大腸直腸癌預後學之研究－著重表觀遺傳學和腫瘤微環境之解析

(1)

國立臺灣大學醫學院腫瘤醫學研究所博士論文

Graduate Institute of Oncology College of Medicine

National Taiwan University Doctoral Dissertation

大腸直腸癌預後學之研究－

著重表觀遺傳學和腫瘤微環境之解析

Studies on the Prognostics of Colorectal Cancer－

with Emphasis on Epigenetics and Tumor Microenvironment

陳國興 Kuo-Hsing Chen

指導教授：葉坤輝博士 Kun-Huei Yeh Ph.D

中華民國 112 年 1 月

(2)

(3)

誌謝

修習博士學位過程並非一路順遂，其中甘苦點滴在心頭。若是沒有師長、同好及家人的支持，是不可能完成的。

首先我要感謝指導教授葉坤輝教授。從入學開始，我的每項研究計畫都經過葉教授的指導，內容包括研究主題、實驗技術、計畫和論文的撰寫等。葉教授也

鼓勵我勇敢地把想法化為行動，例如CIMP-high 大腸直腸癌危險因子的研究，就

是在老師的鼓勵下開始收案。另外也感謝葉教授對於研究計畫執行面上的支持，

對於年輕研究者來說是相當重要的。

接下來，我要感謝論文口試委員們：趙毅教授、林亮音教授、莊曜宇教授、

徐志宏教授從每年的論文指導委員會開始到口試當天，不斷地給予許多研究和論文寫作的建議。而陳立宗教授在口試時的指導也相當受用。您們的指導使得我的博士論文更加完善。再來則是感謝鄭安理教授。鄭教授是我當腫瘤內科研修醫師時的指導老師。在那個階段鄭教授的教誨讓我們對於癌症研究有了初步的認識，

也對於研究生涯有憧憬。鄭教授的許多金玉良言我到現在仍是牢記心中，比如

「寫論文是作結論的過程」、「做研究一定要交朋友」、「要Plan for success」

等。另外，我也要感謝這幾年一路上幫忙我的師長朋友們。梁逸歆醫師不管在實驗上和研究思考上都給我很多的指點、許家郎博士在大數據分析上的協助、袁章祖醫師在病理實驗上的幫忙、林本仁醫師和蘇祐立醫師在臨床研究收案上的大力相助。這印證了鄭教授所言「做研究一定要交朋友」。

最後，我要感謝我最重要的家人。我太太、我的兩個寶貝女兒和我的爸媽。

我太太為了給兩個女兒最好的成長環境，選擇自己照顧和兼職工作的做法，相當辛苦。我的爸媽也在照顧我女兒的方面幫忙很多。我的女兒們則總是帶給我滿滿

(4)

論文摘要

CpG 島甲基化表現型(CpG island methylator phenotype) 大腸直腸癌是一種在 DNA 啟動子許多 CpG 島有廣泛性高度甲基化的一種亞型，和許多臨床病理特徵有高度相關，例如：老年、女性、右側大腸癌、BRAF V600E 突變和偶發性錯配

修復缺陷大腸直腸癌。CpG 島甲基化表現型的預後角色仍存有爭議。為了探究

CpG 島甲基化表現型的預後角色，我們先前針對 450 位大腸直腸癌病人的檢體分

析CpG 島甲基化表現型等生物標記。結果顯示 CpG 島甲基化表現型在早期癌並

不是預後因子，但是在第四期的轉移性癌症中，則病人顯著地有較短的存活期。

我們也發現在「年輕型」大腸直腸癌(小於 50 歲診斷癌症)中，CpG 島甲基化表現

型大腸直腸癌的比率達14.3%，明顯較西方族群的比例(約 5%)高。

為了進一步驗證CpG 島甲基化表現型在台灣「年輕型」大腸直腸癌的高發生

率，以及探究大腸直腸癌危險因子與台灣CpG 島甲基化表現型大腸直腸癌的相關

性，我們於2016-2019 年間「前瞻性」收集大腸直腸癌病人的腫瘤檢體、臨床病

理資料和大腸直腸癌危險因子。研究結果顯示，小於50 歲和大於等於 50 歲的病

人中各有15.7% (14/89)和 15.2% (31/204)是 CpG 島甲基化表現型大腸直腸癌。另

外，我們也發現CpG 島甲基化表現型大腸直腸癌在小於 50 歲的病人中，與較多

的第四期診斷、BRAF V600E 突變、和高身體質量指數 (≥ 27.5 kg/m²)有相關。在

多變相分析中，只有高身體質量指數 (≥ 27.5 kg/m²)這個危險因子和 CpG 島甲基

化表現型在小於50 歲的大腸直腸癌病人有顯著相關。這結果驗證了在台灣年輕

型大腸直腸癌有高比例是CpG 島甲基化表現型，而且這些腫瘤與高身體質量指數

(≥ 27.5 kg/m²)有顯著相關。

許多研究指出 BRAF V600E 突變型在轉移癌與較差的預後有關。我們團隊之

前的研究也發現雖同屬腫瘤細胞基因型BRAF V600E 突變型，在「晚期大腸直腸

(5)

癌」與存活期較短相關，但在「早期大腸直腸癌」則與預後無關。有鑑於腫瘤微

環境中的免疫組成也可以是大腸直腸癌的預後因子，我們假設BRAF V600E 突變

的大腸直腸癌腫瘤內免疫組成有預後影響。因為轉移性BRAF V600E 突變大腸直

腸癌的免疫組成很少被研究，我們納入54 個第四期未接受過治療 BRAF V600E 突

變的轉移性大腸直腸癌的病人，收集他們的檢體和臨床資料。這些檢體的RNA

被萃取出來後利用PanCancer Immune Panel (nanoString)進行免疫基因表現分析。

我們的研究結果顯示，許多腫瘤內補體基因的表現與病人預後有關。我們也發展出「補體分數」這個工具。「補體分數」高的腫瘤相較於「補體分數」低的腫瘤

有較短的疾病無惡化存活期和總存活期。我們發現「補體分數」也與C4d (一個

補體路徑活化的生物標記)染色的強度有顯著正相關，驗證了 RNA 分析的結果。

最後，我們發現，「補體分數」與腫瘤微環境中的M2 型巨噬細胞所代表的基因

表現也是有顯著相關性。因為M2 型巨噬細胞被認為有促進腫瘤生長的作用。這

也許是「補體分數」高的腫瘤有比較差的預後的可能原因之一。

綜合以上，我們的研究首次點出東西方大腸直腸癌在CpG 島甲基化高表現型

大腸直腸癌有不同的發生率，也提醒國人適當身體質量指數(< 27.5 kg/m²)對大腸

癌預防的重要性。我們的研究結果亦將有助於未來對於BRAF V600E 突變大腸直

腸癌發展新的治療策略。

關鍵字：大腸直腸癌、預後、CpG 島甲基化表現型、身體質量指數、BRAF

V600E 突變、補體、腫瘤微環境

(6)

Thesis Abstract

The CpG island methylator phenotype-high (CIMP-high) subtype of colorectal

cancer (CRC) is characterized by widespread hypermethylation in the CpG islands of

DNA promoters. Our previous study demonstrated that CIMP-high was not associated

with poor prognosis in early-stage CRC but an independent prognostic factor in Stage 4

CRC in a multivariate analysis model. Among patients with early-onset CRC (EOCRC)

diagnosed before 50 years old (y/o), the frequency of CIMP-high was 14.3%,

significantly higher than that in the Western population (5%). Thus, we collected

specimens from patients with CRC and analyzed the clinicopathologic characteristics

and CRC risk factors of patients during 2016-2019. We analyzed the tumor’s

KRAS/NRAS mutations, BRAF V600E mutation, microsatellite instability (MSI), and

CIMP. The results revealed that CIMP-high tumors represented 15.7% (14/89) in

EOCRC and 15.2% (31/204) in late-onset CRC (LOCRC, diagnosed at 50 y/o or older),

respectively. In addition, we observed that in a multivariate analysis, a high body mass

index (BMI) of ≥ 27.5 kg/m² was significantly correlated with CIMP-high CRC in those

younger than 50 yrs. These findings validated that the frequency of CIMP-high CRC in

EOCRC is high in Taiwan and demonstrated the significant association between these

tumors and a high BMI.

