; Toka-Farkas, T.; Várvölgyi, T.; Kapitány, A.; Szegedi, A.; Emri, G. Primary tumour category, site of metastasis, and baseline serum S100B and LDH are independent prognostic factors for survival in metastatic melanoma patients treated with anti-PD-1. Front. Oncol. 2023, 13, 1237643. [Google Scholar] [CrossRef]
  41. Lourdault, K.; Cowman, A.W.; Hanes, D.; Scholer, A.J.; Aguilar, T.; Essner, R. Are Nomograms Useful for Predicting Sentinel Lymph Node Status in Melanoma Patients? J. Surg. Oncol. 2025, 131, 685–693. [Google Scholar] [CrossRef] [PubMed]
  42. Hosein, S.; Drebin, H.M.; Kurtansky, N.R.; Olofsson Bagge, R.; Coit, D.G.; Bartlett, E.K.; Marchetti, M.A. Are the MIA and MSKCC nomograms useful in selecting patients with melanoma for sentinel lymph node biopsy? J. Surg. Oncol. 2023, 127, 1167–1173. [Google Scholar] [CrossRef]
  43. Bertolli, E.; Micheletti, S.B.; de Camargo, V.P.; da Silva, T.V.; Bacchi, C.E.; Buzaid, A.C. Nomograms versus artificial intelligence platforms: Which one can better predict sentinel node positivity in melanoma patients? Melanoma Res. 2025, 35, 227–231. [Google Scholar] [CrossRef] [PubMed]
  44. Bajaj, S.; Donnelly, D.; Call, M.; Johannet, P.; Moran, U.; Polsky, D.; Shapiro, R.; Berman, R.; Pavlick, A.; Weber, J.; et al. Melanoma Prognosis: Accuracy of the American Joint Committee on Cancer Staging Manual Eighth Edition. J. Natl. Cancer Inst. 2020, 112, 921–928. [Google Scholar] [CrossRef] [PubMed]
  45. Keung, E.Z.; Gershenwald, J.E. The eighth edition American Joint Committee on Cancer (AJCC) melanoma staging system: Implications for melanoma treatment and care. Expert Rev. Anticancer Ther. 2018, 18, 775–784. [Google Scholar] [CrossRef]
  46. Ecker, B.L.; Kaur, A.; Douglass, S.M.; Webster, M.R.; Almeida, F.V.; Marino, G.E.; Sinnamon, A.J.; Neuwirth, M.G.; Alicea, G.M.; Ndoye, A.; et al. Age-Related Changes in HAPLN1 Increase Lymphatic Permeability and Affect Routes of Melanoma Metastasis. Cancer Discov. 2019, 9, 82–95. [Google Scholar] [CrossRef] [PubMed]
  47. Voinea, S.; Blidaru, A.; Panaitescu, E.; Sandru, A. Impact of gender and primary tumor location on outcome of patients with cutaneous melanoma. J. Med. Life 2016, 9, 444–448. [Google Scholar]
  48. Vossaert, K.A.; Silverman, M.K.; Kopf, A.W.; Bart, R.S.; Rigel, D.S.; Friedman, R.J.; Levenstein, M. Influence of gender on survival in patients with stage I malignant melanoma. J. Am. Acad. Dermatol. 1992, 26 Pt 2, 429–440. [Google Scholar] [CrossRef]
  49. Bellenghi, M.; Puglisi, R.; Pontecorvi, G.; De Feo, A.; Carè, A.; Mattia, G. Sex and Gender Disparities in Melanoma. Cancers 2020, 12, 1819. [Google Scholar] [CrossRef]
Cancers 17 03080 g001
Figure 1. Flowchart showing the patients included in this study. N—number.
Figure 1. Flowchart showing the patients included in this study. N—number.
