SEARCH

SEARCH BY CITATION

Keywords:

  • PD-1;
  • PD-L1;
  • B7-H1;
  • melanoma

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

BACKGROUND:

Cancers are known to elude the immune system, for example, by MHC loss, FAS up-regulation, or increased secretion of TGF-beta. Recently, ligands of coinhibitory receptors like programmed cell death ligand–1 (PD-L1, B7-H1) have come to attention for their role in tumor immune escape. Various tumors have been tested for PD-L1 expression, and conflicting results were obtained regarding its correlative impact on patient survival. This study aimed to determine the prognostic relevance of PD-L1 expression for the survival of melanoma patients.

METHODS:

Paraffin-embedded nevi, primary melanoma, and in-transit, lymph node, and distant organ metastases from a set of 63 stages III-IV melanoma patients referred to the Netherlands Cancer Institute between 2000 and 2004 for a sentinel-node procedure or systemic therapy were studied. A large effort was invested in validating specific PD-L1 staining. In addition to immunological factors such as T-cell infiltration (CD8, CD4, and regulatory T cells), TGF-beta and MHC-I expression were assessed.

RESULTS:

Longitudinal analysis revealed no relevant PD-L1 expression on primary melanoma compared with metastatic disease. No significant correlations with prognosis were found regarding immunological factors, whereas known prognostic markers such as Breslow thickness and sex could be confirmed. Analyses of the overall survival of our patient cohort did not reveal a negative association with PD-L1 expression.

CONCLUSIONS:

Correlation of overall survival with PD-L1 expression by melanoma cells remains controversial, and future clinical studies should focus on antibody validation and time of analysis in respect to disease progression. Cancer 2011. © 2010 American Cancer Society.

Melanoma is a highly immunogenic malignancy. Recent efforts have focused on immunotherapeutic approaches, such as coinhibitory molecule targeting (eg, CTLA-4),1 using vaccines,2 or adoptively transferring tumor-specific T cells.3

Despite often achieving surrogate goals such as expansion and differentiation of tumor-specific immune cells, clinical responses (besides CTLA-4 blockade) based on WHO or RECIST criteria or improvement of overall survival have this far rarely been observed or have been difficult to correlate with conventional markers used for immune monitoring.4

The presence of melanoma-specific T cells in the microenvironment of growing tumors5 suggests tumor immune escape or resistance to effecter T cells. Indeed, ex vivo analysis of tumor-infiltrating lymphocytes (TILs) often indicates defective antigen-specific proliferation, cytolytic activity, and cytokine production that can be reversed by in vitro culture.6-8 Mechanisms thought to be involved are T-cell anergy, regulatory T cells, TGF-β, IL-10 and/or defects in FAS/FAS-L, increased indoleamine 2,3-dioxygenase, myeloid-derived suppressor cells (MDSCs), and stromal protection.5, 9-12

Recently, BTLA (B- and T-lymphocyte attenuator) and PD-1 (programmed cell death-1), both T-cell immune- inhibitory receptors of the CD28 family, have been added to the potential mediators of tumor immune escape.13, 14 Two ligands of PD-1, PDL-1 (PD-1, ligand 1; B7-H1) and PD-L2 (PD-1, ligand 2; B7-DC) have been shown to mediate negative immune regulation by engaging the PD-1 receptor.15 PD-L1 is expressed on T- and B cells, dendritic cells, endothelial cells, and the fetomaternal interface, indicating a physiological role of PD-L1 in mediating peripheral tolerance.13, 15, 16 PD-L1/PD-1 interactions result in inhibition of T-cell growth and cytokine production. Moreover, tumor cell-expressed PD-L1 can induce inhibition or apoptosis of tumor-specific T cells.17-21 PD-L1 has been shown to be expressed in various tumors, for example, melanoma,21-23 renal cell carcinoma (RCC),24 non–small cell lung cancer,25 gastric cancer,26 breast cancer,27 ovarian cancer,28 esophageal cancer,29 pancreatic cancer,30 cervical cancer,31 and bladder cancer.32

