Technical Field of the Invention

The present invention relates to methods for stratifying melanoma patients and methods for identifying melanoma patients likely to respond to immunotherapy. Background to the Invention

Melanoma is a type of skin cancer associated with poor patient survival rates. The incidence and mortality of melanoma increases with age ¹ ’ ^{2 3} and approximately 80% of melanoma deaths affect patients who are older than 59 ^{4 5} Older patients more frequently present with high-risk, thicker primary tumours, (AJCC stage IIB- IIC) ⁴ ’ ⁶ , and additional characteristics of poor prognosis, such as ulceration, elevated mitotic rate, and early visceral metastasis are also more common in the elderly ⁴ . However, even though old patients more frequently present with advanced, high-risk primary disease, there is a survival discrepancy between the elderly and young after taking the main prognostic factors into account ^{4 7} . Importantly, age is the most important independent marker of adverse outcome together with primary tumour thickness ^{1 7} .

Until the advent of novel targeted treatments and immunotherapy, the only chemotherapy available to patients with advanced disease did not provide significant survival benefit, so a diminished ability to tolerate treatment does not explain the survival discrepancy.

Novel immunotherapies offer potential treatment options for some patients and such therapies are known to be effective in early and late stages of disease. Encouragingly, recent analysis of immunotherapy and targeted therapy response stratified by age has shown elderly patients clearly stand to benefit from both treatment regimens, and aged patients may present better response rates to checkpoint inhibitors ^8-13 . These preliminary observations suggest the elderly melanoma population, who is at highest risk of death, may gain the most from immunotherapy. Importantly, these novel agents are currently being tested in the adjuvant setting to treat early stage disease ^{14 15} , and a significant challenge is to identify the subset of the melanoma population that is both at highest risk of progression and most likely to respond to therapy.

Large-scale sequencing efforts that catalogue mutations in melanoma have shown individual mutations are poor predictors of survival and therapy response. However, the number, type and frequency of mutations varies according to patient age, and the impact of age-specific mutations on survival and treatment has not been studied systematically. Identifying the molecular changes that define prognostic categories of melanoma at all stages of disease, and responders to immunotherapy, could improve diagnosis, prognosis, care and outcome of melanoma patients.

Screening and early detection remains an important strategy for the management of melanoma. Currently, sentinel lymph node biopsies (SLNB) are used to determine the stage of a patient’s melanoma by establishing whether cancer cells have spread (metastasized) to lymph nodes nearby the original melanoma. SLNB is an invasive procedure and its use and considered effectiveness is debated. Moreover, older patients seem to have a different pattern of disease spread, bypassing the lymph nodes more frequently, that cannot currently be predicted.

The high costs associated with immunotherapy are likely to prohibit its use as a treatment for all cases of melanoma in the future. Therefore there is a need to identify melanoma patients who have a genetically high risk of death and for whom immunotherapy would be a suitable and cost-effective option. Given the increased incidence of melanoma in geriatric populations (>60 years old), there is a particular need to better stratify this population. There is also a need for an alternative, more effective and patient-friendly method to determine melanoma stage and predict patient prognosis.

It is an object of the present invention to address one or more problems associated with stratifying melanoma patients and the selection of such patients for immunotherapy treatment. It would be desirable to identify one or more mutated genes which could be used to stratify melanoma patients and determine their likelihood of responding to immunotherapy.

Summary of Invention

In accordance with a first aspect of the present invention, there is provided a method of stratifying and/or determining patient prognosis of a melanoma patient aged over 55, the method comprising the identification of one or more mutations in one or more of the following genes: NRAS, OR11 H12, POTEM, IDH1 , HIST1 H3D, C15orf23, ACD, OXA1 L, NPPA, BRAF, C9orf171 , NR5A1 , ABCD3, TROAP, GAGE2A. NAA11 , WNT5A and/or CDKN2A.

There method of stratifying and/or determining patient prognosis of a melanoma patient may be in a patient aged 56 or over, 57 or over or 59 or over. Preferably, the method of stratifying and/or determining patient prognosis of a melanoma patient is in a patient aged 60 or over.