Cancers 17 03080 g001
Cancers 17 03080 g002aCancers 17 03080 g002bCancers 17 03080 g002cCancers 17 03080 g002d
Figure 2. Recurrence-free survival is shown for regional lymph node metastasis (A); skin, soft tissue (including muscle) and/or non-regional lymph node metastasis (B); lung metastasis (C); non-central nervous system visceral metastasis (D); brain metastasis (E). Recurrence-free survival was stratified by the primary tumor’s American Joint Committee on Cancer (AJCC) T category. Overall survival is shown, stratified by patient sex (F) and primary tumor AJCC T category (G); the investigated period was from primary tumor to the end of the follow-up/death. Overall survival is shown stratified by metastasis site (H); the investigated period was from the first recurrence to the end of the follow-up/death.
Figure 2. Recurrence-free survival is shown for regional lymph node metastasis (A); skin, soft tissue (including muscle) and/or non-regional lymph node metastasis (B); lung metastasis (C); non-central nervous system visceral metastasis (D); brain metastasis (E). Recurrence-free survival was stratified by the primary tumor’s American Joint Committee on Cancer (AJCC) T category. Overall survival is shown, stratified by patient sex (F) and primary tumor AJCC T category (G); the investigated period was from primary tumor to the end of the follow-up/death. Overall survival is shown stratified by metastasis site (H); the investigated period was from the first recurrence to the end of the follow-up/death.
Cancers 17 03080 g002aCancers 17 03080 g002bCancers 17 03080 g002cCancers 17 03080 g002d
Cancers 17 03080 g003
Figure 3. Nomogram for predicting three-, five-, and ten-year recurrence probabilities for regional lymph node metastasis. T—primary tumor.
Figure 3. Nomogram for predicting three-, five-, and ten-year recurrence probabilities for regional lymph node metastasis. T—primary tumor.
Cancers 17 03080 g003
Cancers 17 03080 g004
Figure 4. Nomogram for predicting three-, five-, and ten-year recurrence probabilities for skin, soft tissue (including muscle) and/or non-regional lymph node metastasis. T—primary tumor.
Figure 4. Nomogram for predicting three-, five-, and ten-year recurrence probabilities for skin, soft tissue (including muscle) and/or non-regional lymph node metastasis. T—primary tumor.
Cancers 17 03080 g004
Cancers 17 03080 g005
Figure 5. Nomogram for predicting three-, five-, and ten-year recurrence probabilities for lung metastasis. T—primary tumor.
Figure 5. Nomogram for predicting three-, five-, and ten-year recurrence probabilities for lung metastasis. T—primary tumor.
Cancers 17 03080 g005
Cancers 17 03080 g006
Figure 6. A nomogram that predicts the three-, five-, and ten-year recurrence probabilities for non-central nervous system visceral metastases. T—primary tumor.
Figure 6. A nomogram that predicts the three-, five-, and ten-year recurrence probabilities for non-central nervous system visceral metastases. T—primary tumor.
Cancers 17 03080 g006
Cancers 17 03080 g007
Figure 7. Nomogram for predicting three-, five-, and ten-year recurrence probabilities for brain metastasis. T—primary tumor.
Figure 7. Nomogram for predicting three-, five-, and ten-year recurrence probabilities for brain metastasis. T—primary tumor.
Cancers 17 03080 g007
Table 1. Characteristics of the study population.
Table 1. Characteristics of the study population.
