A novel gene expression scoring system for accurate diagnosis of basaloid squamous cell carcinoma of the esophagus

  • Authors:
    • Takeshi Tada
    • Reiko Honma
    • Jun-Ichi Imai
    • Zenichiro Saze
    • Michihiko Kogure
    • Shigeru Marubashi
    • Kazuhiro Tasaki
    • Masamitu Unakami
    • Junji Ezaki
    • Hirosumi Tamura
    • Akira Nishikawa
    • Yuko Hashimoto
    • Satoshi Waguri
    • Shinya Watanabe
    • Mitsukazu Gotoh
  • View Affiliations

  • Published online on: July 19, 2017     https://doi.org/10.3892/ijo.2017.4075
  • Pages: 877-886
Metrics: Total Views: 0 (Spandidos Publications: | PMC Statistics: )
Total PDF Downloads: 0 (Spandidos Publications: | PMC Statistics: )


Abstract

Basaloid squamous cell carcinoma of the esophagus (BSCE) is a rare variant of squamous cell carcinoma that is difficult to distinguish from other carcinomas by preoperative endoscopic biopsy because of its histological varieties. Accurate diagnosis is essential for adequate treatment, and the methods proposed so far (e.g., immunohistochemical staining) have limitations. In this study, we tried to identify the characteristic bundles of gene expression in BSCE using comprehensive gene expression analysis (CGEA). Subsequently, we constructed a gene expression scoring system for the proper diagnosis of BSCE. Fifty-seven surgical specimens, including seven BSCEs, obtained from 30 patients who underwent esophagectomy were used for constructing the scoring system. Three hundred and twelve biopsy specimens, including eight BSCEs, obtained from 80 patients and 20 commercially available formalin-fixed paraffin-embedded (FFPE) specimens diagnosed as esophageal cancer, including 13 BSCEs, were used for validation. After our original mathematical extraction algorithm, 75 genes were extracted to distinguish BSCE from non-BSCE. The cumulative converted values (gene expression score) of the respective 75 genes from each specimen were obtained and lined up in ascending order to assess the optimal gene expression cut-off score for a definitive diagnosis of BSCE. The validation of this scoring system showed high prediction of the biopsy specimens [area under the curve (AUC)=0.981; 95% confidence interval (CI): 0.952‑1.000] and the commercially available FFPE specimens (AUC=0.901; 95% CI: 0.750-1.000). In conclusion, using CGEA in a gene expression scoring system helps in differentiating BSCE from non-BSCE with high accuracy and may contribute in improving BSCE treatment.

Introduction

Basaloid squamous cell carcinoma (BSC) was first reported by Wain et al (1) to occur in the head and neck region. BSC may occur in various other sites, including esophagus (2), lung (3), anus (4), uterine cervix (5), penis (6), and urinary bladder (7). BSC of the esophagus (BSCE) is a rare and uncommon variant of squamous cell carcinoma (SCC), with a reported incidence ranging from 1.0 to 8.7% (816) in Japan and 0.4 to 11.3% (2,1724) in other countries.

BSCE has six typical components on histology: solid nest with central necrosis, cribriform pattern, ductal differentiation, microcyst and/or trabecular nests, hyaline-like material deposition, and coexistence of SCC components (16). Because of these histological varieties, BSCE is difficult to distinguish from adenoid cystic carcinoma, small cell carcinoma, poorly differentiated SCC, or adenosquamous carcinoma (2,23,25,26). Furthermore, it is even more difficult to diagnose BSCE based on the histological examination of endoscopic biopsy specimens, with a low diagnostic accuracy of only 0–10% (23,24,27). BSCE has been frequently diagnosed as SCC on endoscopic biopsy specimens, probably because of the fact that BSCE frequently presents as a submucosal tumor-like structure covered with normal epithelium or SCC (14,15). Therefore, sampling multiple and deeper sites is recommended for diagnosis by endoscopic biopsy (24,27). Some diagnostic approaches using immunoreactivity (2,9,12,16,18,19) or polymerase chain reaction (PCR) (28,29) have been reported, but none of these showed high specificity for BSCE.

The prognosis of BSCE is still controversial. Some studies stated no significant difference between BSCE and SCC (2,20), whereas others stated poorer prognosis of BSCE than that of SCC (16,17,23). Some authors specified that BSCE shows a poor degree of differentiation, high proliferative activity (2), aggressive biological behavior (16), high telomerase activity (20), and a worse prognosis than that in SCC in advanced cases (30). Meanwhile, other authors mentioned that the treatment for BSCE is similar to that for SCC of the esophagus (15,23). The rarity of and difficulty in the proper diagnosis of BSCE (15) may be responsible for this diversity. Therefore, proper diagnosis is mandatory for analyzing the outcome and determining the suitable treatment for this disease entity.

We have previously reported the use of comprehensive gene expression analysis (CGEA) to identify some disease-specific genes (31,32). The present study aimed to improve the diagnostic accuracy for BSCE by attempting to extract the genes expressed in it. From CGEA of esophagectomy specimens, we constructed, verified, and evaluated a gene expression scoring system for the proper diagnosis of BSCE.

Materials and methods

Patient selection

We initially enrolled all 113 esophageal cancer patients who underwent esophagectomy and/or endoscopic biopsy at Fukushima Medical University Hospital from January 2008 to July 2015. Among these patients, 14 patients (1 in stage 0, 4 in stage II, 4 in stage III, 4 in stage IVa, 1 in stage IVb and 1 in unknown stage) were not followed in our department, and one surviving patient (stage II) denied to participate in research. These cases were excluded.