The prognostic relevance of PD-L1 expression, however, remains controversial so far. Although a significant negative prognostic value of PD-L1 has been found in RCC patients, opposite results were found in other malignancies.33 The reasons for this disparate data are the small numbers of patient analyzed, the use of different antibodies to stain PD-L1, and the use of frozen versus paraffin-embedded material. Indeed, a decreased frequency of PD-L1-positive RCC was detected in a study that went from analyzing initially only frozen tissues to paraffin-embedded patient samples.24, 34, 35

In the current work, we aimed to establish a reliable protocol for PD-L1 staining on paraffin-embedded material and addressed the following questions: 1) Does the percentage of PD-L1-positive melanomas increase with disease progression? 2) If so, does PD-L1 expression correlate with prognosis? c) Can PD-L1 expression in melanoma be identified as an independent prognostic marker?

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

Patients

Patients with stages III-IV melanoma who had been referred to the Netherlands Cancer Institute—Antoni van Leeuwenhoek Hospital (NKI-AVL) between 2000 and 2004 for a sentinel-node procedure, lymph node dissection, or systemic therapy were retrospectively enrolled in this study. All living patients were asked in a letter explaining the content of the study for permission to analyze their tumor material. All tumor stages were revised at time of referral by the in-house pathologists according to the American Joint Committee on cancer staging system. This study was approved by the ethical board of the NKI-AVL in line with the Dutch regulation for research on human material (WMO).

Immunohistochemistry

Paraffin-embedded blocks of melanoma could be obtained for 63 patients (for 8 patients, inclusion was not possible because no blocks of primary tumors were provided by the external pathologists). Twenty of the 63 primary tumors could not be analyzed because of lack of sufficient embedded material or poor-quality samples. Tissue sections were cut into 4-μm-thick sections and dried at 70°C for 30 minutes. Dried sections were deparaffinized in xylene and dehydrated through graded alcohols to water. For all clones but 5H1, antigen retrieval was achieved by boiling the sections in citrate buffer, pH 6.0, using a microwave. The antigen retrieval for 5H1 staining was performed by EDTA boiling (according to the optimized protocol of Karim et al31). Endogenous peroxidase activity was blocked by incubating the sections with peroxidase-blocking reagent (DakoCytomation, Glostrup, Denmark) for 10 minutes. To reduce nonspecific binding, sections were incubated for 30 minutes at room temperature (RT) with 5% normal goat serum (Sanquin, Amsterdam, The Netherlands). Primary antibody (Ab) incubation was performed overnight at 4°C in a humidified chamber. Consequently the sections were incubated with poly-HRP-linked goat-antimouse or goat-antirabbit (corresponding to primary Ab; ImmunoVision Technologies) for 30 minutes at RT. Positive staining was visualized with a liquid DAB substrate chromogen system (DakoCytomation, Glostrup, Denmark).

Immunohistochemical staining was performed using the following antibodies: MART-1/Melan-A Ab-3 Cocktail (ThermoFisher Scientific, Waltham, Mass), GP-100 (clone HMB45; DakoCytomation, Glostrup, Denmark), TGF-β2 (ab3649 Abcam, Cambridge, UK), MHC-I (HC-10 mAb; kind gift of Jacques Neefjes, Amsterdam, The Netherlands), CD8 (clone C8/144B; DakoCytomation, Glostrup, Denmark), CD4 (clone 4B12; MONOSAN, Uden, The Netherlands), and FOXP3 (clone 236A/E7; Abcam, Cambridge, UK).

All antihuman-PD-L1 antibodies tested were titrated in comparison with the corresponding isotype control, and specific binding was proven by competitive preincubation of the anti-PD-L1 Ab with recombinant human PD-L1/FC chimera (R&D Systems, Minneapolis, MN).

We used tonsil and placental tissue to set up the staining, as these organs have been shown to constitutively express PD-L1.36

Analysis of Stained Sections of Tissues

All sections (nevi, primary melanoma, satellite or in-transit metastases, and distant metastases) were scored in a blinded manner using a light microscope. All sections stained for MART-1/Melan-A Ab-3 cocktail, GP-100, PD-L1, TGF-β, or MHC-I were scored for the percentage of positive tumor cells.

The melanomas were defined as MART-1, gp100, MHC-I, or TGF-β positive if at least 10% of the tumor cells (based on specific detection limit compared with the isotype control) stained positive. Infiltrating T cells (sum of CD4 and CD8) were scored as a percentage of tumor cells. Because we did not look only at primary melanoma, we could not define TIL infiltrates according to a previous definition as described by Busam et al.37-39 We defined tumors to be infiltrated with lymphocytes when the frequency of lymphocyte to tumor reached at least 5% (based on the detection limit of T-cell staining).

FoxP3-positive cells were scored as a percentage of the total number of T cells, and melanomas were defined as Treg-infiltrated if at least 10% of the total number of T-cell (CD4+ and CD8+) infiltrates were FoxP3 positive. Such a cutoff was based on FoxP3 staining of distant nontumor tissue. Tumors were defined as PD-L1 positive if at least 1% of the melanoma cells stained positive (based on a background of isotype control staining).

Statistical Analysis

Overall survival analyses were constructed using the Kaplan-Meier technique and compared using the log-rank test. Overall survival and disease-related survival were calculated from date of first diagnosis to date of death or last follow-up (see also Table 1, status at the end of observation in October 2009).

Table 1. Patient Characteristics
 No. of Patients (%)
  1. Data presented as n (%), where the percentages were rounded up to the closest full number. Mean patient age on the date of first diagnosis was 53 years (±18 years). None of the patients had received systemic treatment prior to referral at the NKI-AVL. Two patients were treated at the NKI-AVL with an experimental therapy (1 with DC vaccination and 1 with anti-CTLA4).

Sex 
 Male36 (57%)
 Female27 (43%)
Histology of first diagnosis 
 Superficial spreading melanoma28 (44%)
 Nodular melanoma25 (40%)
 Unknown primary4 (6%)
 Acral lentiginous melanoma1 (2%)
 Spindle cell melanoma1 (2%)
 Unclear4 (6%)
Ulceration of primary tumor 
 No ulceration18 (29%)
 Ulcerating tumor28 (44%)
 No information available17 (27%)
Presence of lymph angiogenesis in primary tumor 
 Not present15 (24%)
 Present9 (14%)
 No information available39 (62%)
Breslow stage 
 ≤1 mm0 (0%)
 >1 but ≤2 mm12 (20%)
 >2 but ≤4 mm29 (48%)
 >4 mm19 (32%)
Site of most distant organ metastasis 
 Skin/nodal/GI tract21 (33%)
 Lung11 (18%)
 Liver/brain/bone31 (49%)
Status at end of study (10/2009) 
 Alive with disease1 (2%)
 Alive without disease18 (29%)
 Dead without disease2 (3%)
 Dead of disease42 (67%)

Using Cox proportional hazard analysis, the association of PD-L1 with overall survival was adjusted for sex and Breslow thickness.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

Patient Characteristics

In total, 71 patients were potentially eligible for this study. Paraffin-embedded tissue blocks of melanoma could be obtained for 63 patients. The clinical characteristics of the 63 patients are listed in Table 1. The average age of the patients at time of first diagnosis was 53 years (range, 18-91 years; SD, ±18 years), and 36 (57%) were male. Most of the patients were diagnosed with superficial spreading melanoma (44%), followed by nodular melanoma (40%), metastatic melanoma without known primary tumor (6%), acral lentigious melanoma (2%), spindle cell melanoma (n = 1, 2%), and unclear melanoma subtype (6%). Ulceration was described in 44% of the patients (absent in 29% of the patients and not reported in 27%) and vascular invasion in 14% (24%, absent; 62%, not reported). The Breslow thickness was more than 1 mm for all patients because of the referral selection (19%, 1-2 mm; 46%, 2-4 mm; 30%, >4 mm; 5%, no data). Distant metastases were skin/nodal (33%), lung (18%), and liver/brain/bone (49%). The majority of patients was treated with standard chemotherapy consisting of DTIC, and 2 patients received an experimental therapy (1 DC vaccination and 1 anti-CTLA4). At the time of analysis 67% of patients were dead of disease, 3% were dead of other reasons, 29% were alive without any sign of active disease, and 2% were alive with active disease. The median time patients were followed was 207 weeks (minimum, 27 weeks; maximum, 539 weeks). One patient was lost after 2 years of follow-up.

Validation of PD-L1 Staining in Paraffin-Embedded Tissues

A major challenge for correlative studies concerning PD-L1 expression and overall survival is the available antibodies for tissue staining. Whereas frozen samples can be stained very specifically, paraffin-embedded material (which is often the only material available from pathology labs for retrospective analysis) results in staining that is not as strong and often gives a high background.23, 34, 35 Therefore, we tested all those commercially available—2 mAbs provided by Otto Madjic (University of Vienna, Vienna, Austria) and an additional 9 mAbs provided by Alan Korman (Medarex, Princeton, NJ)—for their PD-L1 reactivity in paraffin-embedded material (Table 2).

Table 2. All Tested Antihuman-PD-L1 Antibodies
CloneProviderSpeciesIsotypeResults
Stains on ParaffinIsotype ControlPD-L1 Fc Blocking
  1. pAb, polyclonal antibody; nd, not done; n/a, not applicable. Antibodies were tested for staining of paraffin-embedded tissues, isotype background staining, and eventually for antigen specificity.

MIH1eBioscience, San Diego, CAMouseIgG1NoNo backgroundnd
5-496O. Majdic, University of Vienna Medical School, Vienna, AustriaMouseIgG1Nondnd
2-272O. Majdic, University of Vienna Medical School, Vienna, AustriaMouseIgG1NoNo backgroundnd
27A2MBL International, Woburn, MAMouseIgG2bYesHigh backgroundnd
16E11Medarex, Princeton, NJMouseUnknownNondnd
9A6Medarex, Princeton, NJMouseIgG3Nondnd
16A4Medarex, Princeton, NJMouseIgG1NoNo backgroundnd
6H3Medarex, Princeton, NJMouseIgG1NoNo backgroundnd
ETM-79Medarex, Princeton, NJRabbitpAbYesNo backgroundPartial block
ETM-80Medarex, Princeton, NJRabbitpAbYesNo backgroundPartial block
1105Medarex, Princeton, NJHumanIgG4Nondnd
25C8E8.F8Medarex, Princeton, NJHumanIgG1Nondnd
24B10.G6.D7Medarex, Princeton, NJHumanIgG1Nondnd
5H1L. Chen, Johns Hopkins University, Baltimore, MDMouseIgG1High backgroundNo backgroundnd
#4059ProSci, Poway, CARabbitpAbYesNo backgroundNear full block

We found that the clone 27A2 (MBL), commonly used for paraffin material,28 gave a high background equivalent to the isotype mAb control (provided by the same company) and could thus be falsely interpreted as positive staining, unless using proper isotype analysis (Fig. 1b). All antibodies tested, except the rabbit polyclonal anti-PD-L1 antibody from ProSci (catalogue no. 4059; Poway, CA), either did not stain paraffin-embedded tissues (Fig. 1a), gave a high background staining (Fig. 1b,c), or were not blocked by PD-L1 fusion protein competition (Fig. 1d). Only the ProSci Ab was found to fulfill all requirements in an acceptable manner (Fig. 1e). The indirectly postulated loss of PD-L1-positive tumor samples when staining paraffin-embedded material instead of frozen samples34 was proven by direct comparison of paraffin with frozen samples. Consistent with previous data, we found that PD-L1 staining in paraffin material underestimated the number of PD-L1-positive melanomas and the percentage of tumor cells positive for PD-L1.21, 23, 28, 35 In detail, 8 of the 12 frozen tumor samples we tested were PD-L1-positive (with 10%-50% positive tumor cells), whereas only 3 of the 12 paraffin counterparts stained positive (with 1%-10% positive cells; data not shown).

thumbnail image

Figure 1. Testing available antihuman-PD-L1 antibodies for paraffin immunohistochemistry is shown. Antibody testing on the indicated tissue of (a) clone MIH1 (representative of staining with clones 5-496, 2-272, 16E11, 9A6, 16A4, 6H3, 1105, 25C8E8.F8, and 24B10.G6.D7), (b) clone 5H1, (c) clone 27A2, (d) ETM-80 (stained similarly to ETM-79), and (e) #4059 is shown. The inserted black bars indicate 50 μm.

Download figure to PowerPoint

Characterization of Melanoma Cohort: MART-1, gp100, and MHC-I Expression and T-Cell Infiltration

Ninety-eight percent and 95% of primary melanomas and 90% and 86% of the metastases were MART-1 and gp100 positive, respectively. MHC-I expression was observed in 45% of the primary melanomas and in 52% of metastases. In both primary and metastatic melanoma, we found 40% of tumor samples containing TIL infiltrates. Analysis for CD8, CD4, and FoxP3 revealed that 52% had a majority of CD8-positive TILs, whereas only 3% of the tumors had mainly CD4-positive TILs. The remaining 45% of tumors showed comparable CD4 and CD8 T-cell numbers. Of these TIL infiltrates, the majority had a relevant subgroup of FoxP3-positive cells (72%). Sixty-three percent of primary and 46% of metastatic melanoma were found to express TGF-β. Representative staining of samples defined as positive or negative are shown in Figure 2.

thumbnail image

Figure 2. Histological staining of paraffin-embedded tissues is shown. Depicted are representative tissue samples that stained either positive or negative for MART-1, GP-100, PD-L1, MHC-I, TGF-β, CD8, CD4, or FoxP3. The inserted black bars indicate 100 μm.

Download figure to PowerPoint

PD-L1 Expression During Tumor Progression

To address the association of PD-L1 expression with patient survival in a longitudinal way (during disease progression), PD-L1 expression was analyzed in benign nevi (n = 10), primary melanomas (n = 43), satellite metastases (n = 8), in-transit metastases (n = 20), lymph node metastases (n = 21), and distant organ metastases (n = 22). We found that 20% of the benign nevi samples were scored PD-L1 positive, whereas only 5% of primary melanomas were PD-L1 positive.

A higher frequency of PD-L1-positive staining was observed in satellite metastases (25%) and in in-transit metastases (40%). Lymph node and distant organ metastases stained PD-L1+ less frequently, in 14% and 18%, respectively (Fig. 3).

thumbnail image

Figure 3. PD-L1 expression varied during disease progression. PD-L1 staining was scored for the percentage of positive tumor cells. (benign nevi, n = 10; primary tumor, n = 43; satellite meta, n = 8; in-transit meta, n = 20; lymph node meta, n = 21; distant organ meta, n = 22).

Download figure to PowerPoint

Patients who expressed PD-L1 at any time and for whom at least 2 samples were available during follow-up were analyzed for the intensity of PD-L1 expression in primary melanoma, in-transit or satellite metastasis, and lymph node or distant organ metastasis. Only 2 of 11 patients (Mel015 and Mel029) showed absent PD-L1 expression in the distant metastases compared with the earlier metastases. In all other patients, PD-L1 expression increased with disease progression (data not shown).

PD-L1 Expression and Patient Overall Survival

To allow an analysis of PD-L1 expression despite variable expression during melanoma progression (see also Fig. 3), we grouped our patient population as patients in whom PD-L1 expression could never be detected (never group) or patients in whom PD-L1 was detected at least 1 time or at 1 site (ever group) and analyzed the correlation of PD-L1 expression with overall survival from date of first diagnosis.