Variables, N (%)CategoriesPatients
N = 2003		Training Set
N = 1402		Validation Set
N = 601		p-Values
Age (years)<40265 (13.2)172 (12.3)93 (15.5)0.193
40–49291 (14.5)196 (14.0)95 (15.8)
50–59388 (19.4)273 (19.5)115 (19.1)
60–69476 (23.8)344 (24.5)132 (22.0)
≥70583 (29.1)417 (29.7)166 (27.6)
Sexmale941 (47.0)649 (46.3)292 (48.6)0.346
female1062 (53.0)753 (53.7)309 (51.4)
Localization of primary tumor head and neck338 (16.9)241 (17.2)97 (16.1)0.167
upper extremities324 (16.2)243 (17.3)81 (13.5)
lower extremities435 (21.7)303 (21.6)132 (22.0)
trunk850 (42.4)579 (41.3)271 (45.1)
occult56 (2.8)36 (2.6)20 (3.3)
Histological subtypeSSM1105 (55.2)772 (55.1)333 (55.4)0.525
LMM133 (6.6)99 (7.1)34 (5.7)
NM292 (14.6)197 (14.1)95 (15.8)
MM473 (23.6)334 (23.8)139 (23.1)
Clark levelI-II-III1260 (62.9)869 (62.0)391 (65.0)0.149
IV394 (19.7)274 (19.5)120 (20.0)
V160 (8.0)122 (8.7)38 (6.3)
occult56 (2.8)36 (2.6)20 (3.3)
unknown133 (6.6)101 (7.2)32 (5.3)
AJCC 8th edition T categoryT1a768 (38.3)540 (38.5)228 (37.9)0.239
T1b-T2a375 (18.7)268 (19.1)107 (17.8)
T2b-T3a219 (10.9)142 (10.1)77 (12.8)
T3b-T4a267 (13.3)190 (13.6)77 (12.8)
T4b276 (13.8)205 (14.6)71 (11.8)
T056 (2.8)36 (2.6)20 (3.3)
unknown42 (2.1)21 (1.5)21 (3.5)
Metastasis during follow-upno1147 (57.3)801 (57.1)346 (57.6)0.856
yes856 (42.7)601 (42.9)255 (42.4)
Death during follow-upno1153 (57.6)803 (57.3)350 (58.2)0.690
yes850 (42.4)599 (42.7)251 (41.8)
N—number; SSM—superficial spreading melanoma; LMM—lentigo maligna melanoma; NM-nodular melanoma; MM—unclassified malignant melanoma or histological subtype other than SSM, LMM or NM or no evidence of primary tumor; AJCC—American Joint Committee on Cancer.
Table 2. Univariate and multivariate Cox regression analyses of melanoma recurrence in regional lymph nodes.
Table 2. Univariate and multivariate Cox regression analyses of melanoma recurrence in regional lymph nodes.
VariablesCategoriesUnivariateMultivariate
HR [95% CI]p-ValueHR [95% CI]p-Value
Age (years)40–49/<401.64 [1.19; 2.26]0.0021.54 [1.11; 2.12]0.009
50–59/<401.47 [1.08; 2.01]0.0151.20 [0.88; 1.64]0.249
60–69/<401.91 [1.43; 2.57]<0.0011.24 [0.92; 1.67]0.158
≥70/<401.56 [1.16; 2.10]0.0030.92 [0.68; 1.25]0.589
Sexmale/female1.53 [1.31; 1.79]<0.0011.34 [1.13; 1.58]0.001
Localization of primary tumor upper extremities/head-neck1.11 [0.81; 1.52]0.4991.05 [0.76; 1.45]0.769
lower extremities/head-neck1.91 [1.46; 2.50]<0.0011.63 [1.23; 2.15]<0.001
trunk/head-neck1.37 [1.06; 1.77]0.0161.22 [0.93; 1.61]0.148
occult/head-neck6.94 [4.70; 10.26]<0.0014.86 [3.02; 6.72]<0.001
Histological subtypeLMM/SSM0.56 [0.31; 1.00]0.0511.15 [0.62; 2.15]0.661
NM/SSM4.93 [3.94; 6.17]<0.0011.79 [1.38; 2.32]<0.001
MM/SSM5.38 [4.44; 6.53]<0.0011.47 [1.15; 1.87]<0.001
Clark levelIV/I-II-III4.31 [3.55; 5.23]<0.0011.39 [1.11; 1.74]0.004
V/I-II-III 7.01 [5.51; 8.92]<0.0011.60 [1.20; 2.13]0.001
occult/I-II-III11.22 [7.92; 15.88]<0.0011.54 [1.13; 2.10]0.007
AJCC 8th edition T categoryT1b-T2a/T1a3.97 [2.83; 5.58]<0.0013.39 [2.39; 4.81]<0.001
T2b-T3a/T1a9.39 [6.70; 13.17]<0.0015.92 [4.02; 8.71]<0.001
T3b-T4a/T1a13.88 [10.08; 19.11]<0.0018.26 [5.57; 12.23]<0.001
T4b/T1a20.67 [15.11; 28.27]<0.00110.66 [7.13; 15.93]<0.001
T0/T1a13.22 [8.09; 21.62]<0.0016.70 [3.82; 11.74]<0.001
Significant results are in bold. HR [95% CI]—hazard ratio with 95% confidence intervals; LMM—lentigo maligna melanoma; SSM—superficial spreading melanoma; NM—nodular melanoma; MM—unclassified malignant melanoma or no evidence of primary tumor; AJCC—American Joint Committee on Cancer; T—primary tumor.
