International Journal of Clinical Oncology https://doi.org/10.1007/s10147-018-1238-y
ORIGINAL ARTICLE
Differences in histological features and PD‑L1 expression between sporadic microsatellite instability and Lynch‑syndrome‑associated disease in Japanese patients with colorectal cancer Rin Yamada1 · Tatsuro Yamaguchi2,3,4 · Takeru Iijima3 · Rika Wakaume3 · Misato Takao4 · Koichi Koizumi5 · Tsunekazu Hishima1 · Shin‑ichiro Horiguchi1 Received: 1 November 2017 / Accepted: 31 December 2017 © Japan Society of Clinical Oncology 2018
Abstract Background The field of immunotherapy has recently focused on cancers with microsatellite instability (MSI). These cancers include both Lynch-syndrome-associated tumors, which are caused by mismatch repair (MMR) germline mutations, and sporadic MSI tumors, which are mainly attributed to MLH1 promoter methylation. The present study aimed to clarify differences in the histological and PD-L1 expression profiles between these two types of MSI cancers in Japanese patients. Methods Among 908 cases of colorectal cancer treated via surgical resection from 2008 to 2014, we identified 64 MSI cancers, including 36 sporadic MSI and 28 Lynch-syndrome-associated cancers, using a BRAF V600E mutation analysis and MLH1 methylation analysis. Of the latter subgroup, 21 (75%) harbored MMR germline mutations. Results The following were more frequent with sporadic MSI than with Lynch syndrome associated cancers: poor differentiation (50.0 vs. 7.1%, P = 0.0002), especially solid type (30.6 vs. 3.6%, P = 0.0061); medullary morphology (19.4 and 0%, P = 0.015), Crohn-like lymphoid reaction (50.0 vs. 25.0%, P = 0.042), and PD-L1 expression (25.0 vs. 3.6%, P = 0.034). However, the groups did not differ in terms of the mean invasive front and intratumoral CD8-positive cell densities. In a logistic regression analysis, PD-L1 expression correlated with poor differentiation (odds ratio: 7.65, 95% confidence interval: 1.55–37.7, P = 0.012), but not with the difference between sporadic MSI cancer and Lynch-syndrome-associated cancer (odds ratio: 4.74, 95% confidence interval: 0.50–45.0, P = 0.176). Conclusions Therefore, compared with Lynch-syndrome-associated cancers, sporadic MSI cancers are more frequently solid, poorly differentiated medullary cancers that express PD-L1. Keywords Microsatellite instability · Colorectal cancer · Lynch syndrome · PD-L1
* Tatsuro Yamaguchi
[email protected] 1
Department of Pathology, Tokyo Metropolitan Cancer and Infectious Diseases Center Komagome Hospital, Tokyo, Japan
Department of Clinical Genetics, Tokyo Metropolitan Cancer and Infectious Diseases Center Komagome Hospital, 3‑18‑22, Honkomagome, Bunkyo‑ku, Tokyo 113‑8677, Japan
2
3
Hereditary Tumor Research Project, Tokyo Metropolitan Cancer and Infectious Diseases Center Komagome Hospital, Tokyo, Japan
4
Department of Surgery, Tokyo Metropolitan Cancer and Infectious Diseases Center Komagome Hospital, Tokyo, Japan
5
Department of Gastroenterology, Tokyo Metropolitan Cancer and Infectious Diseases Center Komagome Hospital, Tokyo, Japan
13
Vol.:(0123456789)
Introduction The tumorigenesis of colorectal cancer has been well studied over the last two decades. In particular, microsatellite instability (MSI) cancers have recently received attention in terms of both immunotherapy and hereditary cancer syndrome [1, 2]. Approximately 15% of all colorectal cancers exhibit MSI, and these tumors result from the accumulation of frameshift mutations in target genes caused by a failure of the mismatch repair system. Two types of MSI cancers have been identified: those associated with Lynch syndrome, which is caused by germline mutations in DNA mismatch repair genes such as MLH1, MSH2, and MSH6, and sporadic tumors that result mainly from methylation of the MLH1 promoter, which leads to an accumulation