(7)

We previously demonstrated the significant prognostic role in late-stage BRAF

V600E mutant CRC but not in early-stage tumors. We hypothesized that immune

contextures in the tumor microenvironment of BRAF V600E mutant CRC have

prognostic implications in CRC. We enrolled 54 patients with untreated, metastatic

microsatellite-stable BRAF V600E-mutated CRC and analyzed the expression of

immune-related genes in these tumors. The results showed that many complement genes

were associated with patient’s survival outcomes. We developed a complement score

and observed that BRAF V600E-mutated CRC with a high complement score was

associated with shorter progression-free survival and overall survival compared to that

with a low complement score. Finally, we identified that complement scores were

significantly associated with M2 macrophage signatures. This may contribute to the

phenomenon that tumors with a high complement score are associated with poor

survival.

Therefore, our study indicates different incidences of CIMP-high CRC in Eastern

and Western populations, and reminds the importance of a proper BMI (<27.5 kg/m²)

for Taiwanese population. The findings in our study could provide insight into

developing novel treatments for BRAF V600E mutant CRC.

(8)

Key words: Colorectal cancer; prognosis; CpG island methylator phenotype; body

mass index; BRAF V600E mutation; complement; tumor microenvironment

(9)

List of Figures

Figure 1. Age-adjusted incidence rates (ASR) of colorectal cancer (ICD 10 codes C18-

C20) for both gender in Taiwan using the WHO standard population 2000…………...54

Figure 2. Flow chart of patient enrollment in this study……….55

Figure 3. The survival probability in patients with de novo metastatic MSS BRAF V600E mutant CRC by Kaplan–Meier analysis……….…………...56

Figure 4. The prognostic implication of complement genes expression………57

Figure 5. The prognostic implication of complement genes expression………59

Figure 6. The prognostic values of complement score………...61

Figure 7. The correlation between complement score and the specific metastatic sites………...…...63

Figure 8. The survival probability by Kaplan–Meier analysis in patients with low and high complement scores………..…………..…..64

Figure 9. The survival probability by Kaplan–Meier analysis in patients with a high complement score compared with those with a low complement score in TCGA dataset. ………65

Figure 10. The correlation of the complement score and C4d density in de novo

(12)

metastatic MSS BRAF V600E mutant CRC…...………..…………...67

Figure 11. The analysis of immune cell abundance revealed a strong association

between the complement score and M2 macrophages………68

(13)

List of Tables

Table 1. The primers used for CpG Island Methylator Phenotype testing in MethyLight

study……….………69

Table 2. Distribution of CIMP-high or CIMP-low/negative according to

clinicopathologic variables. Logistic regression was used for statistical analysis……..70

clinicopathologic variables stratified by age. Chi-square test, and Fisher’s exact test

were used for statistical

analysis………72

Table 4. Distribution of CIMP-high or CIMP-low/negative according to risk factors of

CRC in overall patients and stratified by age. Logistic regression was used for statistical

analysis………...….74

Table 5. Multivariate analysis for the association between CIMP-high CRC and

clinicopathological variables in overall patients, patients with age < 50y and age ≥

50y………...………76

clinicopathologic variables stratified by gender. Chi-square test, and Fisher’s exact test

(14)

were used for statistical analysis……….…77

Table 7. Distribution of CIMP-high or CIMP-low/negative according to risk factors of

CRC stratified by gender. Logistic regression was used for statistical analysis……….78

Table 8. Clinicopathological features in patients with BRAF V600E-mutated colorectal

cancer………..……….80

Table 9. Immune cells in Cox proportional hazard model for hazard ratios of

progression-free survival and overall survival in patients with de novo metastatic MSS

BRAF V600E-mutated CRC………...……….81

Table 10. Immune checkpoints link to B and T cell signaling in Cox proportional

hazard model for hazard ratios of progression-free survival and overall survival in

patients with de novo metastatic MSS BRAF V600E-mutated CRC………...……82

Table 11. Complement genes in the Cox proportional hazards model for hazard ratios

of progression-free survival and overall survival in patients with de novo metastatic

MSS BRAF V600E mutant CRC………...…..83

Table 12. Univariate and multivariate analyses through a Cox proportional hazards

model for hazard ratios of progression-free survival and overall survival in patients with

de novo metastatic MSS BRAF V600E mutant CRC………...84

(15)

List of Abbreviations

ASR Age-adjusted incidence rate

BMI Body mass index

BM BRAF mutant classification

CI Confidence interval

CIMP CpG island methylator phenotype

CMS Consensus molecular subtypes

CRC Colorectal cancer

DM Diabetes mellitus

DMG Differentially methylated gene

dMMR Mismatch repair deficiency

EOCRC Early-onset colorectal cancer

FFPE Formalin-fixed paraffin embedded

HDI Human Development Index

HR Hazard ratio

HRT Hormone replacement therapy

IBD Inflammatory bowel disease

(16)

KCGMH Kaohsiung Chang Gung Memorial Hospital

LOCRC Late-onset colorectal cancer

MSI-high Microsatellite instability-high

MSS Microsatellite stable

NSAID Non-steroidal anti-inflammatory drug

NTUH National Taiwan University Hospital

NTUH-iMD National Taiwan University Hospital-integrated Medical Database

ORR Objective response rate

OS Overall survival

PFI Progression-free interval

PFS Progression-free survival

PMR Percentage of methylated reference

TAM Tumor-associated macrophage

TCGA The Cancer Genome Atlas

TME Tumor microenvironment

TLS Tertiary lymphoid structure

(17)

Chapter I. Background

1.1 The Epidemiology and Carcinogenesis of Colorectal Cancer

Globally, colorectal cancer (CRC) ranks third in cancer incidence and second in

cancer mortality¹. There were estimated more than 1.9 million new CRC cases and

930,000 deaths due to CRC in 2020. CRC is more common in countries with very

high/high Human Development Index (HDI) than those with low/medium HDI. The

highest incidence of CRC is noted in European countries, Northern America, Australia,

and New Zealand followed by Eastern Asia. CRC incidence is still increasing steadily

in most countries but has declined for years in the United States, Canada, Australia, and

New Zealand, probably related to different screening programs². In Taiwan, CRC

incidence has increased in the past 30 years, and CRC has been the leading type of

cancer in newly diagnosed cases for more than 14 years (Figure 1)^3,4. In 2019, more

than 17,000 patients were diagnosed with CRC, among whom 6,400 died.

Based on epidemiologic studies, risk factors of CRC include family histories of

patients with CRC, inflammatory bowel disease (IBD), diabetes mellitus (DM), obesity,

sedentary lifestyle, cigarette smoking, and excessive consumption of red meat and

alcohol ⁵. These risk factors may present together and have different relative risks for

(18)

CRC. The highest relative risk is related to the histories of patients with CRC in their

first-degree relatives and people with IBD (relative risk more than 2). Taylor et al.

analyzed a population-based resource with 2,327,327 people included in three or more

generations of family histories⁶, among whom 10,556 had CRC. The relative family risk

for CRC in those with affected first-degree relatives was 2.05. Jess et al. performed a

meta-analysis of 8 population-based studies to calculate the relative risk of CRC in

patients with ulcerative colitis (UC)⁷. They demonstrated that UC increases the risk of

CRC up to 2.4 folds. Other risk factors are generally modifiable and contribute to lower

relative risk (mostly between 1.2-2.0) but are associated with a bigger disease burden in

the world⁵.

1.1.1 Carcinogenesis and Epigenetics in CRC

Many risk factors, such as DM and obesity, are associated with metabolic

dysregulation attributed to aberrant epigenetic regulation of gene expression, which can

be adaptive and responsive to environmental exposures⁸. In the Dutch Hunger Winter

cohort, the investigators recruited the residents who experienced World War II in

different regions of Dutch and followed up on the development of cancer⁹. The

individuals who lived in the Western urban regions experienced the highest energy

(19)

restriction (rationing < 700 kcal per day, 1944-45) in their childhood; their adolescence

was associated with less CRC incidence than those in other regions. The molecular

subtype that was substantially decreased in CRC incidence was CpG island methylator

phenotype (CIMP), which was characterized by diffuse hypermethylation in CpG island

throughout the genomes¹⁰. One study analyzed genome-wide methylation in the whole

blood of two groups of baboons living in different areas and having different resources

(wild feeding vs. lodge feeding). Famine exposure was associated with the most

differentially methylated regions in the key gene, phosphofructokinase platelet (PFKP),

in the glycolysis pathway between these two groups of baboons¹¹. Inspired by these two

studies, we collected CRC tumors and the adjacent normal colon tissues and analyzed

their differentially methylated genes (DMGs) by genome-wide methylation array. We

focused on the genes in the glycolysis pathway and compared our results to three open

datasets: GSE42752, GSE25062, and the Cancer Genome Atlas (TCGA). We found that

multiple loci were involved in Hexokinase containing domain 1 (HKDC1)

(hypomethylation) and Aldehyde dehydrogenase 1 family member A3 (ALDH1A3)

(hypermethylation)¹². The results warrant further study to explore the roles of these two

DMGs in CRC development.

(20)

1.2 The Prognostics of CRC: Tumor Cells

CRC is a heterogeneous disease associated with many genetic and epigenetic

alterations. Among these molecular alterations, some have prognostic implications. The

common examples of molecular prognostic markers in CRC are listed below.

KRAS and NRAS mutation

The RAS-RAF-MEK-ERK pathway is one of the most common deregulated

pathways in human cancers¹³. It mediates many important cellular signals, such as

growth, proliferation, and senescence. Activating RAS point mutations could activate

the RAS-RAF-MEK-ERK pathway and occur in about 30% of human cancers¹⁴. There

are three human RAS genes: KRAS, NRAS, and HRAS. According to several studies,

mutations in KRAS exon 2-4 and NRAS exon 2-4 accounted for 50% of metastatic CRC

and were both negative predictors for the efficacy of anti-EGFR antibody^15-17. RAS

mutations also have prognostic implications in metastatic CRC. In the TRIBE study,

which compared the combination of bevacizumab and FOLFOXIRI and the

combination of bevacizumab and FOLFIRI as the first-line treatment in metastatic

CRC, the median overall survival (OS) in the group of patients with RAS mutant was

shorter than that in the group of patients with RAS/BRAF wild type ¹⁸. In another study,

(21)

Modest et al. performed a pooled analysis of five randomized clinical trials and enrolled

1,239 patients with metastatic CRC, among whom 462 patients had KRAS mutations, 39

had NRAS mutations, and 74 had BRAF mutations. Patients with KRAS mutations had

significantly shorter OS and progression-free survival (PFS) than those with RAS/BRAF

wild-type tumors¹⁹. Tumors with KRAS G12C and KRAS G13D were significantly

correlated with shorter OS than tumors with RAS/BRAF wild type, and KRAS G12D and

KRAS G12S did not have a prognostic association. The prognostic roles of NRAS

mutations are still controversial. NRAS mutations in tumors had no significant

prognostic impact in Modest’s study but were associated with shorter OS compared to

the wild-type NRAS in tumors in another study^19,20.

BRAF mutation

BRAF is also involved in the RAS-RAF-MEK-ERK pathway²¹. Mutant BRAF

proteins contribute to elevated kinase activity independent of the upstream RAS

signaling. Somatic BRAF mutations in human cancers were first reported in 2002, and

in that report, BRAF mutations accounted for 12% of CRC²¹. BRAF mutations represent

about 4.7–8.7% of metastatic CRC; this type of tumor is significantly associated with

more proximal colon cancer, high microsatellite instability, peritoneal metastases, and

(22)

distal lymph node metastases but fewer lung metastases^22-24. Many studies reported that

BRAF mutations significantly predicted poor survival in patients with metastatic CRC,

and those with BRAF mutant CRC had a median OS of 11–14 months^23,24; V600E

mutation was the most frequent and contributed to poor prognosis. Non-V600E BRAF

mutations accounted for 2.2% of metastatic CRC, and patients with non-V600E BRAF

mutations had an excellent prognosis²⁵. In Taiwan, Tsai et al. reported that there was a

high frequency (11/59, 19%) of BRAF V600E mutations in very young patients (< 30

y/o) with CRC and patients with BRAF V600E mutations had shorter survival than

patients with wild type BRAF²⁶. Our team previously demonstrated that BRAF

mutations had a paradoxical role in the prognosis of the early and late stages of CRC²⁷.

In the late stage of CRC, BRAF V600E mutation was an independent prognostic factor,

while in the early stage of CRC, BRAF V600E mutation had no impact on prognosis.

The mechanism contributed to the paradoxical role in the prognosis of different stages

of CRC with the BRAF V600E mutation is unknown. We hypothesize that the immune

contexture in the tumor microenvironment (TME) may play an important role. The

prognostic implication of the immune contexture in TME will be described in detail in

the next section.

(23)

Mismatch repair deficiency and microsatellite instability-high

Mismatch repair deficiency (dMMR) represents defects in DNA mismatch repair

function due to germline mutations or hypermethylation of one of the mismatch repair

genes (MLH1, PMS2, MSH2, MSH6)²⁸. CRC with dMMR is characterized by many

insertion and deletion mutations in the genome and a high mutational burden^29,30.

