Purpose: Inflammatory marker expression in stage III melanoma tumors was evaluated for association with outcome, using two independent cohorts of stage III melanoma patients' tumor tissues.

Experimental Design: Fifteen markers of interest were selected for analysis, and their expression in melanoma tissues was determined by immunohistochemistry. Proteins associating with either overall survival (OS) or recurrence-free survival (RFS) in the retrospective discovery tissue microarray (TMA; n = 158) were subsequently evaluated in an independent validation TMA (n = 114). Cox proportional hazards regression models were used to assess the association between survival parameters and covariates, the Kaplan–Meier method to estimate the distribution of survival, and the log-rank test to compare distributions.

Results: Expression of CD74 on melanoma cells was unique, and in the discovery TMA, it associated with favorable patient outcome (OS: HR, 0.53; P = 0.01 and RFS: HR, 0.56; P = 0.01). The validation data set confirmed the CD74 prognostic significance and revealed that the absence of macrophage migration inhibitory factor (MIF) and inducible nitric oxide synthase (iNOS) was also associated with poor survival parameters. Consistent with the protein observation, tumor CD74 mRNA expression also correlated positively (P = 0.003) with OS in the melanoma TCGA data set.

Conclusions: Our data validate CD74 as a useful prognostic tumor cell protein marker associated with favorable RFS and OS in stage III melanoma. Low or negative expression of MIF in both TMAs and of iNOS in the validation set also provided useful prognostic data. A disease-specific investigation of CD74's functional significance is warranted, and other markers appear intriguing to pursue. Clin Cancer Res; 22(12); 3016–24. ©2016 AACR.

Translational Relevance

Although gene mutations are now providing predictive information for targeted therapy, molecular markers to inform prognosis remain elusive. It is also known that chronic inflammation associates with melanoma patient disease progression, and resolution of inflammation in experimental models leads to development of adaptive immunity as well as improved outcomes to chemotherapy. Therefore, we tested 15 selective molecular inflammation-related markers in melanoma stage III patient tumor cells for association with outcomes and identified three markers of importance. Protein expression of two (inducible nitric oxide synthase and MIF) had predicted statistical association with poor outcome; surprisingly, CD74 associated strongly with a favorable patient outcome; this is intriguing as CD74 has been hypothesized to be linked to development of specific adaptive immune response as it is known as the MHC class II invariant chain. The presence and positive role of CD74 is novel, and we believe that our findings will lead to approaches for improved personalized treatment.

At this time, the validated prognostic factors in patients with stage III melanoma are clinical and pathologic features: (1) specifically, the pattern of regional metastases (lymph node and/or in-transit), the size and number of regional metastases, and the ulceration status of the primary tumor (2). As outcomes remain widely variable for stage III melanoma patients, it is critical to identify clinically useful molecular markers that may improve our risk models, thereby increasing our ability to provide better clinical care.

Previous studies suggested that intrinsic tumor cell inflammatory mediators are associated with poor prognosis, yet no molecular tumor markers have been proven to be prognostic. To date, one tumor-infiltrating lymphocyte (TIL) molecular marker (CD2; T and natural killer cells identified) was identified as associating with favorable outcomes, heralding an era of molecular marker usage (3–5) in melanoma. In our own studies, we have previously observed impaired outcomes in patients with constitutive expression of inducible nitric oxide synthase (iNOS) and associated production of nitric oxide (NO) in melanoma cells (6, 7). From the work of us and many others, a cohesive model is emerging in which poor clinical outcome is associated with varying concentrations of cellular NO in many cancers (8–14), including stage III melanoma (12, 15–16). Currently, IHC evaluation of iNOS is routinely performed at the Clinical Laboratory Improvement Amendments–approved laboratory in our institution (MD Anderson Cancer Center) to inform clinical trials; however, determining the power of this marker compared with that of other markers is a priority at this time. Therefore in this study, we applied the REporting recommendations for tumor MARKer prognostic studies (REMARK) criteria (17) to test the prognostic value of a set of inflammatory markers, using separate discovery and validation tissue microarrays (TMA) associated with known long-term survival data. Our analyses have determined that iNOS and MIF expression in patient melanoma cells had varying statistical association with poor outcome; however, a strong favorable patient outcome was indicated by the expression of CD74, which we now report as a new prognostic marker for patients with stage III melanoma.

Patient samples and data sets

Discovery and validation data sets. Two independent TMAs of stage III melanoma (regional metastases) were employed. Under the Institutional Review Board–approved protocols, patients were identified whom archival material was available and those had clinical follow-up of at least 5 years for surviving patients. Patient data and primary tumor characteristics were collected for all patients. Patient records were reviewed for clinical events, including development of local recurrence, development and pattern(s) of distant metastasis, and overall survival. Patient demographics and survival status are detailed in Table 1. In accordance with REMARK criteria, the TMA from the John Wayne Cancer Institute (JWCI) served as the discovery set in which candidate biomarkers were identified. This TMA was prepared from formalin-fixed paraffin-embedded lymph node tumors and contained two core samples from each specimen, tumors from 158 patients with defined clinical outcomes, and follow-up data (15).The associations between markers and survival outcomes were then evaluated in an independent and more contemporary TMA of sequential stage III patients' tumors (n = 114) from MD Anderson Cancer Center (MDACC). The MDACC TMA served as the validation (17) and was prepared as triplicate core samples from patients who had undergone standard-of-care lymphadenectomy for stage IIIB/C melanoma and for whom clinical data were available. While MDACC patients often later received several adjuvant therapies, including ipilimumab, BRAF inhibitors, vaccine, and biochemotherapy, JWCI patients predominantly received additional surgeries or an experimental vaccine.

Table 1.