of mutations in genes involved in colorectal carcinogenesis [3, 4]. Generally, MSI cancers are right-sided tumors with a good prognosis [5]. Sporadic MSI cancers predominantly occur in elderly women, and are almost always right-sided. In contrast, Lynch syndrome-associated cancers tend to occur in younger patients, with no sex predominance [6]. Biomarker analyses have detected the BRAF V600E mutation in most sporadic MSI cancers [7]. By contrast, this mutation is rare among Lynch-syndrome-associated cancers, which tend to harbor KRAS mutations [8, 9]. A few reports have described the histological differences between sporadic MSI cancer and Lynch-syndromeassociated cancer [6, 10–13]. It was recently reported that programmed death-ligand 1 (PD-L1), which is a major ligand of immunosuppressive receptor PD-1, is expressed on the surface of MSI colorectal cancer [14–20]. However, previous studies have not clarified the difference between sporadic MSI and Lynch-syndrome-associated cancers in terms of the expression of PD-L1. Therefore, the present study aimed to clarify the histological differences between sporadic MSI and Lynch syndrome-associated cancers in a sample of Japanese colorectal cancer patients.
Materials and methods Patients We consecutively selected 908 colorectal cancer patients who underwent surgical resection at the Tokyo Metropolitan Cancer and Infectious Diseases Center Komagome Hospital from 2008 to 2014 and had provided informed consent for study inclusion. This study was approved by the institutional review board of the hospital (approval number: 1925). Patients with a known history
13
International Journal of Clinical Oncology
of familial adenomatous polyposis and Lynch syndrome were included. For cases involving two or more resected colorectal tumors, the more advanced tumor was selected for analysis. We identified 64 MSI cancers among the 908 cases, including 36 with the BRAF V600E mutation and/or MLH1 methylation (mMLH1) and 28 with neither the BRAF V600E mutation nor mMLH1. These cases were diagnosed as sporadic MSI cancers and Lynch syndrome-associated cancers, respectively. Of the 28 patients with Lynch syndrome-associated cancers, 23 underwent MMR gene testing, and 21 were found to harbor germline pathogenic mutations. Clinicopathological data were obtained from clinical and pathological reports. Cancer staging was determined according to the Tumor, Node, Metastasis Classification of Malignant Tumors, 8th edition, published by the Union for International Cancer Control (UICC).
Microsatellite instability analysis The colorectal cancers and corresponding normal tissues were stored at − 80 °C immediately after resection. Genomic DNA samples were then extracted from the tissues using the QIAamp DNA mini kit (QIAGEN, Valencia, CA, USA). Polymerase chain reactions (PCRs) were performed to amplify sequences from cancer and corresponding normal DNA, as described previously [21]. The microsatellite status was determined using two microsatellite markers (BAT25 and BAT26) and defined as MSI (at least one unstable informative marker) or MSS (no unstable markers), as per the National Cancer Institute guidelines for MSI testing [22].
BRAF gene mutation analysis BRAF mutations in exon 15 (codon 600) were detected either by PCR–restriction fragment length polymorphism or highresolution melting analysis, as described previously [21, 23].
MLH1 methylation analysis Bisulfite conversion and bisulfite-converted DNA recovery were performed using the Zymo EZ DNA methylation kit (Zymo Research, Irvine, CA, USA) according to the manufacturer’s instructions. A MethyLight DNA methylation analysis of each sample was performed as described previously [21], and the methylation status was considered positive at a PMR value of > 4 (a cutoff value based on validated data) [24].
