Host-cell DNA methylation biomarkers in cervical cancer screening towards all molecular screening on self-collected samples Lisanne Verhoef Host-cell DNA biomarkers in cervical cancer screening towards all molecular screening on self-collected samples Lisanne Verhoef


Printing of this thesis was financially supported by: • ABN Amro Bank • Afdeling Pathologie, Amsterdam UMC, locatie VUmc • Bridea Medical B.V. • Cancer Center Amsterdam • Chipsoft B.V. • Copan Italia • Female Cancer Foundation • Methylomics B.V. • Rovers Medical Devices B.V. • Self-screen B.V. • Vrije Universiteit Amsterdam DOI http:/ ISBN/EAN 978-94-6473-464-5 Cover design Lars Vrooman, Lay-out Rogier Vrooman & Lisanne Verhoef Printing Ipskamp printing | © Copyright 2024. Lisanne Verhoef All rights reserved. No part of this thesis may be reproduced, stored or transmitted in any way or by any means without the prior permission of the author, or when applicable, of the publishers of the scientific papers.

VRIJE UNIVERSITEIT HOST-CELL DNA METHYLATION BIOMARKERS IN CERVICAL CANCER SCREENING TOWARDS MOLECULAR SCREENING ON SELF-COLLECTED SAMPLES ACADEMISCH PROEFSCHRIFT ter verkrijging van de graad Doctor aan de Vrije Universiteit Amsterdam, op gezag van de rector magnificus prof.dr. J.J.G. Geurts, in het openbaar te verdedigen ten overstaan van de promotiecommissie van de Faculteit der Geneeskunde op donderdag 30 mei 2024 om 11.45 uur in een bijeenkomst van de universiteit, De Boelelaan 1105 door Lisanne Verhoef geboren te Utrecht

promotoren: dr. D.A.M. Heideman prof.dr. J. Berkhof copromotoren: dr. M.C.G. Bleeker prof.dr. R.D.M. Steenbergen promotiecommissie: prof.dr. M.A. van de Wiel prof.dr. M. van Engeland prof.dr. G.A. Meijer prof.dr. E. Schuuring dr. N.E. van Trommel dr. I. de Kok


TABLE OF CONTENTS 40 8 62 86 108 124 144 164 192 202 214 CHAPTER 1 General introduction and thesis outline CHAPTER 2 Evaluation of six methylation markers derived from genome-wide screens for detection of cervical precancer and cancer CHAPTER 3 Performance of DNA methylation analysis of ASCL1, LHX8, ST6GALNAC5, GHSR, ZIC1 and SST for the triage of HPV-positive women: results from a Dutch primary HPV-based screening cohort CHAPTER 4 Evaluation of DNA methylation biomarkers ASCL1 and LHX8 on hrHPVpositive self-collected samples from primary HPV-based screening CHAPTER 5 Direct bisulphite conversion of cervical samples for DNA methylation analysis CHAPTER 6 Colposcopy referrals and CIN3+ risk after triage of HPV-positive women with low-grade cytology by FAM19A4/miR124-2 methylation, ASCL1/LHX8 methylation and/or HPV genotyping CHAPTER 7 Low methylation marker levels among HPV-vaccinated women with cervical high-grade squamous intraepithelial lesions CHAPTER 8 General discussion and future perspectives CHAPTER 9 English summary CHAPTER 10 Nederlandse samenvatting | Dutch summary CHAPTER 11 Appendices

1 General introduction and thesis outline Chapter 1 This chapter provides a general introduction, describing the aetiology of cervical cancer, HPV and its role in cervical carcinogenesis, strategies for cervical cancer prevention and the introduction of host-cell DNA methylation as a potential molecular biomarker for cervical cancer and its precursors.

CHAPTER 1 General introduction and thesis outline

12 Chapter 1 1.1 CERVICAL CANCER 1.1.1 EPIDEMIOLOGY Cervical cancer is the fourth most common cancer among women worldwide, after breast, colorectal and lung cancer 1, 2. In 2020, an estimated 341,831 women died from cervical cancer worldwide. Cervical cancer disproportionately affects women in lowmiddle income countries (LMIC). Almost 85% of cervical cancers occur in women living in less developed countries, with the highest incidence rates in Eastern, Southern, and Middle Africa (Figure 1.1), where it is responsible for nearly 12% of all female cancers 3. Cervical cancer is most common in relatively young women with age between 35 and 45. The 5-year survival rate depends greatly on the FIGO stage. The global survival rate was 98% for FIGO stage I, 64% for FIGO stage II, 38% for FIGO stage III and 7% for FIGO stage IV 4. 1.1.2 ANATOMY OF THE UTERINE CERVIX Cervical cancer originates in the uterine cervix, which is the lower part of the uterus, partly protruding into the vagina (Figure 1.2). The cervix consists of the ectocervix (the outer part ≥ 24.6 Not applicable No data 15.8–24.6 12.0–15.8 7.1–12.0 2.1–7.1 Figure 1.1 Global cervical cancer incidence rates in 2022. Estimated age-standardised incidence rates (ASR) per 100,000 per country. Adapted from Globocan 2022 3. All rights reserved. The designations employed and the presentation of the material in this publication do not imply the expression of any opinion whatsoever on the part of the World Health Organisation / International Agency for Research on Cancer concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. Dotted and dashed lines on maps represent approximate borderlines for which there may not yet be full agreement.