Patient demographics and survival status

Discovery set (JWCI) measureAll patients (N = 158)Validation set (MDACC) measureAll patients (N = 114)
Gender, n (%)  Gender, n (%)  
 Male 90 (57.0)  Male 79 (69.3) 
 Female 68 (43.0)  Female 35 (30.7) 
Age at surgery (years), median (range) 51.9 (11.5–94.7) Age at surgery (years), median (range) 55.4 (25.0–85.4) 
Positive lymph node, n (%)  Positive lymph nodes, n (%)  
 1 53 (33.5)  0–1 23 (20.2) 
 2–3 34 (21.5)  2–3 34 (29.8) 
 ≥4 71 (45.0)  ≥4 57 (50.0) 
Ulceration, n (%)  Ulceration, n (%)  
 Yes 31 (29.5)  Yes 64 (79.0) 
 No 74 (70.5)  No 17 (21.0) 
 Missing 53  Missing 33 
Breslow thickness (mm)  Breslow thickness (mm)  
N 115 N 95 
 Median (range) 2.0 (0.35–15.00)  Median (range) 3.0 (0.25–33.00) 
OS, n (%)  OS, n (%)  
 Alive 52 (32.9)  Alive 30 (26.3) 
 Dead 106 (67.1)  Dead 84 (73.7) 
Recurrence-free survival, n (%)  Recurrence-free survival, n (%)  
 Alive with no disease 22 (13.9)  Alive with no disease 12 (10.5) 
 Dead and/or disease progressed 136 (86.1)  Dead and/or disease progressed 102 (89.5) 
Discovery set (JWCI) measureAll patients (N = 158)Validation set (MDACC) measureAll patients (N = 114)
Gender, n (%)  Gender, n (%)  
 Male 90 (57.0)  Male 79 (69.3) 
 Female 68 (43.0)  Female 35 (30.7) 
Age at surgery (years), median (range) 51.9 (11.5–94.7) Age at surgery (years), median (range) 55.4 (25.0–85.4) 
Positive lymph node, n (%)  Positive lymph nodes, n (%)  
 1 53 (33.5)  0–1 23 (20.2) 
 2–3 34 (21.5)  2–3 34 (29.8) 
 ≥4 71 (45.0)  ≥4 57 (50.0) 
Ulceration, n (%)  Ulceration, n (%)  
 Yes 31 (29.5)  Yes 64 (79.0) 
 No 74 (70.5)  No 17 (21.0) 
 Missing 53  Missing 33 
Breslow thickness (mm)  Breslow thickness (mm)  
N 115 N 95 
 Median (range) 2.0 (0.35–15.00)  Median (range) 3.0 (0.25–33.00) 
OS, n (%)  OS, n (%)  
 Alive 52 (32.9)  Alive 30 (26.3) 
 Dead 106 (67.1)  Dead 84 (73.7) 
Recurrence-free survival, n (%)  Recurrence-free survival, n (%)  
 Alive with no disease 22 (13.9)  Alive with no disease 12 (10.5) 
 Dead and/or disease progressed 136 (86.1)  Dead and/or disease progressed 102 (89.5) 

Biomarker assessments

A set of 15 inflammatory protein markers were selected in part based on their involvement with cancer inflammation by us or others (13, 18–19). The final list was also dependent on availability of antibodies optimized for use in IHC analysis in this study, having been previously tested and optimized for specificity and validated with use of tumor tissue and controls (Supplementary Material S1). Therefore, this is not an inclusive nor in any means a complete list of such markers, and we continue to test and validate additional antibodies for use in IHC analysis to expand similar testing.

The protein markers assessed for this report were arginase, IL1α, IL1β, IL24, COX-2, CD70, CD74, iNOS, macrophage migration inhibitory factor (MIF), CXCL9, CXCL10, CXCR3, and Ki-67. NO-driven posttranslational modifications were also tested since antibodies to these have recently become available, and are of much interest to our work on iNOS; these were nitrotyrosine (NT) and S-nitroso-cysteine and therefore serve as surrogate markers of NO production and ONOO- availability.

Immunolabeling was scored separately for two variables: (i) percentage of positively stained cells and (ii) intensity of staining of the positive cells as considered standard by many pathologists and used by us previously (20). The complete absence of protein markers was defined as <5% of tumor cell (or infiltrating lymphocytes) expression when performed in parallel with positive expression in controls, such as endothelial cells in the same tissues. Thus, expression of a given protein marker was considered positive if the average positivity was at least 5% and evident intensity of the staining of which quantifications detailed below. For each patient sample, two (JWCI) or three (MDACC) independent cores were read on the respective TMA, and for each of these cores, two independent readings were performed, and results were averaged. Percentage scoring was considered positive if the average positivity was at least 5%. Intensity scoring staining was defined as follows: “0,” none; “1,” light; “2,” moderate; and “3,” intense, and reported as positive if an average score of at least 1 was obtained.

All IHC analyses were independently interpreted by two readers without knowledge of the clinical data. Discrepancies in scoring were reconciled by a consensus reading between the readers.

Public The Cancer Genome Atlas (TCGA) data repositories (https://tcga-data.nci.nih.gov and http://gdac.broadinstitute.org/) were used to obtain Skin Cutaneous Melanoma sample data. Expression of mRNA for the selected markers and the associated patient clinical data were analyzed (21). Normal tissue gene expressions (2,921 samples) were downloaded from the GTEx Portal (http://www.gtexportal.org). We converted the expression values to transcripts per million, which was found to be the most comparable metric across samples and experiments (22).

Statistical analysis

Two time-to-event outcomes were considered: overall survival (OS) and recurrence-free survival (RFS). Both endpoints were calculated from the date of lymph node surgery. For OS, patients who remained alive were censored at the date of their last follow-up. For RFS, disease-free patients were also censored at the date of last follow-up.