Microsatellite instability-high (MSI-high) is a genetic fingerprint of dMMR, Generally

detected by the loss of expression of mismatch repair proteins in immunohistochemical

stains. MSI-high is determined by PCR amplification of the Bethesda panel, which

consists of five microsatellite loci (two mononucleotides and three dinucleotides), or the

Promega panel, which includes five mononucleotides. Clinically, MSI-high CRC

represents about 15% of CRC, and patients with MSI-high CRC have a better prognosis

than those with microsatellite stable (MSS) CRC^31,32. Gryfe and colleagues also

demonstrated that MSI-high CRC was associated with less frequent regional lymph

nodes and distant organ metastases than MSS CRC³¹. A meta-analysis of the prognostic

impact of MSI-high in CRC enrolling 32 clinical studies and 7,642 patients with CRC

(1,277 patients with MSI-high CRC); it demonstrated that MSI-high was an

independent factor for a favorable prognosis³². The underlying mechanism for patients

(24)

with MSI-high CRC having a better prognosis than those with MSS CRC is still

undefined. One mechanism might be due to the fact that more immunoreactive cells

infiltrated in the TME of MSI-high tumors than in MSS tumors^33,34.

Recently, dMMR was identified as a predictive biomarker for anti-PD-1 antibodies

in metastatic CRC and other cancer types. Le et al. reported that in patients with dMMR

CRC, the objective response rate (ORR) for anti-PD-1 antibody was 40% (4/10), but

there was no responder in the group with proficient mismatch repair CRC (0/18, 0%). In

addition, the ORR was also high (4/7, 57%) in CRC with non-dMMR ³⁵. Other studies

further supported the predictive role of dMMR in response to anti-PD-1 antibodies in

metastatic CRC and other cancers^36,37. These results led to the FDA approval of

pembrolizumab and nivolumab to treat metastatic dMMR/MSI-high CRC and

pembrolizumab as a tissue/site agonistic therapy in metastatic dMMR or MSI-high solid

tumors³⁸. A further randomized controlled study proved that pembrolizumab was

superior to the standard of care in first-line therapy for metastatic CRC with

dMMR/MSI-high³⁹. The high response of anti-PD-1 therapy in dMMR or MSI-high

tumors may be attributed to high indel mutational load, increased neoantigens amount,

DNA sensing, and vigorous infiltrative CD8+ T cells 35,36,40,41. Recently, anti-PD-1

(25)

antibodies with and without an anti-CTLA4 antibody were investigated in nonmetastatic

CRC, and the results also showed favorable clinical and pathological response rates^42,43. 1.3 The Prognostics of CRC: Epigenetics

In addition to genetic alterations, some studies also revealed that epigenetic

alterations might be prognostic factors⁴⁴. CIMP-high is one of the common epigenetic

alterations reported as a prognostic factor.

CIMP

CIMP-high CRC was originally recognized by Toyota and colleagues in 1999.

They used methylated CpG island amplification to identify several cancer-specific

clones (type C methylation)⁴⁵ and identified a subgroup of CRC with a high level of

type C methylation; those tumors were characterized by proximal colon cancer and

MSI-high. Thus, a new pathway in CRC was identified, and a new CRC carcinogenesis,

mixed with genetic alterations and inactivation of tumor suppressor genes by

hypermethylation, was proposed⁴⁶. During the following years, several panels were

developed to determine CIMP-high CRC, including the classic panel (MINT1, MINT2,

MINT31, p16, and MLH1), Weisenberger’s panel (CACNA1G, IGF2, NEUROG1,

RUNX3, and SOCS1), and others^47-50. Although these studies used different panels, they

(26)

revealed common clinicopathologic features of CIMP-high CRC: old age, female,

proximal colon cancer, BRAF V600E mutation, and MSI-high. Recently, a methylation

array was used to classify CIMP CRC and showed that CIMP-high CRC was enriched

in hypermutated tumors, probably due to the overlap of hypermutated tumors and MSI-

high tumors²⁹.

The prognostic role of CIMP in CRC has been extensively investigated, but the

results were inconsistent^51,52. Several confounding factors exist in previous studies: first,

the overlap of the other prognostic factors with CIMP-high, such as MSI-high and

BRAF V600E mutation; second, the contribution of anti-cancer therapy to prognosis;

finally, a paradoxical role in prognosis with a molecular marker. For example, we

previously demonstrated that BRAF V600E mutation had a paradoxical role in the early

and late stages of CRC²⁷. Thus, we investigated the prognostic role of CIMP in different

stages (Stages 1-4) of CRC in a retrospective cohort⁵³. We performed multivariate

analysis to adjust the prognostic effects of molecular factors and anti-cancer therapies.

The results revealed that in patients with Stage 1-3 CRC, CIMP-high did not predict

poor prognosis; however, in those with Stage 4 CRC, CIMP-high was significantly

associated with shorter survival. The impact of CIMP-high was still significant in a

(27)

multivariate analysis. Thus, we concluded that CIMP-high was an independent

prognostic marker for Stage 4 CRC.

1.4 The Prognostics of CRC: TME

In addition to genetic and epigenetic alterations, immune contexture in TME has a

prognostic role in many types of cancer^54,55. The immune contexture consists of many

types of immune cells, chemokines, cytokines, and special structures, such as tertiary

lymphoid structure (TLS). TLS is an ectopic lymphoid aggregate adjacent to the tumor

bed. It contains B cells, follicular helper T cells, and follicular dendritic cells and is

usually generated during chronic inflammation. The study of immune contexture

includes the analysis of the density, location, and functions of different types of immune

cells through histopathologic or bioinformatic tools. During the past two decades, the

clinical impact of immune contexture has been extensively investigated, and the results

are quite diverse in different types of immune cells. For example, CD8+T cells and TLS

are commonly correlated to good prognosis; in contrast, Treg cells and M2

macrophages are associated with poor prognosis in many cancer types⁵⁴.

Pages et al. first showed the infiltrating memory CD45RO+ cells were negatively

associated with early metastatic invasions and advanced stages of CRC and survival of

(28)

patients with this disease⁵⁶. Then Galon and colleagues found that tumors with Th1

adaptive immunity (CD3, CD8, GZMB, and CD45RO) had a favorable prognosis⁵⁷.

Combination analysis of the type, density, and location of these immune cells and their

associated molecules had a prognostic value that was independent to and superior to the

UICC-TNM stage. These findings led to the development of the immunoscore, which

was determined by the density of CD3+ and CD8+ T cells at the core and the invasive

front of the tumor⁵⁸. The scoring system ranges from low (Immunoscore 0) to high

(Immunoscore 4). Many studies demonstrated that immunoscore outperformed the

TNM stage in predicting PFS and OS in early-stage CRC^58-60. In addition to CD8+ T

cells, many other immune cells have been reported to be a good prognostic factor in

CRC, such as Th1 cells, Tfh cells, B cells, NK cells, and TLS, and M2 tumor-associated

macrophages (TAMs) are correlated to a poor prognosis^55,61-63. Understanding the

impact of immune contexture on clinical outcomes may provide insight into developing

novel treatments for CRC.

Our previous study identified that CIMP-high was correlated to a poor prognosis of

CRC, and the frequency of CIMP-high tumors (14.7%) in early-onset CRC (EOCRC,

diagnosed at age < 50 y/o) seemed to be higher than that in Western populations (5%).

(29)

Thus, in this study, we are interested in the actual frequency and risk factors of CIMP-

high CRC in Taiwan. Additionally, we demonstrated that BRAF V600E mutation

predicted a poor prognosis in late-stage CRC. Because immune contexture is a

prognostic factor in CRC, they may have prognostic implications in BRAF V600E

mutant CRC. Thus, we hypothesize that immune contexture in TME of metastatic

BRAF V600E-mutated CRC is a prognostic factor. In the following two chapters, we

focus on these two research topics.