General introduction and thesis outline 13 1 of the cervix on the vaginal side), which is lined with multi-layered squamous epithelium and the endocervix (the inner part of the cervical canal on the uterine side), which is covered with single layered glandular epithelium. The squamocolumnar junction (SCJ) is where these two types of epithelia meet. Hormonal changes induce the replacement of a portion of the endocervical columnar epithelium by squamous metaplastic epithelium and is regulated by oestrogen levels. Over time, the location of the SCJ gradually shifts. During birth, the SCJ is found at the external os, while during puberty, the SCJ moves further up the endocervical canal towards the internal os. Postmenopausal the SJC moves back down the endocervical canal. The dynamic and macroscopically visible region between the former and new SCJ is called the transformation zone (TZ). Different types of stem cells maintain the epithelial layers of the cervix. The squamous epithelium of the ectocervix is maintained by ‘conventional’ stem-like cells in the basal epithelial layer 6. The cells in the transformation zone and endocervix are maintained and Increasing epithelial differentiation Ectocervix Endocervical gland Reserve cell Superficial layers Suprabasal layers Parabasal layer Basal layer Basement membrane Epithelial stem or stem-like cell SCJ Endocervix Transformation zone Infectious virion Cervix Cuboidal cell Figure 1.2 Schematic representation of the anatomy of the uterine cervix with detailed illustration of the cervical transformation zone and the squamocolumnar junction. Cervical cancer usually arises from the cervical transformation zone, a region of the cervix where endocervical columnar epithelial cells are replaced by squamous metaplastic cells. Specific epithelial stem cells (reserve and cuboidal cells) are thought to be particularly vulnerable to HPV-associated neoplastic transformation, and infection of these cells is thought to initiate cervical carcinogenesis. Adapted from Schiffman et al., Nat Dis Prim, 2016 5. Abbreviations: SCJ, squamocolumnar junction

14 Chapter 1 different candidates qualify as stem cells 7. Different target cells of lesion formation and transformation towards cervical cancer have been described, but it has been suggested that specific cell populations of epithelial stem cells (reserve cells, located at the transformation zone, and cuboidal cells, located at the SCJ) are particularly vulnerable to neoplastic transformation by the human papillomavirus (HPV; further detailed in section 2) and are considered to be the cells of origin for cervical cancer 5, 8. 1.1.3 CERVICAL CANCER AND ITS PRECURSOR LESIONS Cervical cancer is classified into different histological types, with squamous cell carcinoma (SCC) and adenocarcinoma (ACC) being the two main histotypes. SCC accounts for about 80% of cervical cancers, while ACC accounts for about 15%. The remaining 5% of cervical cancers comprise rare types, including neuro-endocrine carcinoma and HPVindependent adenocarcinoma (i.e., clear-cell, mesonephric, gastric or endometrioid type) 2, 9-11. Cervical SCC and ACC develop through precursor lesions. The development of SCC is preceded by premalignant lesions called cervical intraepithelial neoplasia (CIN). CIN can be histologically classified into three grades based on the extent of immature dysplastic cells in the epithelium (Figure 1.3): CIN1 (mild dysplasia), in which atypical cells are limited to the lower one third of the epithelial layer, CIN2 (moderate dysplasia), in which the lower two thirds of the epithelium contain atypia and CIN3 (severe dysplasia), in which atypical cells affect more than two-third of the epithelial layer. When atypical squamous cells extend through the basal membrane, a lesion is defined as cervical cancer. In the United States, an alternative two-tiered system, the Lower Anogenital Squamous Terminology (LAST), is used, which recognises low-grade squamous intraepithelial lesions (LSIL), equivalent to HPV-related epithelial changes without CIN (CIN0) and CIN1, and high-grade squamous intraepithelial neoplasia (HSIL), equivalent to CIN2 and CIN3 12. Clinical management of CIN depends on the classification of the lesion. Treatment of CIN aims to prevent cancer. In general, a conservative approach is recommended for CIN1, as the majority of these lesions will spontaneously regress (~80%). Treatment is indicated for women with a CIN3 lesion due to their high risk for progression into cancer (~31%) 13. Management guidelines for CIN2 lesions vary, as these lesions represent a heterogeneous group of disease with a highly variable clinical course. Premalignant lesions of ACC are less well-defined 14. The clinical significance of minor endocervical glandular dysplasia in the development of ACC is not known and the diagnosis is poorly reproducible. The only reproducible precursor stage of ACC is

General introduction and thesis outline 15 1 adenocarcinoma in situ (AIS), which is considered the glandular counterpart of CIN3 and, in contrast to CIN, is not further graded 4, 15. 1.2 HPV-MEDIATED CARCINOGENESIS 1.2.1 HUMAN PAPILLOMAVIRUS HPV is a sexually transmitted virus. Persistent infection with HPV is the main aetiologic factor in the development of cervical cancer 16-18. The prevalence of HPV infections is highest among young women (18 – 24 years), soon after the onset of sexual activity, and decreases gradually with age 19-21. The life-time risk of acquiring at least one genital HPV infection is around 80% in sexually active women 22, 23. However, most HPV infections are transient and are cleared spontaneously by the immune system within two years 18, 24. Only a small proportion of HPV infections persists and may lead to cancer. Thus, cervical cancer is a rare complication of a rather common viral infection 25. Risk factors for cervical cancer involve an increased likelihood of acquiring and maintaining an HPV infection and include having multiple sex partners, co-infection with human immunodeficiency virus (HIV), tobacco smoking, long-term use of hormonal contraceptives and high parity. Since the discovery of the link between HPV and cervical cancer in 1983 by professor Normal cervix Cervical intraepithelial neoplasia Invasive cancer Infectious viral particles Squamous epithelium Nuclei with episomal viral DNA Normal nuclei Nuclei with integrated viral DNA Overexpression of E6 and E7 Expression of early and late genes Dermis Basal layer Superficial zone Midzone Basement membrane Grade 1 Grade 2 Grade 3 Figure 1.3 Development of cervical cancer through cervical intraepithelial neoplasia. After entering the basal cells of the cervical epithelium, human papillomavirus infection can give rise to premalignant abnormalities represented by cervical intraepithelial neoplasia grade 1, 2 and 3. These lesions can further progress towards invasive cervical cancer, a process that can take up to 20 to 30 years. Adapted from Woodman et al., Nat Rev Cancer, 2007 26.

