Tumor-infiltrating lymphocytes as a prognostic and predictive factor for Melanoma

1.1. Background

Globally, melanoma is notable among the spectrum of skin cancers for its increasing incidence and lethal potential, accounting for nearly 73% of global skin cancer-related deaths [1-3]. Outcomes in melanoma vary widely depending on the stage at diagnosis; localized thin melanomas exhibit a 5-year survival rate approaching 98%, while in the context of regional lymph node involvement or distant metastases, the 5-year survival rates decrease to approximately 74% and 35%, respectively [4]. This variability in outcomes underscores the importance of validated prognostic and predictive factors for clinicians to appropriately risk stratify patients along the disease spectrum.

Melanoma diagnosis, staging, and management is guided by standardized data-driven guidelines determined by the American Joint Committee on Cancer (AJCC) and National Comprehensive Cancer Network (NCCN). The current 8th edition AJCC staging system determines melanoma stage using Breslow tumor thickness and ulceration (T category), lymph node metastasis or satellites or in-transit disease (N category), and anatomic site of distant metastatic disease (M category). Additional prognostic factors, though not part of the formal staging classification, have been suggested as adjuncts in the risk stratification of patients with melanoma, with variable degrees of evidence [5,6]. For example, mitotic rate is strongly recommended as an important prognostic factor, as it is associated with increased risk of sentinel lymph node metastasis, and decreased melanoma-specific survival. Other primary tumor factors have less clear consensus regarding their clinical use as prognostic factors but have been suggested as potentially important adjunct pathologic variables, including Clark level of invasion, radial growth-phase regression, lymphovascular invasion, neurotropism, and – the topic of this review – tumor-infiltrating lymphocytes (TILs) [7-9].

1.2. Introduction to TILs in melanoma

Tumor infiltrating lymphocytes (TILs) have been identified as a possible prognostic factor for outcomes in various cancers, including melanoma, since the early stages of cancer research [10]. The term ‘tumor-infiltrating lymphocyte’ was first described by Wallace H. Clark and colleagues in a 1989 paper using clinicopathologic variables to develop a prognostic model for clinical stage I cutaneous melanoma. Clark defined TILs as lymphocytes infiltrating and disrupting tumor cells of the vertical growth phase (the vertical development of a lesion) as assessed on histology [7,11,12]. TILs were categorized into grades based on their density and distribution within the vertical growth phase: TILs were defined as ‘brisk’ when infiltrating the entire tumor or the entire base of the tumor, ‘non-brisk’ when noted in distinct foci of the vertical growth phase, and ‘absent’ when present but not infiltrating the melanoma. Clark et al. found that TILs were one of several important independent prognostic factors incorporated into an 8-year survival prognostic model [11].

Since Clark’s sentinel study in 1989, a growing understanding of the role of immune cells within melanoma over time has led to significant advances in immunotherapy as melanoma treatment [13]. Most notably within the last 15 years, immune checkpoint inhibitors including antibodies targeting CTLA-4, PD-1, and more recently LAG-3, have become first-line treatments for unresectable or metastatic melanoma [5,13-15]. In this setting, there has been renewed interest in tumor infiltrating lymphocytes not only as a prognostic factor for survival outcomes, but also as a predictive factor for treatment response to immune checkpoint blockade (ICB).

1.3. Purpose and scope of review

This article will review the role of TILs in cutaneous melanoma and report on the current literature describing their potential prognostic and predictive value. The content of this review is based on papers covering the topic of tumor-infiltrating lymphocytes in melanoma available in PubMed and Web of Science from their inception to 1 December 2023, supplemented by additional papers found within the citations of relevant papers. This included reviews, systematic reviews, meta-analyses, and original research published between 1989 and 2023, with emphasis placed on works published within the last 5 years (i.e. 2019–2023). The primary purpose of this review is to provide a foundation for a diverse array of clinicians to understand the evaluation and significance of TILs in melanoma. It discusses the existing evidence supporting TILs as a prognostic factor for different outcomes and as a predictive factor for the response to immunotherapy in melanoma. Additionally, this review addresses key considerations for prospective clinical use of TILs in melanoma treatment.