Univariate Cox proportional hazards regression models were fit to determine the association between each marker and each survival outcome. Multivariable Cox models were fit for each marker that adjusted for disease and demographic covariates. The Kaplan–Meier method was used to estimate the distribution of survival times, and the log-rank test was used to compare distributions. Statistical analyses were performed with use of R v. 3.1.1 and SAS version 9.3 for Windows. All statistical tests used a significance level of 5%. No adjustments for multiple testing were made.

Patient characteristics

Table 1 details the characteristics of the two patient cohorts used to prepare the discovery and validation TMAs from JWCI and MDACC, respectively. RECIST criteria (23) were used to define progression after therapy for both cohorts. The cohorts were similar; in the MDACC set, 69% of the patients were male and the median age at surgery was 55 years, in the JWCI set, 57% were male with a median age of 52 years. Ulceration status of the primary tumor was lacking in approximately one third of both sets, and mutated BRAF status was available only from the MDACC cohort. Fortunately, both cohorts had significant follow-up periods of more than 10 years, which provided a rich data set, with 86% of JWCI patients progressing or dead as were 90% of the patients in the MDACC cohort. Although both TMAs were prepared from tumor biopsy specimens from stage III patient, the MDACC TMA consisted primarily of stage IIIB/C patients for which larger tumor samples were available. As a result, the MDACC cohort contained more patients with advanced disease, overall.

Univariate analysis

The strategy and the flow of samples for analyses are presented in Supplementary Material S2. Results of Cox proportional hazards regression models for the associations of the 15 markers with OS and RFS in the JWCI discovery set are shown in Fig. 1A (intensity) and Supplementary Fig. S1A (percentage). Tumor CD74 staining intensity (HR, 0.45; P < 0.001) and staining percentage (HR, 0.46; P ≤ 0.001) both correlated significantly with a lower risk of death and disease recurrence or death (HR, 0.58; P = 0.002 for intensity; HR, 0.51; P = 0.002 for percentage). Tumor cell MIF expression intensity ((HR, 1.57; P = 0.06) and percentage (HR, 1.63; P = 0.05) were marginally associated with poorer OS and significantly associated with worse RFS (HR, 1.70; P = 0.01 for intensity; HR, 1.60; P = 0.026 for percentage; Fig. 1A; Supplementary Fig. S1A).

Figure 1.

Univariate analyses of survival models with individual markers in discovery and validation data sets. Forest plots of OS and RFS HRs and associated 95% confidence intervals for specific protein expression from univariate Cox proportional hazards regression models presented as intensity of staining in discovery (A) and validation (B) datasets.

Figure 1.

Univariate analyses of survival models with individual markers in discovery and validation data sets. Forest plots of OS and RFS HRs and associated 95% confidence intervals for specific protein expression from univariate Cox proportional hazards regression models presented as intensity of staining in discovery (A) and validation (B) datasets.

Close modal

The expression of CD74 and of MIF in tumor was then assessed for association with OS and RFS in the validation TMA (Fig. 1B; Supplementary Fig. S1B). As in the discovery set, CD74 expression in tumor cells correlated with both improved OS (HR, 0.32; P < 0.001 for intensity; HR, 0.33; P < 0.001 for percentage) and RFS (HR, 0.49; P = 0.008 for both). MIF expression in tumor cells was also significantly associated with poorer OS (HR, 3.39; P = 0.009 for intensity; HR, 4.86; P < 0.001 for percentage) and RFS (HR, 2.05; P = 0.035 for intensity; HR, 1.92; P = 0.038 for percentage). Staining percentage and intensity of CD74 and MIF in all samples are summarized in Supplementary Table S1.

As we previously observed significant associations of tumor iNOS expression and disease progression in an earlier cohort of MDACC melanoma patients (12, 13), it was noted that such an association was observed in the MDACC TMA (data not shown); iNOS percentage and intensity in tumor cells were again significantly associated with OS (HR, 3.18; P = 0.004 for intensity; HR, 3.74; P = 0.002 for percentage) but not with RFS (HR, 1.51; P = 0.165 for intensity; HR, 1.64; P = 0.103 for percentage). However, iNOS percentage and intensity in tumor cells were not significant in the JWCI discovery TMA. We concluded that iNOS remains of interest, but clearly not as a powerful prognostic marker of either CD74 positivity or MIF negativity.

Multivariable analysis

Multivariable Cox models were fit to assess the association of OS and RFS for all patients in the JWCI cohort, with variables of gender, age, number of positive lymph nodes, primary tumor Breslow thickness, and CD74 and MIF intensity (Table 2). CD74 intensity in tumor remained a significant predictor of OS (HR, 0.53; P = 0.01) and RFS (HR, 0.56; P = 0.01). MIF intensity in tumor was not significantly associated with OS (P = 0.08), but it was significantly associated with RFS (HR, 1.85; P = 0.02). Among the other characteristics, only having ≥4 lymph nodes involved was significantly associated with OS (HR, 2.04; P = 0.01).

Table 2.