16 Chapter 1 Harald zur Hausen and his team 27, HPV has been the subject of extensive research. HPV is a small, non-enveloped deoxyribonucleic acid (DNA) virus that belongs to the family of Papillomaviridae 28. The circular, double-stranded viral genome is approximately 8000 base pairs in length (Figure 1.3) 28, and consists mainly of three functional parts: an early region that encodes for proteins of E1, E2, E4–E7 genes required for viral replication; a late region that encodes for proteins of L1 and L2 genes necessary for viral assembly; and an upstream regulatory region (URR) between E6 and L1, important for viral gene transcription regulation and DNA replication. HPVs are strictly epitheliotropic and can be subdivided into mucosal and cutaneous types according to their affinity for one of these epithelia 29. So far, more than 400 different types of HPV have been identified on the basis of DNA sequence data showing genomic differences 30. Based on the nucleotide sequences of E6, E7 and L1, papillomavirus (PV) types are defined. Genotypes are classified into phylogenetic genera and numbered species 5. Approximately 40 HPV types are known to infect the genital mucosa 31. These mucosal HPV types typically cluster to the alpha-HPV genus. Mucosal HPVs can be further subdivided into low-risk (lr) and high-risk (hr) types, according to their oncogenic potential. The International Agency for Research on Cancer (IARC) Working Group defined HPV types as to its carcinogenicity to humans based on frequency of occurrence in cases of cervical cancer and available biological data: IARC Group 1 carcinogenic to humans, IARC Groups 2A and 2B probably and possibly carcinogenic to humans, respectively, and IARC Group 3 not classifiable as to carcinogenicity in humans. HPV16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58 and 59 are carcinogenic (Group 1), and HPV68, 26, 30, 34, 53, 66, 67, 69, 70, 73, 82, 85 and 97 are probably/possibly carcinogenic (Group 2A/2B) 32. Group 2A/2B accounts for a minority of cervical cancers (~3%) 33. Oncogenic potential differs greatly between hrHPV types. HPV16 is the most common and most carcinogenic HPV type, responsible for over 60% of cervical carcinomas worldwide; followed by HPV18, which accounts for another 10% 34. HPV18 is especially common in ACC 35, 36. HPV31, 33, 45, 52 and 58 account for an additional 19% of cervical cancers 37. Besides being responsible for cervical carcinogenesis, hrHPV types, particularly HPV16, also play a causal role in the development of a subset of other types of cancers (in particular vulvar and vaginal, anal, penile and oropharyngeal) 38-43. On the other hand, lrHPV types, such as HPV6 and HPV11, are associated with the development of benign lesions like genital warts and cervical low-grade premalignant lesions.

General introduction and thesis outline 17 1 1.2.2 THE CLINICAL OUTCOMES OF HRHPV INFECTIONS The various outcomes of exposure of cervical epithelial cells to hrHPV are a transient infection, a productive infection and a transforming infection 25. Around 80% of the genital HPV infections are transient, cleared within months following infection and do not result in cervical pathology. Most of the remaining hrHPV infections are productive infections. Productive infections are characterised by the release of new viral particles through shedding of terminally differentiated cells in the upper layer of the epithelium. The viral life cycle relies heavily on the differentiation process of the infected squamous epithelium. The viral gene expression is regulated by multiple promoters and complex mRNA splicing patterns that enable the expression of different proteins at different stages of the viral life cycle 44, 45. A productive infection with hrHPV begins when virions enter basal keratinocytes of differentiating cervical epithelium through defects in the epithelial covering 46. Viral entry requires binding of virions to heparin sulphate proteoglycans (HSPGs) on the cell surface, resulting in a conformational change, followed by site-specific enzymatic cleavage of the L2 protein by furin, a cell-encoded proprotein convertase 47, 48. The conformational change resulting from the proteolytic cleavage facilitates virus internalisation. The infected basal keratinocytes or stem cells become a reservoir of infection, where the viral genome enters the cell nucleus and the episome is maintained at low copy numbers and expressed at low levels. Infected daughter cells are produced as the cells divide and migrate towards the epithelial surface, where different viral proteins are expressed at different stages of epithelial differentiation (Figure 1.4). In the lower layers of the epithelium, viral proteins E6 and E7 activate the cell cycle machinery to accomplish HPV replication and drive cells through cell cycle and stimulate cell division (cycling cells marked with red nuclei). In the middle layers, viral genome amplification is enhanced by the combined action of viral proteins E1, E2, E6 and E7. Indirectly, the amplification success is dependent on the modifying effect of E4 and E5 on the cellular environment in the S or G2 phases of the cell cycle (presence of E4 is marked in purple, with red nuclei indicating replication competence). In the upper layers, the productive viral life-cycle is completed, as the cells leave the cell cycle. Viral proteins L1 and L2 are produced in a subset of the E4-positive cells, allowing the amplified viral genomes to be packaged and released as viral particles 49. Productive hrHPV infections may give rise to mild to moderate cellular abnormalities, which are histologically recognisable as CIN1 or CIN2 lesions. These so-called productive CIN lesions tend to regress spontaneously within 1-2 years 25. The expression of the HPV E4 protein during the final stage of the viral life cycle is a marker of productive

18 Chapter 1 HPV infections 50, 51. HPV E4 expression is mainly observed in CIN1 and productive CIN2 lesions, with only a small proportion of CIN3 lesions exhibiting E4 expression 52, 53. Immunohistochemical analysis of E4 expression in cervical disease may facilitate the detection and monitoring of low-grade lesions 52, 54. A minority of hrHPV infections (~10-20%) are able to evade immune surveillance, persist during many years and may lead to the development of high-grade cervical premalignant lesions, classified as CIN2 and CIN3 and also cervical cancer. These clinical manifestations are associated with so-called transforming infections, in which the normal viral life cycle is aborted and the viral early genes E6 and E7 are expressed in proliferating cells 25, 55. HPVinduced transformation is initiated with this deregulated expression of the viral E6 and E7 genes, and is characterised by the absence of viral progeny. In the context of dividing cells, viral proteins E6 and E7 serve as oncoproteins by interaction with tumour suppressor gene products p53 and Retinoblastoma protein (pRB), respectively 56. The interaction of E6 with p53 leads to the degradation of p53 through the ubiquitin-dependent proteolysis pathway. Consequently, p53-mediated apoptosis and cell cycle arrest are inhibited, resulting in cell cycle progression despite the presence of DNA damage. E6 also activates telomerase, contributing to the immortalisation of host-cells 57, 58. The E7 oncoprotein targets and degrade pRB, thereby facilitating entry into the S-phase of the cell cycle, leading to uncontrolled proliferation 59, 60. The combined disruption of cell cycle control Virus release Virus assembly / Virus release Genome amplification Genome maintenance / Cell proliferation Genome maintenance Infectious virion PE PAE PL PAE PL PAL Transformation zone E6, E7 E1, E2, E4, E5 viral DNA E4 L2 L1 Figure 1.4 The life cycle of high-risk human papillomavirus. HPVs infect epithelial cells and depend on epithelial differentiation for completion of their life cycle different types of viral proteins are expressed at different phases of the viral life cycle of the human papillomavirus. Adapted from Doorbar et al., Vaccines, 2012 49. Abbreviations: PE, Early promoter; PAE, Position of the early polyadenylation site; PL, Late promoter; PAL, Position of the late polyadenylation site