Table 3. Univariate and multivariate Cox regression analyses of melanoma recurrence at sites classified as AJCC M1a stage.
Table 3. Univariate and multivariate Cox regression analyses of melanoma recurrence at sites classified as AJCC M1a stage.
VariablesCategoriesUnivariateMultivariate
HR [95% CI]p-ValueHR [95% CI]p-Value
Age (years)40–49/<401.56 [1.06; 2.28]0.0231.18 [0.80; 1.73]0.408
50–59/<401.67 [1.17; 2.40]0.0051.26 [0.88; 1.81]0.213
60–69/<402.03 [1.44; 2.86]<0.0011.20 [0.84; 1.71]0.315
≥70/<402.27 [1.62; 3.18]<0.0011.75 [1.23; 2.48]0.002
Sexmale/female1.46 [1.23; 1.74]<0.0011.21 [1.01; 1.47]0.049
Localization of primary tumor upper extremities/head-neck0.80 [0.56; 1.14]0.2160.80 [0.55; 1.16]0.242
lower extremities/head-neck1.74 [1.31; 2.32]<0.0011.19 [0.88; 1.61]0.249
trunk/head-neck1.06 [0.80; 1.39]0.7000.90 [0.67; 1.21]0.488
occult/head-neck7.01 [4.70; 10.45]<0.0018.21 [4.77; 14.14]<0.001
Histological subtypeLMM/SSM0.58 [0.30; 1.10]0.0930.99 [0.49; 1.97]0.969
NM/SSM4.64 [3.61; 5.94]<0.0011.30 [0.98; 1.73]0.066
MM/SSM4.70 [3.79; 5.83]<0.0011.19 [0.91; 1.54]0.201
Clark levelIV/I-II-III4.43 [3.56; 5.51]<0.0011.25 [0.97; 1.61]0.091
V/I-II-III 7.30 [5.57; 9.56]<0.0011.49 [1.08; 2.06]0.017
occult/I-II-III13.48 [9.43; 19.28]<0.0011.35 [0.94; 1.93]0.102
AJCC 8th edition T categoryT1b-T2a/T1a3.55 [2.44; 5.18]<0.0012.17 [1.45; 3.24]<0.001
T2b-T3a/T1a7.89 [5.43; 11.46]<0.0012.67 [1.71; 4.16]<0.001
T3b-T4a/T1a12.61 [8.91; 17.85]<0.0013.31 [2.11; 5.18]<0.001
T4b/T1a16.55 [11.75; 23.30]<0.0013.48 [2.19; 5.53]<0.001
T0/T1a11.33 [6.54; 19.61]<0.0012.98 [1.56; 5.70]0.001
Regional lymph node metastasisyes/no11.64 [9.44; 14.36]<0.0015.70 [4.48; 7.25]<0.001
Significant results are in bold. HR [95% CI]—hazard ratio with 95% confidence intervals; LMM—lentigo maligna melanoma; SSM—superficial spreading melanoma; NM—nodular melanoma; MM—unclassified malignant melanoma or no evidence of primary tumor; AJCC—American Joint Committee on Cancer; T—primary tumor; M1a—metastases to skin, soft tissue (including muscle), and/or non-regional lymph nodes.
Table 4. Univariate and multivariate Cox regression analyses of melanoma recurrence in the lung.