2.1. Identification of TILs via histopathology

As previously described, TILs were first identified by Clark et al. on traditional histopathologic evaluation of melanoma specimens on hematoxylin and eosin (H&E) slides using light microscopy [11]. Histopathologic evaluation of H&E slides from primary melanoma specimens remains the standard method for identification of TILs, with guidelines and recommendations released by various groups including the College of American Pathologists (CAP), American Academy of Dermatology Task Force, National Comprehensive Cancer Network (NCCN), and the International Immuno-Oncology Biomarkers Working Group in an effort to standardize the inclusion of TIL evaluation in melanoma pathology and methods for evaluation of TILs in melanoma [5,16-18]. A 2019 meta-analysis by Fu et al. found that current worldwide literature is roughly divided between two prominent TIL grading systems: (a) 1. absent, 2. non-brisk, and 3. brisk TILs (Figure 1), as recommended by the pathology governance body, CAP [11,16], and (b) 1. absent, mild or scanty, 2. moderate, and 3. marked TILs [19]. While well-standardized, the identification and grading of TILs is dependent on pathologist evaluation, and TILs have been noted to have a worse level of concordance between pathologists in comparison to other secondary histopathologic variables in primary melanoma. A 2012 study assessing concordance between initial pathology reports from referring pathologists and final pathology reports by the Western Australia Melanoma Advisory Service found that TILs had a notably worse level of concordance compared to other secondary histopathologic variables, finding only slight concordance (κ = 0–0.2) for TILs, compared to moderate concordance (κ = 0.41–0.60) for vascular invasion, regression, predominant cell type and histological type, and substantial concordance (κ = 0.61–0.80) for Clark level, mitotic rate, completeness of excision and neurotropism [20]. With the advent of artificial intelligence, automated digital TIL analysis and other deep learning-based digital TIL analyses hold potential as tools to further improve consistent assessment of TILs in primary melanoma specimens [21,22].

2.2. Characterization of TIL subtypes

3.1. Tils in cancer

TILs have been identified in a variety of cancers, including ovarian cancer, colorectal cancer, non-small cell lung cancer, renal cell cancer, bladder cancer, and melanoma [10,37]. In each of these settings, tumor-specific antigens are identified by the host immune system, inciting an anti-tumor response comprising innate and adaptive immunity, analogous to an immune reaction to infection [38]. It follows that TILs have been characterized as a diverse population of immune cells with different roles, including natural killer cells [39,40], macrophages [41], CD4+ (helper) T lymphocytes, CD8+ (cytotoxic) T lymphocytes, FOXP3+ (regulatory) T lymphocytes, and CD20+ (B) lymphocytes [24,42,43], as detailed in Table 2.

3.2. Tils in the melanoma tumor microenvironment

TILs play a complex role in the melanoma tumor microenvironment, contributing to a process known as immune editing. Melanoma is known to generate neoantigens through chromosomal rearrangement or genetic polymorphisms that can mimic ‘foreign’ infection, contributing to its characteristic strong immunogenicity [38]. This strong immune response to melanoma-associated antigens can contribute to the recruitment of a diverse array of TILs to the melanoma tumor microenvironment, where interactions with adjacent tumor cells follow a trajectory of 3 E’s (Elimination, Equilibrium, and Escape) that ultimately contribute to immune editing, a process analogous to natural selection [44]. Early in the disease process, melanoma cells are eliminated by immune cells (Elimination). Subsequently, the melanoma cells continue to mutate and develop ‘immune-resistant’ traits, such as downregulation of tumor antigens, downregulation of MHC expression, and secretion of immunosuppressive cytokines. Melanoma cells have been shown to express immune signaling molecules such as programmed cell death protein 1 (PD-1) and cytotoxic T lymphocyte antigen-4 (CTLA-4) to down-regulate TILs, and Fas ligand to induce T-cell apoptosis [45]. As the ‘immune-resistant’ clones expand, the tumor begins to form but initially remains in check by elimination of ‘immune-sensitive’ clones (Equilibrium). Eventually, the tumor cells exhibit a sufficient degree of immune resistance to proliferate beyond the capacity of immune control (Escape). Further contributing to melanoma immune escape is the phenomenon of T-cell exhaustion, whereby T cells in a chronic environment fail to eliminate tumor cells due to high expression levels of inhibitory receptors such as PD-1, CTLA-4, and lymphocyte activation gene 3 protein (LAG-3) [38,46]. Melanoma immune editing and near-complete immune exclusion from invasive melanoma have been shown to coexist within millimeters of each other within specimens at the tumor-stromal boundary [25], illustrating that the process of melanoma immune editing truly occurs at the cellular level around TILs.

TILs therefore play a complex role in melanoma, facilitating both tumor eradication and immune escape through dynamic interactions with melanoma tumor cells, including via the expression of various immune signaling molecules. Given their important role in the pathogenesis of melanoma tumor development, TILs have emerged as an intriguing prognostic and predictive factor in melanoma.

4.1. Association between TILs and other primary tumor factors

TILs have been studied for their potential association with different primary tumor factors with variable prognostic utility, such as Breslow tumor thickness and presence of tumor regression on pathology. Breslow tumor thickness has been demonstrated to be the most reliable predictor of SLN positivity and is among the most important characteristics independently predictive of survival outcomes for patients with melanomas [5,47]. Multiple studies have demonstrated an inverse relationship between TIL grade/abundance (i.e. brisk TILs compared to absent or non-brisk TILs) and Breslow depth [48-51]. In contrast, studies examining the association between TILs and regression in melanoma demonstrate widely variable results, noting that tumor regression is itself somewhat inconsistently and variably identified as a prognostic factor, being associated with both positive clinical outcomes (e.g. negative SLNB, improved survival) in some studies and negative outcomes (e.g. regional metastasis, unchanged or worse survival) in others [5,48,52-54]. Multiple studies have noted an association between TILs and higher rates of tumor regression [48,55,56]. On a more granular level, cytotoxic T cells have been found to potentially contribute to melanoma regression given their amplification at the site of a regressing lesion [57]. TILs expressing immunophenotypes consistent with T cell exhaustion (e.g. CD4/CD25, FOXP3/CD4, or PD1/CD4) have been negatively associated with tumor regression [58]. In spite of the evidence of an association between TILs and tumor regression, it is difficult to interpret the prognostic significance of this relationship due to contradictory findings as to whether concurrent presence of TILs and regression is associated with better or worse survival outcomes [59-61]. While the associations between TILs and various primary tumor factors such as Breslow tumor thickness and regression vary in degree of clinical evidence, their prognostic significance does not appear to add significant value above the previously established prognostic value of these primary tumor factors.

4.2. Association between TILs and sentinel lymph node status

The association between TILs and sentinel lymph node (SLN) metastasis is of particular interest due to the broad clinical implications of SLN status in melanoma. Presence or absence of SLN metastasis is widely considered one of the most important factors in melanoma staging and prognostication for localized melanoma. SLN status also carries important clinical implications in determining therapeutic and clinical recommendations such as active nodal basin imaging surveillance or completion lymph node dissection and decisions for adjuvant therapy [5,50,62-64]. TILs have been shown to be independently associated with sentinel lymph node status in numerous studies in primary cutaneous melanoma [56,64-66]. Specifically, increasing TIL grades have been associated with decreased SLN positivity rates [67]. Conversely, a paucity of TILs carries an unfavorable prognosis and increased likelihood of regional lymph node metastasis in melanoma patients [68,69].

In the clinical setting, TILs hold potential as a prognostic factor for sentinel lymph node positivity in cases where the determination of risk versus benefit of SLNB is ambiguous; per the NCCN, a SLNB should be considered in cases where risk of positivity is between 5% (below which SLNB is not recommended) and 10% (above which SLNB is recommended) [5,70]. While primary tumor TIL status could thus theoretically refine the risk assessment for SLN positivity, it has not been generally adopted by clinical guidelines nor is it prominent in common web-based SLN risk calculators (the MSKCC calculator and Melanoma Institute of Australia risk calculator [71,72]. Moreover, the association between TILs and SLN status may vary by clinical factors such as sex, though the relationship is unclear. One recent study investigating this question found that TILs are associated with SLN status in women but not men, while another found the opposite results [73,74].

4.3. Association between TILs and survival

A large body of early research established that the presence of TILs is associated with enhanced survival outcomes in melanoma. Clark et al. established TILs as independent predictors of 8-year melanoma overall survival, associating increased TIL grade with better prognosis [11]. These findings were corroborated by Clemente et al. in 1996, who observed TILs to be significant, independent histologic prognostic determinants for 5- and 10-year overall survival [75]. Several subsequent studies have reinforced the correlation between higher TIL grade (i.e. brisk TILs) and improved survival metrics, including melanoma-specific survival, recurrence-free survival, post-recurrence survival, and overall survival [19,50,67,76,77]. In spite of this literature supporting the association between TILs and survival in patients with melanoma, TIL is not a prognostic factor impacting clinical decisions, eg, surveillance strategies or consideration of adjuvant therapies. This may in part be secondary to the variable reporting of TIL by pathologists making clinical standardization challenging, but also by the variable results regarding the prognostic role of TIL in melanoma. One study found that SLN status and TIL grade to both be independent predictors of melanoma-specific and recurrence-free survival, but SLN status to be superior predictor of both outcomes [50]. Further, several studies have found that while TILs were independently predictive of SLN metastasis, TILs were not an independent predictive factor for survival on multivariable analysis [51,64,78], raising the possibility that the prognostic value of TILs is a second-order effect of their prognostic value for sentinel lymph node status. One notable limitation in all of these studies, however, is the relatively simple grading of TILs as brisk, non-brisk, or absent, in spite of our growing recognition that TILs comprise a diverse set of cell types and functions.

Though not as extensively documented, there is a growing body of literature attempting to characterize the immune subtypes of TILs and parse out their unique prognostic significance in melanoma survival. In a series of patients with stage II melanoma, van Houdt et al. found that activated cytotoxic (CD8+) TILs and CD4+ TILs were detected with a significantly higher frequency in patients with more favorable outcomes [79]. A meta-analysis by Fu et al. in 2019 showed that TIL subpopulations marked by CD3+, CD4+, CD8+, FOXP3+, and CD20+ were all associated with increased overall survival in melanoma [80]. In spite of these promising studies, many other studies with TIL immunotyping have not yielded useful prognostic survival data. In one study, CD20+ B cell aggregates in primary melanoma samples detected using immunohistochemistry were not significantly associated with 5-year survival, though they existed at a higher prevalence in thicker tumors [81]. Another study observed that employing specific immunohistochemical markers CD3, CD45, and FOXP3 for TIL quantification did not confer additional prognostic value beyond simple histologic TIL grading with respect to survival, though they did note that melanomas with brisk TILs exhibited a specific immunostimulatory gene expression profile, which correlated with a more favorable prognosis compared to melanomas with non-brisk or absent TILs [82]. A study on survival outcomes in acral melanoma similarly found that high TIL content was linked to better survival outcomes, irrespective of immunophenotype (including FOXP3, CD4, CD8, and PD-1) [66]. Cumulatively, these investigations suggest that the potential for TILs to serve as useful prognostic factor for melanoma-specific survival may be contingent on further research and understanding of the complex roles of different TIL subtypes in the tumor microenvironment.