The genes may comprise one or more of any of the above 18 genes, or alternatively combinations of a smaller number of these genes can be used to predict outcome in the elderly. Specifically, the presence of a single gene mutation in one or more of the following combination: BRAF, NRAS, IDH1 and/or CDKN2A have been determined by the inventors to be advantageously representative, powerful combination of genes that can be used to predict outcome. The present inventors have found that these four genes are strongly associated with poor outcome in older patients in localised and metastatic melanoma. The 4 gene driver mutation signature can provide a robust and simple analysis which can be easily incorporated into clinical practice to identify patients at genetic high risk of disease progression.

In accordance with a second aspect of the present invention, there is provided a method of stratifying and/or determining patient prognosis of a melanoma patient aged 55 or less, the method comprising the identification of one or more mutations in one or more of the following genes: BRAF, NRAS, CDK4, VN1 R4, IGLL5, KRTAP1 Cl 8, SAG, BEGAIN, ZFAND2A, TIGD4 and/or RAC1. All these genes have been identified by the inventors as associated with a better patient outcome when found mutated singly or mutated in combination with one another.

Advantageously the present inventors have identified an association between patient prognosis with driver mutations in 11 key driver genes in younger patients. The present inventors identified 11 driver mutations in young patients (<=55) of which BRAF was identified as the primary driver of outcome.

The genetic signature can stratified the aged melanoma patient at any stage of melanoma according to AJCC criteria. The melanoma patient may have a primary or a late-stage melanoma. The melanoma patient may have a melanoma of any invasive Breslow.

Advantageously the gene mutation signatures identified by the present inventors can predict poor disease specific outcomes in melanoma patients. Identifying the mutation status of the genes at any stage of the disease offers a greater prognostic power than current staging guidelines. Patients who are genetically predisposed to progression can therefore be distinguished. The method can be incorporated into clinical practice to identify patients at genetic high risk of progression and thus select suitable candidates for therapy and specific types of therapy.

The method of either aspect may further comprise stratification of the melanoma patient into disease progression risk groups. For instance if a mutation of any of the specified genes is identified the patient may be stratified into a prognosis group.

In accordance with a third aspect of the present invention, there is provided a method of identifying whether a melanoma patient aged over 55 (and preferably 60 or over) is likely to respond to immunotherapy, the method comprising identification of a high tumour mutation burden in combination with the identification of one or more mutations in one or more of the following genes: BRAF, NRAS, IDH1 and/or CDKN2A.

Advantageously the present inventors have shown that identifying this particular combination powerfully predicted those patients most likely to respond to immunotherapy. In particular those who would respond to immune checkpoint inhibitor therapies. It has been shown that although TMB is a predictor of survival in immunotherapy-treated patients across all cancer types it is not a useful predictor of immunotherapy response in melanoma when used alone.

In accordance with a fourth aspect of the present invention, there is provided a method of selecting a melanoma patient aged 60 or over for immunotherapy treatment, the method comprising identification of a high tumour mutation burden in combination with the identification of one or more mutations in one or more of the following genes: BRAF, NRAS, IDH1 and/or CDKN2A.

The genes may be BRAF, NRAS, IDH1 and CDKN2A. The melanoma patient may have a tumour mutation burden in the highest 30% centile or top tercile of the melanoma patient population in any given test.

The immunotherapy treatment may be one or more of adjuvant, neoadjuvant and/or late stage immunotherapy. Specifically the immunotherapy treatment may be immune checkpoint inhibitor treatment. Advantageously the method can guide the selection of immunotherapy. Integrating tumour mutation burden with the four-gene signature identifies melanoma patients with a better response to immune checkpoint inhibitors. The four gene signature and tumour mutation burden is a powerful tool to improve outcome prediction and is a step towards personalised therapy in melanoma.

In accordance with a fifth aspect of the present invention, there is provided a kit for conducting the method of any previous claim.

The kit may comprise use in the identification of an individual who would be likely to respond to immunotherapy treatment, the kit comprising:

a) oligonucleotide primers directed towards one or more mutations of the genes listed above; and

b) reagents and/or vessels for undertaking amplification and/or sequencing so that a mutation can be detected and direct the physician to administer a immunotherapeutic if required.