Table 4. Univariate and multivariate Cox regression analyses of melanoma recurrence in the lung.
VariablesCategoriesUnivariateMultivariate
HR [95% CI]p-ValueHR [95% CI]p-Value
Age (years)40–49/<401.77 [1.14; 2.77]0.0121.23 [0.78; 1.93]0.373
50–59/<401.75 [1.14; 2.68]0.0101.32 [0.85; 2.04]0.214
60–69/<402.17 [1.44; 3.27]<0.0011.39 [0.91; 2.11]0.125
≥70/<402.06 [1.36; 3.12]0.0011.35 [0.88; 2.07]0.163
Sexmale/female1.82 [1.47; 2.25]<0.0011.30 [1.03; 1.63]0.028
Localization of primary tumor upper extremities/head-neck0.73 [0.48; 1.09]0.1240.77 [0.50; 1.17]0.214
lower extremities/head-neck0.99 [0.69; 1.40]0.9931.11 [0.81; 1.54]0.120
trunk/head-neck1.07 [0.78; 1.46]0.6750.89 [0.63; 1.25]0.483
occult/head-neck5.91 [3.67; 9.53]<0.0012.61 [0.95; 2.87]0.061
Histological subtypeLMM/SSM0.85 [0.46; 1.58]0.6131.11 [0.66; 1.61]0.440
NM/SSM3.91 [2.90; 5.27]<0.0011.17 [0.69; 1.36]0.864
MM/SSM3.83 [2.96; 4.95]<0.0011.20 [0.60; 1.31]0.221
Clark levelIV/I-II-III3.60 [2.77; 4.67]<0.0011.17 [0.86; 1.59]0.331
V/I-II-III 5.94 [4.29; 8.24]<0.0011.25 [0.84; 1.85]0.269
occult/I-II-III11.71 [7.59; 18.07]<0.0011.28 [0.85; 1.94]0.243
AJCC 8th edition T categoryT1b-T2a/T1a2.85 [1.89; 4.30]<0.0011.94 [1.24; 3.02]0.004
T2b-T3a/T1a5.25 [3.43; 8.04]<0.0012.49 [1.50; 4.13]<0.001
T3b-T4a/T1a7.78 [5.24; 11.53]<0.0012.93 [1.73; 4.94]<0.001
T4b/T1a15.24 [10.55; 22.01]<0.0015.53 [3.29; 9.30]<0.001
T0/T1a6.27 [3.04; 12.93]<0.0012.32 [1.03; 4.00]0.043
Regional lymph node metastasisyes/no8.98 [7.07; 11.41]<0.0012.95 [2.18; 4.00]<0.001
Distant metastasis at sites M1ayes/no7.66 [6.13; 9.57]<0.0012.63 [2.02; 3.43]<0.001
Significant results are in bold. HR [95% CI]—hazard ratio with 95% confidence intervals; LMM—lentigo maligna melanoma; SSM—superficial spreading melanoma; NM—nodular melanoma; MM—unclassified malignant melanoma or no evidence of primary tumor; AJCC—American Joint Committee on Cancer; T—primary tumor; M1a—metastases to skin, soft tissue (including muscle), and/or non-regional lymph nodes.
Table 5. Univariate and multivariate Cox regression analyses of melanoma recurrence at non-central nervous system visceral sites.
Table 5. Univariate and multivariate Cox regression analyses of melanoma recurrence at non-central nervous system visceral sites.
