There were an estimated 325,000 new cases of melanoma diagnosed worldwide in 2020, causing an economic burden and placing enormous pressure on healthcare systems
[1]. Cutaneous melanoma arises when melanocytes, the cells that produce melanin and provide ultraviolet (UV) protection to the skin
[2][3], undergo malignant transformation. The transformation of melanocytes to melanoma has been extensively described
[4]. Melanoma may arise from single melanocytes or clusters of melanocytes known as naevi, with or without
BRAF or
NRAS mutation
[5][6][7][8]. Vast evidence has suggested that exposure to UV radiation results in mutations in DNA, in particular UV-induced C > T transitions that are implicated in melanoma development
[9][10]. Both UVA and UVB radiation can induce DNA damage responses through production of pyrimidine dimers and generation of oxidative responses, although at different rates via different mechanisms
[11]. Various genetic factors are correlated with the progression of melanoma. In general, melanoma can be classified into four subtypes based on the mutation carried in the tumour, i.e.,
BRAF,
NRAS,
NF1 or triple wild type
[4]. Approximately 50% of all cutaneous melanoma cases exhibit
BRAFV600E mutations, while
NRAS mutations are found in up to 20% of the cases
[12][13].
TERT promoter mutations have been detected in intermediate lesions, while
NF1 mutation is observed in melanoma in situ
[14][15]. In the advanced stages of melanoma, mutations in
CDKN2A,
PTEN and
TP53 were observed
[15]. Melanoma has the highest mutation frequency among human cancers
[16][17][18][19]. The increasing mutational burden as the disease progresses is thought to contribute to increased heterogeneity. Heterogeneity in all cancers, including haematological and solid tumours, is now a well-recognised barrier, particularly in looking for the best treatment for cancer patients
[20]. In general, heterogeneity can be categorised into interpatient heterogeneity, intertumoural heterogeneity and intratumoural heterogeneity (
Figure 1A,B). While interpatient heterogeneity describes the differences in tumours between patients, intertumoural heterogeneity refers to differences between lesions in the same patient. It is now known that populations of cancer cells within a given tumour may also be different from each other, described as intratumoural heterogeneity. Melanoma tumour heterogeneity has been reviewed previously
[21][22].
Figure 1.
Melanoma heterogeneity. (
A
) Intertumoural heterogeneity refers to differences between lesions in a patient, while (
B
) intratumoural heterogeneity describes the differences between cells or subpopulations of cells within a tumour. Advancement in technology has revealed (
C
) different cell states in melanoma tumours.
2. Intertumoural Heterogeneity
Advancements in molecular biology and sequencing techniques have made detection of intertumoural heterogeneity easier. Sakaizawa and colleagues investigated mutational diversification between circulating tumour cells in patients where paired primary and metastatic lesions were available. Their data revealed discrepancies in the
BRAF mutation status in 33% (3/9) of the paired samples and
KIT mutations in 75% (3/4)
[23]. Similarly, another study showed that mismatched
BRAF/NRAS mutations were found in 15% (15/99) of the paired primary and metastatic lesions
[24]. This discordant
BRAF mutation status has similarly been identified in multiple studies and poses a significant problem for treatment selection and response
(see Section 5) [25][26][27][28][29][30]. Moreover, differential methylation was detected between three paired primary and metastatic cell lines, where at least 7220 differential methylation fragments were identified
[31]. Subsequent data revealed that
EBF3 promoter methylation causes elevated
EBF3 expression, which leads to metastatic characteristics in melanoma cells. Aberrant expression of methyltransferases in melanoma has also been proposed as a mechanism of malignant progression
[32]. More detailed studies have subsequently been conducted to understand the phylogenetic relationship between the primary tumour and metastases. The work demonstrated that metastases are genetically diverse in comparison to the primary tumour or other metastases from the same patient. Whole-genome sequencing (WGS) analysis indicated that metastases can emerge from either identical or distinct cell populations from the primary tumour
[33][34].
Overall, it is conceivable that the metastases are subpopulations disseminated from a primary tumour or tumour cells that have gone through further adaptations and/or selection to become genetically or phenotypically distinct from the primary tumour or a regressed primary tumour. It is clear that mutations accumulate throughout the development of metastases and metastases within a single patient do not share the same mutations. Cancer cells gain mutations, resulting in additional mutational clusters, hyperploidy and whole-genome doubling, whereas cancer cells also lose mutations, causing a loss in heterozygosity. This intertumoural divergence, or better described as intertumoural evolution, likely enhances survival and promotes metastasis.
3. Intratumoural Heterogeneity
Intratumoural heterogeneity describes the differences between cells or subpopulations of cells within a tumour. Heterogeneous cells within a tumour can display remarkable variability in phenotypic traits. A cross-sectional analysis that screened 1165 tumour exomes from 12 different cancers revealed that intratumoural heterogeneity was found in all cancer types, with 86% of the samples consisting of at least two subpopulations, with melanoma being the most heterogeneous [18]. Furthermore, a study has shown that 6/9 (67%) melanoma tumours exhibited intratumoural heterogeneity [25].
3.1. Genetic Contribution to Intratumoural Heterogeneity
Genetic heterogeneity in melanoma can be explored by examining allelic variations using DNA microsatellite loci analysis
[35][36]. Alternatively, analyses, such as sequencing, high-resolution melting analysis, immunohistochemistry (IHC) and mutation-specific PCR, have also been utilised to reveal the presence of cells with either wild type or mutant copies of BRAF in the same lesion/metastasis
[25][37][38]. Busam and colleagues found heterogeneous BRAF
V600E staining as determined by IHC analysis with a mutation-specific antibody in 27% (6/22) of the genetically confirmed BRAF
V600E-positive cases
[39], suggesting either heterogeneity of the BRAF mutation in the lesions or a lack of protein expression. In a case study, subpopulations carrying
KITWT and its mutant
KITL576P were identified in circulating DNA of a metastatic vaginal mucosal melanoma patient
[40]. While it is common to find multiple mutations within a tumour, mutations that are mutually exclusive in different tumour cell populations within a single tumour have also been identified. Sensi and colleagues discovered that 75% (75/100) of the clones of a short-term melanoma cell line expressed
BRAFV600E/NRASWT, while the remaining expressed
BRAFWT/NRASQ61R [41]. Subsequent investigations revealed that the two subpopulations of cells exhibit different phenotypes. For example, while the
BRAFV600E/NRASWT cells remained viable under an ultra-low attachment condition, the
BRAFWT/NRASQ61R cells were more proliferative and tumourigenic. Further, subpopulations
NRASWT and
NRASG13R were found in the same BRAFi-resistant tumour in a melanoma patient
[42]. In a sophisticated study, a metastatic lesion from a patient that had undergone multiple treatment regimens was divided into several regions in which deep targeted DNA sequencing was performed
[43]. The data showed that only 28% (15/53) of the somatic mutations identified were prevalent in all 41 regions sequenced, while 30% of the mutations were confined to a single region. Further, copy number alteration analysis not only revealed modifications, such as gains of chromosome 6p and 20q and losses of chromosome 6q and 9p, that happened across all 41 regions but also alterations that only occurred in certain regions, such as chromosome 7 gain in four regions, whole-chromosome 10 loss in five regions, 10p loss in one region and chromosome 13 gain in four regions. Taken together, there are large amounts of genetic heterogeneity in melanoma, resulting in subpopulations within a tumour that can possess certain survival advantages and, therefore, possibly affect the treatment outcome.
3.2. Intratumoural Heterogeneity from Non-Genetic Sources
3.2.1. Epigenetic Heterogeneity
Epigenetic mechanisms can contribute to melanoma development via DNA methylation, chromatin remodelling, such as histone modification, and non-coding RNA regulation
[44]. Changes in methylation status between subpopulations of cells in melanoma tumours have been observed across multiple studies. For instance, heterogeneous hypermethylation of at least one of the promoters of various tumour-related genes (
RASSF1A,
CDKN2A,
DAPK,
MGMT and
RB1) was observed in 70% (16/23) of the patient samples
[45]. Of note, expression of RASSF1A was observed at the periphery but not in the core of the tumours, indicating that varied degrees of hypermethylation can occur in different areas of a tumour. Differential methylation was observed in clones derived from a melanoma lesion, which then resulted in heterogeneous expressions of MAGE-A3, a previously identified immunotherapeutic target in melanoma
[46]. This could dampen the therapeutic efficacy in melanoma patients. In addition, H3K27 hypermethylation in the core of solid tumours caused by low glutamine can lead to cell dedifferentiation and reduced sensitivity to BRAF inhibitor treatment
[47]. JARID1B, a member of the jumonji/ARID1 (JARID1) histone 3 K4 demethylase family, was expressed in a slow-cycling population and identified as being essential for tumour maintenance and metastatic progression
[48]. Roesch and colleagues also found that melanoma cells maintain heterogeneous expressions of JARID1B. Pure JARID1B+ or JARID1B− cell populations were generated by cell sorting; however, the populations became heterogeneous for JARID1B after just 72 h in culture, showing that expression of JARID1B is dynamically regulated
[49]. Epigenetic alterations are also often associated with drug resistance. Shaffer and colleagues found more than 10,000 differentially accessible sites from the assay for transposase-accessible chromatin with sequencing (ATAC-seq) by comparing pre-resistant cells to drug-induced resistant cells
[50], indicating that drug treatment promotes epigenetic alterations, which lead to treatment resistance and cancer cell survival.
3.2.2. Phenotypic Heterogeneity
It is possible for a tumour to display dynamic expression of certain markers in different regions in response to microenvironmental cues, disease development, therapeutic treatment or progression without going through genetic alterations. This has been described as phenotypic heterogeneity. For instance, a single human melanoma cell can grow into a heterogeneous tumour in an immunocompromised mouse
[51]. In addition, heterogeneous expression of various markers (such as ABCB5, NGFR, CD54, KIT, etc.) has been identified in tumour samples, although there are no significant differences in the tumourigenicity or metastatic properties of the subpopulations with differing expression
[51][52]. Hoek and colleagues initially proposed a reversible proliferative and invasive model based on their observations
[53], which coincides with the occurrence of heterogeneity in melanoma. Many studies have identified gene signatures associated with reversible proliferative or invasive states in melanoma, where genes that were normally expressed in the proliferative state were often inversely correlated in cells in the invasive state and vice versa
[54][55][56]. Remarkably, cells in a proliferative state have been shown to behave similarly to invasive cells when the two populations were co-cultured together
[57][58]. Subsequent findings revealed that heterogeneous populations of proliferative and invasive cells are more metastatic and invasive compared to single proliferative populations
[57][58], suggesting cooperation between heterogeneous populations in driving metastasis.
Heterogeneous expression of microphthalmia-associated transcription factor (MITF) has been observed extensively in melanoma development and progression. Importantly, MITF
high and MITF
low subpopulations co-exist in different regions of patient tumour samples
[59][60][61][62]. While MITF
high expression is known to promote differentiation and proliferation, MITF
low expression is associated with an invasive and stem-cell-like phenotype
[54][63]. The interaction between MITF
high and MITF
low subpopulations was elegantly shown in a zebrafish melanoma xenograft model. When invasive MITF
low (WM266-4) cells and less invasive MITF
high (501mel) cells were co-injected into zebrafish embryos, the MITF
high cells phenocopied the MITF
low cells without changing the MITF expression and became invasive
[57]. When the mixed populations of MITF
low (75%) and MITF
high (25%) were co-injected into mice, the proliferation rate of the tumour was higher compared to single populations and the MITF
high population increased from 25% to 50%
[64]. Subsequent RNA sequencing results revealed that even the heterogeneous tumours expressed higher MITF as compared to the MITF
low tumours; the tumours were enriched for markers of epithelial mesenchymal transition (EMT) and more invasive, less proliferative phenotypes, showing that there is phenotype adaptation that happens due to cooperativity between MITF
high and MITF
low cells in enhancing metastasis. Notably, a MITF
low zebrafish model has been established to address a subgroup of melanoma patients with low MITF having a poor prognosis. The researchers first incorporated a
BRAFV600 mutation into the MITF
low zebrafish and found that the zebrafish developed nevi and melanoma
[65]. Subsequently, a
TP53 mutation was introduced into the model and the zebrafish first developed black pigmented lesions that then progressed into superficial and nodular melanoma
[66]. As the disease progressed, the expression of MITF shifted from low levels in superficial melanoma to high levels in nodular melanoma. Interestingly, by integrating
BRAFV600 and/or
TP53 mutation(s) in the MITF
low zebrafish model, melanoma onset is triggered and, over time, the disease progresses by alternating MITF expression. Collectively, this evidence has shown that MITF is heterogeneously expressed in tumours, and the role of MITF in promoting melanoma heterogeneity cannot be disregarded.
The co-existence of the MITF
high subpopulation with other subpopulations can result in melanoma progression. BRN2, a transcription factor encoded by the gene
POU3F2, was found heterogeneously expressed in mouse xenografts
[67]. BRN2
high cells were migratory and less pigmented, suggesting they potentially expressed low levels of MITF. Intriguingly, MITF and BRN2 were found mutually exclusively expressed in two distinct subpopulations of cells
[68][69][70]. Mechanistically, BRN2 repression is mediated by miR-211, which itself is under transcriptional control of MITF
[71]. Conversely, BRN2 can promote melanoma invasion via repressing MITF expression by binding to the MITF promoter or through activation of Notch signalling, followed by activation of miR-222/221 promoter
[68][69][72]. Heterogeneous expression of PAX3 in melanoma also contributes to MITF regulation by increasing binding of BRN2 to the MITF promoter
[73]. When cells are BRN2
high/PAX3
low, the MITF expression is mainly regulated by BRN2, while, when BRN2
low/PAX3
high in cells, the MITF expression is driven by PAX3. Additionally, MITF and BRN2-expressing cells act jointly in melanoma progression by first stimulating tumour growth with MITF
high cells and then activating tumour invasion with BRN2
high cells
[70]. Moreover, expression of both MITF and BRN2 proved indispensable for melanoma to efficiently metastasise
[74]. Overall, the complex interactions between MITF and BRN2 are essential in regulating each other, creating a heterogeneous tumour and in driving melanoma progression.
The expression level of MITF may be associated with melanoma cell behaviour in conjunction with other factors. Of note, MITF
high/NF-κB
low cells can transition to MITF
low/NF-κB
high/AXL
high cells during acquisition of resistance, suggesting the importance of low MITF expression in treatment resistance
[75][76]. Heterogeneous expression of AXL is commonly observed in melanomas
[77]. It has been suggested that AXL has a pertinent role in the development of acquired resistance to BRAFi and MEKi. Similarly, a shift from ZEB2
high/SNAI2
high/ZEB1
low/TWIST1
low primary melanoma to ZEB2
low/SNAI2
low/ZEB1
high/TWIST1
high metastatic disease was observed in patient samples
[78]. Follow-up experiments suggested that ZEB2 and SNAI2 act as tumour suppressors but ZEB1 and TWIST1 facilitate BRAF signalling in melanoma transformation by downregulating MITF. Thus, a loss of MITF not only correlated with treatment resistance but also melanoma metastasis.
A subpopulation of melanoma cells has been described as slow-cycling due to a slower proliferation rate than other cells under the same conditions. The population, however, could be much more aggressive or invasive compared to other cells due to expression of ABCB5, a transmembrane glycoprotein from the ATP-binding cassette (ABC) superfamily
[79][80]. In the study, the ABCB5+ subpopulation exhibited self-renewing and differentiating phenotypes in addition to playing significant roles in melanoma growth, tumourigenicity and metastasis
[79][80]. The data also showed that the ABCB5+ subpopulation included slow-cycling cells and was potentially involved in drug resistance mechanisms
[81]. Furthermore, a stem cell marker that is frequently found co-expressed with ABCB5, CD133, was also expressed in tumourigenic, tumour-initiating cells
[82][83]. CD133+ cells were also more proliferative, able to survive ultra-low attachment conditions and have an immunosuppressive phenotype compared to the CD133− subpopulation in a mouse xenograft model
[83][84]. Moreover, CD133+ cells are invasive and metastatic when tested in both in vitro and in vivo (zebrafish and athymic mice) systems
[85]. Evidence has shown that ABCB5- and CD133-expressing cells are co-localised in the perivascular niche where vasculogenic mimicry takes place in melanoma xenografts. This suggests the involvement of ABCB5+ and CD133+ cells in tumour growth and progression by promoting vasculogenic mimicry and morphogenesis of tumour microcirculation
[86]. A NGFR+ subpopulation has also been identified as cancer initiating cells due to their capability of forming tumours and metastasising in mice compared to NGFR− subpopulations
[87][88]. Interestingly, only NGFR+ xenografts formed in nude and NOD/SCID mice, not NSG mice, were able to phenocopy parental tumours. This was contrary to earlier work demonstrating no change in the tumourigenic potential of cells on the basis of ABCB5, CD133 or NGFR expression
[51][52]. However, changes in the expression of these proteins during tumour formation in mice is unknown, and, therefore, it is possible that heterogeneous populations are regenerated as tumours develop. Nonetheless, Su and colleagues observed a trajectory of phenotypic changes in patient-derived cell lines upon BRAFi treatment from day 0 until a stably drug-resistant population is established
[89]. The cells initially expressed MART-1, a melanocytic marker, then transitioned into a slow-cycling (MART-1
neg/NGFR
high) phenotype after 2 weeks, followed by a mesenchymal (MART-1
neg/NGFR
neg) phenotype at day 62. This has showcased the phenotypic changes in a process of developing resistance to BRAFi treatment. Increased expression of SerpinE2, JARID1B or PKH26 was associated with a slow-cycling phenotype
[90]. Further, ALDH+ melanoma cells are more tumourigenic compared to ALDH−cells and differentiation capabilities
[91]. In recent studies, NME1 demonstrated interesting characteristics depending on its expression level in melanoma cells
[92][93]. High expression of NME1 promoted cell proliferation, growth of melanoma spheres and stem-cell-like features, whereas NME1low cells were more aggressive and metastatic while exhibiting neural crest-like features.
Peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC1α), a transcriptional coactivator responsible for mitochondrial biogenesis and respiration
[94], was heterogeneously expressed in melanoma
[95][96]. PGC1α
high populations showed a proliferative phenotype, while PGC1α
low populations displayed a more metastatic phenotype
[96]. More recently, a study showed that melanoma cells with higher metastatic potential have elevated levels of the MCT1 transporter, which takes in lactate. Both MCT1
high and MCT1
low cells were able to form tumours in NSG mice; however, MCT1
high cells formed more metastases compared to MCT1
low cells
[97]. Emerging studies indicate that metabolic changes in melanoma cells are playing a part in melanoma metastasis. However, more studies need to be conducted to obtain enough information in determining how metabolic changes affect melanoma metastasis.
3.2.3. Intratumoral Heterogeneity from Other Perspectives
Tumour Microenvironment
Bidirectional interactions between tumour cells and their microenvironment can also cultivate intratumoural heterogeneity. The tumour microenvironment consists of an extracellular matrix (ECM), stromal cells, fibroblasts and endothelial cells, amongst others, which play vital roles in tumour progression and metastasis. Tirosh and colleagues found that cancer associated fibroblasts (CAFs) expressed high levels of AXL in melanoma tumours that also expressed high AXL but apparent low levels of MITF, both markers of therapeutic resistance. In addition, genes expressed by CAFs were strongly correlated to infiltration of T cells into the tumour
[61], which may impact patient response to immune checkpoint inhibitor therapy.
Immune Heterogeneity
Heterogeneous expression of melanoma differentiation antigens and immune infiltrates (CD4+ and CD8+ IHC staining) was observed in 11 melanoma samples
[98]. Tissue microarray analysis was conducted on 21 samples (2–5 regions from each sample), where 11 samples showed a diverse presence of cytotoxic lymphocytes
[99], indicating that there is no preferential localisation of immune cells within the tumour. Similarly, a diversified distribution of immune cells was observed in a tumour sample excised from a patient who was resistant to anti-PD-1 therapy
[43]. Differential expression in immune profile (such as neoantigens, CD4, CD8 and Treg) was detected in a study of 33 synchronous metastases from 15 patients
[100]. Of note, less than 8% of the total T-cell clones were shared between synchronous tumours, indicating significant immune heterogeneity observed intratumourally. It was found that highly heterogeneous tumours correlated with reduced anti-tumour immune cells, such as CD8+ T cells, follicular helper T cells and pro-inflammatory M1-like macrophages, and more immune suppressing cells, such as alternatively activated M2-macrophages
[101]. Exposing B2905 murine melanoma cells to UVB irradiation, the main etiological cause of melanoma, considerably increased mutational loads from 2745 to 5479 exonic mutations, as well as increased the number of C > T transversions, an increase in the ultraviolet radiation mutation signature and immune heterogeneity
[102]. The irradiated cells were more tumorigenic, but single-cell clones isolated from irradiated cells were less tumourigenic and more immunogenic. This could suggest that a reduction in the tumour mutational burden could reduce tumour aggressiveness and improve the treatment response.
4. Impact of Heterogeneity on Treatment Responses
The overall survival rate for metastatic melanoma patients has significantly improved owing to the development and advancement in targeted therapies and immunotherapies
[103]. However, there is a large proportion of patients who do not benefit from these new treatments due to differential responses resulting from tumour heterogeneity. Resistance to BRAFi treatment is often associated with reactivation of the MAPK pathway through mechanisms such as
RAS mutation,
BRAF amplification and alternative splicing
[104]. For example, a patient developed resistance to BRAFi treatment after subpopulations in the tumour switched from
NRASWT/BRAFV600E to
NRASG13R/BRAFV600E, thereby reactivating the MAPK pathway
[42]. In a rare case of transplacental transmission of metastatic melanoma with
BRAFV600E mutation, the mother was refractory to BRAFi treatment, whereas the baby was responsive to therapy
[105]. Genomic analyses showed that there were at least two related but distinct subpopulations causing resistance, which later led to relapse. Notably, single-cell analysis of a panel of 16 melanoma cell lines and a primary cell line (epidermal melanocyte) found the status of markers, such as MITF, AXL and NGFR, impacted the BRAF/MEKi treatment
[106]. AXL
high cells were more resistant to various treatments in comparison to MITF
high or NGFR
high cells. Resistant AXL
high cells exhibited partial MAPK pathway inhibition with high Ki67 expression, while both MITF
high and NGFR
high cells demonstrated considerable inhibition of the MAPK pathway, suggesting that the MAPK pathway is crucial for AXL
high cells but not MITF
high and NGFR
high cells. Co-treatment of BRAF/MEKi with different epigenetic modulators in the cells revealed selective sensitivities presented by different populations of cells
[106]. Several slow-cycling subpopulations, such as MITF
low/NF-κB
high/AXL
high, are also correlated to reactivation of the MAPK pathway
[75][76]. Importantly, elevated expression of AXL was observed following BRAFi and MEKi (either single or combination) treatments in patient-derived xenografts (PDXs)
[77]. To better understand the mechanisms behind resistance, Shaffer and colleagues compared the expression of resistance markers in pre-resistant cells and BRAFi-treated cells
[50]. Interestingly, the number of resistance markers expressed increased from 72 in the pre-resistant cells to 600 (of the 1456 genes tested) in the post-drug-treatment cells by transcriptome analysis. Only 33 differentially accessible sites were discovered by comparing the non-resistant and pre-resistant cells versus more than 10,000 differentially accessible sites between pre- and post-drug-treatment using ATAC-seq, suggesting minute variations between the non-resistant and pre-resistant cells, and that the treatment induced phenotype switching resistance. Yet, whether the 33 differentially accessible sites between non-resistant and pre-resistant cells are sufficient to provoke a remarkable effect in drug resistance is arguable. In another study, resistance-conferring genetic alterations were evidenced in PDXs established from treatment-naïve patient samples, suggesting the existence of intrinsic resistant populations in melanoma patients
[107]. Nevertheless, acquired resistance was observed in some PDXs too, indicating variabilities in causations of resistance to treatments. Meanwhile, other slow-cycling populations, such as JARID1B
high cells, CD133+, ABCB5+ and NGFR+ cells, have also been implicated in resistance to BRAFi and other anti-cancer drugs
[49][83][85][108][109]. Furthermore, the PIK3CA
E545K mutation was found to confer resistance to combined MEKi + CDK4i in a melanoma patient
[110]. Nonetheless, aberrant expression of methyltransferases in melanomas could affect the sensitivity of melanoma cells to chemotherapy
[111].
In addition, not only did pre-existing NGFR
high melanoma cells show reduced sensitivity to the combination of BRAFi and MEKi but the cells were also resistant to antigen-specific cytotoxic CD8+ T cells
[112]. Consistent with this finding, patients whose tumours are NGFR
high were not responsive to anti-PD-1 and/or a combination of anti-PD-1 and anti-CTLA-4 therapy
[112]. Furthermore, acquired
JAK1 and
JAK2 loss-of-function mutations were found in tumours after patients developed resistance to anti-PD-1 therapy
[113]. The mutations were discovered by comparing the whole-exome sequencing data of tumour samples before treatment and after relapse. The cells with
JAK1 mutations were resistant to interferon α, β and γ-induced growth arrest, while the cells with
JAK2 mutations were insensitive to interferon γ-induced growth arrest. The same study also revealed the mutation of beta-2-microglobulin (B2M) in another patient’s baseline and matched relapse samples, where the loss of B2M reduced the immune-cell recognition of cancer cells. Nonetheless, there are ambiguous responses to immunotherapy due to tumour heterogeneity. For instance, data from a mouse study suggested that tumours with a low level of intratumoural heterogeneity may be more immunogenic and, hence, have an improved overall survival rate
[102]. Lin and colleagues found that tumours with high heterogeneity were associated with decreased immune cell infiltration, immune response activation and overall patient survival
[101]. In contrast, a positive correlation was observed between high mutational load and enhanced T cell infiltration in patient samples
[100], indicating that tumours with a high mutational load may be more immunogenic. Despite the inconsistencies observed, it is possible that the identity of the genes mutated is potentially more pertinent in determining the treatment response regardless of the amount of mutational load
[114].
Altogether, studies have shown that drug treatment could trigger secondary mutations in tumour cells, which lead to resistance. It is also evident that some tumour cells harbour pre-existing resistance, conferring genetic properties that reduce sensitivities to treatments and later lead to treatment resistance. The heterogeneous tumour environment similarly contributes to immunotherapy treatment resistance. The development of immunotherapy has greatly transformed the treatment for advanced melanoma and dramatically increased the overall survival rate for melanoma patients. However, there are still nearly 40% of advanced melanoma patients that do not respond initially to immunotherapies, with adverse side effects observed in 60% of the patients
[115]. The reason why some patients respond well while others do not respond at all is still unclear, although it is clear that heterogeneity at the tumour and immune levels plays an important role.
5. Conclusions and Implications
Heterogeneity within melanoma was initially described by Norris, an attentive general practitioner, in 1820
[116]. Since that time, the researchers have gained significantly more discerning information regarding melanoma development and tumour heterogeneity, particularly at the molecular level. However, many aspects of the development and role of heterogeneity remain unclear. For example, the true identity of each of the subpopulations within the patient tumour remains to be clarified. Further, the specific roles of communication between different subpopulations within the tumour and the various biological components within the tumour microenvironment (ECM and immune cells) are not well characterised. As more subpopulations develop in the tumour, the more challenges patients must face, such as tumour growth, metastasis, efficacy of the therapeutic treatments and, finally, resistance to treatment. These open questions must serve as a reminder that much more needs to be done in order to fill in the missing pieces and truly understand the role of heterogeneity in melanoma metastasis and treatment resistance in order to effectively tackle the disease. Effectively targeting the tumour heterogeneity in melanoma could lead to better therapeutic options and better outcomes for patients.