Biomarkers predictive of survival and response to immune checkpoint inhibitors (ICI) in melanoma

Abstract

Immunotherapy has revolutionized the treatment of melanoma. Targeting of the immune checkpoints cytotoxic T-lymphocyte-associated protein 4 and programmed cell death protein 1 has led to improved survival in a subset of patients. Unfortunately, the use of immune checkpoint inhibitors is associated with significant side effects and many patients do not response to treatment. Thus, there is an urgent need both for prognostic biomarkers to estimate risk and for predictive biomarkers to determine which patients are likely to respond to therapy. In this review, prognostic and predictive biomarkers that are an active area of research are outlined. Of note, certain transcriptomic signatures are already used in the clinic, albeit not routinely, to prognosticate patients. In the predictive setting, programmed cell death protein ligand 1 expression has been shown to correlate with benefit but is not precise enough to be used as an exclusionary biomarker. Future investigation will need to focus on biomarkers that are easily reproducible, cost-effective, and accurate. The use of readily available clinical material, such as serum or hematoxylin and eosin-stained images, may offer one such path forward.

1. Introduction

In the United States, melanoma is the fifth most common type of cancer; an estimated 96,480 new cases of melanoma will be diagnosed in 2019.¹ Incidence rates for melanoma are still increasing,¹ and as such an increased focus on both prevention and treatment of the disease is urgent. The development of immunotherapy has revolutionized the treatment of melanoma. In particular, cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) and programmed cell death protein 1 (PD-1) immune checkpoint inhibitors (ICI) such as ipilimumab, nivolumab, and pembrolizumab have shown significant survival benefits in patients with advanced melanoma. In patients with unresectable stage III or IV melanoma, treatment with the CTLA-4 checkpoint blockade agent ipilimumab led to improved overall survival (OS) compared to treatment with gp100 peptide vaccine (10.1 months and 6.4 months, respectively).² For patients with previously untreated BRAF wild type stage III melanoma, treatment with the PD1 checkpoint blockade agent nivolumab resulted in a 40.0% objective response rate (ORR) and progression-free survival (PFS) of 5.1 months, compared to chemotherapy with 13.0% ORR and 2.2 month PFS.³ Pembrolizumab, another PD-1 inhibitor, has shown prolonged PFS relative to ipilimumab in patients with unresectable stage III or IV melanoma, with a 6-month estimated PFS of 47.3% using biweekly pembrolizumab and 26.5% using ipilimumab.^4,5 Such findings have led to the FDA approval of both pembrolizumab and nivolumab for the treatment of patients with unresectable or metastatic melanoma and, more recently, for the adjuvant treatment of stage III disease.⁶ Ipilimumab has also been FDA approved for the treatment of unresectable or metastatic melanoma in patients 12 years and older and in the adjuvant setting.⁷ In 2015, the combination of nivolumab and ipilimumab was also FDA approved in patients with unresectable or metastatic melanoma based on data indicating that such combination treatment led to a 61% ORR as opposed to an 11% ORR with ipilimumab alone.⁸

Despite the success of ICI in the treatment of melanoma, these treatments are also associated with serious adverse events such as colitis.⁹ For example, a phase 3 study evaluating combined ipilimumab with nivolumab against single agent nivolumab and single agent ipilimumab found that 59%, 21%, and 28% of patients experienced a grade 3 or 4 treatment-related adverse event, respectively.¹⁰ Thus, the development of biomarkers is critical to helping clinicians determine whether patients’ possible benefits from immunotherapy outweigh potential toxicities. Such biomarkers may play a key role in the adjuvant setting in identifying which patients have a low risk of recurrence and should be spared treatment.

Biomarkers for cancer can be subdivided into two categories: prognostic and predictive. While prognostic biomarkers aim to estimate a patient’s prognosis (e.g., whether that patient will recur) regardless of therapy, predictive biomarkers are meant to determine a patient’s response to a specific therapy, such as ICI.^11,12 This review summarizes current prognostic biomarkers and predictive biomarkers for ICI in melanoma and highlights biomarkers that are an active focus of research.

Traditional prognostic biomarkers for melanoma include the histological factors involved in disease staging, such as depth of invasion, ulceration, and presence of microsatellite lesions.¹³ The Tumor, Node, Metastasis (TNM) classification in particular is used clinically to assess a patient’s risk category based on tumor thickness and ulceration (T category), lymph node involvement and presence of microsatellites (N category), and presence of sites of distant metastases (M category).¹³ Tumor depth and ulceration are both inversely correlated with survival, with thicker tumors and ulcerated tumors being associated with lower rates of survival. Tumor thickness and ulceration also act as independent predictors of metastases to the sentinel lymph node (SLN).^14,15 SLN metastases remain the most important predictors of recurrence in patients with primary T2 and T3 tumors, and thus SLN biopsies are routinely used in these patients to determine nodal involvement.^16,17 In thin melanomas, guidelines from the American Society for Clinical Oncology (ASCO) recommend that SLN biopsy be considered in patients with high-risk features, such as ulceration and high mitotic rate.¹⁸ However, given the cost of biopsy, its potential morbidity, and the low incidence of SLN metastasis in patients with thin melanoma, a more concrete understanding of predictors of SLN positivity would be useful in clinical decision-making.^19–22

2. Determining risk: prognostic biomarkers

A number of biomarkers have been examined to prognosticate patients with melanoma aside from SLN positivity. Many have been extensively validated and may thus prove useful in stratifying patients into risk groups. Several of the biomarkers discussed below, such as tumor infiltrating lymphocytes (TILs), osteopontin, and lymphatic invasion, have also been examined for correlation to SLN positivity. Such biomarkers may be of particular use in thin melanomas (under 1 millimeter), where the SLN positivity rate is only around 5%²³ and where stratification of patients could help avoid unnecessary costs and potential morbidity of SLN biopsy.

2.1. Tissue-based biomarkers

2.1.1. Pathology-based biomarkers

TILs, which are visible via hematoxylin and eosin (H&E) staining, have been extensively examined both as prognostic biomarkers and as biomarkers predictive of SLN positivity. TIL patterns are classified as grades: absent, with no lymphocytes in the tumor; non-brisk, with a foci or multiple foci of lymphocytes within the tumor; and brisk, with diffuse lymphocyte infiltration within the tumor.²⁴ In early stage melanoma, the presence of brisk TILs in vertical growth phase tumors is associated with the most favorable disease-specific survival (DSS) and OS, followed by non-brisk and absent TILs, as demonstrated by Clark et al. in 1989.^24,25 More recently, a study of 1,865 patients with a single melanoma greater than or equal to 0.75 mm in thickness found that TILs were independent predictors of SLN status.²⁶ The study demonstrated a clear inverse relationship between TIL grade and SLN status, with a positive SLN biopsy in 27.8% of patients with absent TILs, as opposed to a positive SLN biopsy in 5.6% of patients with brisk TILs.²⁶

Melanomas have significant morphologic diversity. A melanoma can be classified as superficial spreading, nodular, lentigo maligna, or acral lentiginous.²⁷ For the most part, melanoma histotypes have not been found to be prognostic when considered independently from tumor thickness and are thus not included in the American Join Committee on Cancer (AJCC) staging system.^13,28,29 However, a recent paper has challenged this belief by finding that nodular melanoma was an independent predictor for recurrence and death from melanoma.³⁰

Most recently, deep learning-based computational methods have been developed to predict a patient’s likelihood of recurrence based on morphological features in digitized pathology images. Some of these methods are intended to assess tumor subtype,³¹ while others have used deep learning to accurately grade a tumor.³² In melanoma, we have developed one such deep learning-based biomarker that stratifies patients with stage I, II, and III melanoma into risk groups.³³ This biomarker correlates with DSS in two independent validation cohorts, and with further validation may be used to accurately predict prognosis in patients with early stage melanoma.³³

2.1.2. Protein biomarkers

Cell adhesion proteins are important in the development of metastatic disease, and as such have been examined as potential prognostic biomarkers. Melanoma cell adhesion molecule (MCAM), the predominant cell adhesion marker in melanoma, is expressed in over 80% of metastatic tumors but is rarely expressed on benign nevi.³⁴ MCAM positive melanoma is associated with a significantly worse 5-year survival than MCAM negative melanoma, and survival decreases with increased expression of this marker.^35,36