General introduction and thesis outline 19 1 and prevention of apoptosis provides a mechanism of malignant transformation. Through a series of interactions and pathways, activation of E6 and E7 leads to the upregulation of Ki-67 and increased p16INK4A levels. Ki-67 is a marker of proliferation and has been suggested as a sensitive biological indicator of CIN progression 61. p16INK4A is involved in regulation of the cell cycle and is used to guide CIN grading as expression increases with increasing severity of disease 62. The College of American Pathologists and the American Society for Colposcopy and Cervical Pathology recommended in the LAST criteria the use of p16INK4A as an adjunct to morphologic assessment for differentiation between HSIL and a disease state mimicking HSIL. CIN2 lesions can be further differentiated based on their p16INK4A status, where p16INK4A-positive CIN2 lesions are classified into HSIL and p16INK4A-negative CIN2 lesions into LSIL 12, 63, 64. The concept of cervical carcinogenesis is summarised in Figure 1.5. Viral infection Original concept TZ Productive CIN hrHPV New concept Type of hrHPV infection Morphological appearance Normal CIN1/2 CIN2/3 Cancer Onset of cervical carcinogenesis Viral persistence • E ▪ E6 and E7 expression in proliferating cells ▪ Viral transformation Development of precancerous lesions Progression to invasive cancer Transforming CIN Cancer Transient (latent) infection Productive (permissive) infection Transforming (non-permissive) infection 20-30 years Productive CIN SCJ cells Ectocervix or TZ Transforming CIN Cancer Figure 1.5 The concept of HPV-mediated cervical carcinogenesis. Schematic representation of several outcomes of a high-risk (hr) human papillomavirus (HPV) infection in cervical epithelial cells. Most HPV infections (~80%) are transient and are cleared by the immune system without causing cellular abnormalities. If HPV persists, the (reversible) progression of a productive CIN1/2 towards a transforming CIN2/3 reflects the onset of cervical carcinogenesis. Transforming infections are characterised by deregulated E6 and E7 expression in proliferating cells. Transforming CIN is a heterogeneous disease with variable duration of lesion existence and includes both progressive and regressive lesions. Only a very small proportion of infections will eventually lead to the development of cervical cancer, which can take up to 20-30 years. Adapted from Steenbergen et al., Nat Rev Cancer, 2014 25. Abbreviations: TZ, transformation zone; CIN, cervical intraepithelial neoplasia; SCJ, squamous columnar junction; hr, high-risk; HPV, human papillomavirus

20 Chapter 1 1.2.3 MOLECULAR HOST-CELL ALTERATIONS IN CERVICAL CARCINOGENESIS The combined disruption of cell cycle control and prevention of apoptosis by viral oncoproteins E6 and E7 in the context of a transforming infection induces genome instability. The gradual accumulation of genetic and epigenetic changes in the host-cell genome, affecting both oncogenes and tumour suppressor genes, along with a persistent hrHPV infection, is considered as a crucial driving force necessary for the progression to cervical cancer. In concordance with this, the number of molecular aberrations increases with the severity of cervical lesions, with the highest numbers being found in cervical cancer 55, 57, 65. Genetic host-cell alterations detected in cervical cancer and CIN2/3 lesions include copy number alterations (CNA) and DNA mutations, and epigenetic changes include altered microRNA (miRNA) expression and DNA methylation 25, 66. DNA METHYLATION DNA methylation is a well-studied epigenetic process that plays an important role in several human processes, such as X-chromosome inactivation, suppression of repetitive element transcription, genomic imprinting and transposition. Epigenetic silencing of hostcell genes by DNA methylation has also proven to be essential for cervical carcinogenesis. DNA methylation entails the covalent binding of a methyl (-CH3) group to the carbon-5 position of a cytosine molecule in CpG dinucleotides and is regulated by the activity of DNA methyltransferases (DNMT) proteins (Figure 1.6). During carcinogenesis, there is a general loss of DNA methylation in the genome, resulting in chromosomal instability, while hypermethylation of promoter regions of tumour suppressor genes with high CpG density, occurs leading to transcriptional repression, altogether contributing to the development of cervical cancer. HPV E6 and E7 oncoproteins interact with several proteins that regulate epigenetic processes, including DNMTs, histone-modifying enzymes and chromatin remodelling complex subunits, influencing the host-cell transcription program. For instance, E6 degrades p53, leading to the upregulation of DNMT1. E7 binds directly to DNMT1. Additionally, E7 indirectly binds to pRB, causing the release of E2F. This results in activation of DNMT1, regulation of promoter activity of DNMT1 and ultimately leads to overexpression of DNMT1. E7 has also been shown to increase the protein levels of DNMT3A and DNMT3B. The interaction between HPV and the cellular DNA methylation machinery can influence the epigenetic regulation of both the viral genome and the host genome. This interplay plays a crucial role in the viral life cycle, but also participates in the maintenance of a persistent infection, cell transformation, and development of invasive