Ethics statement

This study was approved by the ethics committee of Fukushima Medical University (approval no. 1953). Written informed consent was obtained from 98 patients.

Commercially available esophageal specimens

From US Biomax Inc. (Rockville, MD, USA), we purchased 20 formalin-fixed paraffin-embedded (FFPE) specimens that were diagnosed to have BSCE components. These FFPE specimens were reviewed by three pathologists before inclusion.

Specimen sampling

Small fractions (7×7 mm) of the cancerous site and normal mucosa were removed from each surgical specimen and were immediately frozen in liquid nitrogen before performing CGEA. Residual tissue specimens were fixed in formalin and then embedded in paraffin before pathological examination.

For the biopsy specimens, tiny fractions (3×3 mm) of the esophageal epithelium, including cancerous and normal sites, were obtained endoscopically; they were immediately frozen separately in liquid nitrogen before performing CGEA. Another specimen from near the first biopsy site was obtained endoscopically; it was fixed in formalin and embedded in paraffin before pathological examination. We made an effort to sample multiple and deeper sites by endoscopic biopsy. We ascertained that the frozen specimens for CGEA and the FFPE specimens for pathological examination had identifiable pathological features.

Pathological review

The surgical, endoscopic biopsy and commercially available FFPE specimens were stained with hematoxylin and eosin and were reviewed by three pathologists (Fig. 1). BSCE was defined using the criteria described by Wain et al (1); the six component histological features reported by Imamhasan et al (16) were also evaluated. In this study, tumors that contained some BSC components within SCC were categorized as BSCE.

Comprehensive gene expression analysis

Frozen specimens were processed for total RNA extraction using Isogen (Nippon Gene Co., Ltd., Tokyo, Japan) and for poly(A)+RNA purification using MicroPoly(A) Purist kit (Ambion, Austin, TX, USA). Commercially available FFPE specimens were processed for total RNA extraction using ISOGEN PB kit (Nippon Gene Co., Ltd.). The human common reference RNA was prepared by mixing equal amounts of total RNA and poly(A)+RNA, which were extracted from 22 human cancer cell lines (A431, A549, AKI, HBL-100, HeLa, HepG2, HL60, IMR-32, Jurket, K562, KP4, MKN7, NK-92, Raji, RD, Saos-2, SK-N-MC, SW-13, T24, U251, U937, and Y79).

The DNA microarray that used poly(A)+RNA was named system 1; a set of synthetic polynucleotides (80-mers) representing 31,797 species of human transcript sequences was printed on a glass slide using a custom arrayer. The DNA microarray that used total RNA was named system 2; a set of synthetic polynucleotides (80-mers) representing 14,400 species of human transcript sequences was printed on a glass slide using a custom arrayer. For RNA of the samples, SuperScript II (Invitrogen Life Technologies, Carlsbad, CA, USA) and Cyanine 5-dUTP (Perkin-Elmer Inc., Boston, MA, USA) were used to synthesize labeled cDNA from 2 µg of poly(A)+RNA in system 1 and 5 µg of total RNA in system 2. Using the same method for the reference RNA, Cyanine 3-dUTP (Perkin-Elmer Inc.) was used to synthesize labeled cDNA from 2 µg of poly(A)+RNA in system 1 and 5 µg of total RNA in system 2.

Hybridization was performed with a Labeling and Hybridization kit (MicroDiagnostic, Tokyo, Japan). Signals were measured using a GenePix 4000B Scanner (Axon Instruments, Inc., Union City, CA, USA) and then processed into the primary expression ratios of the cyanine 5 intensity of each specimen to the cyanine 3 intensity of the human common reference RNA. Each ratio was normalized using GenePix Pro 3.0 software (Axon Instruments, Inc.). The primary expression ratios were converted into log2 values, which were designated as log ratios or converted value. Data were processed using Microsoft Excel software (Microsoft, Bellevue, WA, USA) and MDI gene expression analysis software package (MicroDiagnostic) (33).

Statistical analysis

Clustering analysis was performed using group average method with an Expression View Pro (MicroDiagnostic).

The cut-off score was determined by receiver operating characteristic (ROC) curve analysis with the aim of validating the gene scoring system. The optimal cut-off score for the definitive diagnosis of BSCE was assessed and determined by area under the ROC curve (AUC) analysis of the maximum values of sensitivity and specificity. ROC curve analysis was performed using the software program SPSS version 23 (SPSS, Inc., Chicago, IL, USA).

Refinement steps to identify candidate genes from CGEA of surgical specimens

Step 1: Genes with fluorescence intensity below the detection limit in two or more of the seven BSCE specimens were excluded. Step 2: The genes with a converted value of ≥1 in at least one of the 57 surgical specimens were selected. Step 3: The mean or average of the converted values of the chosen genes were calculated, and the genes that met the following requirement were selected: average value - converted value ≥1. Step 4: Clustering analysis was performed on the chosen genes.

Construction of gene expression scoring system for BSCE

Step 5: The mean of the converted values of the genes that were expressed in six specimens of the BSCE cluster was calculated; genes with an average value of ≥1 were selected. Step 6: Genes with fluorescence intensity below the detection limit were excluded in more than half of the non-BSCE specimens. Step 7: The standard deviation (SD) of the converted values of the genes that were expressed in the non-BSCE specimens were calculated; genes with an SD of <0.5 were selected. Step 8: Genes that met the following requirement were selected: average value of six specimens in the BSCE cluster - average value of non-BSCE specimens of ≥1. Step 9: A t-test was used to compare the average value of six specimens between the BSCE cluster and the non-BSCE specimens; genes with a P-value of <0.01 were selected. Step 10: The converted values of the selected genes from all specimens were added as gene expression scores, which were arranged in ascending order.

Study design

First, using CGEA of the surgical specimens, genes that were characteristically expressed in BSCE were identified; subsequently, a gene expression scoring system was constructed to more accurately diagnose BSCE. Second, the accuracy of our scoring system was validated using biopsy and commercially available FFPE specimens. The study design is shown in Fig. 2.

Results

Pathological diagnosis of the specimens for CGEA

Of the 98 patients, surgical specimens were obtained from 30, endoscopic biopsy specimens from 80, and both from 12. Seven cases of BSCE were included in this study, six cases underwent esophagectomy and one case underwent only endoscopic biopsy. We obtained more than one specimen from each individual, and all specimens were subjected to gene expression analysis. The number of specimens that contained enough amount of RNA for CGEA was 369 (57 surgical and 312 endoscopic) (Table I). The surgical specimens comprised 26 normal esophageal tissues, 23 SCCs, seven BSCEs, and one neuro-endocrine carcinoma (NEC). Biopsy specimens comprised 229 normal esophageal tissues, 51 SCCs, eight BSCEs, one NEC, 21 adenocarcinomas, and two intraepithelial neoplasias. Commercially available FFPE specimens comprised two SCCs, 13 BSCEs, and five NECs. All commercially available FFPE specimens were also subjected to CGEA.

Table I

Sources of specimens for CGEA.

Table I

Sources of specimens for CGEA.

HistologySurgical specimens
(N=57)
Biopsy specimens
(N=312)
Commercially available FFPE specimens
(N=20)
Normal esophageal tissue262290
SCC23512
BSCE7813
NEC115
Adenocarcinoma0210
Intraepithelial neoplasia020

[i] CGEA, comprehensive gene expression analysis; FFPE, formalin-fixed paraffin-embedded; SCC, squamous cell carcinoma; BSCE, basaloid squamous cell carcinoma of the esophagus; NEC, neuroendocrine carcinoma.

Clinicopathological characteristics

Patients with BSCE, including five men and one woman, with a mean age of 63 (range, 55–68) years, underwent esophagectomy; their clinicopathological characteristics are listed in Table II. Only three of six patients (50%) were diagnosed as having BSCE by preoperative endoscopic biopsy. The mean tumor size was 28.6 (range, 12–45) mm. Based on the seventh Union for International Cancer Control tumor-node-metastasis classification of malignant tumors, the pathological stage was stage I in four patients, stage II in one, and stage III in one. Within a mean follow-up period of 35 (range, 7–60) months, two patients died because of BSCE, one remained alive with recurrence, and three remained alive without recurrence.

Table II

Clinicopathologic features of six esophagectomy cases.

Table II

Clinicopathologic features of six esophagectomy cases.

Age, sexBiopsy diagnosisPathologic diagnosis sizeTumor (mm)Tumor typeDepth of invasion (pT)Lymph node metastasis (pN)Lymphatic invasion (ly)Venous invasion (v)UICC stagePrognosis
168, MASCBSCE42Type 2T1N001IA58 months-died (lung metastasis)
255, MSCCBSCE30Type 5T1N002IA60 months-alive
356, MSCCBSCE25Type 3T4N322IIIC19 months-died (lung metastasis)
468, FSCC (first), BSCE (second)BSCE18Type 2T1N001IA37 months-alive
564, MBSCEBSCE12Type 0-IIaT1N100IIB29 months-alive (LN metastasis)
665, MBSCEBSCE, SCC5, 45Type 0-IIcT1N000IA7 months-alive

[i] UICC, Union for International Cancer Control; ASC, adenosquamous carcinoma; SCC, squamous cell carcinoma; BSCE, basaloid squamous cell carcinoma; LN, lymph node.

Comprehensive gene expression analysis of the surgical specimens

Fig. 3 shows the result of CGEA of 57 surgical specimens, including seven BSCE specimens (one specimen from cases 1, 2, 3, 5 and three specimens from case 4); 10,027 genes were selected in step 1; 9,004 in step 2; and 7,379 in step 3.

A two-dimensional hierarchical clustering analysis of 7,379 genes yielded three different clusters: the 1) BSCE cluster, which comprised six of the seven BSCE specimens; 2) SCC cluster, which mainly comprised SCC; and 3) normal cluster, which mainly comprised normal esophageal tissue. It was possible to distinguish BSCE specimens from the others using this analysis.

Gene expression scoring system for BSCE

We selected BSCE-specific candidate maker genes and attempted to construct a gene expression scoring system for more accurate diagnosis of BSCE, and 986, 972, 243, and 100 genes were sequentially selected from steps 5, 6, 7, and 8, respectively. Finally, 75 genes were selected in step 9 (Table III) and subjected to extrapolation of the gene expression score, as described in step 10. Our gene expression scoring system, which set the cut-off score at 56.5, very clearly distinguished the seven BSCE specimens from the non-BSCE specimens (Fig. 4).

Table III

Genes characteristically expressed in BSCE based on CGEA of surgical specimens.

Table III

Genes characteristically expressed in BSCE based on CGEA of surgical specimens.