Another promising protein biomarker is the expression of Ki-67. Ki-67 is a nuclear antigen that is a marker of cellular proliferation during the active stages of the cell cycle.³⁷ In thin melanoma (<1 mm), both Ki-67 expression and increased mitotic rate are associated with increased risk of metastasis.³⁸ Of note, mitotic rate was recently removed from the AJCC staging system for thin melanomas given that tumor thickness and ulceration were stronger predictors of survival, although assessing mitotic rate is still recommended across melanomas of different thicknesses.¹³ However, in thick melanoma, Ki-67 may be a better prognostic indicator than mitotic rate, and its expression is associated with increased melanoma thickness, presence of tumor ulceration, necrosis, increased Clark’s level of invasion, and vascular invasion.³⁹ Additionally, in recurrent melanomas, high Ki-67 expression was found to be independently associated with decreased OS.⁴⁰

The presence of lymphatic invasion has been shown to act as a predictor of locoregional cutaneous recurrence.⁴¹ D2–40 is an antibody to a sialoglycoprotein and is selective for the endothelium of lymphatic vessels.^42,43 In a study of primary melanomas with a thickness equal to or greater than 1mm, lymphatic invasion as assessed by D2–40 staining was associated with SLN metastasis.⁴⁴ The same association between lymphatic invasion and SLN positivity has been reported in a study focusing on thin to intermediate depth melanomas.⁴⁵

Osteopontin, an integrin-binding protein that drives IL-17 expression and that is overexpressed in many malignancies including colorectal, lung, breast, liver, and prostate cancer has been identified as another potential biomarker associated with tumor progression and metastasis.^46–48 In a cohort of 345 primary melanoma patients examined by immunohistochemical (IHC) analysis of a tissue microarray, osteopontin expression was found to correlate with an increased risk of SLN positivity and was an independent predictor of prognosis in melanoma.⁴⁹ Several other studies have also found osteopontin to be overexpressed in melanoma and that its presence at higher levels is associated with disease progression.^50–53

Previous studies have identified macrophages at the leading edge of the tumor to be prognostic in early-stage melanoma.⁵⁴ In addition, in depth analyses of TILs have found that CD8+ cytotoxic T lymphocytes (CTLs) are best associated with survival.⁵⁵ Closer evaluation using quantitative immunofluorescence (qIF) allows for location-specific concatenation of markers. Using qIF, a study of 104 primary stage II-III melanomas combined CTL and macrophage density in the stroma, finding that the CTL/macrophage ratio demonstrated improved prognostic power for both OS and DSS over independent CTL and macrophage density.⁵⁶

2.1.3. Genomic biomarkers

Gene expression profiles can provide additional information for risk stratification in patients with melanoma. The DecisionDx®-Melanoma Prognostic test (Castle Biosciences) is one such gene expression profile that was developed to classify patients with early melanoma (stage I and II) into low-risk (class 1) or high-risk (class 2) groups based on the expression of 31 genes in primary melanoma.⁵⁷ In studies using it in conjunction with the AJCC clinical staging system, DecisionDx®-Melanoma was shown to improve identification of patients with stage I and II disease who have a high risk of developing recurrent or metastatic disease.⁵⁸ However, this gene expression profile has not been widely adopted in melanoma management. Another gene expression profile has been developed specifically for primary stage II-III melanoma, which is a diagnosis of significant clinical uncertainty and thus for which risk stratification is especially important in guiding therapy. Melanoma immune profile (MIP), a 53-immune gene transcriptomic signature score, predicts the risk of distant metastatic recurrence and DSS in patients with stage II-III melanoma.⁵⁹ In creating a system for risk stratification, gene expression profiles provide an important clinical tool in guiding systemic therapy in melanoma.

2.1.4. Genetic biomarkers

Somatic cancer genomes have a large number of mutations within individual tumors and, in one study that used whole-genome sequencing, melanoma was found to be the most frequently mutated tumor.⁶⁰ Mutations that promote cancer progression are known as driver mutations. The most common driver mutations associated with the development of melanoma occur in the BRAF, NRAS, and NF1 genes. These mutations lead to oncogenic activation of the MAPK pathway, which is involved in cell growth, proliferation, and survival.⁶¹

BRAF mutations have been identified in 66% of melanomas, the most common being V600E, which constitutes approximately 80% of those mutations.⁶² NRAS is mutated in approximately 15–25% of melanomas,⁶³ and NF1 is mutated in approximately 14% of these tumors.⁶⁴ Each of these driver mutations impacts disease prognosis.

