If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
UNC Lineberger Comprehensive Cancer Center, Chapel Hill, North CarolinaDepartment of Medicine, University of North Carolina at Chapel Hill, Chapel Hill, North CarolinaDepartment of Genetics, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina
Lymphoepithelioma-like carcinoma of the bladder (LELC-B) is a rare subtype of urothelial carcinoma consisting of undifferentiated epithelial cells within a dense inflammatory cell infiltrate. We set out to molecularly characterize LELC-B through RNA expression profiling as well as immunohistochemistry (IHC) to understand its underlying biology. Sixteen cases of LELC-B were identified at Johns Hopkins University. RNA sequencing was performed on 14 cases. IHC staining for programmed cell death ligand 1 (PD-L1) and mismatch repair proteins MutL homolog 1 (MLH1), MutS homolog 2 (MSH2), MSH6, and PMS1 homolog, mismatch repair system component 2 (PMS2) was performed. Transcriptomic profiling of LELC-B showed that they are enriched in a basal-like phenotype, with 12 of 14 LELC-B cases correlating to the basal centroid of the bladder cancer analysis of subtypes by gene expression 47 (BASE47) predictive analysis of microarrays (PAM) classifier. Gene signature analysis confirmed the lymphocyte infiltration profile consistent with the histomorphology. LELC-B lacked features to explain the robust lymphocytic infiltrate, such as loss of mismatch repair protein expression or expression of Epstein-Barr virus transcripts. Nonetheless, PD-L1 IHC was positive in 93% of LELC cases. Our study demonstrates that LELC-B tumors are enriched in a basal-like molecular subtype and share a high level of immune infiltration and PD-L1 expression, similar to basal tumors. The basal-like phenotype is consistent with the known sensitivity of LELC-B to chemotherapy and suggests that immune checkpoint therapy should be explored in this rare disease.
Lymphoepithelioma-like carcinoma of the bladder (LELC-B) is a rare variant of urothelial carcinoma composed of sheets or nests of undifferentiated cells with ill-defined cytoplasmic borders and large pleomorphic nuclei with prominent nucleoli. The tumor cells are present within a dense infiltrate of predominantly lymphocytes, plasma cells, and other inflammatory cells.
The histologic appearance of LELC-B closely resembles lymphoepithelioma of the nasopharynx, which is defined by both its histologic appearance and presence of Epstein-Barr virus (EBV) positivity. In bladder, LELC was first shown to be EBV negative via absence of EBV staining by in situ hybridization to EBV-encoded RNA (EBER1).
has shown better overall survival compared with tumors displaying <50% LELC, but the mechanism for this finding is not yet understood. On the basis of a systematic review of LELC-B in the literature, Yoshino et al
reported an overall cause-specific survival rate of 83% for LELC-B, with 97% and 95% survival in pure and predominant patterns, respectively, compared with 15% survival in those patients with only focal LELC-B pattern.
Another histologic analog of LELC-B is the medullary phenotype of colorectal carcinoma. These colon tumors show minimal to no glandular growth and are composed of sheets and nests of epithelial cells that have large, pleomorphic nuclei and prominent nucleoli, embedded within a dense lymphocytic infiltrate.
The expression of PD-L1 on tumor cells interacts with its ligand to inhibit T-cell activation. Recently, PD-L1 expression has been the focus of studies evaluating its use as a marker of response on tumor cells and tumor-infiltrating lymphocytes. Consequently, it is believed that the overexpression of PD-L1 leads to worse outcomes in numerous tumor types, including urothelial carcinoma.
The molecular and immunohistochemical (IHC) profiles of urothelial cancers with variant histology are beginning to yield important results. Micropapillary urothelial carcinoma is a particularly aggressive variant and has been assigned a luminal subtype, has unique miRNA expression patterns, and demonstrates high rates of ERBB2 (HER2) amplification and activating mutation.
This loss is thought to explain the tumor's tendency for aggressive spread beyond fascial planes. Partly because of the rarity of LELC-B, molecular characterization of LELC-B has not been explored. There are potential treatment implications to understanding the molecular subtypes of these rare variants, given that large-scale clinical trials are not feasible in this population.
Our aim was to profile LELC-B via immunohistochemistry and RNA sequencing. We investigated the immunohistochemical expression of PD-L1 and MMR proteins MLH1, MSH2, MSH6, and PMS2, as well as use PAM classifiers to classify LELC-B into the BASE47 and University of North Carolina at Chapel Hill (Chapel Hill, NC) subtypes of bladder cancer. We hypothesized that LELC-B shows enrichment in markers of the basal molecular subtype of bladder cancer because of the sensitivity to chemotherapy seen in LELC-B. Immune gene signature analyses were done to address the high immune infiltration in LELC-B.
Materials and Methods
LELC-B Case Selection
The study was performed with approval of the Institutional Review Boards at both University of North Carolina at Chapel Hill and Johns Hopkins (Baltimore, MD). Cases were identified by a search of the Johns Hopkins Pathology Lab Information System for cases containing the terms lymphoepithelioma-like carcinoma and bladder from January 1, 2000, to July 25, 2016. Inclusion criteria included patients with a surgical resection specimen from the bladder or transurethral resection of bladder tumor accessioned at Johns Hopkins, with a final diagnosis of LELC-B. Cases with only focal LELC-B were excluded from the study. LELC-B tumors were classified as pure or predominant using the cutoff of >50% for predominant pattern LELC-B and 100% LELC-B for pure pattern.
Initial hematoxylin and eosin–stained slides (Figure 4A) were reviewed for tumor adequacy and stained with PD-L1 IHC (Cell Signaling, Boston, MA; E1L3N) and MMR proteins (MLH1, MSH2, MSH6, and PMS2; Cell Marque, Rocklin CA) in a single batch. Additional immunohistochemical stains [cytokeratin 5/6 (CK5/6), CK14, forkhead box A 1 (FOXA1), erythroid transcription factor 3 (GATA3), and uroplakins 2 (UPK2)] were performed by restaining the prior slides used for MMR and one hematoxylin and eosin–stained slide. The IHC staining for these five stains was interpreted independently by two pathologists (S.M.J. and S.E.W.) and any cases not reaching consensus were reviewed together to resolve discrepancies. CK5/6, CK14, FOXA1, GATA3, and UPK2 were given both a stain intensity and percentage of tumor cells staining positively. To develop the modified Allred score, these stains were given a score of 0 to 6 for percentage-positive tumor cells (0 indicates 0%; 1, 1% to 4%; 2, 5% to 20%; 3, 21% to 40%; 4, 41% to 75%; 5, 76% to 99%; and 6, 100%) and 0 to 3 for stain intensity (0 indicates none; 1, weak; 2, moderate; and 3, strong). Scores were added for a maximum of 9. The H-score was calculated by multiplying the stain intensity by the continuous percentage-positive tumor cells (possible range, 0 to 300)
(Supplemental Table S1). For MMR IHC, cases were designated as retained expression (any tumor nuclei staining positive at any intensity) or lost expression (negative in all tumor nuclei). For PD-L1, cases were interpreted using the tumor proportion score and combined positive score (CPS) method. CPS was developed for solid tumors stained with the PD-L1 22C3 clone as a way to capture both tumor cell and immune cell staining in one number, calculated as the number of PD-L1–positive cells (tumor cells, lymphocytes, and macrophages) divided by the total number of viable tumor cells, then multiplied by 100.
Although the CPS system is not formally validated for PD-L1 clones other than 22C3, it was used to express nontumor cell expression of PD-L1 in these cases.
RNA Extraction and Data Set Processing
Sixteen cases were identified with formalin-fixed, paraffin-embedded tissue available for study. Fourteen cases had sufficient RNA yield after extraction. RNA was extracted with the Roche High Pure RNA Paraffin kit protocol (Roche Diagnostics, Indianapolis, IN). Fourteen samples generated adequate yield for the library preparation with the Illumina TruSeq RNA Access Library Prep Kit (Illumina, San Diego, CA) and were sequenced on an Illumina NextSeq500. FASTQ files from all cases in The Cancer Genome Atlas (TCGA) bladder cancer (TCGA BLCA) data set were obtained from database of Genotypes and Phenotypes (dbGap; https://www.ncbi.nlm.nih.gov/gap). TCGA BLCA along with 14 LELC-B samples were analyzed in R version 3.5.2
The data were filtered for genes that had expression in at least 80% of the samples (n = 422). The raw expression data were normalized to account for sequencing depth differences via size factor normalization and were log transformed via the DESeq2 package (1.20.0).
The merged data set contained gene expression data from LELC-B (n = 14) and TCGA BLCA (n = 408) aligned to the hg38 genome. Principal component analysis was conducted to check for outliers (Supplemental Figure S2). All data were visualized via the use of ComplexHeatmap (1.20.0) and ggplot2 (3.1.0) vignettes.
Significance was tested by Wilcoxon-Mann-Whitney rank sum method as it does not assume a parametric distribution. Further characterization of LELC-B was performed via a comparison to Urothelial Carcinoma via Gene Set Enrichment Analysis and the MSigDB Hallmark Gene Set Collection
(Supplemental Figure S3B and Supplemental Table S3). Just for the Gene Set Enrichment Analysis, three cases were removed from TCGA BLCA data set (one case each for adenomas/adenocarcinomas, epithelial neoplasm not otherwise specified, and squamous cell neoplasm) to yield a data set of n = 405 urothelial carcinoma samples.
Detection of Epstein-Barr Virus
Epstein-Barr virus (human herpesvirus 4/EBV) was detected using VirDetect.
All reads in the sample FASTQ file that did not align to the hg38 genome were aligned to the viral genome collection. Our LELC samples were assessed for wild-type EBV (accession identifier: NC_007605.1) and EBV type 2 (accession identifier: NC_009334.1) along with cell lines with validated EBV presence.
The alignments of the tumor sequence to the viral genomes were verified with human endogenous retrovirus as a positive control (accession identifier: NC_022518.1). All sequences used can be referenced via the accession numbers through the National Center for Biotechnology Information website (https://www.ncbi.nlm.nih.gov/nuccore).
LELC-B Is Enriched in Markers Corresponding to Basal-Like Bladder Cancer
High-quality RNA was obtained from 14 pure or predominant LELC-B tumors, most of which initially presented with muscle invasion (Table 1). Examination of the LELC-B tumors by BASE47 showed that most of LELC-B has a significant correlation toward the basal centroid, with 12 LELC-B tumors classified as basal-like and 2 LELC-B tumors being classified as luminal-like bladder cancer (Fisher exact P = 0.0117) (Figure 1A and Supplemental Table S2). Eight of 10 cases of pure LELC-B and four of four predominant cases correlated to the basal centroid. The PAM classification results were confirmed by unsupervised clustering on the genes that highlight the subtypes within muscle invasive bladder cancer (Figure 2A).
None of the LELC-B basal tumors were characterized as the claudin-low subtype (Figure 1B). Therefore, although LELC-B tumors exhibit a basal phenotype, they are not claudin-low. The hallmark genes that are representative of bladder cancer subtypes, such as GATA3 and FOXA1, confirm that LELC-B is significantly different from the luminal and claudin-low profile (Figure 2B) and most in keeping with a basal-like subtype.
LELC-B Phenocopies Basal Bladder Cancer via Immune Gene Signatures
The expression of a panel of immune gene signatures that have previously been shown to be representative of differing components of the immune system was examined next.
Representative Z-scores of immune gene signatures in a heat map supervised by University of North Carolina at Chapel Hill subtype and rank ordered within each subtype were used (Supplemental Figure S3A).
Although the LELC-B tumors overall appeared to have heterogeneous expression levels of these immune gene signatures, in general, LELC-B tumors showed a profile of active immune infiltration. To verify this, the expression of T-cell, B-cell, and an immunosuppression signature was specifically examined. In all three signatures, LELC-B tumors showed gene signature levels similar to basal urothelial carcinoma tumors, yet significantly higher and lower than luminal and claudin-low tumors, respectively (Figure 3). These findings are consistent with the notion that LELC-B tumors have RNA expression patterns similar to basal-like bladder tumors.
LELC-B Has High Tumor Cell PD-L1 Expression
To further classify the immune landscape of LELC-B, immunohistochemistry was performed for PD-L1 (Figure 4B). In 15 of 16 total cases (one case was excluded because of inadequate tissue), 14 (93%) were determined to have positive tumor cell staining for PD-L1 (nine demonstrating high expression and five demonstrating low expression) (Table 2). The five cases with low expression were all pure LELC-B, whereas the nine cases with high expression included seven pure and two predominant types. The single PD-L1–negative case was a predominant LELC-B. There was no significant association between pure or predominant LELC-B cases and PD-L1 IHC status via Fisher exact test (P = 0.505). According to the CPS interpretation, all 14 LELC-B tumors that expressed PD-L1 scored ≥10, which is positive overall. The IHC images demonstrate PD-L1 expression in LELC-B tumor cells, with lack of expression in the infiltrating lymphocytes (Figure 4B).
Table 2Summary of LELC-B Immunohistochemistry Results
PD-L1 staining (IHC), n (%)
MMR deficiency (IHC), n (%)
MutL homolog 1 (MLH1)
MutS homolog 2 (MSH2)
MutS homolog 6 (MSH6)
PMS1 homolog 2 (PMS2)
IHC, immunohistochemistry; LELC-B, lymphoepithelioma-like carcinoma of the bladder; MMR, mismatch repair; PD-L1, programmed cell death ligand 1.
All 16 cases of LELC-B uniformly retained MMR protein expression, the only exception being the loss of MSH6 in a single case of a pure pattern LELC-B tumor. Figure 4, C–F, depicts photomicrographs demonstrating representative cases for each IHC marker: MLH1, MSH2, PMS2, and MSH6. There were no significant associations between MMR intact status and PD-L1 expression. Therefore, although MMR deficiency may explain the robust immune infiltrate in medullary colorectal carcinoma, LELC-B does not appear to be MMR deficient, at least when assessed by immunohistochemistry.
LELC-B Does Not Associate with Epstein-Barr Viral Co-Infection
A final potential explanation for the robust immune infiltrate associated with LELC-B is the expression of a pathogenic virus (ie, EBV). Although previous studies have shown that LELC-B tumors do not express EBV by EBER staining, this question was asked in an orthogonal manner by querying the RNA-sequencing data for expression of the EBV genome.
To examine EBV presence, reads were aligned to a masked and curated database of vertebrate genomes from GenBank (Materials and Methods). Alignment to human herpesvirus 4 genome (EBV) sequence showed a lack of EBV transcripts in LELC-B sequencing reads (Table 3). As expected, counts aligning to human endogenous retrovirus K113 were present in all samples tested, validating its use as a positive control. RNA expression from known EBV expressing cell lines was included to confirm that this method detects EBV transcripts (Table 3). These results are consistent with prior studies that used in situ hybridization to EBER1 to detect EBV presence and support in an orthogonal manner that LELC-B tumors do not harbor EBV.
Table 3Viral Genome Quantification through STAR, including EBV (HHV-4)
Many of these variants have distinct natural histories, with micropapillary, plasmacytoid, and sarcomatoid variants associated with poorer outcomes. To our knowledge, this is the first study investigating the molecular characterization and immune profiling of LELC of the bladder.
Molecular characterization of high-grade bladder cancers, specifically intrinsic molecular basal and luminal phenotypes, is increasingly recognized as prognostic and predictive biomarkers.
Finally, because of differential immune infiltration, the subtypes might also be primed for immune checkpoint inhibitor response, although this has yet to be confirmed. On the basis of these observations and others, some groups have molecularly subtyped both variant and divergent histologic subtypes of bladder cancer in hopes of better understanding the underlying biology and potential implications for treatment choice.
Using the BASE47 gene set predictor, LELC-B cases show a basal-like phenotype. The basal subtype is so termed because of its increased expression of basal keratin markers (KRT5, KRT6, and KRT14), leading to a more basaloid/stem cell phenotype.
LELC-B is historically thought to be exquisitely sensitive to chemotherapy and, although merely correlative, the enrichment of LELC-B in the basal subtype is at least consistent with this chemoresponsive phenotype.
Hematoxylin and eosin stains of LELC demonstrate high lymphocytic infiltrate, and previous work has shown that the lymphocytes are predominantly CD8+ and CD4+ T cells.
In our studies, the immune gene signatures were also broadly up-regulated in LELC-B and had levels comparable to the basal-like urothelial carcinoma. Investigation of mechanisms underlying this robust lymphocytic infiltrate suggested no evidence for loss of MMR proteins or the presence of active EBV transcription. Unfortunately, tumor mutational burden and neoantigen load could not be assessed as DNA sequencing could not be performed on the LELC-B samples. Therefore, although these studies confirm the presence of an immune infiltrate within the tumor microenvironment, they do not elucidate the underlying drivers of this phenotype or define the relative clonality of the T cell. Nonetheless, it is tempting to speculate that the lymphocytic infiltrate may represent a robust, but exhausted, antitumor immune response that could be reinvigorated through immune checkpoint therapy.
Immune checkpoint inhibitors are being investigated as second-line therapy after platinum-based chemotherapy in metastatic bladder cancer. In the IMvigor210 clinical trial, atezolizumab, an anti–PD-L1 antibody demonstrated a correlation between PD-L1 expression on tumor-infiltrating lymphocytes with longer overall survival.
According to the PD-L1 IHC results, immune cells in LELC-B do not stain for PD-L1 (Figure 4B). However, the CheckMate275 trial showed that nivolumab, an antibody to programmed cell death protein 1 (PD-1), led to response regardless of high or low PD-L1 CPS scores in metastatic urothelial carcinoma.
With immune checkpoint blockade becoming a prominent second-line therapy, further research is needed to investigate the dense tumor lymphocytic infiltrate seen histologically in LELC-B and any associations with better prognosis of pure and predominant LELC-B.
The limitations of our study are inherent in the use of formalin-fixed, paraffin-embedded archival tissue for RNA analysis. Biases may be present in data as RNA extracted from formalin-fixed, paraffin-embedded slides has lower yields and integrity than flash frozen tissue RNA. The data set processing removes any technical batch effects and minimizes bias. In any future studies, immune cell PD-L1 levels specifically should be investigated to determine whether LELC-B could be a candidate for immune checkpoint inhibitors. Further explanation for the robust immune infiltration should also be examined, such as the somatic mutational load or neoantigen burden. Additional investigation of a larger series of LELC-B cases with clinicopathologic outcomes should be attempted to confirm this study's findings and expand on the results.
In summary, the immune infiltration and immunosuppression profile in LELC is consistent with the low cause-specific mortality rate; however, the immune gene signature expression was lower than expected. With Food and Drug Administration approval of immune checkpoint inhibitors for bladder cancer, the knowledge of the immune profile of the tumor is increasingly pertinent in treatment decision making.
The robust lymphocytic infiltrate in LELC may be predictive as immune checkpoint blockade therapies gain Food and Drug Administration approval as second-line therapies for muscle invasive bladder cancers.
Ongoing studies include RNA sequencing and further profiling of LELC-B tumors.
We thank the members of the Kim Laboratory for continued help and discussion throughout the project; Yongjuan Xia (University of North Carolina at Chapel Hill Translational Pathology Laboratory) for expert technical assistance; and the Department of Pathology at Johns Hopkins Hospitals for providing the formalin-fixed, paraffin-embedded tumor set.
S.E.W. designed the study, performed literature search, collected data, and wrote the manuscript; U.M. generated and analyzed data, performed literature search, generated the figures, and wrote the manuscript; J.K. interpreted and analyzed the data; M.Z. analyzed the data; S.J. generated and analyzed data and reviewed the manuscript; C.B. generated and analyzed data; S.S. generated and analyzed data and reviewed the manuscript; W.Y.K. designed the study and reviewed the manuscript.
A: Processing to generate a normalized transformed data set (NTD) of lymphoepithelioma-like carcinoma of the bladder (LELC-B) and The Cancer Genome Atlas bladder cancer (TCGA BLCA) raw gene expression data. B: Boxplot of all gene expression of LELC and TCGA BLCA merged data set before (top panel) and after (bottom panel) batch effect correction (BEC) via ComBat. n = 14 (A, LELC-B); n = 408 (A, TCGA BLCA). *Normal tissue samples have been filtered out.
Principal component analysis (PCA) shows lymphoepithelioma-like carcinoma of the bladder (LELC-B) grouping with the basal and luminal subtypes of bladder cancer. BLCA, bladder cancer; TCGA, The Cancer Genome Atlas; UNC, University of North Carolina at Chapel Hill.
A: Supervised heat map of the normalized transformed data set across previously identified gene signatures detailing the immune composition of lymphoepithelioma-like carcinoma of the bladder (LELC-B) along with consensus bladder cancer subtypes.
Average linkage was used within each University of North Carolina at Chapel Hill (UNC) subtype for hierarchical clustering. B: Gene Set Enrichment Analysis (GSEA) normalized enrichment scores (NESs) for Hallmark gene sets enriched in urothelial carcinoma (UC) compared with LELC-B, filtered for false discovery rate (FDR) < 0.05. n = 405 (B, urothelial carcinoma); n = 14 (B, LELC-B). NK, natural killer; TCGA, The Cancer Genome Atlas.
Supported by the University Cancer Research Fund (W.Y.K and S.E.W.) and in part by National Cancer Institute grant 5P30CA016086-42 , NIH grant U54-CA156733 , National Institute of Environmental Health Sciences grant 5 P30 ES010126-17 , the University Cancer Research Fund , and North Carolina Biotechnology Center grant 2015-IDG-1007 .