Pathologic assessment Hematoxylin- and eosin-stained sections of formalinfixed, paraffin-embedded tissues were reviewed by two
International Journal of Clinical Oncology
pathologists who were blinded to the results of genetic testing. The reviewers evaluated the following pathological factors on all slides: tumor differentiation, medullary morphology, mucinous component, lymphatic invasion, vascular invasion, and Crohn-like lymphoid reaction (CLR). Vascular invasion was also evaluated using Elastica van Gieson staining. The tumors were graded as well, moderately, or poorly differentiated according to the World Health Organization (WHO) Classification of Tumors of the Digestive System, 4th edition. Poorly differentiated tumors were further classified by cancer cell structure into solid tumors characterized by the presence of solid sheets with expansive growth or non-solid tumors characterized by cords, small nests, or single cells with invasive growth (Fig. 1). Medullary morphology was defined as a tumor that fulfilled all of the following criteria: (1) a solid or nested growth pattern; (2) a circumscribed tumor border; (3) cellular (nondesmoplastic) stroma rich in lymphocytes or other inflammatory cells; (4) uniform and “syncytial” tumor cells frequently exhibiting vesicular nuclei, prominent nucleoli, and abundant eosinophilic cytoplasm; and (5) abundant tumor-infiltrating lymphocytes (TILs) (Fig. 2) [25]. In this study, CLR was evaluated using the Väyrynen–Mäkinen criteria, and CLR density was calculated as the number of lymphoid follicles divided by the length of the tumor invasive front. Cases were classified as high density or low density using a cutoff point of 0.38 follicles/mm [26]. In cases of disagreement, including those involving immunohistochemistry determinations as described below, the slides were reviewed again and a consensus was reached.
Immunohistochemistry For immunohistochemistry, a single formalin-fixed, paraffinembedded tissue sample block with the most deeply invasive edge was selected from each case. Sections (4 μm in thickness) were cut from each block and either stained with hematoxylin or subjected to immunohistochemical staining. The characteristics and assay conditions of the antibodies used are listed in Table 1. Briefly, the sections were deparaffinized in xylene, hydrated through a graded ethanol series, and immersed in 0.3% hydrogen peroxide in methanol for 15 min to inhibit endogenous peroxidase activity. After rinsing in phosphate-buffered saline (PBS), the sections were processed in an autoclave or microwave oven in citrate buffer (10 mM, pH 6.0) for antigen retrieval, as indicated in Table 1. After three rinses in PBS, the tissue sections were incubated with primary antibodies and subsequently processed using the reagent in the commercial Elite ABC kit (Vectastain, Vector Laboratories, Burlingame, CA, USA). A diaminobenzidine solution was added to develop the reaction color. The slides were finally counterstained with Mayer’s hematoxylin for 5 min, dehydrated, and mounted. Negative controls were treated as described above; however, the primary antibody was replaced with nonimmune serum or omitted. Each slide was scored based on the percentage of positively stained tumor cells. When the percentage of PDL1-positive tumor cells exceeded 5%, the sample was considered positive. The area containing the highest density of CD8-positive cells at the invasive front or within the tumor was scanned at a low magnification, and the number of CD8positive cells was counted per high-power field.
Fig. 1 Poorly differentiated tumors were subclassified based on the cancer cell structure. a The solid type was characterized by solid sheets of cells with expansive growth. b The non-solid type was characterized by the presence of cords, small nests, or single cells with invasive growth
13
International Journal of Clinical Oncology
Fig. 2 The features of medullary morphology. a A circumscribed tumor border. b A solid or nested growth pattern and a lymphocyterich cellular stroma. c Uniform and syncytial tumor cells frequently
Table 1 Details of the antibodies used in this study
Antigen
Antibody clone
Species
Dilution
Source
PD-L1
E1L3N
Rabbit monoclonal
1:100
CD8
C8/144B
Mouse monoclonal
1:100
Cell Signaling Technology, Inc., Danvers, MA, USA Nichirei Bioscience, Inc., Tokyo, Japan
Statistical analysis The χ2 test or t test was used to evaluate the relationship between two discrete and dichotomous variables. Fisher’s exact test was used if the expected frequency was