No.IDSymbolName
1NM_001033568.2RHOT1Ras homolog family member T1 (RHOT1), transcript variant 1
2NM_015690.4STK36Serine/threonine kinase 36 (STK36), transcript variant 1
3NM_000915.3OXT Oxytocin/neurophysin I prepropeptide (OXT)
4NM_001409.3MEGF6Multiple EGF-like-domains 6 (MEGF6)
5NM_021197.3WFDC1WAP four-disulfide core domain 1 (WFDC1)
6NM_020796.4SEMA6ASema domain, transmembrane domain (TM), cytoplasmic domain, (semaphorin) 6A (SEMA6A)
7NM_153213.3ARHGEF19Rho guanine nucleotide exchange factor (GEF) 19 (ARHGEF19)
8XM_005261771.3PLA2G6Phospholipase A2, group VI (cytosolic, calcium-independent) (PLA2G6), transcript variant X18
9NM_000933.3PLCB4Phospholipase C, β4 (PLCB4), transcript variant 1
10NM_023110.2FGFR1Fibroblast growth factor receptor 1 (FGFR1), transcript variant 1
11AK055081.1cDNA FLJ30519 fis, clone BRAWH2000859
12NM_000346.3SOX9SRY (sex determining region Y)-box 9 (SOX9)
13NM_025176.4NINLNinein-like (NINL)
14NM_014698.2TMEM63ATransmembrane protein 63A (TMEM63A)
15NM_020870.3SH3RF1SH3 domain containing ring finger 1 (SH3RF1)
16NM_001110514.1EBF4Early B-cell factor 4 (EBF4)
17NR_036481.2FGD5P1FYVE, RhoGEF and PH domain containing 5 pseudogene 1 (FGD5P1), non-coding RNA
18NM_005117.2FGF19Fibroblast growth factor 19 (FGF19)
19NM_032192.3PPP1R1BProtein phosphatase 1, regulatory (inhibitor) subunit 1B (PPP1R1B)
20NM_020659.3TTYH1Tweety family member 1 (TTYH1), transcript variant 1
21NM_145804.2ABTB2Ankyrin repeat and BTB (POZ) domain containing 2 (ABTB2)
22NM_194302.3CCDC108Coiled-coil domain containing 108 (CCDC108), transcript variant 1
23NM_002995.2XCL1Chemokine (C motif) ligand 1 (XCL1)
24NM_001940.3ATN1Atrophin 1 (ATN1), transcript variant 2
25AK021565.1cDNA FLJ11503 fis, clone HEMBA1002113
26NM_006941.3SOX10SRY (sex determining region Y)-box 10 (SOX10)
27NM_003222.3TFAP2CTranscription factor AP-2γ (activating enhancer-binding protein 2γ) (TFAP2C)
28NM_003963.2TM4SF5Transmembrane 4 L six family member 5 (TM4SF5)
29NM_002180.2IGHMBP2Immunoglobulin mu-binding protein 2 (IGHMBP2)
30NM_015696.4GPX7Glutathione peroxidase 7 (GPX7)
31NM_017789.4SEMA4CSema domain, immunoglobulin domain (Ig), transmembrane domain (TM) and short cytoplasmic domain, (semaphorin) 4C (SEMA4C)
32NM_178502.3DTX3Deltex 3, E3 ubiquitin ligase (DTX3), transcript variant 1
33NM_014937.3INPP5FInositol polyphosphate-5-phosphatase F (INPP5F), transcript variant 1
34NM_001380.4DOCK1Dedicator of cytokinesis 1 (DOCK1), transcript variant 2
35NM_007081.2RABL2BRAB, member of RAS oncogene family-like 2B (RABL2B), transcript variant 2
36AK055044.1TARBP1TAR (HIV-1) RNA-binding protein 1 (TARBP1)
37NM_006312.5NCOR2Nuclear receptor corepressor 2 (NCOR2), transcript variant 1
38NM_007270.4FKBP9FK506-binding protein 9, 63 kDa (FKBP9), transcript variant 1
39NM_016162.3ING4Inhibitor of growth family, member 4 (ING4), transcript variant 1
40NM_005937.3MLLT6Myeloid/lymphoid or mixed-lineage leukemia; translocated to 6 (MLLT6)
41AK021700.1cDNA FLJ11638 fis, clone HEMBA1004323
42NM_015662.2IFT172Intraflagellar transport 172 (IFT172)
43NM_032501.3ACSS1Acyl-CoA synthetase short-chain family member 1 (ACSS1), transcript variant 1
44NM_016102.3TRIM17Tripartite motif containing 17 (TRIM17), transcript variant 1
45NM_152753.3SCUBE3Signal peptide, CUB domain, EGF-like 3 (SCUBE3), transcript variant 1
46NM_133455.3EMID1EMI domain containing 1 (EMID1), transcript variant 1
47NM_014640.4TTLL4Tubulin tyrosine ligase-like family member 4 (TTLL4)
48NM_001161616.2RGL3Ral guanine nucleotide dissociation stimulator-like 3 (RGL3), transcript variant 1
49NM_024798.2SNX22Sorting nexin 22 (SNX22), transcript variant 1
50NM_032781.3PTPN5Protein tyrosine phosphatase, non-receptor type 5 (striatum-enriched) (PTPN5), transcript variant 2
51NM_005996.3TBX3T-box 3 (TBX3), transcript variant 1
52NM_000875.4IGF1RInsulin-like growth factor 1 receptor (IGF1R), transcript variant 1
53NM_178238.3PILRBPaired immunoglobin-like type 2 receptor β (PILRB)
54NM_152748.3KIAA1324LKIAA1324-like (KIAA1324L), transcript variant 1
55NM_003505.1FZD1Frizzled class receptor 1 (FZD1)
56NM_173812.4DPY19L2Dpy-19-like 2 (C. elegans) (DPY19L2)
57NM_032447.3FBN3Fibrillin 3 (FBN3)
58NM_001987.4ETV6Ets variant 6 (ETV6)
59NM_017563.3IL17RDInterleukin 17 receptor D (IL17RD)
60NM_032040.4CCDC8Coiled-coil domain containing 8 (CCDC8)
61NM_018257.2PCMTD2 Protein-L-isoaspartate (D-aspartate) O-methyltransferase domain containing 2 (PCMTD2), transcript variant 1
62NM_152730.5TBC1D32TBC1 domain family, member 32 (TBC1D32), transcript variant 1
63NM_152739.3HOXA9Homeobox A9 (HOXA9)
64NM_021156.3TMX4Thioredoxin-related transmembrane protein 4 (TMX4)
65NM_002507.3NGFRNerve growth factor receptor (NGFR)
66NM_004776.3B4GALT5UDP-Gal:betaGlcNAc β 1,4-galactosyltransferase, polypeptide 5 (B4GALT5)
67NM_015544.2TMEM98Transmembrane protein 98 (TMEM98), transcript variant 1
68NM_001852.3COL9A2Collagen, type IX, α2 (COL9A2)
69NM_005247.2FGF3Fibroblast growth factor 3 (FGF3)
70NM_002523.2NPTX2Neuronal pentraxin II (NPTX2)
71NM_001853.3COL9A3Collagen, type IX, α3 (COL9A3)
72NM_001851.4COL9A1Collagen, type IX, α1 (COL9A1), transcript variant 1
73NM_014289.3CAPN6Calpain 6 (CAPN6),
74NM_002336.2LRP6Low-density lipoprotein receptor-related protein 6 (LRP6)
75NM_001692.3ATP6V1B1ATPase, H+ transporting, lysosomal 56/58 kDa, V1 subunit B1 (ATP6V1B1)
Validation of the gene expression scoring system

Using CGEA, we calculated 75 gene expression scores, which were arranged in ascending order (Fig. 5). ROC curve analysis of the gene expression scoring system using biopsy specimens yielded an optimal cut-off score of 40.5, with an AUC of 0.981, sensitivity of 87.5%, and specificity of 99.0% (Fig. 6A).

By the same procedure, ROC curve analysis of the gene expression scoring system using commercially available FFPE specimens, including 13 BSCE, yielded an optimal cut-off score of 34.9, with AUC of 0.901, sensitivity of 92.3%, and specificity of 71.4% (Fig. 6B).

Discussion

Using CGEA of esophagectomy specimens, we identified the 75 genes that were characteristically expressed in BSCE to construct a gene expression scoring system, which made it possible to distinguish BSCE from non-BSCE in biopsy and commercially available FFPE specimens with high sensitivity and specificity. To our knowledge, this is the first report to show an accurate diagnostic modality that could significantly contribute in improving the diagnosis and treatment of BSCE.

Diagnosing BSCE using endoscopic biopsy specimens is difficult, with a reported diagnostic accuracy of only 0% to 10% (23,24,27). We too were unable to accurately diagnose BSCE using preoperative endoscopic biopsy in three of six surgical patients. To overcome this difficulty, immunohistochemical staining was performed. Cytokeratin (CK) subtypes, including CK13 (12), CK14 (16), and CK19 (9,12,19), were attempted, but they failed to show specific properties, as did p53 and Rb protein (12,34,35). On the other hand, a combination of immunohistochemical staining and PCR analysis was previously tested for differential diagnosis: Bcl-2 expression together with c-myc amplification was demonstrated to be more frequent in BSCE than in SCC (28), but the specificity of this test was low (43.5%). In this study, we identified the 75 genes that were characteristically expressed in BSCE. None of these genes were mentioned in previous reports. Of 75 genes, collagen related genes (COL9A2,COL9A3,COL9A1) and fibroblast growth factor related genes (FGF19, FGF3) might be concerning the characteristics of BSCE based on the association with genes identified in this study.

In one study, comprehensive gene expression profiling was performed for endoscopic biopsy specimens of esophageal SCC (36). However, CGEA in our institution is different from that in the other institution. We had previously extracted some disease-specific genes through the CGEA system at our institution (31,32). Our CGEA has three features: 1) it can analyze small samples, like endoscopic biopsy specimens; 2) it can be performed without RNA amplification; and 3) the gene expression ratio of all types of samples can be compared with human common reference RNA. Comprehensive gene expression analysis is done to compare the expression levels of the human common reference RNA which was prepared from 22 human cancer cell lines. Profiles were obtained even from the BSC samples comprising SCC components. A group of genes specific for BSC were selected by comparing BSC containing SCC component with non-BSC (SCC). A group of genes that were expressed in SCC components were eliminated by selection process for the genes specific for BSC. Therefore, the group of genes specific for BSC can distinguish between the BSC and non-BSC even though BSC samples included SCC components. Introducing of the scoring system enabled us to differentiate BSC from non-BSC if the sample contains a limited portion of BSC with SCC components. In this study, we selected 75 genes for constructing a gene expression scoring system for the proper diagnosis of BSCE. There have been no reports on studies in which a gene expression scoring system was constructed to diagnose cancers. This scoring system is a novel, precise, and powerful tool for diagnosing BSCE in both endoscopic biopsy and FFPE specimens.

There are several limitations in this study. First, the biopsy specimens obtained for histology and CGEA were not identical, although we tried as much as possible to choose specimens that were adjacent to each other. In addition, biopsy samples might not hit the component of BSC for accurate diagnosis by pathological and genetical analysis. Thus, biopsy samples should be obtained from multiple sites in deep portion of tumor to obtain histological characteristics of BSC. In contrast, with this method, we are able to diagnose BSC from very small amount of specimens as long as it contains BSC component. Second, the parameters of processing the commercially available FFPE specimens (i.e., interval between resection and fixation, duration of fixation) were not controlled. Third, the cut-off scores varied among the sources (surgical, endoscopic biopsy, and FFPE specimens); this may have affected the attainment of reasonable specificity and sensitivity. In this study, the number of patients was too small to enable comparison of the prognoses between BSCE and SCC. These limitations should be addressed using a larger number of cases in the future. Nevertheless, we believe that this method elucidates the proper diagnosis of very rare cases of BSCE. Furthermore, it may be able to clarify whether the prognosis of patients with BSCE is similar to or poorer than that of patients with SCC. Lastly, recently (December 1, 2016) International Cell Line Authentication Committee released Version 8.0 of database of cross-contaminated or misidentified cell lines (37), in which we found five cell lines (AKI human melanoma, HBL-100 human breast carcinoma, human gastric carcinoma MKN-7, SK-N-MC human neuroblastoma and U937 lymphoma histiocytic cells) among our 22 reference cell lines have been contaminated with HeLa cervical adenocarcinoma cells, human cells of unknown origin, a cell line of unknown origin, human Sarcoma (Ewing's) cells and a cell line of unknown origin, respectively. Even with this condition of reference cell lines it is obvious that our results would not be affected since we only used a relative ratio of BSCE against SCC, but not an absolute ratio to reference cell lines in order to select the responsible genes for the scoring system.

In conclusion, using CGEA of esophagectomy specimens, we identified 75 genes that are characteristically expressed in BSCE; a gene expression scoring system constructed from these data enabled us to distinguish BSCE from non-BSCE with high sensitivity and specificity, even on endoscopic biopsy specimens. We believe that this scoring system can be a novel method that may significantly contribute to improving the diagnostic accuracy for BSCE.

Acknowledgments

This work was partially supported by grants for translational research programs from the New Energy and Industrial Technology Development Organization (Tokyo, Japan) and Fukushima Prefecture. The authors would like to thank Enago (www.enago.jp) for the English language review.

Glossary

Abbreviations

Abbreviations:

BSC

basaloid squamous cell carcinoma

BSCE

BSC of the esophagus

CGEA

comprehensive gene expression analysis

FFPE

formalin-fixed paraffin-embedded

AUC

area under the curve

CI

confidence interval

SCC

squamous cell carcinoma

PCR

polymerase chain reaction

NEC

neuroendocrine carcinoma

CK

cytokeratin

SD

standard deviation

ROC

receiver operating characteristic

References

1 

Wain SL, Kier R, Vollmer RT and Bossen EH: Basaloid-squamous carcinoma of the tongue, hypopharynx, and larynx: Report of 10 cases. Hum Pathol. 17:1158–1166. 1986. View Article : Google Scholar : PubMed/NCBI

2 

Sarbia M, Verreet P, Bittinger F, Dutkowski P, Heep H, Willers R and Gabbert HE: Basaloid squamous cell carcinoma of the esophagus: Diagnosis and prognosis. Cancer. 79:1871–1878. 1997. View Article : Google Scholar : PubMed/NCBI

3 

Brambilla E, Moro D, Veale D, Brichon PY, Stoebner P, Paramelle B and Brambilla C: Basal cell (basaloid) carcinoma of the lung: A new morphologic and phenotypic entity with separate prognostic significance. Hum Pathol. 23:993–1003. 1992. View Article : Google Scholar : PubMed/NCBI

4 

Chetty R, Serra S and Hsieh E: Basaloid squamous carcinoma of the anal canal with an adenoid cystic pattern: Histologic and immunohistochemical reappraisal of an unusual variant. Am J Surg Pathol. 29:1668–1672. 2005. View Article : Google Scholar : PubMed/NCBI

5 

Brainard JA and Hart WR: Adenoid basal epitheliomas of the uterine cervix: A reevaluation of distinctive cervical basaloid lesions currently classified as adenoid basal carcinoma and adenoid basal hyperplasia. Am J Surg Pathol. 22:965–975. 1998. View Article : Google Scholar : PubMed/NCBI

6 

Cubilla AL, Reuter VE, Gregoire L, Ayala G, Ocampos S, Lancaster WD and Fair W: Basaloid squamous cell carcinoma: a distinctive human papilloma virus-related penile neoplasm: a report of 20 cases. Am J Surg Pathol. 22:755–761. 1998. View Article : Google Scholar : PubMed/NCBI

7 

Vakar-López F and Abrams J: Basaloid squamous cell carcinoma occurring in the urinary bladder. Arch Pathol Lab Med. 124:455–459. 2000.PubMed/NCBI

8 

Takubo K, Mafune K, Tanaka Y, Miyama T and Fujita K: Basaloid-squamous carcinoma of the esophagus with marked deposition of basement membrane substance. Acta Pathol Jpn. 41:59–64. 1991.PubMed/NCBI

9 

Abe K, Sasano H, Itakura Y, Nishihira T, Mori S and Nagura H: Basaloid-squamous carcinoma of the esophagus. A clinicopathologic, DNA ploidy, and immunohistochemical study of seven cases. Am J Surg Pathol. 20:453–461. 1996. View Article : Google Scholar : PubMed/NCBI

10 

Koide N, Koike S, Adachi W, Amano J, Usuda N and Nagata T: Immunohistochemical expression of bcl-2 protein in squamous cell carcinoma and basaloid carcinoma of the esophagus. Surg Today. 27:685–691. 1997. View Article : Google Scholar : PubMed/NCBI

11 

Kawahara K, Makimoto K, Maekawa T, Yamamoto S, Shiraishi T, Takahashi S, Shirakusa T, Nakayama Y and Kikuchi M: An immunohistochemical examination of basaloid squamous cell carcinoma of the esophagus: Report of a case. Surg Today. 31:655–659. 2001. View Article : Google Scholar : PubMed/NCBI

12 

Ohashi K, Horiguchi S, Moriyama S, Hishima T, Hayashi Y, Momma K, Hanashi T, Izumi Y, Yoshida M and Funata N: Superficial basaloid squamous carcinoma of the esophagus. A clinicopathological and immunohistochemical study of 12 cases. Pathol Res Pract. 199:713–721. 2003.

13 

Yoshioka S, Tsujinaka T, Fujitani K and Kawahara K: Prognostic analysis of four cases of basaloid cell carcinoma of the esophagus and 60 reported cases in Japan. Jpn J Gastroenterol Surg. 37:290–295. 2004. View Article : Google Scholar

14 

Kobayashi Y, Nakanishi Y, Taniguchi H, Sekine S, Igaki H, Tachimori Y, Kato H, Matsubara H, Okazumi S and Shimoda T: Histological diversity in basaloid squamous cell carcinoma of the esophagus. Dis Esophagus. 22:231–238. 2009. View Article : Google Scholar

15 

Saito S, Hosoya Y, Zuiki T, Hyodo M, Lefor A, Sata N, Nagase M, Nakazawa M, Matsubara D, Niki T, et al: A clinicopathological study of basaloid squamous carcinoma of the esophagus. Esophagus. 6:177–181. 2009. View Article : Google Scholar

16 

Imamhasan A, Mitomi H, Saito T, Hayashi T, Takahashi M, Kajiyama Y and Yao T: Immunohistochemical and oncogenetic analyses of the esophageal basaloid squamous cell carcinoma in comparison with conventional squamous cell carcinomas. Hum Pathol. 43:2012–2023. 2012. View Article : Google Scholar : PubMed/NCBI

17 

Zhang XH, Sun GQ, Zhou XJ, Guo HF and Zhang TH: Basaloid squamous carcinoma of esophagus:a clinicopathological, immunohistochemical and electron microscopic study of sixteen cases. World J Gastroenterol. 4:397–403. 1998. View Article : Google Scholar

18 

Cho KJ, Jang JJ, Lee SS and Zo JI: Basaloid squamous carcinoma of the oesophagus: A distinct neoplasm with multipotential differentiation. Histopathology. 36:331–340. 2000. View Article : Google Scholar : PubMed/NCBI

19 

Huang Z, Shen Y, Liang Y and Wu X: Basaloid squamous cell carcinoma of the esophagus: An immunohistochemical study of 8 cases. Chin Med J (Engl). 114:1084–1088. 2001.

20 

Lam KY, Law S, Luk JM and Wong J: Oesophageal basaloid squamous cell carcinoma: A unique clinicopathological entity with telomerase activity as a prognostic indicator. J Pathol. 195:435–442. 2001. View Article : Google Scholar : PubMed/NCBI

21 

Klaase JM, Hulscher JBF, Offerhaus GJA, ten Kate FJ, Obertop H and van Lanschot JJ: Surgery for unusual histopathologic variants of esophageal neoplasms: A report of 23 cases with emphasis on histopathologic characteristics. Ann Surg Oncol. 10:261–267. 2003. View Article : Google Scholar : PubMed/NCBI

22 

Li TJ, Zhang YX, Wen J, Cowan DF, Hart J and Xiao SY: Basaloid squamous cell carcinoma of the esophagus with or without adenoid cystic features. Arch Pathol Lab Med. 128:1124–1130. 2004.PubMed/NCBI

23 

Chen SB, Weng HR, Wang G, Yang JS, Yang WP, Li H, Liu DT and Chen YP: Basaloid squamous cell carcinoma of the esophagus. J Cancer Res Clin Oncol. 138:1165–1171. 2012. View Article : Google Scholar : PubMed/NCBI

24 

Zhang BH, Cheng GY, Xue Q, Gao SG, Sun KL, Wang YG, Mu JW and He J: Clinical outcomes of basaloid squamous cell carcinoma of the esophagus: A retrospective analysis of 142 cases. Asian Pac J Cancer Prev. 14:1889–1894. 2013. View Article : Google Scholar : PubMed/NCBI

25 

Akagi I, Miyashita M, Makino H, Nomura T, Ohkawa K and Tajiri T: Basaloid squamous cell carcinoma of the esophagus: Report of two cases. J Nippon Med Sch. 75:354–360. 2008. View Article : Google Scholar

26 

Nishimura W, Naomoto Y, Hamaya K, Toda S, Miyagi K and Tanaka N: Basaloid-squamous cell carcinoma of the esophagus: Diagnosis based on immunohistochemical analysis. J Gastroenterol Hepatol. 16:586–590. 2001. View Article : Google Scholar : PubMed/NCBI

27 

Kato T, Morita T, Fujita M, Miyasaka Y, Horita S, Watanabe Y and Kato H: Basaloid-squamous carcinoma of the esophagus: Report of a case. Surg Today. 30:163–167. 2000. View Article : Google Scholar : PubMed/NCBI

28 

Sarbia M, Loberg C, Wolter M, Arjumand J, Heep H, Reifenberger G and Gabbert HE: Expression of Bcl-2 and amplification of c-myc are frequent in basaloid squamous cell carcinomas of the esophagus. Am J Pathol. 155:1027–1032. 1999. View Article : Google Scholar : PubMed/NCBI

29 

Bellizzi AM, Woodford RL, Moskaluk CA, Jones DR, Kozower BD and Stelow EB: Basaloid squamous cell carcinoma of the esophagus: Assessment for high-risk human papillomavirus and related molecular markers. Am J Surg Pathol. 33:1608–1614. 2009. View Article : Google Scholar : PubMed/NCBI

30 

Arai T, Aida J, Nakamura KI, Ushio Y and Takubo K: Clinicopathologic characteristics of basaloid squamous carcinoma of the esophagus. Esophagus. 8:169–177. 2011. View Article : Google Scholar

31 

Miyamoto K, Iwadate M, Yanagisawa Y, Ito E, Imai J, Yamamoto M, Sawada N, Saito M, Suzuki S, Nakamura I, et al: Cathepsin L is highly expressed in gastrointestinal stromal tumors. Int J Oncol. 39:1109–1115. 2011.PubMed/NCBI

32 

Okabe N, Ezaki J, Yamaura T, Muto S, Osugi J, Tamura H, Imai J, Ito E, Yanagisawa Y, Honma R, et al: FAM83B is a novel biomarker for diagnosis and prognosis of lung squamous cell carcinoma. Int J Oncol. 46:999–1006. 2015.PubMed/NCBI

33 

Miura A, Honma R, Togashi T, Yanagisawa Y, Ito E, Imai J, Isogai T, Goshima N, Watanabe S and Nomura N: Differential responses of normal human coronary artery endothelial cells against multiple cytokines comparatively assessed by gene expression profiles. FEBS Lett. 580:6871–6879. 2006. View Article : Google Scholar : PubMed/NCBI

34 

Owonikoko T, Loberg C, Gabbert HE and Sarbia M: Comparative analysis of basaloid and typical squamous cell carcinoma of the oesophagus: A molecular biological and immunohistochemical study. J Pathol. 193:155–161. 2001. View Article : Google Scholar : PubMed/NCBI

35 

Baba Y, Ishimoto T, Harada K, Kosumi K, Murata A, Miyake K, Hiyoshi Y, Kurashige J, Iwatsuki M, Iwagami S, et al: Molecular characteristics of basaloid squamous cell carcinoma of the esophagus: analysis of KRAS, BRAF, and PIK3CA mutations and LINE-1 methylation. Ann Surg Oncol. 22:3659–3665. 2015. View Article : Google Scholar : PubMed/NCBI

36 

Motoori M, Takemasa I, Yamasaki M, Komori T, Takeno A, Miyata H, Takiguchi S, Fujiwara Y, Yasuda T, Yano M, et al: Prediction of the response to chemotherapy in advanced esophageal cancer by gene expression profiling of biopsy samples. Int J Oncol. 37:1113–1120. 2010.PubMed/NCBI

37 

International Cell Line Authentication Committee: Register of Misidentified Cell Lines, version 8.0. December 1–2016, http://iclac.org/databases/cross-contaminations/urisimplehttp://iclac.org/databases/cross-contaminations/.

Related Articles

Journal Cover

September 2017
Volume 51 Issue 3

Print ISSN: 1019-6439
Online ISSN:1791-2423

Sign up for eToc alerts

Recommend to Library

Copy and paste a formatted citation
x
APA
Tada, T., Honma, R., Imai, J., Saze, Z., Kogure, M., Marubashi, S. ... Gotoh, M. (2017). A novel gene expression scoring system for accurate diagnosis of basaloid squamous cell carcinoma of the esophagus. International Journal of Oncology, 51, 877-886. https://doi.org/10.3892/ijo.2017.4075
MLA
Tada, T., Honma, R., Imai, J., Saze, Z., Kogure, M., Marubashi, S., Tasaki, K., Unakami, M., Ezaki, J., Tamura, H., Nishikawa, A., Hashimoto, Y., Waguri, S., Watanabe, S., Gotoh, M."A novel gene expression scoring system for accurate diagnosis of basaloid squamous cell carcinoma of the esophagus". International Journal of Oncology 51.3 (2017): 877-886.
Chicago
Tada, T., Honma, R., Imai, J., Saze, Z., Kogure, M., Marubashi, S., Tasaki, K., Unakami, M., Ezaki, J., Tamura, H., Nishikawa, A., Hashimoto, Y., Waguri, S., Watanabe, S., Gotoh, M."A novel gene expression scoring system for accurate diagnosis of basaloid squamous cell carcinoma of the esophagus". International Journal of Oncology 51, no. 3 (2017): 877-886. https://doi.org/10.3892/ijo.2017.4075