Previously, it was assumed that children with melanoma had similar survival rates to adults. Although stage remains a significant predictor of survival across age groups, current research reveals that childhood melanoma patients may outperform adults with identical cancer stages [6–7]. Lorimer et al. used the National Cancer Database to conduct an analysis and found that children aged 1–10 years (HR 0.11; 95% CI 0.06–0.21) and adolescents aged 11–20 years (HR 0.22; 95% CI 0.19–0.26) had statistically higher overall survival (OS) rates than adults (reference HR 1.0) [6]. A recent SEER study also found that disease-specific 5-, 15-, and 30-year survival rates for children and adolescents with extremities melanoma were 96%, 94%, and 93%, respectively [8]. In this population-based study, 77% of patients showed localized disease, 15% had regional dissemination, and 1% had distant metastases. Furthermore, the 5-year OS rates by disease extent were 99% for localized, 87% for regional, and 50% for distant disease, which was broadly consistent with our findings [9–11].
Established negative prognostic factors in adult melanoma include primary tumor thickness, non-extremity location, advanced age, regional lymph node involvement, and visceral or brain metastases [12–14]. However, prognostic indicators and outcomes in pediatric cases are less well-defined and often based on limited retrospective case series. Notably, in a SEER-based study by Strouse et al., while overall survival differences across age groups were minimal, mortality from melanoma increased with older age in multivariate analysis, highlighting advancing age as an adverse prognostic factor across all disease stages [15]. Regarding clinicopathological characteristics, previous research has indicated that anatomical site significantly impacts melanoma overall survival rates, with the face/ears exhibiting lower survival rates compared to other sites, consistent with our findings [16–17]. Historically, prognosis for patients with distant metastatic melanoma has been bleak, with median survival less than one year and 5-year survival under 10% [18–19]. Although advancements in combination therapies have led to improvements, 5-year overall survival rates for metastatic disease remains low [20].
We identified common independent risk factors in pediatric melanoma patients, including age, sex, race, primary site, stage, and treatment. These factors are readily available and comprehensive in clinical practice. Our nomogram demonstrated enhanced discriminatory power in predicting prognosis, representing the first instance, to our knowledge, of such a tool being developed for pediatric melanoma prognosis. Internal and external validation C-indices exceeding 0.78 indicated robust discriminatory ability, offering personalized prognostic information. Consistently, AUC values indicated strong discrimination capability. Calibration curves demonstrated high consistency in predicted values. Additionally, decision curve analysis (DCA) illustrated significant clinical net benefits across 3-, 5-, and 10-year overall survival rates.
This retrospective study has several limitations. Firstly, our nomogram was developed using data from the SEER database, which may lead to some loss of patient information, thereby limiting the generalizability of our findings to certain populations and racial groups. Future iterations of the predictive model could incorporate these factors. Secondly, we did not account for other potential prognostic factors such as economic status, pathological subtype, and tumor-related genetic markers. Integrating these factors into the nomogram could enhance its accuracy and individualization. Lastly, our patient cohort was divided into two groups, with 70% used for model development and the remaining 30% for validation. While the C-index, AUC, calibration curve, and DCA demonstrated satisfactory performance, external validation of the nomogram through future studies is warranted.
In conclusion, our study introduced and verified a novel model for predicting 3-, 5-, and 10-year OS in pediatric melanoma patients. This model exhibited favorable discrimination and clinical relevance, offering a valuable tool for clinical prediction and decision-making. Further validation from diverse institutions is warranted to confirm its utility.