VariablesCategoriesUnivariateMultivariate
HR [95% CI]p-ValueHR [95% CI]p-Value
Age (years)40–49/<401.49 [1.01; 2.02]0.0481.04 [0.69; 1.55]0.864
50–59/<401.36 [0.93; 1.99]0.1120.99 [0.67; 1.46]0.953
60–69/<401.77 [1.23; 2.54]0.0021.13 [0.78; 1.65]0.514
≥70/<401.61 [1.12; 2.33]0.0101.09 [0.75; 1.60]0.646
Sexmale/female1.83 [1.50; 2.24]<0.0011.29 [1.04; 1.61]0.021
Localization of primary tumor upper extremities/head-neck0.88 [0.60; 1.29]0.5020.84 [0.59; 1.20]0.338
lower extremities/head-neck1.08 [0.77; 1.52]0.6571.08 [0.73; 1.61]0.707
trunk/head-neck1.16 [0.86; 1.57]0.3371.10 [0.79; 1.53]0.583
occult/head-neck5.00 [3.13; 7.98]<0.0013.35 [0.97; 4.05]0.059
Histological subtypeLMM/SSM0.86 [0.48; 1.56]0.0631.18 [0.62; 2.27]0.613
NM/SSM4.11 [3.10; 5.45]<0.0011.10 [0.80; 1.53]0.556
MM/SSM4.04 [3.17; 5.15]<0.0011.19 [0.88; 1.62]0.250
Clark levelIV/I-II-III3.13 [2.44; 4.01]<0.0011.02 [0.73; 1.30]0.862
V/I-II-III 5.51 [4.06; 7.46]<0.0011.18 [0.82; 1.69]0.384
occult/I-II-III8.49 [5.58; 12.92]<0.0011.04 [0.70; 1.54]0.853
AJCC 8th edition T categoryT1b-T2a/T1a2.66 [1.82; 3.88]<0.0011.28 [0.84; 1.95]0.259
T2b-T3a/T1a5.07 [3.44; 7.47]<0.0011.69 [1.05; 2.69]0.029
T3b-T4a/T1a7.24 [5.06; 10.36]<0.0011.93 [1.19; 3.12]0.007
T4b/T1a11.40 [8.09; 16.08]<0.0012.46 [1.51; 4.00]<0.001
T0/T1a8.56 [4.80; 15.27]<0.0012.86 [1.44; 5.66]0.003
Regional lymph node metastasisyes/no10.04 [7.96; 12.67]<0.0012.62 [1.96; 3.51]<0.001
Distant metastasis at sites M1ayes/no8.43 [6.81; 10.43]<0.0012.31 [1.79; 2.97]<0.001
Lung metastasisyes/no10.81 [8.78; 13.30]<0.0014.00 [3.18; 5.04]<0.001
Significant results are in bold. HR [95% CI]—hazard ratio with 95% confidence intervals; LMM—lentigo maligna melanoma; SSM—superficial spreading melanoma; NM—nodular melanoma; MM—unclassified malignant melanoma or no evidence of primary tumor; AJCC—American Joint Committee on Cancer; T—primary tumor; M1a—metastases to skin, soft tissue (including muscle), and/or non-regional lymph nodes.
Table 6. Univariate and multivariate Cox regression analyses of melanoma recurrence in the brain.
Table 6. Univariate and multivariate Cox regression analyses of melanoma recurrence in the brain.
VariablesCategoriesUnivariateMultivariate
HR [95% CI]p-ValueHR [95% CI]p-Value
Age (years)40–49/<401.70 [1.01; 2.87]0.0461.40 [0.82; 2.37]0.218
50–59/<401.70 [1.03; 2.81]0.0371.34 [0.80; 2.22]0.265
60–69/<401.82 [1.12; 2.95]0.0171.28 [0.77; 2.10]0.341
≥70/<401.92 [1.19; 3.12]0.0081.43 [0.87; 2.35]0.159
Sexmale/female2.05 [1.58; 2.65]<0.0011.49 [1.13; 1.96]0.005
Localization of primary tumor upper extremities/head-neck0.87 [0.54; 1.42]0.5820.87 [0.53; 1.43]0.583
lower extremities/head-neck1.11 [0.72; 1.71]0.6370.86 [0.55; 1.35]0.520
trunk/head-neck1.16 [0.79; 1.71]0.4430.98 [0.64; 1.48]0.906
occult/head-neck4.54 [2.51; 8.24]<0.0012.05 [0.98; 3.58]0.058
Histological subtypeLMM/SSM0.58 [0.23; 1.43]0.2350.79 [0.30; 2.08]0.635
NM/SSM3.81 [2.65; 5.49]<0.0011.06 [0.69; 1.62]0.803
MM/SSM4.41 [3.26; 5.98]<0.0011.17 [0.79; 1.72]0.442
Clark levelIV/I-II-III2.93 [2.15; 4.00]<0.0011.11 [0.93; 1.33]0.685
V/I-II-III 4.65 [3.14; 6.90]<0.0011.19 [0.92; 1.46]0.721
occult/I-II-III7.19 [4.23; 12.25]<0.0011.17 [0.97; 1.60]0.920
AJCC 8th edition T categoryT1b-T2a/T1a2.39 [1.44; 3.94]0.0011.53 [0.90; 2.61]0.118
T2b-T3a/T1a5.13 [3.11; 8.45]<0.0012.32 [1.28; 4.20]0.006
T3b-T4a/T1a6.52 [4.06; 10.47]<0.0012.82 [1.54; 5.19]0.001
T4b/T1a14.40 [9.37; 22.12]<0.0015.00 [2.73; 9.18]<0.001
T0/T1a7.47 [3.41; 16.34]<0.0013.29 [1.33; 8.13]0.010
Regional lymph node metastasisyes/no5.72 [4.38; 7.47]<0.0011.34 [0.81; 2.61]0.455
Distant metastasis at sites M1ayes/no6.79 [5.22; 8.83]<0.0012.11 [1.53; 2.92]<0.001
Lung metastasisyes/no8.07 [6.24; 10.45]<0.0012.45 [1.79; 3.36]<0.001
Non–central nervous system visceral metastasesyes/no7.31 [5.65; 9.47]<0.0012.10 [1.53; 2.89]<0.001
Significant results are in bold. HR [95% CI]—hazard ratio with 95% confidence intervals; LMM—lentigo maligna melanoma; SSM—superficial spreading melanoma; NM—nodular melanoma; MM—unclassified malignant melanoma or no evidence of primary tumor; AJCC—American Joint Committee on Cancer; T—primary tumor; M1a—metastases to skin, soft tissue (including muscle), and/or non-regional lymph nodes.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Janka, E.A.; Szabó, I.L.; Toka-Farkas, T.; Soltész, L.; Szentkereszty-Kovács, Z.; Ványai, B.; Várvölgyi, T.; Kapitány, A.; Szegedi, A.; Emri, G. Development of Nomograms to Predict the Probability of Recurrence at Specific Sites in Patients with Cutaneous Melanoma. Cancers 2025, 17, 3080. https://doi.org/10.3390/cancers17183080

AMA Style

Janka EA, Szabó IL, Toka-Farkas T, Soltész L, Szentkereszty-Kovács Z, Ványai B, Várvölgyi T, Kapitány A, Szegedi A, Emri G. Development of Nomograms to Predict the Probability of Recurrence at Specific Sites in Patients with Cutaneous Melanoma. Cancers. 2025; 17(18):3080. https://doi.org/10.3390/cancers17183080

Chicago/Turabian Style

Janka, Eszter Anna, Imre Lőrinc Szabó, Tünde Toka-Farkas, Lilla Soltész, Zita Szentkereszty-Kovács, Beatrix Ványai, Tünde Várvölgyi, Anikó Kapitány, Andrea Szegedi, and Gabriella Emri. 2025. "Development of Nomograms to Predict the Probability of Recurrence at Specific Sites in Patients with Cutaneous Melanoma" Cancers 17, no. 18: 3080. https://doi.org/10.3390/cancers17183080

APA Style

Janka, E. A., Szabó, I. L., Toka-Farkas, T., Soltész, L., Szentkereszty-Kovács, Z., Ványai, B., Várvölgyi, T., Kapitány, A., Szegedi, A., & Emri, G. (2025). Development of Nomograms to Predict the Probability of Recurrence at Specific Sites in Patients with Cutaneous Melanoma. Cancers, 17(18), 3080. https://doi.org/10.3390/cancers17183080

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop