Abstract
Background/Aim: Over-expression of glucose transporters (GLUTs), membrane proteins that facilitate glucose transport, has been implicated in cutaneous melanomas. Our prior studies have demonstrated increased expression of GLUT1 and GLUT3 in melanomas and their association with poorer prognosis. This study aimed to investigate the expression of GLUT isoforms 4 and 8 in melanocytic lesions, examine the co-expression status of multiple GLUTs, and evaluate their prognostic significance. Materials and Methods: We analyzed 171 melanocytic lesions (97 primary melanomas, 19 metastatic melanomas, and 55 nevi) using a tissue microarray and immunohistochemistry using antibodies against GLUT4 and GLUT8. Membranous expression of GLUTs was scored using a semi-quantitative method. A combined GLUT total score was generated by summing scores from GLUT1, GLUT3, GLUT4, and GLUT8 (including data from previous studies). Results: A significant up-regulation of GLUT4 and GLUT8 expression was found in melanomas compared to nevi (p<0.0001 for both). Concurrent over-expression of multiple GLUTs was more prevalent in melanomas compared to nevi (p<0.0001), and it was also more frequent in metastatic melanomas compared to primary melanomas (p=0.047). Importantly, high total GLUT expression scores were significantly correlated with negative prognostic factors, such as ulceration and mitoses (p=0.03 and p=0.008 respectively). Additionally, Kaplan–Meier survival curves revealed that patients with elevated GLUT total score in their melanomas had a lower disease-specific survival (p=0.006). Furthermore, analysis of multiple GLUTs improved diagnostic sensitivity. Conclusion: Similar to GLUT1 and GLUT3, melanoma exhibits up-regulation of GLUT 4 and GLUT8 compared to nevi. Evaluation of multiple GLUT isoforms improves diagnostic and prognostic values.
Malignant melanoma, a highly aggressive form of skin cancer, is the fifth most common malignancy in both man and woman in the United States. This notorious disease presents significant diagnostic challenges, particularly when distinguishing it from benign nevi, due to the intricate similarities in melanocytic lesions. While meticulous examination of cytological and architectural features is crucial, it may not guarantee a definitive diagnosis, even for experienced dermatopathologists, as evidenced by the lack of consensus among expert dermatopathologists for difficult cases (1, 2). This underscores the importance of additional biomarkers to improve diagnostic accuracy, inform prognosis, and guide treatment strategies for malignant melanoma.
A hallmark of malignant neoplasms is their profound reliance on glucose as a primary energy source, manifested in their elevated glucose uptake and aerobic metabolism, a phenomenon known for nearly a century (3). Glucose transporters (GLUTs) play a pivotal role, catalyzing the bidirectional movement of substrates, primarily glucose, across cell membranes. Currently, there are 14 known types of GLUT isoforms, classified into three distinct classes based on sequence similarities. Class I includes better studied isoforms GLUT1, 2, 3, 4, and 14; Class II encompasses GLUT5, 7, 9, and 11; and Class III consists of GLUT6, 8, 10, 12, and 13. However, knowledge of most Class II and III GLUT proteins remains limited, given their recent identification via human genome sequencing (4). The up-regulation of GLUTs, a critical factor regulating glucose metabolism, is identifiable across a wide range of neoplastic processes (5). The over-expression of GLUTs including GLUTs 4 and 8 is prominently observed in various cancers, including breast, lung, head and neck, thyroid, colon and pancreatic, where it is linked to heightened glycolytic activity, contributing to tumor growth and imparting resistance to chemotherapy (6-9). Recent research indicated that cancer cells rely on GLUT4 and GLUT8 for proliferation, survival, and metastasis (10, 11). Our previous investigations have illustrated an increased expression of GLUT1 and GLUT3 in malignant melanoma compared to benign nevi (12, 13). Additionally, patients with melanoma expressing GLUT1 and GLUT3 exhibited lower disease-specific survival rates. Despite their high specificity, GLUT1 and GLUT3 expression was absent in 32-45% of melanomas evaluated, suggesting these tumors might express other GLUT subtypes. This phenomenon underscores the urgent need to expand our understanding of the expression of other GLUT subtypes and their prognostic value in melanocytic lesions. The primary objective of the current study was to evaluate the expression of additional glucose isoforms GLUT4 and GLUT8 in benign and malignant melanocytic lesions, and assess the co-expression status of multiple GLUTs, and their prognostic values.
Materials and Methods
Tissue microarrays (TMAs). Tissues were obtained via an exempted Institutional Review Board protocol (CPHS#21764) with deidentified patient demographic information, pathology, and survival outcomes. TMAs were prepared as described previously (13). A previously constructed TMA comprised 116 melanoma cases (including 97 primary melanomas and 19 metastatic melanomas) and 55 benign nevi. Immunohistochemical results for GLUT4 and GLUT8 were available for 97 and 93 primary melanoma cases, respectively. The partial availability of data is due to the depletion of some tissue cores over time, a common issue with TMAs, especially in melanomas with small sizes. Of the 97 primary melanoma patients tested for GLUT4, 31 developed metastatic disease, 49 had not metastasized at the time of follow-up, and the metastatic status was unknown for 17 patients. Similarly, of the 93 patients tested for GLUT8, 31 developed metastatic disease, 46 remained non-metastatic at follow-up, and 16 had no follow up information. Demographic and clinical characteristics of this cohort are detailed in Table I. Melanoma metastases originated from various anatomic sites, including the skin, lymph nodes, and solid organs. The mean follow-up duration was 9.8 years, with a median of 10 years. Histopathologic and prognostic factors were assessed on sections from initial tumor biopsies or excisional specimens. For smaller tumors, a single core (0.6 mm in diameter) was used for evaluation. In 61.2% of the cases within the TMA, 2 or 3 cores representative of the tumor were included, depending on the lesion size.
Patient demographics of melanocytic lesions in tissue microarrays.
Immunohistochemistry. GLUT4 (mouse monoclonal, Abcam 48547, dilution 1:2,000; Cambridge, MA, USA, positive control: kidney tissue) and GLUT8 (rabbit polyclonal, Abcam191269, dilution 1:100; positive control: testis tissue) immunostaining was performed on 4-μm TMA sections using automatic immunostainer (Biogenex I-6000; Biogenex Laboratories, Ind., Fremont, CA, USA), according to the manufacturer’s instructions. Protein expression was visualized using Leica Bond Polymer Refine Red detection kit with Fast Red Chromogen (Leica Biosystems Inc., Buffalo Grove, IL, USA). Red chromogen was chosen to aid in the visual distinction of GLUT proteins from any potential brown melanin pigment in the sampled tissue. Normal dermis served as a negative internal control for both antibodies. Omission of the primary antibodies was also performed as a negative control.
Evaluation of immunohistochemistry. Each tissue core was analyzed at 400× magnification. Positive membrane staining within dermal melanocytes was considered positive (Figure 1). Positive staining of the junctional melanocytes or lymphocytes was excluded. Scoring included percentage of positive dermal melanocytes (0 points <1%; 1 point=1-10%; 2 points=11-50%; 3 points=51-80%; and 4 points >80%) and intensity of membrane staining (1 point=weak; 2 points=moderate; 3 points=strong). Less than 1% positive cells were considered negative regardless of staining intensity. If a case had more than one tissue core and the cores showed heterogeneity, we chose to use an average of percentage scores and an average of intensity scores. For each case, both sets of points were added, and specimens were assigned to four groups according to their overall score (0-1 points=negative, overall score=0; 2-3 points=weak, overall score=1; 4-5 points=moderate, overall score=2; 6-7 points=strong, overall score=3) (12). A combined score (0 to 12) of four GLUTs (GLUT1, 3, 4, and 8) was generated based on the sum of the individual scores of each GLUT, defined as GLUT total score.
Representative immunohistochemical staining for glucose transporter isoform 4 (GLUT4) and isoform 8 (GLUT8) in melanomas and benign nevi. A) Positive membranous expression of GLUT4 in melanoma. B) Negative expression of GLUT4 in melanoma. C) Negative GLUT4 expression in benign nevus. D) Positive membranous expression of GLUT8 in melanoma. E) Negative expression of GLUT8 in melanoma. F) Negative GLUT8 expression in benign nevus. Magnification for all images: 400×.
Statistical analysis. Statistical Analysis System (SAS 9.2) (SAS Institute Inc., 2008, Cary, NC, USA) software was used for data analyses. All p-values were two-sided, and p≤0.05 was considered statistically significant. A proportional odds model was used to evaluate association between GLUT4 and GLUT8 expression and subgroups of melanocytic lesions. A multivariate Cox model was used to study the hazard of dying from melanoma in patients with primary melanomas and potential risk factors. The Kaplan–Meier method was used to estimate disease-specific survival. The log-rank test was used to compare differences in disease-specific survival between subgroups. Scheffe’s method was used to adjust the p-value in pairwise comparisons if there were more than two subgroups. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of GLUT4 and GLUT8 for diagnosing melanoma were calculated using traditional methods. Using the receiver operating characteristic (ROC) curve and Youden index analysis, we identified the optimal cut-off values for GLUTs. Death not caused by melanoma was treated as censored data. Univariate and multivariate disease-specific survival analyses using Cox’s regression models were conducted to assess the correlation of prognostic factors with survival. Figures were generated using GraphPrism software (version 8.21, La Jolla, CA, USA).
Results
Increased expression of GLUT4 and GLUT8 in cutaneous melanomas compared to nevi. Immunohistochemical assays were employed to evaluate membranous GLUT4 and GLUT8 expression in melanoma and benign nevus specimens on the TMA (Figure 1). The clinical and histological features of this cohort are listed in Table I. For malignant melanomas, 116 (97 primary and 19 metastatic) GLUT4-stained and 112 (93 primary and 19 metastatic) GLUT8-stained sections were available for evaluation. For benign nevi, 55 were available for both GLUT4 and GLUT8 assessment. The majority of the benign nevi were negative for GLUT4 (94.5%) expression and more than half (56.4%) showed negative GLUT8 immunostaining. Moderate to strong GLUT4 and GLUT8 immunostaining was only seen in melanomas except for one benign nevus case showing moderate GLUT4 expression (Table II). Melanomas displayed a significantly higher frequency of GLUT4 and GLUT8 expression compared to benign nevi (p<0.001 and p<0.0001, respectively). Specifically, for GLUT4, 79 (68.1%) melanoma cases exhibited negative expression, while 28 (24.1%) showed weak, 8 (6.9%) moderate, and 1 (0.9%) exhibited strong membranous expression. In contrast, 52 (94.5%) benign nevi were negative for GLUT4, with only 2 (3.6%) and 1 (1.8%) cases demonstrating weak and moderate expression, respectively. For GLUT8, only 19 (16.9%) melanomas exhibited negative expression, with 38 (33.9%) showing weak, 51 (45.5%) moderate, and 4 (3.6%) strong expression, while most benign nevi displayed negative (31 cases, 56.4%) or weak (24 cases, 43.6%) GLUT8 expression, with no instances of moderate or strong expression observed. However, no statistically significant difference in GLUT4 or GLUT8 expression was observed between primary and metastatic melanomas, not between primary melanomas that developed metastasis during follow up and those that did not.
Glucose transporter-4 (GLUT4) and Glucose transporter-8 (GLUT8) expression in melanocytic lesions.
Expression status of multiple glucose transporters (GLUTs, 1, 3, 4, 8) in melanocytic lesions. Co-expression of multiple glucose transporters (GLUTs 1, 3, 4, and 8) was evaluated in melanomas and nevi (Table III). Data for all four GLUT isoforms were available for 110 melanomas (92 primary and 18 metastatic) and 54 nevi. The majority of melanomas displayed co-expression of multiple GLUTs. Negative expression of all four GLUTs was only observed in a minority of melanomas (7 cases, 6.3%) but was more common in benign nevi (23 cases, 42.6%). Only 11 (10.0%) melanomas exhibited expression of just one GLUT, while 40 (36.3%), 36 (32.7%), and 16 (14.5%) melanoma cases showed positive membranous staining of 2, 3, or 4 GLUTs, respectively. Nevi displayed a lower frequency of co-expression, with 22 (40.7%) and 8 (14.8%) cases showing positive staining of one or two GLUTs, respectively, most with weak intensity. Notably, the number of co-expressed GLUTs was significantly higher in melanomas compared to nevi (p<0.0001). Interestingly, metastatic melanomas expressed more GLUTs compared to primary melanoma (p=0.047), while the numbers of GLUTs in primary melanomas did not differ between those that developed metastasis during follow-up and those that did not (p=0.19). As expected, the total combined score of all four GLUTs (Table IV) was significantly higher in melanomas compared to nevi (p<0.0001). Furthermore, no significant difference in GLUT total score was observed between primary and metastatic melanomas (p=0.11), or between primary melanomas that developed metastasis at follow up and those that did not (p=0.55).
Status of co-expression of multiple glucose transporters (GLUTs, 1, 3, 4, 8) in melanocytic lesions.
Glucose transporter (GLUTs, 1, 3, 4, 8) total expression scores in melanocytic lesions.
Clinicopathological correlation. When evaluating the correlation between GLUT expression scores and common clinicopathologic parameters of primary melanoma, we first assessed GLUT4 and GLUT8. These analyses revealed no significant association between the expression levels of either GLUT4 or GLUT8 and a range of clinicopathologic parameters, including tumor thickness, mitotic rate, ulceration, tumor stage, tumor-infiltrating lymphocytes, regression, tumor pigmentation, presence of background nevus, or metastatic status at follow-up (all p>0.05). Additionally, no significant correlation was observed between GLUT4 and GLUT8 expression levels themselves. Interestingly, a composite analysis of the GLUT total score (sum of expression scores of GLUT1, 3, 4, and 8) revealed a significant correlation with both dermal mitotic rate (p=0.008) and ulceration (p=0.03) (Figure 2). However, the GLUT total score did not show a significant correlation with other clinicopathologic parameters.
Correlation between GLUT total score and pathological features of melanoma. A) Association with tumor ulceration. B) Correlation with mitotic rate. C) Relationship with Breslow depth. p<0.05 indicates statistical significance.
Survival analysis. The Kaplan–Meier method revealed no significant difference in disease-specific survival between patients with negative (overall score=0) and those with positive GLUT4 and 8 expression (overall score >0) in their primary melanomas (data not shown). To assess the prognostic value of the GLUT total scores, we used the receiver operating characteristic (ROC) curve and Youden index analysis. A score of 6 was identified as the optimal cut-off. Patients with high GLUT total scores (≥6) in their primary melanomas had significantly lower disease-specific survival compared to those with low scores (<6) (p=0.006, Figure 3A and B). Interestingly, the prognostic value of the score differed based on disease stage. In early-stage melanoma, patients with high GLUT total scores (≥4) had worse survival compared to those with low scores (<4) (p=0.04, Figure 3C). Similarly, in late-stage melanoma, high GLUT total scores (≥7) were associated with poorer survival compared to low scores (<7) (p=0.04, Figure 3D).
Kaplan–Meier survival curves depicting disease-specific survival stratified by GLUT total scores in patients with primary melanomas. A) Significantly lower survival rates for patients with high GLUT scores (≥6) compared to those with low scores (<6). B) Receiver operating characteristic (ROC) curve for the GLUT total score. C) In early-stage melanoma (T stages I and II), lower survival rates associated with high GLUT scores (≥4) compared to low scores (<4). D: In late-stage melanoma (T stages III and IV), lower survival rates correlated with high GLUT scores (≥7) compared to low scores (<7). p<0.05 indicates statistical significance.
A multivariate Cox regression analysis (Table V) further evaluated the effect of these biomarkers on survival probability. Shorter survival was significantly associated with high tumor mitotic count and the presence of ulcer (p<0.001 and p=0.005, respectively), indicating their roles as independent predictors of survival. Tumor thickness showed a trend towards association with survival (p=0.05), suggesting its potential role as a prognostic factor. GLUT expression levels were not predictive of survival (p>0.05 in all GLUTs), indicating that they are not independent prognostic markers.
Multivariable survival analysis of primary melanomas.
Diagnostic value of GLUTs. To evaluate GLUT4 and GLUT8 as diagnostic markers for melanoma, we analyzed sensitivity, specificity, PPV, and NPV using various cut-off scores (Table VI). GLUT4 demonstrated high specificity (94.5% and 98.2%) but low sensitivity (31.9% and 7.8%) for cut-off scores ≥1 and ≥2, respectively. Conversely, GLUT8 showed higher sensitivity (83.0% and 49.1%) but lower specificity (56.4% and 100%) for cut-off scores ≥1 and ≥2, respectively.
Diagnostic value of GLUT4 and GLUT8 in distinguishing melanomas from nevi.
Given that melanomas express multiple GLUT isoforms, we also investigated the diagnostic performance of GLUTs (1, 3, 4, 8) total score in differentiating melanoma from nevi (Table VII). At a cut-off of ≥1, the GLUT total score offered high sensitivity (93.6%) but lower specificity (42.6%). Increasing the cut-off to ≥2 improved specificity (83.3%) at the expense of some sensitivity (89.1%). A cut-off of ≥3 yielded the higher specificity (98.1%) but with lower sensitivity (71.8%). We obtained similar results when analyzing the number of co-expressed GLUTs. A cut-off of ≥1 offered high sensitivity (93.7%) but low specificity (42.6%). Increasing the cut-off resulted in a trade-off between sensitivity and specificity (Table VIII).
Discussion
Increased glucose uptake is a characteristic of cancer, distinguishing it from benign cellular activity (5). Tumor cells undergo a metabolic shift to glycolysis for energy production, converting glucose to lactate in oxygen-rich conditions, necessitating increased glucose uptake through deregulated glucose transporters (14). The GLUT family, also known as the SLC2A gene family, comprises 14 transporters, each featuring 12 transmembrane domains arranged in a pore structure for hexose transport, alongside intracellular N- and C-termini, and two loops, one intracellular and one extracellular (15). Various cancers exhibit elevated GLUT expression, with altered GLUT isoforms observed in lung, breast, thyroid, head and neck, colorectal, ovarian carcinoma, and glioblastoma (16-20). Unlike their ubiquitous expression in other cell types, GLUTs are not consistently detected in melanocytes. Our previous work found that immunohistochemical expression of GLUT1 and GLUT3 is increased in melanomas compared to nevi, correlating with significantly lower survival rates (12, 13). This study furthers our understanding by exploring more glucose transporters’ expression patterns and prognostic value in cutaneous melanoma. We found higher levels of GLUT4 and GLUT8 in melanomas compared to nevi, suggesting their specific roles in melanoma progression, distinct from the more extensively studied GLUT1 and GLUT3. Research indicates that GLUT4 expression is up-regulated in gastric cancer and metastatic pancreatic cancers, proposing GLUT4 as a promising biomarker in solid tumors (21, 22). GLUT4 over-expression can also promote metastasis in head and neck carcinomas (9). Conversely, down-regulation of GLUT4 inhibits melanoma cell proliferation and metastasis (23). Similar up-regulation patterns of GLUT4 and GLUT8 have been observed in multiple myeloma cell lines (11). In endometrial cancer, increased GLUT8 expression is noted across all tumor subtypes compared to atrophic endometrium (24). Furthermore, GLUT8 is crucial for serine biosynthesis in certain KRAS/NRF2 mutant lung cancer cell lines (25).
Cancer cells have high levels of various GLUTs, increasing their glucose uptake and glycolytic activity to meet the energy requirement of rapid growth. Studies have shown that blocking GLUTs can suppress cancer (26, 27); however, knockout of a single GLUT gene did not seem to reduce tumor size, indicating multiple GLUTs involvement in cancers (28). Our study is focused on the expression of individual GLUTs (GLUT 4 and GLUT 8), and the expression patterns of multiple GLUTs (GLUT1, 3, 4, 8) in melanomas. The expression patterns of these four GLUTs range from quadruple negative to co-expression of multiple isoforms. Given the large number of GLUT isoforms and their different degree of involvement in various types of cancers, this process is very complex. Hence, understanding the distinct expression patterns of multiple GLUTs in melanoma might be useful for developing a potentially attractive anticancer strategy by targeting glucose uptake via inhibiting multiple GLUTs.
Histologic diagnosis of difficult melanocytic lesions can be challenging. Identification of useful immunohistochemical biomarkers is therefore valuable in assisting diagnosis. We have previously shown that high GLUT1 and GLUT3 expression was highly specific for distinguishing melanoma from nevi although the sensitivity was low due to heterogenous expression of different GLUT isoforms in melanoma (12, 13). When evaluating both GLUT1 and GLUT3 compared to single GLUT, the sensitivity increased while specificity remained high. In our current study, GLUT4 demonstrated high specificity but low sensitivity in distinguishing melanoma from nevi. In contrast, GLUT8 showed higher sensitivity but lower specificity. When evaluating co-expression status of GLUT1, 3, 4 and 8, the sensitivity of diagnosing melanoma was markedly increased compared to using any single GLUT. The co-expressed status of multiple GLUTs improved the diagnostic performance, with higher cut-off scores yielding higher specificity, but at the cost of decreased sensitivity. Similarly, the overall GLUT expression scores offer higher sensitivity than any single isoform or combination of isoforms, resulting in the most optimal diagnostic performance. These findings indicate that examining expression patterns of multiple GLUTs, may yield valuable diagnostic insights for differentiating melanoma from nevi in clinical practice.
In addition to efforts to identify GLUTs as biomarkers to aid in the diagnosis of melanoma, we also explored their prognostic value. Consequently, we observed that an elevated GLUT total score in primary melanomas strongly correlated with adverse prognostic factors, such as tumor ulceration and mitotic rate, and was significantly associated with decreased disease-specific survival. These findings suggest that measuring expression scores of multiple GLUTs could be helpful for predicting outcomes in melanoma patients. While factors like ulceration, mitotic rate, and tumor depth independently predicted prognosis, the total GLUT expression score was not an independent prognostic factor in multivariable analysis. This is likely due to its strong correlation with these adverse prognostic factors.
Our study extends the growing body of literature on altered glucose metabolism in cancers by providing a more detailed understanding of several GLUTs involved in melanoma. The comprehensive analysis of multiple GLUTs offers a broader perspective on glucose transport within this type of cancer. Nevertheless, our study is limited by its small sample size, the disadvantages of TMA techniques, and the lack of knowledge regarding the expression of other GLUT isoforms and their correlation with other molecular factors. Further validation via whole tissue section, using larger and more diverse cohorts, is necessary. Moreover, the study’s retrospective nature might introduce biases, which could be mitigated in future prospective studies.
Identifying specific GLUTs as prognostic markers in melanoma could have significant impact. These findings could guide the development of targeted therapies and contribute to treatments tailored to individual patients. Understanding the role of GLUTs in melanoma progression may assist in patient stratification for therapy, influencing treatment decisions based on GLUTs expression profiles. Future research could focus on dissecting the molecular mechanisms underlying GLUTs’ roles in melanoma. Longitudinal studies tracking GLUT expression from early nevi transformation to advanced melanoma stages could yield valuable insights into their temporal roles in progression. Experimental studies investigating the impact of GLUT inhibition on melanoma growth and metastasis are crucial for exploring their therapeutic potential. Expanding this research to encompass diverse patient populations is vital for understanding the variability in GLUT expression and its clinical implications.
In conclusion, our study sheds light on the role of multiple GLUTs in melanoma, adding valuable insights to this growing field of research. These findings are valuable for future research on developing targeted therapies and using GLUTs to predict patient outcomes and response to treatment.
Acknowledgements
The Authors wish to acknowledge the support of the Pathology Shared Resource in the section for Clinical Genomics and Advanced Technology of the Department of Pathology and Laboratory Medicine at the Dartmouth Hitchcock Health System and the Dartmouth Cancer Center with NCI Cancer Center Support Grant 5P30 CA023108-37 RRID#SCR_023479.
Footnotes
Authors’ Contributions
Conceptualization, S.Y.; Original draft preparation, writing and review of the manuscript, W.M and S.Y.; Resources and data curation, W.M. Z.L., and S.Y. review and editing, J.W..; final review and supervision, S.Y.; All Authors have read and agreed to the published version of the manuscript.
Conflicts of Interest
The Authors declare no conflicts of interest in relation to this study.
Funding
This research received no external funding.
- Received June 24, 2024.
- Revision received July 14, 2024.
- Accepted July 15, 2024.
- Copyright © 2024 The Author(s). Published by the International Institute of Anticancer Research.
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).