Features, integers, characteristics, compounds, methods, assays and devices described in conjunction with a particular aspect, embodiment or example of the invention are to be understood to be applicable to any other aspect, embodiment or example described herein unless incompatible therewith. All of the features disclosed in this specification (including any accompanying claims, abstract and figures), and/or all of the steps of any method or process so disclosed, may be combined in any combination, except combinations where at least some of such features and/or steps are mutually exclusive. The invention is not restricted to the details of any foregoing embodiments. The invention extends to any novel one, or any novel combination, of the features disclosed in this specification (including any accompanying claims, abstract and drawings), or to any novel one, or any novel combination, of the steps of any method or process so disclosed.

Detailed Description of the Invention

Aspects and embodiments of the present invention will now be illustrated, by way of example, with reference to the accompanying figures. Further aspects and embodiments will be apparent to those skilled in the art. All documents mentioned in this text are incorporated herein by reference.

Figures 1A and 1 B show overall (1A) and progression-free (1 B) survival of patients with metastatic melanoma from TCGA, stratified by age;

Figure 2 shows the mutational signature decomposition for old and young melanoma samples from TCGA. The vast majority of the mutational contribution comes from UV damage, with minor contributions that differ between the two groups;

Figures 3A, 3B and 3C show the molecular characterisation of metastatic melanoma (TCGA). (3A) Top 2 panels show rainfall plots of total mutations across the genome (x) by the distance from the previous mutation (y) illustrating the greater density of mutations in the elderly (left panel). Black arrows on the x axis indicate points of hypermutation. (3B) Results of the oncodriveCLUST algorithm, which identifies 18 driver genes in elderly melanoma, brackets indicate the cluster of mutation for that gene. (3C) Overall survival of patients in TCGA age >59 with metastatic melanoma who have mutations in any of the 4 key driver genes; Figure 4 shows a summary of the types and frequencies of mutations in metastatic melanoma samples from old or young patients in TCGA. Older patients harbour greater numbers of mutations in all categories, but the relative proportions of types of mutations are similar between the groups;

Figure 5 shows OncodriveCLUST analysis of metastatic melanoma in age <55 showing 11 driver genes and their clusters in brackets;

Figure 6 shows the overall survival in metastatic melanoma in TCGA. Upper plot shows the predictive effect of mutations in any of the 18 identified driver mutations in the elderly, the lower plot show the same for the 11 driver mutations in young melanoma;

Figure 7 shows overall survival in melanoma patients of age >59 from high- risk primary melanomas (French cohort) and from early stage melanomas (Spanish cohort);

Figure 8 shows the correlation between age and TMB across the TCGA and MSK-IMPACT cohorts; and

Figure 9 shows overall survival in immunotherapy treated metastatic melanoma from the MSK-IMPACT study stratified by age and the 4 gene driver signature.

This study reveals the molecular drivers of old age melanoma. It identifies the genetic changes of early and late stage melanoma that predict patient outcome in international, independent cohorts. The genetic predictors of prognosis are additionally able to identify immune checkpoint inhibitor responders in the aged melanoma population with a high tumour mutation burden.

The genetic signature of poor outcome stratifies patient prognosis more accurately than the current staging guidelines, making it immediately relevant to clinical practice. The use of age-specific markers can guide the selection of immunotherapy in old patients at late metastatic stages.

Metastatic melanoma samples were analysed to establish the effect of age on prognosis and treatment response. Age-specific molecular differences and driver genes linked to outcome and to response to immunotherapy in early and late stage international independent cohorts were identified.

The molecular events of older melanoma patients, who carry a much higher mutation burden than younger ones, and more loci of hypermutation were anlaysed. Old and young age driver mutations, and old age mutations associated with poor prognosis were identified. A set of four driver genes is strongly associated with poor outcome in older patients in localised and metastatic melanoma. Integrating tumour mutation burden with the four-gene signature identifies aged metastatic melanoma patients with a better response to immune checkpoint inhibitors.

The study defines the driver mutations of old age melanoma that are powerful predictors of poor outcome at all stages of disease. This signature stratifies patients more accurately than the current staging guidelines, making it relevant to immediate clinical practice. Furthermore, integrating the molecular signature with tumour mutation burden identifies the subset of aged patients with a higher rate of response to immune checkpoint inhibitors. A robust tool to distinguish patients who are genetically predisposed to progression and the subset of patients with a higher rate of immunotherapy response is provided.

Patient cohorts

The initial exploratory analysis of the effects of age on survival, as well as the mutational profiles and driver genes in patients of different ages, were undertaken in The Cancer Genome Atlas dataset which represents the most complete collection of available linked molecular and clinical data in cutaneous melanoma. To minimise bias in the data only metastatic melanoma samples were used for this exploration. Clinical and genetic mutation data for all patients with metastases at age >59 or <=55 were included. The experimental age cut-offs were decided a priori to obtain more clear-cut young and old age categories. Mutation data in the form of curated maf files were obtained from the maftools R package ¹⁶ , and clinical data from the GDAC firehose database (doi: 10.7908/C11 G0KM9).

To validate the clinical and molecular findings from the exploratory analysis the data from two separate, unrelated clinical cohorts from France (La Timone Aix- Marseille University Hospital) and Spain (Instituto Valenciano de Oncologia) were studied. The French cohort comprises 197 patients with clinical outcome data and targeted deep sequencing information. 109 patients were older than 59 at the time of diagnosis. The Spanish cohort comprises 72 patients with similar clinical and exome sequencing data, 30 of whom were older than 59 at diagnosis (Table 1 below). Relevant institutional review boards and ethics committees at both institutions approved molecular research. All human participants gave written informed consent.

Table 1A

fable 1 B _ _

Table 1 : (A) Demographic and clinical characteristics of patients in the 4 tested cohorts, and (B) Results of multivariate regression analysis of TCGA patient survival. Age, Breslow = mean + sd. Exact ages and staging information not given in MSK- IMPACT. Breslow thickness and stage not recorded in MSK-IMPACT. Mutation count: all mutations in TCGA, non-synonymous variants from selected gene panel in MSK-IMPACT.

To investigate the effects of the driver genes and tumour mutation burden in immunotherapy treated patients data from the MSK-IMPACT study ¹⁷ was downloaded and reanalysed. The melanoma cohort included 320 patients, 181 of whom were older than 59 at diagnosis, all of whom had been treated with immunotherapy agents and whose sequencing data included the mutation status of the four driver genes. To determine high vs low tumour mutation burden in this cohort the centile groupings within the melanoma group were investigated and the third decile (ie those in the highest 30% tumour mutation burden) used as high. The effect of varying the decile threshold described in the original publication was reproduced ¹⁷ , and minimal differences between deciles 1 -3 in predictive effect (consistent with the original findings) were found so the threshold which included as many patients as possible was used.

Data Analysis and statistics

Data analysis was performed using the R statistical programming language ¹⁸ . Mutation analysis used the maftools package ¹⁶ , survival analysis the survival ¹⁹ and survminer packages and plotting used the ggplot2 ²⁰ and ggscipackages. Statistical testing was performed according to best practice guidelines. Multivariate regression was performed using a cox proportional hazards model in which each non-significant variable was sequentially eliminated from the model to leave a final minimal model containing only significant variables. Table 2 below details the iterations of the regression model, and the variables excluded at each step.

Table 2. Multivariate regression using a cox proportional hazards model. Non- significant variables sequentially eliminated from the model during iterative process. Stg. Stage.

Tumour mutation burden (TMB) was analysed as a log value in Cox regression models. Mutation density and kataegis (hypermutation) analysis was calculated by assessing the average distance between consecutive mutations across all series of six mutations, with hypermutation loci defined as those with six consecutive mutations with an average distance of 1000 kb or less between them ²¹ . Driver gene analysis incorporating background mutation rates and gene length used the oncodriveCLUST algorithm ²² implemented in maftools.

Results

Impact of age on melanoma survival The association between older age and poor prognosis in melanoma is established ⁵ , but the molecular characteristics of older age and their impact on survival have not been studied systematically previously. Data from the Cancer Genome Atlas study was analysed to investigate the impact of older age (>59) on overall survival in melanoma (Figure 1A). To distinguish better the effect of age, the population was divided into old (>59) and young (<=55). In this population (n=324, median follow-up 47.5 months, 137 patients deceased) age >59 was significantly associated with poorer overall survival compared with age <=55 in metastatic melanoma (p<0.0001 , median survival 62 vs 144 months). Multivariate stepwise regression (Table 1 ) showed that age was the single most significant determinant of outcome (p=0.001 ) followed by stage at diagnosis (p=0.025), and importantly, tumour mutational load is not significantly associated with survival in this cohort after multivariate regression. This result might be explained through non-disease related mortality, so the impact of age on progression free survival (PFS, Figure 1 B) was investigated and a similarly striking and significant effect of age (p=0.013, 48 vs 143 months median survival) was found, and confirmed tumour mutational load is not significantly associated with PFS in this cohort. These findings remained consistent when the entire melanoma population was used and compared all patients >59 to patients <=59.

Age-specific melanoma mutations and survival

The primary molecular drivers linked to the striking impact that age has on melanoma survival were determined. The whole exome sequencing data from TCGA was analysed and established that older patients carry a significantly higher TMB than do younger patients (median 335 old, 274 young, p<0.0001 ), but in multivariate analysis, TMB alone was not significantly associated with survival (Table 1 ). Previous work has shown a survival effect of TMB in melanoma ²³²⁴ which was explored in detail. When TMB was considered as in previous work a binary variable (high or low) with a cut-off value of 130 mutations, it was associated with improved survival ²³ ; however when analysed as a continuous variable in multivariate regression (logarithmic or unchanged), TMB was no longer a significant predictor of survival and only age predicted outcome in the full model (Table 1 ).

Having established that total mutational load alone was insufficient to explain the survival difference, it was hypothesised that the nature and distribution of the mutations would be significant, and so was investigated. It was confirmed by mutational signature decomposition that the mutations in all ages were overwhelmingly consistent with UV-mediated DNA damage (Figure 2). By analysing the genomic locations of mutations in the young and old cohorts it was found that regions of hypermutation - that is where multiple consecutive mutations occur in short stretches of the genome - occur more frequently in older than in younger patients (Figure 3A). These data indicate that not only do mutations accumulate to a greater degree in elderly patients, but also that specific regions of the genome, which are more susceptible to ultraviolet radiation induced damage, become more frequently mutated in older patients. Furthermore, it was discovered that the genes affected by these mutations were more numerous and varied than in younger patients (Figure 4).

To test the hypothesis that melanoma development in older patients is driven by accumulation of mutations in specific risk genes, the mutation data was analysed using the oncodriveCLUST algorithm ²² to discover the key driver mutations taking into account the background mutation density and gene length in melanoma of the elderly (Figure 3B). The 18 key driver genes in older patients were different from the 11 driver mutations in young patients (Figure 5) with only BRAF and NRAS found in both groups. Critically it was shown that the group of patients whose melanomas harboured driver mutations in any of the key driver genes in the aged was significantly associated with poor prognosis (p= 0.013, Figure 6). These driver mutations were analysed to identify a simple, powerful combination predictive of survival by systematically excluding genes and only keeping combinations that increased predictive power. A core set of four genes were identified {BRAF, NRAS, IDH1 and CDKN2A) that when mutated singly or in combination powerfully predicts poor outcome in the TCGA cohort (Figure 3C). The mutational signature identified increases in frequency as the melanoma population ages, and it is in the elderly subgroup where its strongest predictive power is demonstrated. Critically, a set of alternative genes can more accurately predict outcome in the young, with BRAF the primary driver of outcome in that group (Figure 6). The core four-gene driver mutation signature identified represents a robust, simple set of analyses that can be easily incorporated into clinical practice to identify patients at genetic high risk of progression.

To be clinically useful a predictive signature should be applicable to all stages of disease, including early stage primary melanomas from an unselected group of patients. To comprehensively assess the validity of the four-gene core signature of poor prognosis two additional multinational cohorts of primary melanoma including patients aged >59 with available sequencing information were interrogated. It was confirmed that the signature significantly and powerfully predicted poor disease- specific outcomes in older patients with primary melanoma in both cohorts including primary melanomas >1 mm diagnosed in France and primary melanomas (Breslow >0mm) diagnosed in Spain (Figure 7). These results show conclusively that identifying the mutation status of the four genes in melanoma samples at any stage of disease in elderly patients offers a greater prognostic power than current staging guidelines ²⁵ , and represents a novel strategy for clinical stratification of melanoma patients.

Age, TMB and the four gene signature predict immunotherapy response

Prediction of response to immunotherapy for high-risk early and late stage melanoma is a pressing clinical requirement and to date there are no reliable pre treatment genomic biomarkers to guide the selection of melanoma patients at highest risk of progression and with a higher likelihood of response. Recent evidence shows a higher TMB leads to a greater likelihood of neo-antigen-driven immunotherapy response in the melanoma population not selected for age ^26-30 ; and additionally, the presence of two of our core signature genes A/RAS ³¹ and CDKN2A mutations ³² in metastatic melanoma identify subgroups of patients with improved rates of response to immunotherapy. However, data from the large-scale MSK-IMPACT study has demonstrated that although TMB is a predictor of survival in immunotherapy-treated patients across all cancer types, it is not a useful predictor of immunotherapy response in melanoma specifically ¹⁷ . Because age and TMB are significantly correlated (Figure 8), the relationship between age, TMB, response to immunotherapy and the 4-gene signature was investigated. Critically, although neither the four gene signature nor the TMB in young or old patients predicted response to immunotherapy when used alone, the combination powerfully predicted response in the immunotherapy-treated, aged cohort. Specifically, patients with both high TMB and the presence of the 4-gene signature showed a 12-month longer median survival compared with patients with low TMB and the four gene signature (Figure 9A-C). In young patients neither TMB nor the signature alone or in combination stratified patient response to immunotherapy (Figure 9D).

Untreated aged melanoma patients have a significantly worse outcome. These analyses delineate the molecular changes that identify aged patients at highest risk of death. Intriguingly, it is precisely this subset of patients who will derive the most benefit from checkpoint inhibitor treatment in the metastatic setting. The core four gene driver mutation signature represents a robust, simple set of analyses that can be easily incorporated into clinical practice to identify patients at genetic high risk of progression, and when combined with TMB, can identify the patients who are most likely to respond to immunotherapy. The four gene signature and TMB is a powerful tool to improve outcome prediction and a genuine step towards personalised therapy in melanoma.

Discussion

For the first time a comprehensive, systematic analysis of the molecular and clinical characteristics of malignant melanoma in older patients has been conducted. Patients who are 60 and older represent a significant bulk of all melanoma cases, and they have a disproportionately poor outcome compared to younger patients. There are no established stratification strategies for prognosis in older patients, leading to a one-size-fits-all clinical approach with no specific guidelines for monitoring, diagnosis or treatment variation by age. The vast genomic differences following age-specific dissection of molecular data supports the view that melanoma in elderly patients comprises a distinct group, and it is shown for the first time that this difference is driven by defined molecular characteristics. The study suggests elderly melanoma patients should be stratified and cared for separately, and specific molecular differences can be used to improve current prognostic categories at all stages of disease. Additionally, the molecular signature identified shows a link with nonsynonymous somatic TMB and checkpoint inhibitor response in aged patients - potentially allowing for targeted therapeutic interventions that are more likely to work in age-specific molecularly defined categories. Older patients are known to respond to ICI ¹¹ - ¹³ ’ ³³ , and the molecular differences described in this work have the advantage of identifying elderly patients at highest risk for progression who will benefit the most from treatments for advanced stage melanoma.

Numerous other tumour characteristics are known to be independent predictors of response to ICIs. On the one-hand these include markers of the tumoral immune-state, such as immune cell infiltration, immune cell exclusion and expression of checkpoint molecules; and on the other hand tumour markers of molecular damage, such as clonality, genotype and increasing TMB ²⁸ ’ ^34-37 . Importantly, there are tumour immune-state ICI response predictors of poor outcome in melanoma that are more prevalent in the young ³³ , further supporting the premise that age classifies melanoma into distinct clinical and therapeutic categories, possibly underpinned by unique pathophysiological drivers. At one end of a simplified spectrum, young patients present lower levels of TMB with an immune phenotype that renders them more immune evasive; and on the other, old patients with high TMB have an immune microenvironment and tumour profile more propitious to ICI.

Increasing levels of accrued DNA damage is an established biomarker of ICI response in other cancer types ¹⁷ , and is in keeping with the emerging data showing a higher burden of mutations leads to more neoantigens driving an antitumour response. The data reveals the same fundamental mechanism likely contributes to melanoma ICI response, but restricted to the elderly population. Integrating TMB with the relevant four gene mutational signature allows for the identification of the subset of patients for whom ICI will be most advantageous. Given the financial cost and toxicities of ICI agents, the study provides a biomarker that can stratify patients, predict response and inform therapy decisions.

References

1 Balch CM, Soong S, Gershenwald JE, et al. Age as a Prognostic Factor in Patients with Localized Melanoma and Regional Metastases. Ann Surg Oncol 2013; 20: 3961-8.

2 Hillen U, Leiter U, Haase S, et al. Advanced cutaneous squamous cell carcinoma: A retrospective analysis of patient profiles and treatment patterns— Results of a non-interventional study of the DeCOG. Eur J Cancer 2018; 96: 34-43.

3 Green AC, Olsen CM. Cutaneous squamous cell carcinoma: an epidemiological review. Br J Dermatol 2017; 177: 373-81.

4 Balch CM. Decreased Survival Rates of Older-Aged Patients with Melanoma:

Biological Differences or Undertreatment? Ann Surg Oncol 2015; 22: 2101-3.

5 Tsai S, Balch C, Lange J. Epidemiology and treatment of melanoma in elderly patients. Nat Rev Clin Oncol 2010; 7: 148-52.

6 Cavanaugh-Hussey MW, Mu EW, Kang S, Balch CM, Wang T. Older Age is Associated with a Higher Incidence of Melanoma Death but a Lower Incidence of Sentinel Lymph Node Metastasis in the SEER Databases (2003-2011 ). Ann Surg Oncol 2015; 22: 2120-6.

7 Balch CM, Soong SJ, Gershenwald JE, et al. Prognostic factors analysis of 17,600 melanoma patients: validation of the American Joint Committee on Cancer melanoma staging system. J Clin Oncol 2001 ; 19: 3622-34.

8 Schadendorf D, van Akkooi ACJ, Berking C, et al. Melanoma. Lancet 2018;

392: 971-84.

9 Ribas A, Wolchok JD. Cancer immunotherapy using checkpoint blockade.

Science 2018; 359: 1350-5.

10 Kugel CH, Douglass SM, Webster MR, et al. Age Correlates with Response to Anti-PD1 , Reflecting Age-Related Differences in Intratumoural Effector and Regulatory T-Cell Populations. Clin Cancer Res 2018; 24: 5347-56.

11 Perier-Muzet M, Gatt E, Peron J, et al. Association of Immunotherapy With Overall Survival in Elderly Patients With Melanoma. JAMA Dermatology 2018; 154: 82.

12 Joshi KP, Atwal D, Ravilla R, et al. Outcomes of immunotherapy in advanced melanoma in relation to age. J Clin Oncol 2018; 36: 187-187.

13 De Rosa F, Ridolfi L, Tanda ET, et al. Anti-PD1 antibodies in late elderly advanced melanoma patients: A retrospective multicentre study. J Clin Oncol 2018; 36: 10038-10038.

14 Eggermont AMM, Robert C, Ribas A. The new era of adjuvant therapies for melanoma. Nat Rev Clin Oncol 2018; 15: 535-6.

15 Eggermont AMM, Blank CU, Mandala M, et al. Adjuvant Pembrolizumab versus Placebo in Resected Stage III Melanoma. N Engl J Med 2018; 378: 1789-801.