In high risk tumors (≥stage T2b), BRAF and NRAS are associated with a significantly lower melanoma specific survival.⁶⁵ NF1 mutations are also associated with a lower DSS and OS.⁶⁶ In addition to providing prognostic information, the identification of patients with BRAF mutations is important in directing therapy. Targeted therapy, including dabrafenib (BRAF inhibitor) plus trametinib (MEK inhibitor), improves outcomes in patients with resected stage III melanoma with the BRAF mutation. Patients treated with a combination of dabrafenib plus trametinib have increased OS and metastasis-free survival, highlighting the importance of identifying these mutations in patients with melanoma.⁶⁷ Finally, activating mutations in the KIT gene, although less common than the mutations described above, arise most often in mucosal or acral lentiginous melanomas and can be treated with imatinib, a kinase inhibitor.^68,69

2.2. Serum biomarkers

Serum biomarkers have the potential to provide a minimally invasive means of determining a patient’s prognosis. Lactate dehydrogenase (LDH) catalyzes the conversion of pyruvate to lactate, and is thus important for tumor metabolism.⁷⁰ Several studies have pointed to high LDH levels as an unfavorable prognostic factor in patients with cancer.^71,72 A meta-analysis of 76 solid tumor studies found that higher serum LDH was associated with poor survival, with melanoma being one of the tumor types in which this association was most significant.⁷³ Studies that have investigated the role of LDH in melanoma prognosis have confirmed this association.^74–76 Serum LDH is included as an independent prognostic biomarker for metastatic melanoma in the AJCC staging system, and until recently high levels of serum LDH upstaged patients with metastatic disease to stage IV M1c, which includes patients with metastases to the lungs and other visceral sites, excluding the central nervous system.^13,77

Similarly, S100B is a soluble calcium-binding protein that is secreted into the bloodstream by malignant melanocytes.⁷⁸ S100B levels in the serum have been shown to be dependent on tumor load and several studies have found that high levels of S100B may act as prognostic biomarkers and evidence of disease progression.^79–81

miRNAs have also been proposed as potential serum biomarkers in melanoma. Ever since their discovery in 2001, micro-RNAs have been extensively studied as post-transcriptional repressors of translation.^82–86 These short, noncoding RNAs have been found to be detectable and stable in exosomes in the serum.⁸⁷ Expression of various miRNAs at abnormal levels has been shown to correlate with disease stage and recurrence.^88,89

Finally, circulating tumor DNA (ctDNA) found in the plasma has been investigated as a means to determine disease status in patients with advanced melanoma.⁹⁰ This ctDNA can both provide insights into total tumor load and into tumor-associated mutations. Studies have shown that evaluation of circulating DNA can assist in detection of BRAF^V600 mutations, thus providing a useful tool for monitoring the efficacy of treatment with BRAF/MEK inhibitors.^91,92 Disease progression and treatment response as assessed by imaging scans have both been shown to correlate with levels of ctDNA that increase or decrease, respectively.^93,94

3. Predicting response to ICI: predictive biomarkers

As discussed above, a number of prognostic biomarkers have been developed in melanoma. Further, mutation of the BRAF gene is a widely used predictive biomarker for treatment with targeted therapy. On the other hand, the development of predictive biomarkers for response to ICI remains in a more experimental phase. No predictive biomarkers are yet in routine use for ICI in melanoma; a selection of promising biomarkers is discussed below.

3.1. Tissue-based biomarkers

3.1.1. Pathology-based biomarkers

As discussed above, the use of artificial intelligence (AI) to gather information from digitized H&E slides is an active area of research.^31,32 This technique has been extended to the development of predictive biomarkers as well. Recent data suggested that AI could potentially be used to predict the likelihood of response to ICI using pre-treatment digitized H&E slides.⁹⁵

3.1.2. Protein biomarkers

Immune infiltration by CD8+ TILs has been proposed as a predictive biomarker for response to immunotherapy in patients with melanoma. In a study of patients with advanced melanoma treated with pembrolizumab, patients who experienced a tumor response were found to have a higher CD8+ T cell density at the invasive tumor margin.⁹⁶ Another study similarly examined the role of CD8+ TILs in response to PD-1 inhibition, finding that pretreatment CD8+ lymphocytic infiltration as assessed by quantitative immunofluorescence was associated with prolonged survival after treatment.⁹⁷

Because ICI such as nivolumab and pembrolizumab target PD-1, which is often expressed on TILs, there has been significant interest in the potential correlation between the expression of PD-L1 on tumor cells and response to these immunotherapies.⁹⁸ Although PD-L1 is also expressed on cells of the myeloid lineage, its presence on tumor cells has garnered the most interest.^99–101 A tumor is generally defined as PD-L1 positive when 5% of its cells express the marker, and several studies have found that patients whose tumors express PD-L1 have a greater benefit from treatment with anti-PD1 therapy.^3,102 One such study found that in melanoma patients treated with nivolumab, those with a PD-L1 positive tumor status had a median PFS of 14.0 months and an ORR of 57.5%, whereas those with PD-L1 negative tumors had a median PFS of 5.3 months and an ORR of 41.3%.⁹⁸ Of note, PD-L1 levels are already commonly assessed in biopsies both in the clinic and in trials. However, treatment with PD-1 inhibitors has been shown to be effective both in patients with and without overexpression of PD-L1,¹⁰³ and the quality of IHC staining for PD-L1 is not reliable enough for clinical application.^11,104 This may in part be due to the fact that tumor heterogeneity may lead to heterogeneous expression of PD-L1, thus complicating interpretation of PD-L1 immunohistochemistry.¹⁰⁵ As a result, PD-L1 expression is not currently used in the clinic as an exclusionary biomarker for immunotherapy treatment.¹⁰⁶

Human Leukocyte Antigen (HLA) molecules are responsible for antigen presentation by both immune cells and cancer cells, and are thus key players in the immune system’s response to cancer.¹⁰⁷ HLA class II molecules, which mediate presentation of foreign peptides and are generally found on antigen-presenting cells (APCs), are frequently expressed on melanoma cells.¹⁰⁸ High expression of HLA-DR, a class II HLA molecule, was found to predict response to anti-PD-1 therapy in patients with metastatic melanoma.¹⁰⁸

Because ICI is thought to target and activate TILs, there has been interest in further phenotyping the TILs that create a tumor environment receptive to ICI.⁹⁶ TCF7, a transcription factor, has garnered interest as a potential predictive biomarker in TILs for immunotherapy.^109,110 The presence of TCF7 has been shown to promote a central memory stem-like phenotype, with CD8+ cells that express higher levels of TCF7 having the capacity to both self-renew and to differentiate into effector cells.^111,112 Further, TCF7-positive cells seem to be those that proliferate after treatment with anti-PD1.¹⁰⁹ In line with this observation, expression of TCF7 was found to correlate with positive clinical outcome in patients with melanoma treated with ICI.¹¹⁰ Thus, with further validation TCF7 expression could be used as a predictive biomarker in melanoma patients treated with ICI, and may potentially be used as a target for treatment.

3.1.3. Genomic biomarkers

Several studies have investigated whether specific gene expression signatures could predict sensitivity or resistance to treatment with ICIs. For example, an IFNɣ-related 18-gene signature has been found to correlate with response to pembrolizumab in patients with melanoma and has been tested in a number of other cancer types including head and neck squamous cell carcinoma (HNSCC) and gastric cancer.¹¹³ This T cell-inflamed gene expression profile has been evaluated in solid tumors treated with pembrolizumab in the KEYNOTE 028 trial, validating that this gene expression pattern is associated with improved ORR.¹¹⁴ Another signature, the innate anti-PD-1 resistance (IPRES) signature, involves upregulation of genes involved in such processes as the mesenchymal transition, angiogenesis, and cell adhesion and is enriched in patients resistant to treatment with anti-PD1 checkpoint inhibition.¹¹⁵

3.1.4. Genetic biomarkers

A tumor’s mutational burden (TMB), which is generally assessed by whole exome sequencing, is a well-established predictive biomarker in cancer types such as non-small cell lung cancer.^116–118 Tumors with a high TMB are thought to be more susceptible to ICI because they express more neoantigens and can thus be more easily recognized and targeted by T cells.¹¹⁹ There has recently been an increasing interest in the potential for the TMB to predict response to immunotherapy in melanoma.^120,121 A study that included 321 patients with melanoma who were treated with ICI found that a high TMB could predict improved survival after treatment. Of note, the TMB was not found to be prognostic, as there was no association between TMB and outcome in patients who were not treated with ICI.¹²²

Specific driver mutations have also been found to correlate with response to treatment. Patients with activating NRAS mutations were found to respond more favorably to immunotherapy than patients without NRAS mutations.¹²³ Further, the CheckMate 067 trial reported that patients with BRAF-mutant melanoma had a 62% 4-year OS with the combination of ipilimumab and nivolumab as opposed to 50% with nivolumab alone and 33% with ipilimumab alone.^10,124 Thus, determining mutation status may inform treatment for patients with melanoma.