(30)

Chapter II. The Epigenetics of CRC ^－ From Prognostics to Carcinogenesis

2.1 Introduction

CIMP-high CRC accounts for 15-20% of all CRC and is characterized by a wide range of DNA hypermethylation in many promoter CpG islands⁵⁰. This type of CRC is associated with clinicopathologic features, such as old age, female, proximal colon cancer, MSI-high, and BRAFV600E mutation, and shorter survival in metastatic disease^47,53,64. The etiology of CIMP-high needs to be clearly defined, but its crosstalk with environmental factors and epigenetic alterations has been well studied^8,65. Thus, understanding the risk factors of CIMP-high CRC is important. The risk factors of CIMP-high CRC have been studied in Western populations. Weisenberger and colleagues analyzed the associations among CIMP-high genotype, risk factors, and clinicopathologic variables in the Colon Cancer Family Registry cohort that recruited patients with CRC from USA, Canada, and Australia. The results showed that the use of non-steroidal anti-inflammatory drugs (NSAIDs) was significantly associated with CIMP-high CRC in all patients, and smoking and obesity were correlated with CIMP- high CRC only in female patients⁶⁴. However, a relationship between CIMP-high CRC and risk factors in an East Asian population remains to be investigated.

In our previous cohort, we found that the frequency of CIMP-high CRC was 16.4%

(N=450), which was in the similar range of the frequency reported in Western

populations⁵³. However, in Taiwan, the frequency of CIMP-high CRC in EOCRC

(14.7%) seemed to be higher than that in the Western populations (5%). Lui et al. also

(31)

described that CIMP-high tumors accounted for 17.9% of EOCRC in a Chinese

cohort⁶⁶. Both studies indicate that CIMP-high CRC was not associated with old

age^53,66.

On the other hand, studies indicate that methylations could be adaptive to

environmental exposures⁸. The baboon study and Dutch Hunger Winter study proved

that methylations were changed after exposure to famine^10,11. Thus, we hypothesized

that a high frequency of CIMP-high tumors in EOCRC in Taiwan, and those CIMP-high

tumors were associated with environmental factors, such as body mass index (BMI).

Therefore, we initiated a prospective cohort study to enroll patients diagnosed with

CRC in 2016 to prove our hypotheses.

2.2 Methods

Patient eligibility

The target number of recruited patients with CRC was 320, and more than 30%

should be EOCRC. Patients who met the following inclusion criteria were eligible for

this study: (1) age ≥ 20 y/o, (2) cytologically or pathologically proven CRC and

adequate staging (American Joint Cancer Committee on Cancer, 7^th edition) by

computed tomography, and (3) undergoing a colectomy surgery. The exclusion criteria

(32)

were that they received systemic chemotherapy or radiotherapy before colectomy.

Written informed consent was obtained from all patients before collecting the tumor

specimens. We also collected the patients’ clinical and pathological characteristics,

including age, sex, histology and grade of tumors, tumor location, the risk factors of

CRC, and BMI. The definition of these clinicopathologic variables is described below.

For age, the patients were categorized as age < 50 y/o and age ≥ 50 y/o. The histology

of tumors included the presence or absence of mucinous components, signet-ring cells,

and medullary histology, which were observed by microscopic examination. Mucinous

carcinoma was designated if more than 50% of the tumor volume was a mucinous

component. The tumor grade was classified into low and high grades. Tumors were

graded as a low grade if ≥ 95% of the tumor has glandular formation or MSI-high.

Otherwise, tumors were graded as a high grade. Tumor location was grouped into right-

sided and left-sided. Right-sided tumors were designated if the primary tumors were

located from the cecum to the splenic flexure of the transverse colon, and left-sided

tumors were located from descending colon to the rectum. The Institute Review Board

of National Taiwan University Hospital (NTUH) approved this study.

Risk factors

(33)

In this study, risk factors include first-degree family history for CRC, history of

colorectal polyp, DM, hyperlipidemia, BMI at diagnosis, pre-diagnosis use of NSAIDs,

and hormonal replacement therapy (HRT). These data were obtained from interviews

with the patients and the survey of the electronic chart of NTUH. The definitions of

these risk factors are described below. Patients with more than one first-degree family

member diagnosed with CRC were grouped into a group with a family history of CRC.

BMI at diagnosis was retrieved from a medical record. We categorized BMI into three

groups based on WHO criteria of 18–24.9, 25–29.9, and ≧30 kg/m². We also

categorized BMI into two groups (< 27.5 and ≧ 27.5 kg/m²) according to the experts’

opinion for appropriate BMI in Asian populations from WHO expert consultation⁶⁷. In

that report, a BMI of ≧ 27.5 kg/m²was regarded as a high risk for public health.

Histories of DM and hyperlipidemia were determined according to self-reports and/or

the identification of the use of medication for these diseases. Pre-diagnosis NSAIDs use

was defined if two or more times per week in one month or longer within one year of

NSAIDs use was reported. Pre-diagnosis of HRT use was determined if six months or

longer of HRT use within one year was reported.

CIMP analysis

(34)

We used a QIAamp DNA formalin-fixed paraffin-embedded (FFPE) tissue kit to

extract genomic DNA from tumor specimens (Qiagen, Santa Clarita, CA, USA). Then

the genomic DNA was treated with sodium bisulfide according to the EZ DNA

Methylation Kit protocol (Zymo Research, Irvine, CA, USA). We evaluated the

methylation status of five loci: MINT1, MINT2, MINT31, p16, and MLH1 (classical

panel) using MethyLight assay, which was a probe-based, methylation-specific real-

time polymerase chain reaction technology. The primers used for the MethyLight study

are listed in Table 1. The methylation-independent Alu repeat was measured for

normalization control reaction. The methylation levels of five loci in tumor samples and

a constant reference sample were measured, and quantification analysis of DNA

methylation was performed. The percentage of methylated reference (PMR) of each

locus was calculated based on the equation proposed in a previous study as the

following: PMR = 100 × (methylated reaction / control reaction)sample / (methylated

reaction / control reaction)M.SssI-Reference. A PMR of > 10 was regarded as

hypermethylated⁴⁹. Finally, CIMP-high tumors were determined if there were three or

more loci with a PMR of >10 identified. Otherwise, the tumors were determined as

CIMP-low/negative.

(35)

RAS/BRAF mutation analysis

We evaluated RAS (KRAS and NRAS) and BRAF mutations by PCR coupled with

Sanger’s sequencing method. The primers in this study covered exons 2, 3, and 4,

including codons 12, 13, and 61 of KRAS and NRAS, and covered exon 15 of BRAF.

After PCR, we used an automated ABI 3730 sequencer (Applied Biosystems, Foster

City, CA, USA) for direct sequencing.

MSI analysis

We used the MSI Analysis System (Promega Corp., Madison, WI, USA), a PCR-

based assay for detecting MSI, in five mononucleotide loci (NR-21, NR-24, MONO-27,

BAT-25, and BAT-26). Tumors were designated MSI-high if abnormal allele length

was identified in two or more loci. Otherwise, tumors were MSS.

