Brain metastases in adult patients with melanoma of unknown primary in the Netherlands (2011–2020)

Although patients with melanoma of unknown primary (MUP) have a better prognosis than similar-staged melanoma patients with known primary, the occurrence of brain metastases (BM) entails a serious complication. This study provides an overview of the incidence, treatment patterns, and overall survival (OS) of adult patients with BM-MUP in the Netherlands. BM-MUP cases were retrieved from the Netherlands Cancer Registry. Patient, disease and treatment-related characteristics were summarised using descriptive statistics. Overall survival (OS) was calculated by the Kaplan–Meier method, and the impact of prognostic factors on OS was assessed using Cox proportional hazard regression analyses. Among 1779 MUP patients, 450 were identified as BM-MUP (25.3%). Of these patients, 381 (84.7%) presented with BM along with other metastases, while 69 (15.3%) had BM only. BM-MUP patients were predominantly male (68.2%), and had a median age of 64 years at diagnosis (interquartile range 54–71 years). Over time, the proportion of BM along other metastatic sites increased, and the occurrence of BM decreased (p = 0.01). 1-Year OS improved for the total population, from 30.0% (95% confidence interval (CI): 19.8–40.9%) in 2011–2012 to 43.6% (95%CI: 34.5–52.3%) in 2019–2020, and median OS more than doubled from 4.2 months (95%CI: 3.3–6.2 months) to 9.8 months (95%CI: 7.0–13.2 months). Patient’s age, localisation of BM, presence of synchronous liver metastasis and treatment were identified as independent predictors of OS. Notwithstanding the progress made in OS for patients with BM-MUP in the past decade, their overall prognosis remains poor, and further efforts are needed to improve outcomes.


Introduction
Melanoma of unknown primary (MUP) accounts for around 3% of all melanoma cases [1]. MUP is defined as the presence of melanoma in subcutaneous tissue, lymph nodes, or visceral organs without a cutaneous, mucosal or ocular primary site [2]. Aside from an unknown primary, it is suggested that MUP have a different tumour biology, as patients were shown to achieve better outcomes than similar-staged melanoma patients with known primary (MKP) even before the advent of novel systemic therapies in 2011 [3,4]. This benefit remained in the new era following the introduction of immune checkpoint inhibitors (ICIs, immunotherapy) and BRAF and/or MEK inhibitors (targeted therapy), often despite of more unfavourable prognostic characteristics in MUP patients compared to those with MKP [5][6][7][8]. Advances in the genomic profiling of tumours enables further identification of molecular differences between MUP and MKP [9].
Notwithstanding the progress that has been made, prognosis for MUP patients remains far from favourable, particularly if metastases occur in the brain. Brain metastases (BM) entail a common and serious complication in melanoma. BM occur in up to 75% of patients with metastatic melanoma at some point in the course of their disease [10], and are a direct cause of death in 60-70% of patients [11]. The occurrence of BM may be associated with the use of new treatments, as melanoma patients on BRAF inhibitor therapy showed an increased risk of progression in the brain [12,13]. Some reports have suggested an increased risk of BM in MUP patients [14,15], but few studies have focused on describing the population of patients with BM-MUP.
In a recent survey on BM in cancer of unknown primary, we showed that the proportion of BM-MUP patients increased by almost 50% in a decade [16]. In the current study, we aimed to provide a real-world overview of the incidence, treatment patterns, and overall survival (OS) of adult patients diagnosed with BM-MUP in the Netherlands in 2011-2020.

Study population
The NCR database collects information on all newly diagnosed malignancies in the Dutch population. Data is subtracted from hospital patient records, and include patient, tumour and treatment-related characteristics. The main modalities are registered for therapies provided as part of the initial treatment plan, which means no information was available on second-or higher line treatment. Follow-up information on patients' vital status was obtained through linkage with the Dutch Municipal Personal Records Database, which for this study was updated until January 31st, 2022. The study design, data abstraction process, and storage protocols were approved by the national supervisory committee of the NCR.
For this study, we queried all diagnoses of BM-MUP made in 2011-2020 from the NCR database. Cases were selected on the basis of topography (C80.9) and histology codes (M8720/3-M8790/3) of the International Classification of Diseases for Oncology ( Fig. 1) [17]. Patients aged below 18 years at diagnosis of BM-MUP were excluded.

Statistical analysis
Baseline characteristics were presented with frequencies for categorical variables, and median numbers and interquartile ranges (IQR) for continuous variables. Differences between patients with BM along extracranial metastatic sites (synchronous BM) and those with BM only were evaluated with chi-square tests. Time trends in characteristics were assessed by calculating a chi-squared test statistic for Pearson's correlation coefficient. OS was calculated by the Kaplan-Meier method, and log-rank tests were used to compare 1-year survival rates between subgroups. Rates and median survival times (in months) were presented with corresponding 95% confidence intervals (95%CI). Cox proportional hazards regression analyses were used to assess the impact of independent predictors on OS. Variables were selected for the multivariable regression model following a p-value of 0.10 in the univariable analysis, and the definitive model was developed using backward selection on the basis of likelihood ratio tests. Lastly, the assumption of proportional hazards underlying this model was tested using the Schoenfeld residuals. Overall, test results with a p-value of 0.05 or lower were considered statistically significant. All statistical analyses were performed using Stata (version 17.0, StataCorp, College Station, Texas).

Survival
Over Survival also appeared better for younger patients, for patients with BM confined to a single supratentorial site, and for those with BM only (Fig. 4). The presence of other metastatic sites appeared associated with poorer survival, with the exception of metastasis to lymph nodes. Overall, patients who were able to receive treatment had better survival than those who were not. The difference in median OS was most obvious for surgery (13.6 months versus 5.0 months, p < 0.01), immunotherapy (4.8 months versus 15.0 months, p < 0.01) and targeted therapy (5.3 months versus 11.8 months, p < 0.01). Although less pronounced, benefit was also observed for the provision of radiotherapy (5.1 months versus 8.0 months, p < 0.01).
Following adjustment for confounding effects, patient's age at diagnosis, the localisation of BM, presence of liver metastasis, and treatment with surgery, radiotherapy, targeted therapy or immunotherapy proved independent predictors of OS (Table 3). Compared to BM confined to a single supratentorial site, OS was worse for BM in multiple Fig. 2 Involvement of metastatic sites other than the brain in adult patients with BM-MUP. MUP melanoma of unknown primary, BM brain metastasis sites (HR 1.50, p = 0.01) and for unspecified sites of BM (HR 1.53, p = 0.01). The presence of liver metastasis also entailed a significantly worse OS (HR: 1.42, p = 0.01). The multivariable analyses confirmed the significant benefits of treatment with immunotherapy (HR 0.31, p < 0.01), targeted therapy (HR 0.38, p < 0.01), surgery (HR 0.49, p < 0.01) and radiotherapy (HR 0.69, p < 0.01). No violation of the proportional hazards assumption underlying the multivariable model was determined.

Discussion
Although a body of knowledge has accumulated on MUP over the past decade, BM-MUP is still scarcely studied. With our population-based survey, we were able to provide information on the basic characteristics, global treatment patterns and prognosis of 450 BM-MUP patients diagnosed in 2011-2020. We found that the introduction of immuno-and targeted therapy-alongside surgery and  radiotherapy-accounted for significant survival benefit in these patients, which aligns with the progress that was observed in high-stage melanoma in general. Still, median survival in case of BM-MUP remained below one year even in the most recent time period. Patients with BM-MUP are unlikely to enrol in clinical studies on new therapeutic interventions. Until a few years ago, many would be considered ineligible based on the presence of BM by itself, in addition to their generally poor performance status and other unfavourable parameters. While the responsiveness of MUP to ICI's and/or targeted therapy is well-established by now, with some studies even suggesting more benefit for MUP compared to MKP [18], few outcomes have been reported for patients with BM in this population.
The role of new therapies in the management of melanoma BM has also become the focus of extensive study [19][20][21][22][23]. Given their potential to achieve an intracranial OS Overall survival  response in addition to extracranial control, they are increasingly applied, sometimes as first-line treatment [24], regularly combined with local treatment [25,26]. This is reflected in our study results on the most recent years, which show that surgery or radiotherapy (as whole brain radiation or stereotactic radiosurgery) formed the mainstay of treatment of patients with BM only, most often supplemented with immunotherapy. Several issues warrant caution in the interpretation of our results. Given the observational set-up of the study, associations of outcomes with treatment should be evaluated while considering the impact of confounding by indication. Unfortunately, more detailed data on patients' performance status and tumour characteristics that would have enabled adjustment for this effect were unavailable to us. Despite these limitations, this study provides one of the first overviews of a hitherto poorly described patient population about which more information is needed.

Conclusion
In line with the overall improvement in prognosis for highgrade melanoma patients, mostly due to the introduction of immuno-and targeted therapy, survival has improved for patients with BM-MUP in the past decade. Nevertheless, their overall prognosis is still poor, with median OS remaining below one year. More efforts are needed to further improve outcomes, such as longer follow-up that has become even more relevant in the modern era to establish the added value of new treatments in managing the long-term disease course, including the development of brain metastases.