HLA class I molecules are used by cells to present ‘self,’ and therefore are one means by which tumor cells would present tumor-derived neoantigens to the immune system. However, many cancer cells lose their expression of HLA class I molecules due to selective pressure by the immune system, thus allowing them to evade detection.¹²⁵ Because ICI relies on immune response to the presentation of tumor neoantigens,¹²⁶ downregulation of HLA-I expression or lack of HLA-I heterozygosity (and thus presentation of a lower diversity of antigens) has been proposed as a predictive biomarker for ICIs.^127,128 In an analysis of 1,535 patients with cancer, over 500 of whom had melanoma, HLA class I homozygosity in at least one locus was found to correlate with reduced survival in response to ICI.¹²⁷

3.2. Serum biomarkers

In addition to its value as a prognostic biomarker, serum LDH has been found to be a predictive blood-based biomarker in melanoma patients receiving ICI. As mentioned above, high serum LDH levels have been widely investigated as a poor prognostic factor in many cancers including melanoma.^{13,75,129–131} Recently there has been increasing interest in the potential of serum LDH to act as a predictive biomarker, as well. In a univariable analysis of patients treated with anti-PD-1 monotherapy, elevated LDH was found to be associated with a decreased response to anti-PD1 therapy.¹³² Other blood biomarkers such as absolute monocyte counts, myeloid-derived suppressor cells, and regulatory T cell frequency were also found to be associated with outcome in response to ipilimumab.¹³³ Gene signatures predictive of response to immunotherapy have also been developed using peripheral blood, such as a four-gene signature predictive of survival in patients treated with tremelimumab, an anti-CTLA-4 agent.¹³⁴

4. Discussion

The field of prognostic and predictive biomarkers in melanoma is very promising and it is anticipated that in the near future one or several of the biomarkers under development reviewed above may be integrated into standard clinical care algorithms.

In the prognostic setting, multiple biomarkers have been developed to help guide the clinician in melanoma risk assessment, especially in cases where the tumor thickness is less than 1 millimeter. However, no standardized recommendations exist at this time on how to best utilize such biomarkers together to assess disease risk. We believe that genomic biomarkers in particular hold the potential to change clinical management of early stage melanoma. For example, the DecisionDx®-Melanoma test, which is commercially available for stage I-II melanoma, improves identification of early-stage melanoma patients at high risk for developing metastatic disease. Thus, the use of such tests in addition to conventional staging methods may allow clinicians to adjust screening intervals and interventions based on an individual patient’s disease risk. These biomarkers may especially benefit from additional validation on robust sample sets such as tissues from E1697, the historical cooperative group trial of adjuvant interferon. Doing so may help establish their prognostic ability and allow for their integration into standard AJCC staging. Other biomarkers, such as protein-based biomarkers, will still require further investigation, while genetic biomarkers are firmly established in their ability to separate high-risk from low-risk patients.

In the predictive setting, while PD-L1 staining does generally correlate with benefit, it is not sufficiently accurate for wide scale clinical use due to problems such as tumor heterogeneity, and limited effective options are available beyond PD-1 blockade. However, several other biomarkers such as the tumor mutational burden, the IFNɣ-related gene expression profile, and driver mutations can be implemented using widely available and robust tests. Such biomarkers may be easily incorporated into clinical practice to help predict whether a patient will respond to immune checkpoint inhibition. Other biomarkers mentioned in this review such as LDH, CD8+ TILs, and HLA heterozygosity will need more testing standardization and evaluation, although they hold significant promise as treatment predictors in anti-PD-1 therapy.

Further investigation into biomarkers for melanoma will need to prioritize sensitivity, specificity, accuracy, precision, and cost. Utilizing biomarkers that rely on readily available material such as blood-based biomarkers or H&E-based biomarkers may offer a path forward as a means of systematically estimating patient risk and response in a time-efficient and easily reproducible manner.

Key points.

Biomarkers for melanoma are urgently needed to determine which patients are at highest risk for recurrence and which patients will most benefit from treatment with immune checkpoint inhibitors.

The development of prognostic and predictive biomarkers is an active focus of research in the melanoma field, but these biomarkers have yet to be implemented into routine clinical care.

Biomarkers that use readily available clinical material such as blood-based biomarkers or hematoxylin and eosin-based biomarkers may provide a quick, cost-effective way of developing biomarkers for melanoma.