Statistical analysis

The patient’s clinicopathological characteristics were presented as percentages and

frequencies. The association between CIMP-high CRC and clinicopathological

variables was estimated by logistic regression (Table 2), Fisher’s exact test, and Chi-

square test (Table 3 and Table 6). The association between risk factors and CIMP-high

CRC was also evaluated by logistic regression (Table 4 and Table 7). Finally, we used a

(36)

multivariate logistic regression model to evaluate the odds ratio of each variable for

CIMP-high tumors. Only a 2-sided p-value of ≤ 0.05 was considered statistically

significant. Only the variables with a p-value less than 0.1 in univariate analysis (Table

3 and Table 4) were put into a multivariate logistic regression model. All the statistical

analyses were performed using SAS statistical software (version 9.4, SAS Institute,

Cary, NC, USA).

2.3 Results

From March 2016 to June 2019, we enrolled 320 patients, among whom 99

(30.9%) were younger than 50 y/o. The clinicopathological characteristics of these

tumors and the risk factors of CRC were well-collected. Among collected tumor

samples, 293 (91.6%) had enough genomic DNA for CIMP analysis, and 288 (90%),

286 (89.4%), and 318 (99.4%) were adequate for RAS mutation, BRAF V600E mutation,

and MSI analyses, respectively. Finally, 293 tumors were enrolled in the primary

analysis, among which 89 (30.4%) were EOCRC.

In the primary analysis, CIMP-high CRC accounted for 15.4% of all CRC. The

frequencies of CIMP-high CRC in patients aged < 50 y/o and aged ≥ 50 y/o were 15.7%

(14/89) and 15.2 % (31/204), respectively. The distribution of CIMP-high and CIMP-

(37)

low/negative tumors in clinicopathological variables is presented in Table 2. The key

findings are described briefly here. CIMP-high CRC was significantly associated with

BRAF V600E mutation (p < 0.01), MSI-high genotype (p < 0.01), and right-sided

primary tumor (p = 0.03). In contrast, it was not associated with gender (p = 0.49) and

old age (p = 0.91). Then we further evaluated the distribution of CIMP-high and CIMP-

low/negative tumors in clinicopathological variables in two age groups (age < 50 y/o

and age ≥ 50 y/o), and the results are shown in Table 3. In EOCRC, CIMP-high tumors

were significantly associated with BRAF V600E mutation (p < 0.01) and stage IV at

diagnosis (p < 0.01). In contrast, CIMP-high tumors were significantly associated with

right-sided primary tumor (p = 0.02), BRAF V600E mutation (p < 0.01), and MSI-high

(p < 0.01) in late-onset CRC (LOCRC, diagnosed at age ≥ 50 y/o).

The distributions of CIMP-high and CIMP-low/negative CRC associated with

various risk factors in overall patients and patients aged < 50 y/o and aged ≥ 50 y/o are

shown in Table 4. In patients younger than 50 yrs, CIMP-high CRC was likely

associated with a BMI of ≥ 27.5 kg/m² (p = 0.09). In contrast, in those 50 yrs or older,

CIMP-high CRC was significantly associated with colorectal polyp (p = 0.03). In the

multivariate logistic regression model, we found BMI of ≥ 27.5 kg/m² (p = 0.03) and

(38)

Stage 4 disease at diagnosis (p < 0.01) were independent factors for early-onset CIMP-

high CRC; however, only MSI-high (p < 0.01) was an independent factor for late-onset

CIMP-high CRC (Table 5). We also performed the statistical analysis of CIMP-high

CRC, clinicopathological features, and risk factors by gender, and the results are

presented in Tables 6 and 7, respectively.

2.4 Discussion

In this prospective study, we showed that the high frequency (15.7%) of CIMP-

high CRC in EOCRC was consistent with the result from our previous study (14.3%,

11/77)⁵³. These findings might not be related to the classical panel because a similar

frequency of CIMP-high genotype (17.9%) in EOCRC was found in a Chinese cohort,

and CIMP-high tumors were determined by ≥ 3/5 of the Weisenberger’s panel in that

study⁶⁶. In addition, one study demonstrated no significant difference between these two

panels in the correlation to clinicopathological features⁶⁸. Thus, there may be different

frequencies of CIMP-high tumors in EOCRC between East Asian and Western

populations; this warrants further large-size studies to confirm this molecular

epidemiology difference.

In addition to the frequency of CIMP-high CRC, we also found significant

(39)

associations between CIMP-high CRC and BRAF V600E mutation, MSI-high, right-

sided primary tumor, or stage 4 at diagnosis. These findings are consistent with previous

analyses^53,64,69. We demonstrated that CIMP-high CRC had no association with old age

and females, different from the data from Western populations. Moreover, we showed

different associations among risk factors, clinicopathological features, and a CIMP-high

genotype in EOCRC and LOCRC. In multivariate analysis, a BMI of ≥ 27.5 kg/m² was

the only independent risk factor for CIMP-high CRC in patients younger than 50. The

WHO expert consultation had added the thresholds of 23, 27.5, 32.5, and 37.5 kg/m² for

the operation of public health in Asian populations in 2004 because the prevalence of

cardiovascular diseases and type II DM were increased and the average BMI in these

populations was below 25 kg/m²⁶⁷. People with a BMI of ≥ 27.5 kg/m²represented the

high-risk groups. Our study indicates the importance of maintaining an adequate BMI,

especially in people younger than 50. In Weisenberger’s study, the investigators did not

evaluate the association between risk factors and CIMP-high CRC in different age

groups. Still, they found a higher BMI was significantly associated with CIMP-high

CRC in female patients⁶⁴. However, our study did not identify this association (Table

7). So further study is needed to clarify these controversial findings.

(40)

Why was CIMP-high CRC significantly correlated to a high BMI in younger

patients? An analysis of the Dutch Hunger Winter cohort revealed that early-life

exposure to famine was significantly associated with decreased incidence of CRC,

specifically, the CIMP-high CRC⁷⁰. In an animal model, the caloric restriction could

delay methylation drift during aging⁷¹. Thus, the correlation between a high BMI and

CIMP-high CRC may be related to calorie overload-induced biologic effects. On the

other hand, epigenetic aging may also play a role. Epigenetic aging is proposed to be the

accumulating work done by an epigenetic maintenance system⁷². Epigenetic aging can

be evaluated by genome-wide DNA methylation array, and accelerated epigenetic aging

is associated with CRC, especially CIMP-high CRC⁷³. Obesity also accelerates

epigenetic aging in multiple kinds of human tissues^72,74. Thus, the excessive

accumulated aging-related methylations in obese or overweight patients may contribute

to hypermethylated cancers. The next question is why the significant association

between CIMP-high CRC and high BMI was only observed in younger patients aged <

50 y/o but not in older patients aged ≥ 50 y/o. A recent report provided potential

explanations. Nevalainen’s study showed the correlation of an increased BMI with

accelerated epigenetic age only in middle-aged individuals (40-49 y/o) but not in

(41)

young-aged (15-24) and very old (> 90 y/o) individuals⁷⁵. Instead, they observed a high

BMI was correlated with a lower epigenetic age in very old individuals. Based on these

findings, we hypothesize that a higher BMI accelerates epigenetic aging, and it should

have a long enough time to let obesity-induced epigenetic aging contribute to CRC in

patients with middle age. But for older patients with CRC with a high BMI, the etiology

may be different because they have more time (chances) to acquire multiple epigenetic

and genetic alterations, so obesity-induced accelerated aging has less influence.

There are several limitations in the present study. First, the sample size is relatively

small, which may confound the results of the associations between CIMP-high CRC and

variables in different age and gender groups. Second, we lack reliable data on smoking

and alcohol consumption and could not explore the association between them and

CIMP-high CRC. Many studies indicate the dose-dependent relationship between

smoking and CIMP-CRC, so the absence of the smoking analysis would be

influential^64,69. Regarding the association between alcohol consumption and CIMP-high

CRC, the results were conflicting^64,76. This study has strengths. This study is a

prospective cohort study, so we can easily collect clinical data and risk factors like BMI.

Our study first explored the link between CRC risk factors and CIMP-high CRC in an

(42)

East Asian country. Our findings indicate that the association between CIMP-high CRC

and some risk factors (age, sex, and BMI) may differ in East Asian and Western

populations despite similar associated clinicopathological characteristics.

(43)

Chapter III. The Study of TME in Metastatic BRAF V600E mutant CRC

3.1 Introduction

BRAF mutation is present in 4-8% of metastatic CRC, and BRAF V600E mutation

is the most common variant of BRAF mutations^23,24. Right-sided primary tumors,

peritoneal and distant lymph node metastases, and MLH1 hypermethylation are

significantly associated with BRAF mutations in CRC, while KRAS mutation is

mutually exclusive with BRAF mutations^22,77. The consensus molecular subtypes (CMS)

of CRC have recently been proposed based on pooled transcriptomic analyses⁷⁸. There

are four CMS subtypes (CMS 1-4); among them, CMS1 (MSI-Immune) tumors are

characterized by more MSI-high, CIMP-high, BRAF V600E mutation, and infiltrative

immune cells. BRAF mutations are also categorized into two BRAF mutant

classifications (BM)⁷⁹. Activation of the KRAS/AKT pathway, dysregulation of

mTOR/4EBP, and epithelial-mesenchymal transition signaling occur in the BM1

subtype, and cell-cycle dysregulation is noted in the BM2 subtype. Kopetz et al.

recently presented that 48.8%, 2.9%, 11.8%, and 36.5% of BRAF V600E mutant CRC

corresponded to CMS1, CMS2 (canonical), CMS3 (metabolic), and CMS4

(44)

(mesenchymal) subtype in the BEACON study⁸⁰. They also showed that 33.1% and

66.9% of BRAF V600E mutant CRC belonged to BM1 and BM2 subtypes, respectively.

In addition, CMS4 and BM1 subtypes were correlated with a higher objective response

to cetuximab, encorafenib and binimetinib treatment. These findings revealed that there

was heterogeneous gene expression in the TME of BRAF V600E mutant CRC,

contributing to different efficacies of targeted therapy.

BRAF V600E mutation has been recognized as a poor prognostic factor in

metastatic CRC, and patients with this genotype had approximately 11-14 months of

OS, much shorter than the survival time in those with RAS/BRAF wild type and those

with RAS mutations^18,81-83. The mechanisms underlying the poor prognosis of BRAF

V600E mutant CRC remain unknown. As mentioned in the earlier chapter, immune

contexture in TME correlates with the clinical outcomes in patients with CRC. It is

reasonable to hypothesize that the immune contexture may also have prognostic

implications in patients with metastatic BRAF V600E mutant CRC. Currently, the

transcriptomic data of BRAF V600E mutant CRC was mainly derived from early-stage

tumors. Still, Mlecnik et al. showed the expression of immune-related genes in primary

CRC tumors differed between metastatic and non-metastatic diseases. Thus, it is

(45)

important to analyze the immune contexture in metastatic BRAF V600E mutant CRC.

The development of anti-PD-1/PD-L1 antibodies has changed the paradigms in

many cancer types⁸⁴. dMMR (or MSI-high) and high tumor mutational burden are the

predictive biomarkers for anti-PD-1/PD-L1 antibodies^36,84,85. According to previous

studies, 42-48% of BRAF V600E mutant CRC belongs to CMS1 (Immune), which holds

great potential for novel immunotherapy^78,80. Indeed, clinical trials showed good

responses in CRC concomitant with BRAF V600E mutation and MSI-high; however, for

MSS BRAF V600E mutant CRC, anti-PD-1 monotherapy was ineffective, and the

combination of targeted therapy and anti-PD-1 therapy is recently under investigation^86-

89. We further hypothesize that a novel treatment for metastatic MSS BRAF V600E

mutant CRC can be developed after illustrating the prognostic roles of immune

contexture of TME in these tumors. Thus, we designed this study to recruit patients with

de novo Stage 4 MSS BRAF V600E mutant CRC and perform the transcriptomic

analysis on their primary tumors.

3.2 Methods

Patient enrollment

(46)

Patients who met the following inclusion criteria were eligible for this study: (1)

they had pathologically confirmed de novo metastatic BRAF V600E mutated CRC,

which referred to distant metastases at diagnosis; (2) they had colectomy for CRC; and

(3) they had MSS CRC. Patients who received systemic chemotherapy or targeted

therapy before colectomy were ineligible for this study. We enrolled patients from two

medical centers: NTUH and Kaohsiung Chang Gung Memorial Hospital (KCGMH,

Kaohsiung, Taiwan). At NTUH, we searched the medical records on the BRAF mutation

examination from 2010 to 2017. The codes for BRAF mutation analyses included

000W0406 for expanded RAS+BRAF mutation, 000W0389 for KRAS and BRAF

mutations, and 000W0404 for BRAF mutation alone. Then we could identify patients

with de novo metastatic BRAF V600E mutated CRC. In addition, we included the

patients who met the inclusion/exclusion criteria from our previous cohort (2005-2013).

Finally, another cohort from KCGMH was also enrolled. We collected patients’

clinicopathological characteristics, in terms of age, sex, histology, tumor grade, disease

stage at diagnosis, primary tumor location, systemic chemotherapy, targeted therapy,

and survival time from electronic medical records. Patients were recognized to have

right-sided tumors if the primary tumors were located from the cecum to the splenic

(47)

flexure of the transverse colon. Patients had left-sided tumors if the primary tumors

were located from the descending colon to the rectum. The definition of first-line triplet

chemotherapy was the anti-cancer therapy consisting of oxaliplatin, irinotecan, and one

of the fluoropyrimidines. The first-line doublet referred to the anti-cancer therapy

consisted of one of the fluoropyrimidines and either oxaliplatin or irinotecan. This study

was approved by the Institute Review Board of NTUH and KCGMH.

BRAF V600E mutational analysis

The methods for detecting BRAF V600E mutation at NTUH include PCR and

direct sequencing (ABI 3730 sequencer (Applied Biosystems, Foster City, CA, USA).

The primer pairs (5’-TCA TAATGCTTGCTCTGATAGGA-3’ and 5’ - GGCCA

AAAATTTAATCAGTGGA-3’) covering BRAF exon 15 were used. On the other hand,

BRAF V600E mutation was identified by Roche BRAF/NRAS Mutation Test (Roche

Diagnostics, IN, USA) at KCGMH following the manufacturer’s instructions. We

performed BRAF mutation subtype classification (BM1 and BM2) by linear regression

classifier as described in a previous study⁷⁹.

MSI analysis

The method for MSI testing and the definitions of MSI-high and MSS have been

(48)

described in chapter 2.