Multivariable cox analyses on survival models in discovery set (N = 103)

Multivariable cox regression modelOSRecurrence-free survival
ParameterLevelHR (95% CI)P valueHR (95% CI)P value
CD74 tumor intensity Positive 0.53 (0.32–0.88) 0.01 0.56 (0.35–0.89) 0.01 
MIF tumor intensity Positive 1.67 (0.94–2.96) 0.08 1.85 (1.10–3.10) 0.02 
Breslow thickness 0–1 — — — — 
 >1–2 0.68 (0.35–1.34) 0.27 0.62 (0.34–1.13) 0.12 
 >2–4 1.44 (0.74–2.79) 0.29 0.82 (0.45–1.50) 0.51 
 >4 0.9 (0.46–1.75) 0.75 0.83 (0.46–1.52) 0.55 
Positive lymph nodes — — — — 
 2–3 1.29 (0.62–2.70) 0.49 0.98 (0.50–1.92) 0.95 
 ≥4 2.04 (1.19–3.50) 0.01 1.71 (1.05–2.79) 0.03 
Age at surgery Continuous, per year 1.02 (1.001–1.03) 0.03 1.01 (0.99–1.02) 0.35 
Gender Male 1.15 (0.69–1.94) 0.59 1.16 (0.73–1.84) 0.54 
Multivariable cox regression modelOSRecurrence-free survival
ParameterLevelHR (95% CI)P valueHR (95% CI)P value
CD74 tumor intensity Positive 0.53 (0.32–0.88) 0.01 0.56 (0.35–0.89) 0.01 
MIF tumor intensity Positive 1.67 (0.94–2.96) 0.08 1.85 (1.10–3.10) 0.02 
Breslow thickness 0–1 — — — — 
 >1–2 0.68 (0.35–1.34) 0.27 0.62 (0.34–1.13) 0.12 
 >2–4 1.44 (0.74–2.79) 0.29 0.82 (0.45–1.50) 0.51 
 >4 0.9 (0.46–1.75) 0.75 0.83 (0.46–1.52) 0.55 
Positive lymph nodes — — — — 
 2–3 1.29 (0.62–2.70) 0.49 0.98 (0.50–1.92) 0.95 
 ≥4 2.04 (1.19–3.50) 0.01 1.71 (1.05–2.79) 0.03 
Age at surgery Continuous, per year 1.02 (1.001–1.03) 0.03 1.01 (0.99–1.02) 0.35 
Gender Male 1.15 (0.69–1.94) 0.59 1.16 (0.73–1.84) 0.54 

Abbreviation: CI, confidence interval.

Multivariable Cox proportional hazards regression analysis was also performed for these same markers in the MDACC cohort (Table 3). Positive CD74 expression in tumor cells was again significantly associated with OS (HR, 0.27; P < 0.001) and RFS (HR, 0.41; P = 0.005). Tumor cell MIF intensity was also significantly associated with both inferior OS (HR, 3.02; P = 0.023) and RFS (HR, 3.60; P = 0.003). Furthermore, iNOS intensity in tumor was now significantly associated with OS (data not shown; HR, 2.82; P = 0.024). No other clinical or tumor factors were significantly associated with OS or RFS in this cohort. BRAF mutation data were available only for the MDACC validation cohort, so in an exploratory analysis, we also reported that BRAF mutation status was not significantly associated with either OS or RFS (Supplementary Table S2).

Table 3.

Multivariable cox analyses on survival models in validation set (N = 79)

Multivariable cox regression modelOSRecurrence-free survival
ParameterLevelHR (95% CI)P valueHR (95% CI)P value
CD74 tumor intensity Positive 0.27 (0.14–0.51) <0.001 0.41 (0.22–0.76) 0.005 
MIF tumor intensity Positive 3.02 (1.16–7.83) 0.02 3.6 (1.57–8.26) 0.003 
Breslow thickness 0–1 — —   
 >1–2 0.81 (0.31–2.11) 0.66 0.73 (0.29–1.84) 0.51 
 >2–4 0.99 (0.44–2.25) 0.99 0.61 (0.27–1.39) 0.24 
 >4 1.12 (0.48–2.57) 0.8 1.02 (0.47–2.19) 0.96 
Positive lymph nodes 0–1 — —   
 2–3 1.24 (0.49–3.09) 0.65 0.74 (0.31–1.77) 0.5 
 ≥4 1.46 (0.68–3.14) 0.37 1.26 (0.62–2.55) 0.52 
Age at surgery Continuous, per year 1.01 (0.99–1.03) 0.47 1 (0.99–1.02) 0.81 
Gender Male 1.24 (0.68–2.29) 0.48 1.49 (0.79–2.80) 0.22 
Multivariable cox regression modelOSRecurrence-free survival
ParameterLevelHR (95% CI)P valueHR (95% CI)P value
CD74 tumor intensity Positive 0.27 (0.14–0.51) <0.001 0.41 (0.22–0.76) 0.005 
MIF tumor intensity Positive 3.02 (1.16–7.83) 0.02 3.6 (1.57–8.26) 0.003 
Breslow thickness 0–1 — —   
 >1–2 0.81 (0.31–2.11) 0.66 0.73 (0.29–1.84) 0.51 
 >2–4 0.99 (0.44–2.25) 0.99 0.61 (0.27–1.39) 0.24 
 >4 1.12 (0.48–2.57) 0.8 1.02 (0.47–2.19) 0.96 
Positive lymph nodes 0–1 — —   
 2–3 1.24 (0.49–3.09) 0.65 0.74 (0.31–1.77) 0.5 
 ≥4 1.46 (0.68–3.14) 0.37 1.26 (0.62–2.55) 0.52 
Age at surgery Continuous, per year 1.01 (0.99–1.03) 0.47 1 (0.99–1.02) 0.81 
Gender Male 1.24 (0.68–2.29) 0.48 1.49 (0.79–2.80) 0.22 

Abbreviation: CI, confidence interval.

While MDACC patients received several different adjuvant therapies, such as, ipilimumab, BRAF inhibitors, vaccine, and biochemotherapy (Supplementary Table S3), JWCI TMA patients predominantly received additional surgery or an experimental vaccine. Even when adjuvant therapy was included in our multivariable analyses, CD74 and MIF intensity persisted as significant. We choose not to present this data due to the variation of the treatment selections between two TMAs and smaller numbers of groups receiving each treatment. The association of these markers with response and prognosis after various immunotherapies and targeted therapies, or their combinations, remains of great interest and will be pursued.

Detailed analysis of CD74 expression

As described above, CD74 expression in melanoma cells correlated with improved OS and RFS in both cohorts, in univariate and multivariate analyses. Analysis of both the quartiles of percentage and intensity staining demonstrated a near dose-dependent difference in CD74 expression with OS and RFS (Supplementary Fig. S2). Patients with >75% of their tumor cells expressing CD74, as well as those with the higher intensity of staining generally had the best outcomes, suggesting a quantitative relationship. The difference between OS or RFS and CD74 tumor cell intensity categories (P < 0.001 and P = 0.013, respectively) and tumor cell percentage categories in the validation data set (P = 0.011 for both) was statistically significant.

As an additional finding, we further noted that CD74 was positive in the TIL. Analysis of this TIL data indicated that CD74 expression on TIL was also associated with improved OS (HR, 0.54; P < 0.0001 for percentage; HR, 0.61; P = 0.004 for intensity) and RFS (HR, 0.65; P = 0.002 for percentage; HR, 0.68; P = 0.001 for staining intensity). Although our studies did not measure CD2, we assume all TILs are CD2 positive, which also will be tested in further studies. Therefore, CD74 protein expression in both tumor cells and TIL predicts better OS and RFS, indicating that most patients with higher CD74 percentage or intensity of protein expression had longer survival.

Survival analysis of the markers of interest in stage III melanoma patients

Figure 2 presents Kaplan–Meier plots of OS and RFS by a classifier that uses both CD74 and MIF intensity, depicting the major outcome of this study. A statistically significant relationship between OS and combined markers intensity (P < 0.001) as well as RFS and combined markers intensity (P < 0.001) was determined in our discovery data set. Similarly, there was a statistically significant relationship between OS or RFS and CD74+/MIF intensity in tumor cells in the MDACC validation data set. Patients with CD74+/MIF intensity in melanoma cells had the longest OS (P < 0.001) and RFS (P = 0.003) among patients with stage III melanoma (P = 0.003). We also observed similar patterns of significance for OS and RFS by combined CD74 and MIF as the percentage of expression in tumors (Supplementary Fig. S3).

Figure 2.

Clinical outcomes by combined markers of CD74 and MIF in tumor cells in stage III melanoma patients in discovery (A) and validation (B) data sets. Kaplan–Meier analyses are shown for OS and RFS for their expression by intensity of staining.

Figure 2.

Clinical outcomes by combined markers of CD74 and MIF in tumor cells in stage III melanoma patients in discovery (A) and validation (B) data sets. Kaplan–Meier analyses are shown for OS and RFS for their expression by intensity of staining.

Close modal

TCGA results confirm CD74 association with good survival

Recently, the cutaneous melanoma TCGA data set has become available (21), and CD74 expression was interrogated. Our analysis was restricted to patients for whom the TCGA tumor biopsy was collected at the first diagnosis of stage III disease (n = 98). Publicly available RNAseq data were used to categorize tumors as having low (lowest quartile) versus high (top three quartiles) CD74 mRNA expression. Including all stage III patients, longer OS was observed in patients with high CD74 (n = 73; median, 73 months) compared with patients with low CD74 expression (n = 25; median, 27.6 months; P = 0.03; Fig. 3A). Considering that this difference might have been due to cumulative CD74 expression in tumor plus TIL, the same analysis was also performed using data only from patients with low lymphocyte scores (21). This analysis again identified longer OS in patients with high CD74 expression (n = 37; median, 69 months) compared with patients with low CD74 (n = 12; median, 16.8 months; P = 0.003; Fig. 3B). These data suggest that tumor cell CD74 expression correlates with OS in the TCGA cohort. Of note, in comparing tumor CD74 mRNA expression in cutaneous melanoma samples (n = 470) with normal tissue expression (n = 2,921), only the spleen and lymphocytes had a higher median expression than melanoma, yet the highest individual expressing samples were in melanoma (Fig. 3C).

Figure 3.

Expression characteristics and survival outcomes of CD74 mRNA expression in TCGA database. Survival analysis of all stage III samples (A) and a group of patients with stage III disease and low lymphocyte infiltration in their tumors (B), as well as tumor CD74 mRNA expression in normal tissues (C) in comparison with melanoma samples.

Figure 3.

Expression characteristics and survival outcomes of CD74 mRNA expression in TCGA database. Survival analysis of all stage III samples (A) and a group of patients with stage III disease and low lymphocyte infiltration in their tumors (B), as well as tumor CD74 mRNA expression in normal tissues (C) in comparison with melanoma samples.

Close modal

Improved risk models of patients with stage III melanoma will improve the treatment of patients with this disease; however, molecular markers are lacking. Our analysis of independent cohorts (two for protein in TMA, and one for mRNA in TCGA) of patients' tumors strongly implicates increased CD74 expression as a new marker of good prognosis in stage III melanoma. Previously, we had considered inflammatory markers to be associated with poor prognosis, which was the case for MIF in both cohorts, and for iNOS in the MDACC cohort. The surprising finding of higher expression of CD74 having a strong association with good prognosis is most intriguing, and elucidating the mechanism involved is likely to open new avenues for melanoma research.

The functional role of CD74 is not well understood. Historically, CD74 was known primarily as the MHC class II invariant chain and functions in the molecular processing of MHC II through the Golgi (24). It has a potential role in the antitumor immune response (25), and also been shown to be expressed in inflammatory disorders and other inflammation-associated cancers (26). Our unexpected finding of high expression of CD74 in a subset of samples in both the discovery and validation TMAs of melanoma patients from two different institutions, along with highly significant associations with improved survival in both univariate and multivariable analyses of both TMAs, strongly suggests a previously unappreciated biologic role for this curious molecule. Irrespective of the mechanism(s) involved, CD74 now meets the criteria as a validated marker of improved prognosis in stage III melanoma patients. Similar findings that high expression of CD74 is an independent prognostic factor for OS were recently reported in mesothelioma (27).

Numerous inflammation markers have been identified in melanoma cell lines and patient samples with use of gene expression array methods; these markers include iNOS, arginase, VEGFA, CXCL10, IL8, IL1α, IL1β, and TNFSF9 (19, 28). More recent results from our laboratory have continued to support association of iNOS protein with NT, COX-2, pSTAT3, and arginase, consistent with the report of Johansson and colleagues (18). Because iNOS expression and NO production lead to tumor progression, in this study, we asked whether we could identify other markers or even a signature of common inflammatory markers. Although many markers have been noted extensively in the literature, in melanoma, we have identified inflammatory cytokines IL1α, IL1β, and IL6 (29) but none of these proved prognostic in our controlled and parallel studies shown here. In a previous functional study, we observed that an MIF-CD74 autocrine interaction upregulated by exogenous IFNγ promoted the phosphorylation of AKT Ser473 and drove expression of such inflammatory cytokines (30), presenting a potentially protumor circuitous or alternative function when MIF is present, which may be likely in a subset of tumors after exposure to IFNγ. These findings could be specifically important as we observed this effect regardless of the BRAF and NRAS mutation status. This implies that the augmentation of CD74-MIF signaling by IFNγ is a universal phenomenon in melanoma. In that functional study, we were able to show all the melanoma lines expressed MIF and the inhibition of its binding to CD74 by using ISO-1 suppressed the phosphorylation of AKT Ser 473, as well as the expression of BCL-2, IL6, and IL8. Another important finding in that study was the expression characteristics of CD74 in a melanoma progression array. We reported that CD74 expression in primary and metastatic lesions was significantly higher than that in nevi (P < 0.0001). These results suggested that melanoma cell lines express CD74 and that expression is associated with the progression from benign nevi to clinically evident melanoma. However, the TMA we used in that study was not designed to evaluate CD74's role in melanoma prognosis. Therefore, we concluded that CD74 expression was associated with melanoma progression; however from our new data, it represents a good prognostic marker if expressed in the absence of MIF in the microenvironment. Another important feature is that its role in MHC processing potentiates antigen expression, and its association with immunotherapy responses is now a high priority to resolve.

Low MIF was validated as a good prognostic finding in this report. In an attempt to understand possible interactions, it is known that MIF is a pleiotropic proinflammatory protein for which the effects include the promotion of cytokine expression and inhibition of apoptosis (31). It was recently shown that MIF could also play a role as a dominant chemotactic cytokine by physical interaction with its many receptors (CXCR2, CXCR4, and CD74) in human mesenchymal stem cell recruitment to tumors (32). CD74 has been reported to bind MIF and to be required for MIF-dependent cellular responses, such as PGE2 production and proliferation (33). Although the role of surface-expressed CD74 in chemokine-mediated signaling has not been well studied, we originally hypothesized that positive expression of both MIF and CD74 expressions in tumor cells would predict poor survival of melanoma patients via activation of tumor survival pathways. Our results of CD74 in the absence of MIF providing value as a good prognostic marker was most significant, so that low MIF and high CD74 provide another intriguing model for functional analysis. CD74 appears to have at least two functions: (i) originally recognized as the MHCII invariant chain, loading antigen into MHC for cell surface expression; and (ii) as a cell surface receptor for MIF in the presence of IFNγ (19). The intracellular portion of CD74 lacks signaling domains, but has been predicted to interact with downstream signaling molecules (33). Another consideration is that CD74 is a homotrimer, and MIF engagement of CD74 may affect the oligomerization of the intracellular domain that is necessary for downstream signaling; alternatively, there may be unknown variants of CD74 unique to melanoma. Regardless of the mechanisms yet to be revealed, it is clear from our data that CD74 expression in both the melanoma cell and infiltrating lymphocytes indicates a survival advantage for the patient. Studies are underway to assess whether the CD74 expression alone would predict a favorable response to immunotherapies; conversely, if the IFNγ–driven product of MIF is present in the tumor site, perhaps other means to inhibit them or targeted therapy approaches might prioritized. Although this scenario is hypothetical, we currently envision that these or other inflammatory markers are likely to be useful beyond their prognostic value and may help make predictive treatment decisions.

In conclusion, we have identified and validated the protein expression of CD74 and of MIF as providing survival information, and proposed that CD74+ and MIF together be considered to form a “signature” for stage III melanoma prognosis. Our data provide a clear justification for including these markers for testing in larger stage III data cohorts to address their value in refining the current substaging data for stage III melanomas, and for testing these markers in prospective clinical trials to determine whether this signature adds value to either targeted or immunotherapy treatments. It is our hope that these two-protein markers (and others to be discovered), either together as a “signature” or individually, will help clinicians make better personalized treatment suggestions and help patients make better treatment decisions.

J.E. Gershenwald is a consultant/advisory board member for Merck. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S. Ekmekcioglu, V.G. Prieto, D.L. Morton

Development of methodology: S. Ekmekcioglu, D.S. Hoon

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Ekmekcioglu, K. Tanese, V.G. Prieto, J.E. Gershenwald, D.L. Morton

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Ekmekcioglu, M.A. Davies, K. Tanese, J. Roszik, M. Shin-Sim, R.L. Bassett Jr, D.R. Milton, S.E. Woodman, V.G. Prieto, J.E. Gershenwald, D.L. Morton, E.A. Grimm

Writing, review, and/or revision of the manuscript: S. Ekmekcioglu, M.A. Davies, K. Tanese, M. Shin-Sim, R.L. Bassett Jr, D.R. Milton, V.G. Prieto, J.E. Gershenwald, D.L. Morton, D.S. Hoon, E.A. Grimm

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Ekmekcioglu, K. Tanese, E.A. Grimm

Study supervision: S. Ekmekcioglu, E.A. Grimm

The authors thank the members of Melanoma Informatics, Tissue Resources and Pathology Core, who provided the deidentified tissues for the validation TMA used in this project as well as maintained accurate patient information available for our studies. They also thank Sandra A. Kinney for her excellent technical assistance in our IHC studies and Tamara K. Locke for her editorial assistance.

This work was supported by the MD Anderson Cancer Center SPORE in Melanoma P50 CA093459 (to S. Ekmekcioglu, M.A. Davies, R.L. Bassett Jr, V.G. Prieto, J.E. Gershenwald, and E.A. Grimm) funded from the NIH, the MD Anderson's Cancer Center Support Grant P30- CA016672 (to E.A. Grimm) funded from the NIH, the Dr. Miriam and Sheldon G. Adelson Medical Research Foundation (to K. Tanese, D.L. Morton, D.S. Hoon, and E.A. Grimm), The University of Texas, M.D. Anderson Cancer Center Melanoma Moon Shots Program (to M.A. Davies, J.E. Gershenwald, and E.A. Grimm), the Miriam and Jim Mulva Foundation, and AIM at Melanoma Foundations.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Balch
CM
,
Gershenwald
JE
,
Soong
SJ
,
Thompson
JF
,
Atkins
MB
,
Byrd
DR
, et al
Final version of 2009 AJCC melanoma staging and classification.
J Clin Oncol
2009
;
27
:
6199
206
.
2.
Balch
CM
,
Gershenwald
JE
,
Soong
SJ
,
Thompson
JF
,
Ding
S
,
Byrd
DR
, et al
Multivariate analysis of prognostic factors among 2,313 patients with stage III melanoma: comparison of nodal micrometastases versus macrometastases
.
J Clin Oncol
2010
;
28
:
2452
9
.
3.
Harcharik
S
,
Bernardo
S
,
Moskalenko
M
,
Pan
M
,
Sivendran
M
,
Bell
H
, et al
Defining the role of CD2 in disease progression and overall survival among patients with completely resected stage-II to -III cutaneous melanoma
.
J Am Acad Dermatol
2014
;
70
:
1036
44
.
4.
Sivendran
S
,
Chang
R
,
Pham
L
,
Phelps
RG
,
Harcharik
ST
,
Hall
LD
, et al
Dissection of immune gene networks in primary melanoma tumors critical for antitumor surveillance of patients with stage II-III resectable disease
.
J Invest Dermatol
2014
;
134
:
2202
11
.
5.
Bogunovic
D
,
O'Neill
DW
,
Belitskaya-Levy
I
,
Vacic
V
,
Yu
YL
,
Adams
S
, et al
Immune profile and mitotic index of metastatic melanoma lesions enhance clinical staging in predicting patient survival
.
Proc Natl Acad Sci U S A
2009
;
106
:
20429
34
.
6.
Schetter
AJ
,
Heegaard
NH
,
Harris
CC
. 
Inflammation and cancer: interweaving microRNA, free radical, cytokine and p53 pathways
.
Carcinogenesis
2010
;
31
:
37
49
.
7.
Grimm
EA
,
Sikora
AG
,
Ekmekcioglu
S
. 
Molecular pathways: inflammation-associated nitric-oxide production as a cancer-supporting redox mechanism and a potential therapeutic target
.
Clin Cancer Res
2013
;
19
:
5557
63
.
8.
Thomsen
LL
,
Lawton
FG
,
Knowles
RG
,
Beesley
JE
,
Riveros-Moreno
V
,
Moncada
S
. 
Nitric oxide synthase activity in human gynecological cancer
.
Cancer Res
1994
;
54
:
1352
4
.
9.
Cobbs
CS
,
Brenman
JE
,
Aldape
KD
,
Bredt
DS
,
Israel
MA
. 
Expression of nitric oxide synthase in human central nervous system tumors
.
Cancer Res
. 
1995
;
55
:
727
30
.
10.
Gallo
O
,
Masini
E
,
Morbidelli
L
,
Franchi
A
,
Fini-Storchi
I
,
Vergari
WA
, et al
Role of nitric oxide in angiogenesis and tumor progression in head and neck cancer
.
J Natl Cancer Inst
1998
;
90
:
587
96
.
11.
Reveneau
S
,
Arnould
L
,
Jolimoy
G
,
Hilpert
S
,
Lejeune
P
,
Saint-Giorgio
V
, et al
Nitric oxide synthase in human breast cancer is associated with tumor grade, proliferation rate, and expression of progesterone receptors
.
Lab Invest
1999
;
79
:
1215
25
.
12.
Ekmekcioglu
S
,
Ellerhorst
J
,
Smid
CM
,
Prieto
VG
,
Munsell
M
,
Buzaid
AC
, et al
Inducible nitric oxide synthase and nitrotyrosine in human metastatic melanoma tumors correlate with poor survival
.
Clin Cancer Res
2000
;
6
:
4768
75
.
13.
Ekmekcioglu
S
,
Ellerhorst
JA
,
Prieto
VG
,
Johnson
MM
,
Broemeling
LD
,
Grimm
EA
, et al
Tumor iNOS predicts poor survival for stage III melanoma patients
.
Int J Cancer
2006
;
119
:
861
6
.
14.
Borghese
F
,
Clanchy
FI
. 
CD74. an emerging opportunity as a therapeutic target in cancer and autoimmune disease
.
Expert Opin Ther Targets
2011
;
15
:
237
51
.
15.
Haq
R
,
Yokoyama
S
,
Hawryluk
EB
,
Jönsson
GB
,
Frederick
DT
,
McHenry
K
, et al
BCL2A1 is a lineage-specific antiapoptotic melanoma oncogene that confers resistance to BRAF inhibition
.
Proc Natl Acad Sci U S A
2013
;
110
:
4321
6
.
16.
Granados-Principal
S
,
Liu
Y
,
Guevara
ML
,
Blanco
E
,
Choi
DS
,
Qian
W
, et al
Inhibition of iNOS as a novel effective targeted therapy against triple-negative breast cancer
.
Breast Cancer Res
2015
;
17
:
25
.
17.
McShane
LM
,
Altman
DG
,
Sauerbrei
W
,
Taube
SE
,
Gion
M
,
Clark
GM
, et al
Reporting recommendations for tumor marker prognostic studies (REMARK)
.
J Natl Cancer Inst
2005
;
97
:
1180
4
.
18.
Johansson
CC
,
Egyházi
S
,
Masucci
G
,
Harlin
H
,
Mougiakakos
D
,
Poschke
I
, et al
Prognostic significance of tumor iNOS and COX-2 in stage III malignant cutaneous melanoma
.
Cancer Immunol Immunother
2009
;
58
:
1085
94
.
19.
Tanese
K
,
Grimm
EA
,
Ekmekcioglu
S
. 
The role of melanoma tumor-derived nitric oxide in the tumor inflammatory microenvironment: its impact on the chemokine expression profile, including suppression of CXCL10
.
Int J Cancer
2012
;
131
:
891
901
.
20.
Prieto
VG
,
Mourad-Zeidan
AA
,
Melnikova
V
,
Johnson
MM
,
Lopez
A
,
Diwan
AH
, et al
Galectin-3 expression is associated with tumor progression and pattern of sun exposure in melanoma
.
Clin Cancer Res
2006
;
12
:
6709
15
.
21.
The Cancer Genome Atlas Network. Genomic classification of cutaneous melanoma
Cell
161
:
1681
96
.
22.
Wagner
GP
,
Kin
K
,
Lynch
VJ
. 
Measurement of mRNA abundance using RNA-seq data: RPKM measure is inconsistent among samples
.
Theory Biosci
2012
;
131
:
281
5
.
23.
Eisenhauer
EA
,
Therasse
P
,
Bogaerts
J
,
Schwartz
LH
,
Sargent
D
,
Ford
R
, et al
New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1)
.
Eur J Cancer
2009
;
45
:
228
47
.
24.
Karakikes
I
,
Morrison
IE
,
O'Toole
P
,
Metodieva
G
,
Navarrete
CV
,
Gomez
J
, et al
Interaction of HLA-DR and CD74 at the cell surface of antigen-presenting cells by single particle image analysis
.
FASEB J
2012
;
26
:
4886
96
.
25.
Basha
G
,
Omilusik
K
,
Chavez-Steenbock
A
,
Reinicke
AT
,
Lack
N
,
Choi
KB
, et al
A CD74-dependent MHC class I endolysosomal cross-presentation pathway
.
Nat Immunol
2012
;
13
:
237
45
.
26.
Beswick
EJ
,
Reyes
VE
. 
CD74 in antigen presentation, inflammation, and cancers of the gastrointestinal tract
.
World J Gastroenterol
2009
;
15
:
2855
61
.
27.
Otterstrom
C
,
Soltermann
A
,
Opitz
I
,
Felley-Bosco
E
,
Weder
W
,
Stahel
RA
, et al
CD74: a new prognostic factor for patients with malignant pleural mesothelioma
.
Br J Cancer
2014
;
110
:
2040
6
.
28.
Qin
Y
,
Deng
W
,
Ekmekcioglu
S
,
Grimm
EA
. 
Identification of unique sensitizing targets for anti-inflammatory CDDO-Me in metastatic melanoma by a large-scale synthetic lethal RNAi screening
.
Pigment Cell Melanoma Res
2013
;
26
:
97
112
.
29.
Qin
Y
,
Ekmekcioglu
S
,
Liu
P
,
Duncan
LM
,
Lizée
G
,
Poindexter
N
, et al
Constitutive aberrant endogenous interleukin-1 facilitates inflammation and growth in human melanoma
.
Mol Cancer Res
2011
;
9
:
1537
50
.
30.
Tanese
K
,
Hashimoto
Y
,
Berkova
Z
,
Wang
Y
,
Samaniego
F
,
Lee
JE
, et al
Cell surface CD74-MIF interactions drive melanoma survival in response to interferon-gamma
.
J Invest Dermatol
2015
;
135
:
2775
84
.
31.
Fan
H
,
Hall
P
,
Santos
LL
,
Gregory
JL
,
Fingerle-Rowson
G
,
Bucala
R
, et al
Macrophage migration inhibitory factor and CD74 regulate macrophage chemotactic responses via MAPK and Rho GTPase
.
J Immunol
2011
;
186
:
4915
24
.
32.
Lourenco
S
,
Teixeira
VH
,
Kalber
T
,
Jose
RJ
,
Floto
RA
,
Janes
SM
. 
Macrophage migration inhibitory factor-CXCR4 is the dominant chemotactic axis in human mesenchymal stem cell recruitment to tumors
.
J Immunol
2015
;
194
:
3463
74
.
33.
Leng
L
,
Metz
CN
,
Fang
Y
,
Xu
J
,
Donnelly
S
,
Baugh
J
, et al
MIF signal transduction initiated by binding to CD74
.
J Exp Med
2003
;
197
:
1467
76
.

Supplementary data