Although a trend toward better survival if PD-L1 was ever detected was suggested, the association of PD-L1 with overall survival was not statistically significant (P = .2; Fig. 4).

thumbnail image

Figure 4. Patient overall survival did not correlate with PD-L1 expression. Patients were grouped based on the presence (= ever, >1% of tumor cells staining PD-L1 positive) or the absence (= never) of PD-L1 expression at any point during disease progression. Numbers at the bottom of the graph represent remaining patients.

Download figure to PowerPoint

TIL Infiltration, Treg Infiltration, TGF-β, MHC-I Expression, and Overall Survival

Aside from PD-L1 expression, we analyzed additional immunological markers thought to correlate with a better prognosis (TIL infiltration) or worse prognosis (Treg infiltration, TGF-β expression, and MHC loss) in our melanoma patient cohort. None of these immunological parameters showed any association with patient outcome in our sample (Fig. 5).

thumbnail image

Figure 5. Patient overall survival did not correlate with immunological parameters such as MHC loss (a), TGF-β expression (b), TIL infiltration, (c) and presence of infiltrating regulatory T cells (d), but it correlated with Breslow thickness (e) and sex (f). Ever MHC loss corresponds to patients for whom >90% MHC loss was observed in at least 1 stage throughout disease progression. Ever TGF-β corresponds to the detection of >10% positive tumor cells in at least 1 stage of disease progression. Ever TIL defines patient samples for which at least 5% T-cell infiltration was detected in at least 1 stage of disease progression. Never Treg corresponds to the grouping of never TIL and ever TIL with less than 10% FoxP3 staining among total T cells at any given stage of disease progression.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

PD-L1 (B7-H1) is expressed in various tumor types, and its interaction with PD-1 has been shown to inhibit T-cell cytokine production and proliferation.15, 33 PD-L1/PD-1 blockade has also been associated with improved tumor growth (reviewed in Keir et al15 and Blank et al33). Because of its role in tumor development, the prognostic value of PD-L1 has been studied in several instances and found to be associated with poor prognosis in non–small cell lung cancer, gastric carcinoma, and renal cell carcinoma.25, 26, 35 Despite several reports of PD-L1 expression in melanoma,21-23, 40 its prognostic relevance in melanoma patients has been suggested only recently.41 In line with this recent study, the current work addressed whether, in the melanoma setting, PD-L1 expression increases during disease progression and whether its expression correlates with other immunological factors thought to mediate tumor immune escape based on preclinical animal models.5

First, a large effort was invested to obtain reliable PD-L1 staining in paraffin-embedded material, because in our hands the antibody that has been claimed to stain best in paraffin, PD-L1,28, 29 gave high background staining. A broad array of commercially available mAbs, as well as mAbs provided by the University of Vienna (Vienna, Austria) and clones provided by Medarex (Princeton, NJ), were tested for their staining patterns. Only 1 antibody gave no background staining (compared with its isotype control) and was competitively blocked by the addition of PD-L1Fc protein (polyclonal anti-PD-L1 Ab, #4059; ProSci, Poway, CA) in 2 PD-L1-expressing tissues: tonsil and placenta (Fig. 1, Table 2).

Our study cohort comprised a poor-prognosis population of malignant melanoma patients (stages III-IV) that had been referred to our center between 2000 and 2004. Contrary to Hino et al41 and in line with our previous data from frozen material,23 PD-L1 expression was observed in only a minority of primary tumors, in contrast with PD-L1 expression in metastases. Interestingly, a higher percentage of PD-L1-expressing metastases was observed in satellite and in-transit metastases, compared with lymph node and distant organ metastases, suggesting a preferential kinetic of PD-L1 expression during disease progression. However, the lack of sufficient patients available for complete longitudinal analysis did not permit the defining of a reliable pattern of PD-L1 expression (Fig. 3). Yet primary tumors consistently expressed less PD-L1 than did metastases, and PD-L1 expression was variable during disease progression, which is an important factor to take into account during future clinical studies.