General introduction and thesis outline 21 1 cancer by considerably deregulating tumour suppressor genes and oncogenes. Cellular, or so-called host-cell, DNA methylation markers have been recognised as sensitive biomarkers for transforming CIN lesions 25. Early lesions are characterised by low levels of DNA methylation, while advanced lesions with a high cancer progression risk are characterised by high/increased levels of DNA methylation. 1.3 CERVICAL CANCER PREVENTION Cervical cancer is highly preventable and treatable if detected early and managed effectively. Despite this, it remains a leading cause of cancer-related deaths among women worldwide. To combat cervical cancer, the World Health Organisation (WHO) CH3 Unmethylated Promoter region Promoter region Cytosine DNMTs 5-Methylcytosine Methylated Unmethylated Methylated Tumour suppressor gene Tumour suppressor gene CH3 CH3 CH3 CH3 CH3 CH3 Figure 1.6 DNA methylation-mediated silencing of tumour suppressor genes. DNA methylation is an epigenetic process that controls gene expression and plays a significant role in the development of cervical cancer. The process of DNA methylation is regulated by DNA methyltransferases (DNMTs), which binds a methyl (-CH3) group to the carbon-5 position of a cytosine molecule in CpG dinucleotides. This process generally leads to the repression of gene transcription, particularly in promoter regions of tumour suppressor genes that have a high density of CpG sites and is commonly observed in cancer cells. Abbreviations: DNMTs, DNA methyltransferases; CH3, methyl group; C, cytosine; Me, methylation

22 Chapter 1 issued a global call in May 2018 for the implementation of evidence-based interventions, including HPV vaccination, cervical cancer screening and management of disease, with the goal of eliminating cervical cancer as a public health problem. In August 2020, the World Health Assembly adopted the Global Strategy for cervical cancer elimination, with the aim of achieving and maintaining an incidence rate of less than four per 100,000 women in all countries 67. To achieve this, countries must meet the 90-70-90 targets by 2030 to be on the path towards cervical cancer elimination: 1. 90% of girls fully vaccinated with the HPV vaccine by the age of 15; 2. 70% of women screened with a high-performance test by the age 35 and again by the age of 45; 3. 90% of women identified with cervical disease receive treatment, with 90% of women with precancer treated, and 90% of women with invasive cancer managed. 1.3.1 PRIMARY PREVENTION: PROPHYLACTIC VACCINES Prophylactic HPV vaccines are designed to prevent initial HPV infection and thereby HPVassociated lesions. These vaccines consist of virus-like particles (VLPs) that resemble native virus particles, but are non-infectious and lack viral DNA 68. Currently, three highly efficacious licensed HPV vaccines are available: 1. The bivalent Cervarix® vaccine (GlaxoSmithKline) targeting HPV16 and HPV18 69, 70; 2. The quadrivalent Gardasil® vaccine (Merck), directed against HPV16, HPV18, HPV6 and HPV11 71, 72; 3. The nonavalent Gardasil-9® vaccine (Merck) targeting HPV16, HPV18, HPV31, HPV33, HPV45, HPV52, HPV58, HPV6 and HPV11 73. While type-specific, these vaccines, in particular the bivalent vaccine, also offer some crossprotection 74, 75. HPV vaccines have been available since 2006, and in the Netherlands, HPV vaccination with the bivalent vaccine was introduced in the Dutch National Immunisation program in 2009 for girls aged 12, with a coverage rate of 54.8% in 2021 76-78. From 2022 onwards, the Dutch HPV vaccination program is gender-neutral, vaccinating children at the age of 9 years 79. In 2023, a catch-up campaign was launched for unvaccinated boys and girls up to age 26. All strategies aim to increase HPV vaccine uptake and thereby contributing to herd immunity to protect the population against HPV-related cancer. 1.3.2 SECONDARY PREVENTION: CERVICAL SCREENING The goal of population-based cervical screening is to reduce mortality from cervical cancer by detecting and treating early cancers and cervical precursor lesions. Due to the long

General introduction and thesis outline 23 1 latency period (i.e., from the onset of a transforming hrHPV infection it may take another 20 to 30 years before cervical cancer may develop 13), there is a window of opportunity to identify lesions early and effectively treat them to prevent cancer development 80. Cervical screening programs have been implemented worldwide since the mid-20th century. For several decades, cytology-based screening has been the cornerstone of cervical cancer prevention relying on microscopic evaluation of cervical cells. Different classification systems are in use to define the severity of cytological abnormalities found in cervical cytology preparations, for example the Bethesda 2001 and CISOE-A classification (Table 1.1) 81. Cytology-based cervical screening programs have led to a notable decrease in cervical cancer incidence and mortality in many countries 82-86. Cervical cancer screening revolutionised by the discovery of HPV as the main causative agent of the disease. Over the last decade, many countries have transitioned from cytology-based screening to primary HPV-based screening, to obtain better detection of cervical cancer and its high-grade precursor lesions 87-92. Large randomised controlled trials (RCTs) have shown that HPV screening leads to earlier detection of clinically relevant lesions compared to cytology and provides a 60 – 70% better protection against invasive cervical cancer 87, 89, 90, 92-96. The high accuracy and reproducibility of HPV testing has led to the replacement of cytology, along with an extension of screening intervals for women with a negative screening result 87, 97-101. Description* Inadequate Normal Borderline dyskaryosis Mild dyskaryosis Moderate dyskaryosis Severe dyskaryosis Carcinoma in situ Carcinoma CISOE-A C0 S1, O1, E1-2 S1, O2, E1-2 S2-3, O3, E3 S4, E4-5 S5, O4-5 S6, O6, E6 S7, E7 S8-9, O7-8, E9 Pap-score Pap 0 Pap 1 Pap 2 Pap 3a1 Pap 3a2 Pap 3b Pap 4 Pap 5 Bethesda 2001 Unsatis- factory for evaluation NILM Atrophy ASC-H HSIL or >BMD SCC BMD ASC-US LSIL AGC favour neoplastic CIS / AIS AC Abbreviations: CISOE-A: C, composition; I, inflammation; S, squamous epithelium; O, other abnormalities and endometrium; E, endocervical columnar epithelium; A, adequacy; BMD, borderline or mild dyskaryosis; CIS, carcinoma in situ; Pap, Papanicolaou; NILM, negative for intraepithelial lesion or malignancy; ASC-H, atypical squamous cells cannot exclude HSIL; HSIL, high-grade intraepithelial lesion; ASC-US, atypical squamous cells of undetermined significance; AGC, atypical glandular cells; LSIL, low-grade intraepithelial lesion; HSIL, high-grade intraepithelial lesion; AIS, endocervical adenocarcinoma in situ; SCC, squamous cell carcinoma; AC, adenocarcinoma Table 1.1 Classification of cervical cytology (adapted from Bulk et al., 2004 81).