16 Mayakonda A, Lin D-C, Assenov Y, Plass C, Koeffler HP. Maftools: efficient and comprehensive analysis of somatic variants in cancer. Genome Res 2018; 28: 1747-56.

17 Samstein RM, Lee C-H, Shoushtari AN, et al. Tumour mutational load predicts survival after immunotherapy across multiple cancer types. Nat Genet DOI: 10.1038/S41588-018-0312-8.

18 R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R- project.org/. 2013. T T. A Package for Survival Analysis in S. version 2.38. https://CRAN.R- project.org/package=survival. 2015.

Wickham H. ggplot2. New York, NY: Springer New York, 2009 DOI: 10.1007/978-0-387-98141 -3.

Nik-Zainal S, Van Loo P, Wedge DC, et al. The Life History of 21 Breast Cancers. Cell 2012; 149: 994-1007.

Tamborero D, Gonzalez-Perez A, Lopez-Bigas N. OncodriveCLUST: exploiting the positional clustering of somatic mutations to identify cancer genes. Bioinformatics 2013; 29: 2238-44.

Gupta S, Artomov M, Goggins W, Daly M, Tsao H. Gender Disparity and Mutation Burden in Metastatic Melanoma. J Natl Cancer Inst 2015; 107: djv221.

Trucco LD, Mundra PA, Hogan K, et al. Ultraviolet radiation-induced DNA damage is prognostic for outcome in melanoma. Nat Med 2018; published online Dec 3. DOI: 10.1038/s41591 -018-0265-6.

Gershenwald JE, Scolyer RA, Hess KR, 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: 472-92.

Snyder A, Makarov V, Merghoub T, et al. Genetic Basis for Clinical Response to CTLA-4 Blockade in Melanoma. N Engl J Med 2014; 371 : 2189-99.

Johnson DB, Frampton GM, Rioth MJ, et al. Targeted Next Generation Sequencing Identifies Markers of Response to PD-1 Blockade. Cancer Immunol Res 2016; 4: 959-67.

McGranahan N, Furness AJS, Rosenthal R, et al. Clonal neoantigens elicit T cell immunoreactivity and sensitivity to immune checkpoint blockade. Science 2016; 351 : 1463-9.

Goodman AM, Kato S, Bazhenova L, et al. Tumour Mutational Burden as an Independent Predictor of Response to Immunotherapy in Diverse Cancers. Mol Cancer Ther 2017; 16: 2598-608.

Rizvi NA, Hellmann MD, Snyder A, et al. Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer. Science 2015; 348: 124-8.

Johnson DB, Lovly CM, Flavin M, et al. Impact of NRAS mutations for patients with advanced melanoma treated with immune therapies. Cancer Immunol Res 2015; 3: 288-95.

Helgadottir H, Ghiorzo P, van Doom R, et al. Efficacy of novel immunotherapy regimens in patients with metastatic melanoma with germline CDKN2A mutations. J Med Genet 2018; : jmedgenet-2018-105610.

Kugel CH, Douglass SM, Webster MR, et al. Age Correlates with Response to Anti-PD1 , Reflecting Age-Related Differences in Intratumoural Effector and Regulatory T-Cell Populations. Clin Cancer Res 2018; 24: 5347-56.

Topalian SL, Hodi FS, Brahmer JR, et al. Safety, Activity, and Immune Correlates of Anti-PD-1 Antibody in Cancer. N Engl J Med 2012; 366: 2443- 54.

Rooney MS, Shukla SA, Wu CJ, Getz G, Hacohen N. Molecular and Genetic Properties of Tumours Associated with Local Immune Cytolytic Activity. Cell 2015; 160: 48-61.

Chowell D, Morris LGT, Grigg CM, et al. Patient HLA class I genotype influences cancer response to checkpoint blockade immunotherapy. Science 2018; 359: 582-7.

Balachandran VP, tuksza M, Zhao JN, et at. Identification of unique neoantigen qualities in long-term survivors of pancreatic cancer. Nature 2017; 551 : 512.