Transcriptomic analysis

We extracted total RNA from the FFPE samples of primary colon tumors derived

from colectomy using Qiagen’s miRNeasy FFPE Kit (Qiagen, Valencia, CA, USA).

Macrodissection of tumor slides was performed before RNA extraction. Then we used

NanoDrop 2000 at 220–320 nm (Thermo Scientific) to measure 260/280 ratios to

determine RNA quality and quantity. Bioanalyzer (model 2100, Agilent Technologies,

Santa Clara, CA, USA) was used to measure nucleic acid fragmentation. Finally, the

gene expression was evaluated by nCounter PanCancer Immune Profiling Panel

(NanoString, Seattle, WA, USA). This panel consists of 770 genes and is widely used to

investigate TME. We processed raw data as previously described⁹⁰. Gene counts for

each sample were normalized first by trimmed mean of M-values method⁹¹ and then

converted to log2-transformed counts per million later for clinical analysis.

Development of complement scores

We calculated pairwise expression correlation in complement genes according to

the similarity metric of Pearson’s correlation coefficient⁹². The results were regarded as

an input for hierarchical clustering. Then the complement genes were classified into

(49)

three co-expression modules, and the clustering results were shown by a heatmap

symmetrical along the diagonal line. The complement score of each sample was

determined as the average gene expression in module 2. A high complement score was

defined when the complement score was higher than the median score; in contrast, a

low complement score was defined when the complement score was lower than the

median score.

Evaluation of immune cell abundance

The immune cell abundance in the TME of each tumor was measured according to

its expression profiles. The simple average of marker genes’ expression values of each

immune cell was calculated as cell-type scores. The reference marker genes were

determined by Danaher et al.⁹³. We also used the CIBERSORT algorithm with the

LM22 signature matrix to measure cell abundance⁹⁴. The phenotypes of macrophages

(M1 and M2) were defined based on the mean expression values of gene signatures⁹⁵.

The genes correlated to M1 macrophage consist of CD40, CD86, CCL5, CCR7, CXCL9,

CXCL10, CXCL11, IL1A, IL1B, IL6, IRF1, IRF5, IDO1, and KYNU, and the M2

macrophage-associated genes are CD276, CCL4, CCL13, CCL18, CCL20, CCL22,

(50)

CTSA, CTSB, CTSC, CTSD, CLEC7A, FN1, IL4R, IRF4, LYVE1, TGFB1, TGFB2,

TGFB3, TNFSF8, TNFSF12, MMP9, MMP14, MMP19, MSR1, VEGFA, VEGFB, and

VEGFC.

External cohorts

The first external cohort was colon and rectal adenocarcinomas in the TCGA

dataset⁹⁶. Genomic and transcriptomic data of these tumors were obtained from

Genomic Data Commons through the R package of TCGA bio-links, and these tumors’

microsatellite status was obtained according to the method described preciously⁹⁷. The

second external cohort was GSE39582 dataset⁹⁸. Complement scores of all these CRC

tumors from two external cohorts were calculated, and tumors were determined to have

high or low complement scores based on the abovementioned definitions.

Immunohistochemical staining

The collected FFPE tumor tissues were cut into 5-µm sections, and then these

sections were deparaffinized by EZ prep (Ventana Medical Systems, Inc., Tucson, AZ,

USA). These sections were also pretreated with Cell Condition 1 solution for 32

minutes (Ventana Medical Systems, Inc.). We incubated the slides with anti-human C4d

(clone A24-T, DB Biotech) and used the automated Ventana Benchmark XT for 1:200

(51)

titration (Ventana Medical Systems, Inc.). Labeling was measured using the Optiview

DAB Detection Kit (Ventana Medical Systems, Inc.). All sections were counterstained

with hematoxylin in Ventana reagent, and the positive controls were the human FFPE

tonsil tissues.

C4d is generated after C4 splitting during the activation of the complement

pathway. C4d is a surface-bound protein that binds to tissue near the activation site

because its covalent bond does not break spontaneously⁹⁹. Thus, C4d is recognized as a

marker of complement activation. We categorized CRC into three groups based on the

percentages of C4d deposits on the tumor cell membrane. C4d-IHC 0 (weak) referred to

<1% of non-neoplastic cells, C4d-IHC 1 (intermediate) was defined as 1%–30% of non-

neoplastic cells, and C4d-IHC 2 (high) was defined as >30% of non-neoplastic cells.

Semi-quantification was evaluated by two independent pathologists.

Statistical analysis

Patients’ clinical and pathological data were summarized as percentages. Survival

analyses among groups, including PFS, OS, and progression-free interval (PFI), were

evaluated using the Kaplan-Meier method. A p-value of less than 0.05 in the log-rank

test was considered significant. We developed a Cox proportional hazards regression

(52)

model for survival analysis with the coxph function implemented in the R package. PFS

in this study was defined as the period from the date of diagnosis to the date of

radiographically confirmed progression of disease or death during first-line therapy. OS

in this study was defined as the time from diagnosis to death. PFI was specifically used

in TCGA and GSE39582 datasets to evaluate clinical outcomes^96,98. PFI was defined as

the time from the date of diagnosis to the date of the first evidence of disease

progression, locoregional recurrence, new primary tumor, distant metastasis, or death.

Only variables with p < 0.1 were included in the multivariate analysis. We used

Benjamini and Hochberg multiple-testing correction method to adjust multiple tests.

The statistical analyses and visualization in this study were performed in an R

environment (version 3.6.1).

3.3 Results

A total of 54 patients with de novo metastatic CRC with MSS BRAF V600E were

included in this study, among whom 31 were enrolled from the National Taiwan

University Hospital-integrated Medical Database (NTUH-iMD), five from our previous

cohort²⁷, and 18 from KCGMH. The detailed consort diagram of the enrollment of

patients is presented in Figure 2. The patient’s clinicopathological features are presented

(53)

in Table 8. Briefly, the median age of patients was 51 years old; among these patients,

79.6% had low-grade tumors, 63% had right-sided primary tumors, and 42% had

peritoneal metastases. No patients had concomitant BRAF and KRAS mutations. Two

patients did not receive anti-cancer therapy after surgery and thus were not included in

the PFS analysis; their median PFS was 5.1 months, and their median OS was 16.8

months (Figure 3).

After profiling the transcriptome of all tumors, the association between immune

cell subtypes and patient survivals revealed that CD4+ T cell and B cell signatures were

significantly correlated with longer PFS (Hazard ratio, HR= 0. 61, p = 0.005 for CD4+

T cell and HR = 0.72, p = 0.039 for B cell), and no immune cell types were significantly

associated with patient OS (Table 9). In the analyses of the association between immune

gene expression and patient survival, we found some checkpoints responsible for

linking B cell and T cell signaling, such as ICOS (inducible T cell costimulatory) (PFS,

HR = 0.0.71, p = 0.035), ICOSLG (inducible T cell costimulatory ligand) (OS, HR =

0.69, p = 0.034) and CD40LG (CD40 ligand) (PFS, HR = 0.62, p = 0.010), were

associated with longer survival time. In contrast, there was no association between

CD40, CD80, or CD86 and prognosis (Table 10). Besides, there were prognostic

大腸直腸癌預後學之研究－ 著重表觀遺傳學和腫瘤微環境之解析

國立臺灣大學醫學院腫瘤醫學研究所 博士論文