5.1. Immunotherapy in melanoma

Over the past 10 years, clinical guidelines for the management of metastatic and high-risk melanoma have dramatically changed with the advent of ICB and targeted therapies [5]. Factors such as clinical/pathologic stage and presence of specific markers like the BRAF V600 activating mutation are primarily used to guide clinical recommendations which include pembrolizumab or nivolumab (anti-PD1) monotherapy or combination therapy with ipilimumab (anti-CTLA-4), combination immunotherapy regimens such as nivolumab plus relatlimab (anti-LAG-3), and targeted therapy such as dabrafenib (BRAF inhibitor) plus trametinib (MEK inhibitor). In the setting of a now wide array of immunotherapy regimens options to treat locally advanced as well as metastatic disease, there is now a pressing need for factors that can be used to predict treatment response at the level of individual patients. Given the mechanisms of action of these drugs in the tumor microenvironment, TILs are a natural candidate to act as predictive factors for ICB treatment outcomes.

5.2. Pre-treatment TILs as a predictive factor for immunotherapy response

5.3. Post-treatment TILs as a biomarker for immunotherapy response

Oftentimes, patients with melanoma undergo the procurement of melanoma tumor specimens at various intervals prior to and while on treatment. The serial evaluation of TILs in these tumor specimensfor example, before and after neoadjuvant immunotherapy, opens the potential to use changes in TIL quantity, distribution, and genotypic expression as a potential biomarker for immunotherapy response. Many recent studies and clinical trials evaluating the efficacy of different immunotherapy regimens have found an association between changes in TIL response, TIL T-cell receptor (TCR) expression, and T cell clonality and treatment response [92-97].

In their 2014 paper, Tumeh et al. lay important foundation work for how to utilize post-treatment TIL response as a predictive factor for melanoma treatment response [92]. This investigation employed quantitative immunohistochemistry, multiplex immunofluorescence, and NGS to analyze samples from 46 metastatic melanoma patients before and during pembrolizumab (anti-PD1) therapy. After finding a direct correlation between proliferation of intra-tumoral CD8+ T cells and tumor size reduction after anti-PD1 (pembrolizumab) treatment, the authors formulated and subsequently validated a predictive model for pembrolizumab response based on CD8 expression at the invasive margin and inside melanoma tumors. This innovative study illuminates the importance of a nuanced understanding of the TILs in the tumor immune microenvironment in order to translate post-treatment TIL response into a usable model to predict melanoma treatment responses in other contexts.

While other studies do not go so far as to develop this type of predictive model, several studies exist reporting an association between TIL response and subsequent disease response to different immunotherapies. In the context of BRAF targeted inhibitor treatment, Wilmott et al. found that treatment with dabrafenib or vemurafenib led to a pronounced increase in CD4+ and CD8+ T-cell infiltration, which was in turn associated with a post-treatment reduction in tumor size and increase in tumor necrosis [93]. In a distinct clinical population of high-risk resectable stage III/IV melanoma, Huang et al. found that a single dose of neoadjuvant pembrolizumab significantly increased the presence of brisk TILs in melanomas – a finding which was also associated with complete or major pathologic response in melanoma tumors [94]. Riaz et al. found that TCR sequencing and IHC staining both demonstrated an increase in TILs among ipilimumab-naïve patients who achieved disease control (i.e. complete response, partial response, or stable disease) with nivolumab therapy compared to those who did not, though the same effect was not observed in those who had previously experienced disease progression with ipilimumab [95]. Roh et al. found that increased T cell repertoire clonality was predictive of response to PD-1 blockade, but not CTLA-4 blockade [96]. Daud et al. found that CD8+ T cells expressing high levels of CTLA-4 and PD-1 by flow cytometry, functionally consistent with partially exhausted T-cell phenotypes, strongly correlated with response to anti-PD-1 therapy and progression-free survival in metastatic melanoma [97].

These studies underscore the potential of TIL response to immunotherapy as a predictive biomarker for tumor response to a given immunotherapy, though this may prove challenging to interpret in patients who have failed prior ICB therapy.

5.4. Tils as a predictive factor for immune-related adverse events

While not a primary immunotherapy outcome, immune-related adverse events (irAEs) are important considerations in risk stratification of different treatment options for melanoma patients. The first immune checkpoint inhibitor to be approved for the treatment of advanced melanoma, ipilimumab, was approved by the FDA in 2011 following clinical trials that showed a significant improvement in overall survival while exhibiting a low rate of treatment-related deaths (from 0%–2.4%) [14,84,98]. The most commonly documented irAE among patients receiving ipilimumab is a diffuse maculopapular rash (47%–68% of patients), though it is rarely severe (4% of patients). Even so, irAEs associated with immune checkpoint inhibitors for the treatment of melanoma have a range from mild rashes and colitis to life-threatening toxic epidermal necrolysis and fulminant myocarditis [98,99].

Given the mechanism of action of immune checkpoint inhibitors, TILs may be a considered as a potential predictor of irAEs, which have been associated with longer overall survival in melanoma patients receiving immunotherapy [100]. Recently, Stephens et al. found no significant association between TIL status in primary melanoma biopsy specimens and development or severity of cutaneous irAEs among melanoma patients undergoing ICB therapy [101]. Other studies have used NGS to analyze the T-cell repertoire from peripheral blood mononuclear cell (PBMC) samples among melanoma patients who developed irAEs from ICB therapy, finding an association between increased PMBC clonality and irAEs [102,103]; similar findings have been reported in patients who received ICB therapy prostate cancer [104]. While a stronger association has been established between PMBCs and irAEs, the current literature has not yet demonstrated a consistent association between TILs specifically and irAEs.

6.1. Use of TILs for patient risk stratification and management

While TILs are not currently prominent in any clinical guidelines, they hold potential for application in the clinical setting for melanoma risk stratification and treatment determination. In particular, ambiguous clinical cases with a combination of low and high-risk pathologic or clinical features may benefit from the additional consideration of TILs as a prognostic factor. For example, the NCCN does not recommend SLNB for patients with a <5% risk of SLN positivity, and recommends ‘discussing and considering’ it in patients with a 5–10% risk of SLN positivity [5,70]. In these cases, consideration of TIL status may help clinicians better assess an individual’s risk versus benefit of SLNB [56,64-69,105,106]. Similarly, TIL status may help predict the presence of non-SLN disease in a nodal basin, which may influence the decision for adjuvant therapy and or frequency of nodal surveillance [107]. In one study, Sinnamon et al. developed a risk score for likelihood of non-SLN positivity identified at completion lymph node dissection following a positive SLNB; they found four factors to be independently associated with positive CLND, including presence of TILs at the primary tumor (OR 0.33) [107]. TILs may also help inform decisions for receipt of adjuvant therapy particularly in earlier stage (stage II) melanoma, where overall risk of recurrence is lower. Finally, as further research continues to explore TILs as a potential predictive factor for response to immunotherapy as discussed in section 5 of this review, TILs may 1 day be used as a factor in determining optimal immunotherapy regimens for individual patients.

6.2. Tils as cancer immunotherapy

Though not the primary focus of this review, it should be noted that TILs have not only been explored for their potential as prognostic and predictive factors for melanoma outcomes and treatment, but also as a form of personalized treatment for melanoma and other cancers. Adoptive T-cell therapy encompasses several techniques to develop an infusion of T cells that identify tumor antigens as targets. Among these is a technique in which TILs are isolated from a tumor specimen, amplified, and processed to create a TIL infusion, exploiting a patient’s existing anti-tumor immunity as a form of systemic therapy [108]. Another form of adoptive T-cell therapy is Chimeric Antigen Receptor (CAR) T-cells, which are engineered using T-cell receptors with tumor recognition capacity grafted onto autologous T-cells. CAR-T cells have been studied in hematologic malignancies, but have not yet been applied in solid tumors [109]. While still in its early stages, several groups have reported developments in the application of adoptive t-cell therapy for melanoma, including evidence of durable clinical responses, and strategies for improvement in response such as assessing intensity of lymphodepletion prior to TIL transfer [109-112]. A current commercially developed personalized cellular TIL therapy (Lifileucel, Iaovance Biotherapeutics) is well underway seeking clinical approval [113]. Developed at the intersection of immunology and oncology, these TIL-based therapies are poised to make an important impact on the therapeutic landscape of melanoma.