24 Chapter 1 HPV-BASED CERVICAL SCREENING IN THE NETHERLANDS In 2011, the Dutch Health Council advised the Ministry of Health to replace cytology with primary HPV testing in cervical screening 101. Consequently, the Netherlands became the first country to switch to a nationwide primary HPV-based screening program in 2017, followed by Australia and other countries 102, 103. Triage testing of hrHPV-positive women is required to identify women with clinically relevant infections and to avoid over-referral. One of the most common triage tests internationally is cytology 104, 105. In the new HPVbased screening program triage of hrHPV-positive women was performed by (repeat) cytology to achieve an acceptable CIN3+ risk for triage-negative women 101, 106. Women with low-grade cytology (atypical squamous cells of undetermined significance (ASC-US); or low-grade squamous intraepithelial lesions (LSIL)), either at baseline or at six months follow-up, were referred to a gynaecologist for colposcopy. Women with an hrHPVpositive, normal cytology test result at baseline were offered repeat cytology because they have only a 4% risk of CIN2/3+. This risk is too low to recommend immediate colposcopy referral and by repeating cytology after six months, a transient infection may resolve in the meantime. The first results of the new Dutch cervical cancer screening program showed that more women with clinically relevant findings (CIN2+) were discovered, but screen-positivity, referral rates and low-grade lesion (<CIN1) rates were also markedly higher than in the former, cytology-based program 107. The increase in colposcopy referrals was mainly caused by the direct referral of women with ASC-US/LSIL cytology, who often do not have CIN2/3. In 2021, the Dutch Health Council therefore recommended that referrals to gynaecologists should be subject to more specific criteria. Additional triage of hrHPV-positive women with ASC-US/LSIL cytology might reduce the number of women referred for colposcopy, while maintaining clinical sensitivity. This could be achieved by incorporating hrHPV genotyping to discriminate hrHPV types with clearly increased risk, such as HPV16 and HPV18, and hrHPV-types with a moderately increased risk. Since mid-2022, women with ASC-US/LSIL cytology are therefore only directly referred to a gynaecologist if HPV16 and/or -18 are present. In case of a non-HPV16/18 type, women are invited for repeat testing. Furthermore, extending the interval for repeat cytology from 6 to 12 months might reduce the number of clinically unnecessary referrals and was introduced mid-2022. The performance of HPV genotyping in differentiating between a transient infection and a persistent infection, that is associated with CIN2/3 and cancer, is only moderate 108. Therefore, alternative (secondary) triage strategies are warranted.

General introduction and thesis outline 25 1 Self-collection of cervicovaginal specimens can be used for HPV testing (i.e., HPV selfsampling), and has the advantage that it can lower barriers to participation and increase attendance rates 109-117. In the Netherlands, HPV self-sampling was implemented in the cervical screening program in 2017 for women who were not willing to participate by clinician-collection of a cervical sample 101. Recent studies support the use of HPV selfsampling as a primary screening test for all women invited for cervical cancer screening 118. Based on this evidence, the Dutch Health Council recommended in 2021 that selfsampling kits should be offered to all women who are invited to undergo population screening for cervical cancer, and these kits should be automatically included with the invitations. By offering women the option to choose between two equivalent screening methods (i.e., clinician-based or self-collection), the Dutch Health Council expects that participation rates will increase. In July 2023, this was introduced in the Netherlands, where women participating in the population screening for the first time will receive a self-sampling kit immediately. For all other women, they will have the option to express their preference after receiving the invitation. If no response is received, they will receive a self-sampling kit. A drawback of self-sampling is that cytology cannot be accurately performed, requiring women who test positive for hrHPV on a self-collected sample to visit a general practitioner for a clinician-collected cervical sample for cytology-triage. This can result in women being lost to follow-up (i.e., 7-40%) and women experience delays in the diagnostic track 110, 116, 119. To improve follow-up adherence for women who test hrHPV-positive on self-collected samples, alternative triage methods that are applicable directly to self-collected samples are needed. Another aspect to consider for screening in the near future is that vaccinated women will gradually enter the screening program, i.e., in the Netherlands the first vaccinated women are screened in 2023. Although the vaccines protect against most hrHPV types that can cause cervical cancer, suboptimal vaccine coverage means that screening will remain a crucial part of cervical cancer prevention for the next decades. Nonetheless, adjustments to the screening strategy need to be anticipated as the risk of CIN3+ and the predictive value of a positive screening result (PPV) will drop 120. With the decreasing prevalence of HPV types with high oncogenic potential (i.e., HPV16 and HPV18 in the Netherlands given the use of the bivalent vaccine), the majority of positive screening findings threaten to be false positive findings. As a consequence, improved screening and/or triage tests are required for better risk stratification.

26 Chapter 1 Over recent years, several other viral and cellular biomarkers have been evaluated as a triage test for HPV-positive women, such as HPV E6 and/or E7 mRNA, HPV proteins, HPV methylation, p16INK4A/Ki-67 dual-stain cytology (CINtec® PLUS cytology test), assessment of DNA copy number alterations, analysis of miRNA expression, and host-cell DNA methylation 133. Molecular tests have the advantage over morphology-based strategies in that they comprise objective, highly reproducible techniques which are automatable and applicable to both clinician-collected cervical samples and self-collected samples. This thesis focuses on host-cell DNA methylation markers. Host-cell DNA methylation markers may offer a specific molecular approach to detect advanced CIN lesions requiring treatment and could meet the needs of cervical cancer screening in the post-vaccination era. 1.4 DNA METHYLATION-BASED BIOMARKERS FOR CERVICAL CANCER The ‘fifth base of DNA’, known as methylated cytosine, is gaining increased attention as a potential novel biomarker for cervical cancer. DNA methylation is an early and common molecular change in cervical carcinogenesis, and methylation levels increase as the disease progresses. In particular, high methylation levels have been found in cervical cancer. Nearly all cervical cancers are methylation-positive 121. DNA methylation markers have the ability to differentiate clinically relevant CIN lesions with high risk of progression to cervical cancer 122, highlighting its potential as a biomarker for screening and CIN management. Multiple studies have consistently demonstrated that DNA methylation markers have high accuracy in detection of clinically relevant cervical lesions 123-125. 1.4.1 GENOME-WIDE DISCOVERY OF HOST-CELL DNA METHYLATION MARKERS Genome-wide techniques, such as array-based methods and next generation sequencing (NGS), have enabled comprehensive studies of DNA methylation events associated with cervical cancer development 126, 127. Aberrant methylation patterns have been found in numerous genes and various host-cell DNA methylation markers have been derived thereof, including those that are further discussed in this thesis: FAM19A4 (TAFA-4, Family with sequence similarity 19 [chemokine (C-C motif)-like, member 4A]), miR124-2 (microRNA 124-2), ASCL1 (Achaete-scute family BHLH transcription factor 1), LHX8 (LIM homeobox 8) and ST6GALNAC5 (ST6 N-Acetylgalactosaminide Alpha-2,6-Sialyltransferase 5), GHSR (Growth Hormone Secretagogue receptor), SST (Somatostatin), ZIC1 (Zinc finger protein ZIC1). FAM19A4 was discovered using methylation-specific digital karyotyping of

General introduction and thesis outline 27 1 HPV16E6E7-transduced primary foreskin keratinocytes and is associated with disease progression 128. Downregulation of miR124-2 was identified through genome-wide studies on miRNA expression and has been linked to increased promoter methylation in cervical cancer 129. ASCL1, LHX8 and ST6GALNAC5 were discovered through an array-based methylation discovery screen (Infinium 450K array) on hrHPV-positive self-collected samples 126, while methylation markers GHSR, SST and ZIC1 were identified through a methyl-binding domain-enriched DNA-based discovery screen (MBD-seq) on high-risk HPV-transformed cell lines and cervical tissue specimens 127. Methylation levels of these genes increase progressively with the severity of underlying disease and are significantly higher in CIN3 and cervical cancer lesions compared to control women without evidence of CIN2+ 123, 126, 127. 1.4.2 DNA METHYLATION MARKER ANALYSIS For validation studies and diagnostic purposes, a targeted and more sensitive method to detect DNA methylation is preferred over genome-wide techniques. Multiplex quantitative methylation-specific PCR (qMSP) is an efficient and reproducible technique that can detect multiple methylated genes and a reference gene in a single assay. This method enables multiple methylation targets to be analysed using a single sample aliquot, which can reduce target-to-target variations and allow for the monitoring of sample adequacy for PCR purposes by an internal reference gene 130. Additionally, this saves material, time, and costs while improving quality control. The method can also be automated, making it suitable for high-throughput settings. The qMSP technique starts with bisulphite treatment of DNA to convert cytosine to uracil, while leaving methylated cytosines unchanged. The converted DNA is then amplified by PCR using specific primer pairs for methylated DNA and fluorescent probes, with methylation status quantified by fluorescence monitoring. 1.4.3 CLINICAL IMPLICATIONS OF DNA METHYLATION ANALYSIS In recent years, the use of DNA methylation analysis of host-cell genes as a clinical tool for detecting cervical cancer and precancerous lesions has gained considerable attention 131. Analysis of DNA methylation can be performed on different sample types, including clinician-collected cervical samples, self-collected samples, cervical tissue specimens and even urine. Methylation assays can be used for several purposes 132, 133: 1. As primary triage of hrHPV-positive women to detect CIN3+; 2. As secondary triage for women with minor cytological abnormalities to identify those with the highest risk of CIN3+;

28 Chapter 1 3. To support post-colposcopy management; 4. To support post-treatment management of CIN. Meta-analyses have reported that methylation markers have good diagnostic ability 131, 134. In the setting of HPV triage, methylation sensitivity for CIN2+ and CIN3+ were reported to be 68.6% (95% CI: 62.9-73.8%) and 71.1% (95% CI: 65.7-76.0%), respectively, at a fixed specificity of 70%, and PPV were 53.4% (95% CI: 44.4-62.1%) and 35.0% (95% CI: 28.941.6%), respectively. Among hrHPV-positive women, the relative sensitivity of methylation for CIN2+ was found to be 0.81 (95% CI: 0.63-1.04) when compared to cytology (threshold borderline; atypical squamous cells of undetermined significance (ASC-US), or worse (ASC-US+) and 1.22 (95% CI: 1.05-1.42) when compared to HPV16/18 genotyping, while relative specificity was 1.25 (95% CI: 0.99-1.59) and 1.03 (95% CI: 0.94-1.13), respectively. Additional risk-stratification of hrHPV-positive women with borderline/low-grade cytology (ASC-US/LSIL) by methylation analysis has shown that it can reduce direct colposcopy referral rate with 60%, while retaining high CIN3+ sensitivity 108. Host-cell methylation analysis has been reported as a good tool to rule out cervical cancer whether caused by hrHPV infection prior to vaccination, hrHPV genotypes not covered by the current prophylactic vaccines or characterised by the absence of hrHPV 121. For the management of women with CIN2/3, host-cell methylation analysis has shown valuable in predicting regression, i.e., a methylation-negative test was associated with more clinical regression as compared to women with a methylation-positive result (74.7% versus 51.4% respectively; p value = 0.013) 122. Though highly promising, it is of note that most studies that have examined the use of DNA methylation analysis as a method for (secondary) triage of hrHPV-positive women in screening have been conducted on smaller or selected series from screening or referral populations, such as self-sampling studies focused on non-attendee populations. This highlights the need for additional research using samples from women participating in routine HPV-based screening, which is the focus of this thesis 126, 127, 131, 135-139. 1.5 THESIS OUTLINE The primary objective of this thesis is to evaluate the role of DNA methylation markers in cervical cancer screening. With the global introduction of HPV-based screening in various countries, there is a need for effective triage strategies. These strategies aim to identify hrHPV-positive women with the highest risk of developing cervical cancer while simultaneously minimising unnecessary referrals and avoiding over-treatment of

General introduction and thesis outline 29 1 women without clinically relevant infections. Several strategies have been proposed, yet a consensus on the optimal triage test remains elusive. Each proposed strategy carries its own set of strengths and limitations, both in terms of clinical performance and screeningrelated burden. Although current triage methods, such as cytology and HPV16/18 genotyping, demonstrate acceptable clinical performance, there exists a demand for alternative methods applicable to both clinician-collected and self-collected cervical screening samples. This need is underscored by contemporary challenges, including reducing over-referral and overtreatment, the implementation of self-sampling as a primary screening tool, the call for automation, and the inclusion of vaccinated women in the screening program. DNA methylation markers emerge as promising objective biomarkers capable of identifying women with clinically relevant cervical disease, offering potential solutions to the aforementioned challenges. Chapter 2 evaluates six novel host-cell DNA methylation markers derived from two genome-wide discovery screens. The assessment focuses on the detection of cervical cancer and CIN3 using hrHPV-positive clinician-collected cervical samples. Building on the findings of Chapter 2, Chapter 3 validates the host-cell DNA methylation markers, for detection of CIN3+ on clinician-collected cervical samples. The validation is conducted on a large and independent cohort of hrHPV-positive women who participated in primary HPV-based screening. Chapter 4 delves into an analysis of the best-performing markers in Chapter 3, namely ASCL1 and LHX8, for the detection of CIN3+ in self-collected samples from hrHPV-positive women who underwent primary HPV self-sampling for cervical cancer screening. Chapter 5 centres on the evaluation of a bisulphite conversion protocol directly applicable to clinician-collected cervical samples. This protocol is designed to eliminate the need for prior DNA isolation, aiming to establish an automated laboratory workflow for increased efficiency. Chapter 6 explores the utilisation of FAM19A4/miR124-2 methylation, ASCL1/LHX8 methylation and HPV genotyping analysis as a secondary triage tool in hrHPV-positive women with ASC-US/LSIL cytology, aiming to enhance the efficiency of the triage process. Chapter 7 explores FAM19A4/miR124-2 methylation in a pilot series of cervical samples from young women who have been vaccinated against HPV. Chapter 8 provides a general discussion, summarising the results of this thesis in light of existing evidence. It discusses the clinical implications of the findings and explores potential avenues for further research. Finally, Chapter 9 offers a concise summary of the thesis, providing a quick overview of the key contributions and findings presented in each chapter.

30 Chapter 1 1. Gage JC, Schiffman M, Katki HA, Castle PE, Fetterman B, Wentzensen N, et al. Reassurance against future risk of precancer and cancer conferred by a negative human papillomavirus test. J Natl Cancer Inst 2014; 106(8). 2. Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2018; 68(6): 394-424. 3. Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021; 71(3): 209-49. 4. Oncoline: Guidelines Oncologcal Care. 2015. Guideline CIN, AIS and VAIN (in Dutch). 5. Schiffman M, Doorbar J, Wentzensen N, de Sanjose S, Fakhry C, Monk BJ, et al. Carcinogenic human papillomavirus infection. Nat Rev Dis Primers 2016; 2: 16086. 6. Clayton E, Doupe DP, Klein AM, Winton DJ, Simons BD, Jones PH. A single type of progenitor cell maintains normal epidermis. Nature 2007; 446(7132): 185-9. 7. Martens JE, Arends J, Van der Linden PJ, De Boer BA, Helmerhorst TJ. Cytokeratin 17 and p63 are markers of the HPV target cell, the cervical stem cell. Anticancer Res 2004; 24(2B): 771-5. 8. Pierce GB. Neoplasms, differentiations and mutations. Am J Pathol 1974; 77(1): 103-18. 9. Wang SS, Sherman ME, Hildesheim A, Lacey JV, Jr., Devesa S. Cervical adenocarcinoma and squamous cell carcinoma incidence trends among white women and black women in the United States for 19762000. Cancer 2004; 100(5): 1035-44. 10. Pirog E, Wright T, Ronnett B, Kurman R. Carcinoma and Other Tumors of the Cervix. 2018: 1-61. 11. Hohn AK, Brambs CE, Hiller GGR, May D, Schmoeckel E, Horn LC. 2020 WHO Classification of Female Genital Tumors. Geburtshilfe Frauenheilkd 2021; 81(10): 1145-53. 12. Darragh TM, Colgan TJ, Cox JT, Heller DS, Henry MR, Luff RD, et al. The Lower Anogenital Squamous Terminology Standardization Project for HPVAssociated Lesions: background and consensus recommendations from the College of American Pathologists and the American Society for Colposcopy and Cervical Pathology. J Low Genit Tract Dis 2012; 16(3): 205-42. 13. McCredie MR, Sharples KJ, Paul C, Baranyai J, Medley G, Jones RW, et al. Natural history of cervical neoplasia and risk of invasive cancer in women with cervical intraepithelial neoplasia 3: a retrospective cohort study. Lancet Oncol 2008; 9(5): 42534. 14. McCluggage WG. Premalignant lesions of the lower female genital tract: cervix, vagina and vulva. Pathology 2013; 45(3): 214-28. 15. Zaino RJ. Symposium part I: adenocarcinoma in situ, glandular dysplasia, and early invasive adenocarcinoma of the uterine cervix. Int J Gynecol Pathol 2002; 21(4): 314-26. 16. Walboomers JM, Jacobs MV, Manos MM, Bosch FX, Kummer JA, Shah KV, et al. Human papillomavirus is a necessary cause of invasive cervical cancer worldwide. J Pathol 1999; 189(1): 12-9. 17. zur Hausen H. Papillomaviruses and cancer: from basic studies to clinical application. Nat Rev Cancer 2002; 2(5): 342-50. REFERENCES