The prognostic value of PD-L1 in the current study cohort was analyzed. Our observations did not support PD-L1 expression being a poor prognostic factor for malignant melanoma.41 In fact, the tumors expressing PD-L1 in our series showed an insignificant trend toward better prognosis (Fig. 4).

Such discrepancies concerning the prognostic value of PD-L1 in different studies may result from the use of different staining protocols or paraffin for embedding. Also, it is unclear whether other groups performing such studies have tested their antibodies against isotype controls and PD-L1Fc protein blockade to ensure lack of background or unspecific signals are not positive signal interpreted.

Another likely explanation for the conflicting results between the recently published data41 and our data may be that small series of fewer than 100 melanoma patients from diverse origin were analyzed. The most noticeable difference between the Asian melanoma patient population and the 2 European cohorts is tumor type. In our cohorts, superficial and nodular melanomas were predominantly diagnosed, whereas the cohort of Hino et al mainly consisted of patients with acral lentigious melanomas. The latter have been shown to preferentially express c-kit amplifications or mutations, whereas the melanomas from the European cohort are likely to more often express BRAF mutations.42, 43 Moreover, our cohort contained a majority of women, in whom PD-L1 was detected more often (63% versus 38%) than in the male counterpart. Women are known to have a better prognosis in melanoma44 and may have thus influenced the prognostic analysis of PD-L1. Whether melanoma in female patients indeed expresses PD-L1 more often deserves further investigation. Although PD-L1/PD-1 interaction has been shown to play a role in tumor development,21, 45, 46 other immunological phenomena could explain the lack of prognostic value of PD-L1. PD-L1 is not constitutively high in antigen-presenting or tumor cells, but it is up-regulated on exposure to interferon.17, 45 Thus, our observation of PD-L1 expression in the tumors of patients with better prognosis could indicate a preexisting tumor inflammatory environment as a consequence of IFNγ-producing tumor infiltrating lymphocytes (TILs). Although the prognostic value of TIL infiltration in melanoma is still under debate because of contradictory results,47 the results of a recent study analyzing a group of 294 patients argues in favor of a correlation between the presence of TILs and improved survival.48 Finally, although preclinical models show that PD-1/PD-L1 blockade improves tumor control,21, 45, 46 recent data indicate that PD-L1-mediated inhibition can occur not only during tumor-effecter T-cell interaction, but also during APC–T-cell interaction and regulatory T-cell expansion. In addition, PD-L1 is found in monocytes and myeloid-derived suppressor cells.15, 49, 50 Therefore, the positive effect of PD-L1 blockade on tumor growth control could evolve from alterations in compartments other than that of the tumor cell.

Although the well-established clinical parameters of Breslow thickness and sex did significantly correlate with overall survival (P = .00343 and P = .00095, respectively; Fig. 5e,f), in the multivariate analysis including sex, Breslow thickness, and PD-L1 expression, we found no significant difference concerning PD-L1 expression and overall survival (hazard ratio, 0.72; P = .45). When studying other immunological parameters, we found that TIL infiltration, TGF-β expression, regulatory T-cell infiltration, and MHC down-regulation did not correlate with patient prognosis, possibly because of small cohort size (Fig. 5a-d).

In summary, no significant association of PD-L1 expression with outcome in melanoma patients was found. In future studies, the time of PD-L1 expression assessment should be carefully considered, and frozen material should be the primary source of data (unless new antibodies performing better in paraffin become available). While these conditions are not met, one should be careful about concluding that PD-L1 has negative clinical prognostic value in melanoma

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

We thank Alan Korman (Medarex, Princeton, NJ) and Otto Madjic (University of Vienna, Germany) for providing various anti-PD-L1 antibodies.

CONFLICT OF INTEREST DISCLOSURES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

Supported by the Dutch Cancer Society (KWF funds NKI2008γ3988).

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES