CHAPTER 38 Premalignant disease of the genital tract

Esther Moss, Charles W. Redman, Raji Ganesan

INTRODUCTION 566

THE CERVIX 566

THE VAGINA 575

THE VULVA 576

THE UTERINE CORPUS 577

THE OVARY 579

KEY POINTS 580

Introduction

The key to the effective management of premalignancy is to understand its aetiology, pathogenesis and natural history. Our knowledge and understanding of premalignant disease of the genital tract has progressed from descriptive morphology to a molecular biological level; a shift that is reflected by the evolution of clinical management from the detection and surgical removal of premalignant lesions to their prevention.

The tissues of the female genital tract differ in terms of embryological origin, responsiveness to intrinsic factors such as hormones, and exposure to external mutagens. Cellular transformation, leading to the development of cancer, is brought about by a variety of events, in part responding to various agents, acting alone or in combination, termed ‘carcinogens’. In the female genital tract, these may be viruses, chemicals, hormones or ionizing radiation. It follows that a single chapter under the heading of ‘Premalignant disease of the female genital tract’ will necessarily cover a variety of issues that are best considered by anatomical site.

The Cervix

In 1886, Sir John Williams described eight cases of cervical cancer, one of which was equivalent to carcinoma in situ:

’this is the earliest condition of undoubted cancer of the portio vaginalis that I have met with, and it is the earliest condition that is recognisable as cancer. It presented no distinct symptoms and was discovered accidentally.’

This described the cardinal features of cervical premalignancy; it is asymptomatic, clinically undetectable and has malignant potential. Its recognition had profound therapeutic significance as detection and treatment offered the potential to prevent cancer. The challenge has been to realize this therapeutic goal effectively with as little collateral damage as possible.

The terminology of cervical premalignancy

Most cervical cancers are squamous in origin, but a significant, and increasing, proportion arise from cervical glandular epithelium (estimates ranging from 5% to 30% of the precancerous lesions found on the cervix). Both cancer types have premalignant lesions.

The premise that cervical squamous premalignancy was a continuum underpinned the concept of cervical intraepithelial neoplasia (CIN), which was suggested by Richart (1967). The usefulness of this system was limited by significant interobserver variability in the diagnosis and grading of CIN, particularly in differentiating CIN1 from human papilloma virus (HPV) lesions, and separating CIN1 from CIN2 lesions. Furthermore, there is no clear evidence that CIN3 arises from earlier lesions.

These considerations were addressed by the Bethesda classification which was initially introduced as a cervical cytological grading system. Cervical lesions are classified into high- and low-grade squamous intraepithelial lesions; the latter group includes HPV lesions. This has the potential advantage of simplicity and tends to reflect the way in which clinicians practise, in that high-grade lesions are thought to be genuinely premalignant and should be treated.

Currently, whilst CIN terminology is used in the UK, clinical practice reflects the Bethesda classification in that CIN1 is regarded as a low-grade lesion, which tends to be managed conservatively, and CIN2 and CIN3 are regarded as high-grade lesions which are treated.

Attempts to classify premalignant glandular lesions have been bedevilled by a number of problems. In the UK, they are termed ‘cervical glandular intraepithelial neoplasia’ (CGIN) and are classified as high- or low-grade lesions. Low-grade CGIN lesions are difficult to distinguish from a number of benign and reactive changes. The biological behaviour of low-grade CGIN is not understood.

Pathology of cervical premalignancy

Cervical intraepithelial neoplasia

CIN relates to lesions that are confined to the squamous epithelium. The diagnosis of CIN is based upon the architectural and cytological appearances of the squamous epithelium. It is characterized by abnormal cellular proliferation, abnormal epithelial maturation and cytological atypia. Grading depends on the level in the epithelium to which abnormal changes extend.

The proportion of the thickness of the epithelium showing differentiation is a useful feature to be taken into account when deciding the severity of a CIN; roughly lower-, middle- and upper-third involvement equate to CIN1, CIN2 and 3, respectively. It is not the most important criterion despite the fact that it is one of the easiest to assess. In CIN1, at least the upper half of the epithelium usually shows good differentiation and stratification, whereas in CIN3, differentiation may be very slight or even absent.

CIN may affect the gland crypts as well as the surface epithelium (Figure 38.1). It is recognized that the degree and depth of crypt involvement increases with the grade of CIN. Histological assessment of crypt involvement in women with CIN3 has shown a mean depth of 1–2 mm, with a maximum of 5.22 mm and a mean ±3 standard deviations (99.7%) of 3.8 mm (Anderson and Hartley 1980). These figures suggested that treatment of ectocervical lesions to a depth of 7 mm should be sufficient to eradicate most CIN.

Figure 38.1 The cervical transformation zone shows the metaplastic squamous epithelium replaced by full-thickness lack of maturation amounting to cervical intraepithelial neoplasia 3 (long arrow). This is seen extending down the gland cleft (block arrow). The inset shows brisk epithelial mitotic activity. Original magnification ×100, inset ×200.

Cervical glandular intraepithelial neoplasia

CGIN is characterized by columnar cells with hyperchromatic nuclei and stippled chromatin (Figure 38.2). The nuclei show increased stratification and abnormal mitotic figures with loss of normal mucin. In some cases, the whole of a gland may be involved, but the lesion often occurs as a sharply demarcated area. It may be multifocal. CGIN is often associated with goblet cells or intestinal metaplasia of the endocervical cells. In two-thirds of cases, there are associated squamous abnormalities, and CGIN is often serendipitously discovered in the management of these abnormalities.

Figure 38.2 High-grade cervical glandular intraepithelial neoplasia, characterized by hyperchromatic glandular cells with pseudostratification, relative paucity of mucin, brisk mitotic activity and abrupt junction with normal epithelium (long arrow). Intestinal metaplasia (short arrow) is noted. Original magnification ×40.

Histological problems

The histological diagnosis of these lesions is based largely upon subjective criteria and is dependent on the adequacy of sampling. It is not surprising to find substantial inter- and intraobserver variation in the grading of CIN and the identification of CGIN. Variation is greatest at the lower end of the spectrum.

These considerations have important implications. Firstly, understanding of the risk of progression of CIN and its subgroups, on which treatment strategies are based, is inevitably flawed by bias of one form or another. Furthermore, in any given patient, there will be a spectrum of disease so that the reliability of the histological diagnosis is dependent on the quality and extent of sampling.

Pathogenesis of cervical premalignancy

The development of squamous precancerous abnormalities is intimately associated with the region of the cervical squamocolumnar junction (SCJ) and the changes that occur at puberty and adolescence.

The position of the SCJ is influenced by the hormonal changes that occur during a woman’s life (Figure 38.3). With the onset of puberty, the uterus enlarges and the cervix swells with a resultant eversion, exposing columnar epithelium to the acid environment of the vagina. This induces metaplasia in the exposed columnar epithelium, resulting in the development of metaplastic squamous epithelium. This area of transformation is termed the ‘cervical transformation zone’ (cervical TZ), and it is within this area that preneoplastic changes can occur with the development of CIN. It is thought that these dysplastic changes occur at the time of metaplasia, indicating that this is the time when the cervix is most vulnerable to potential carcinogenic factors, such as HPV and other cofactors.

Figure 38.3 The different stages in the development and involution of the cervix. (A) Before puberty, the ectocervix is covered with squamous epithelium, and columnar epithelium is usually confined to the endocervical canal. (B) The cervix enlarges and everts when oestrogen levels rise. This exposes columnar epithelium on the ectocervix. (C) The columnar epithelium on the ectocervix is replaced with squamous epithelium by a process of metaplasia. (D) Following the climacteric when oestrogen levels fall, the cervix shrinks, drawing the squamocolumnar junction up the canal.

These considerations have important clinical significance with regards to detecting and treating CIN, as these changes occur within the cervical TZ and it is accessible. Furthermore, the recognition that the development of precancerous changes can occur early in a woman’s sexual and reproductive life provided some indication that these events were potentially associated with some form of environmental exposure related to sexual activity. In other words, most of the identified risk factors for CIN are thought to be largely surrogate markers of HPV infection (Table 38.1).

Table 38.1 Risk factors for cervical intraepithelial neoplasia

Malignant potential of cervical premalignant lesions

Cervical intraepithelial neoplasia

The malignant potential of CIN3 was shown by McIndoe et al (1984) in a crucial paper. This indicated that approximately 30% of women with CIN3 would develop invasive cancer over a 20-year period.

The authors’ rationale for treating CIN3 is based upon this paper, which implies that when CIN3 is discovered, it should be treated. However, there are caveats. Firstly, the patients in this study had large carcinoma in situ lesions, and can we be sure that the undoubted premalignant potential for the lesions managed in this paper are shared by patients with smaller lesions that have small foci of CIN3? For example, it has been estimated that perhaps one-third of cases with CIN3 regress (Östör 1993).

Overall, it is thought that 1% of all CIN progresses to invasive cancer. Studies indicate that approximately 45–60% of CIN1 regress, 22–45% persist without progression and 10–16% progress to CIN3; whereas 28–40% of CIN2 regress, 20–40% persist and up to 50% progress to CIN3. However, all these studies are blighted by the difficulty in accurately determining the grade of the initial lesion and methodological variations.

These data indicate that treatment should always be considered for CIN3, but not necessarily for lesser grades of CIN. However, given the shortcomings in accurate assessment and the risk of progression of these cases, careful follow-up of untreated cases is mandatory.

Cervical glandular intraepithelial neoplasia

The relative rarity of this condition and classification problems have made this a more difficult area to study compared with squamous lesions. However, there is good evidence to support the concept that high-grade CGIN is a premalignant condition.

Aetiology of cervical premalignancy

The epidemiological risk factors for both squamous and glandular cervical premalignant lesions are similar and include young age at first intercourse and multiple sexual partners. It is now well established that infection with oncogenic high-risk HPV types is the central causal factor in the development of cervical neoplasia (Walboomers et al 1999).

The fact that virtually all cervical malignancies contain HPV DNA illustrates the central role of HPV in cervical carcinogenesis. It has now been proven beyond reasonable doubt that HPV infection is a necessary prerequisite for cervical carcinogenesis.

HPVs are small double-stranded DNA viruses which have an icosahedral protein capsid (Figure 38.4). They are typed according to the DNA sequence homology in particular genes, specifically L1 (which codes for the viral capsid) and E6 and E7 (which have important carcinogenic functions). Nearly 30 HPV types can infect the genital tract and can be classified into high-, intermediate- and low-risk oncogenic types. HPV types 16 and 18 are by far the most common high-risk types, accounting for 60% of HPV-positive invasive cervical cancers.

Figure 38.4 Model of human papilloma virus showing the arrangement of capsid proteins.

HPV infection can lead to integration of viral DNA into the host’s genome, with expression of the viral oncogenes E6 and E7 which produce proteins that interfere with tumour-suppression genes controlling the cell cycle. As a result, the cell loses the ability to repair DNA damage and to undergo apoptosis, becoming susceptible to additional mutations and genomic instability. It is therefore postulated that HPV integration can lead to carcinogenesis.

HPV infection is very common. It can be detected in up to 20% of sexually active women in the reproductive age group, and approximately 80% of women will, at some point, be infected. In most cases, genital HPV infections are transient with only a small proportion developing persistent infection, but the risk of subsequent development of CIN increases substantially in this group. HPV alone is not thought to lead to neoplastic change, and other cofactors are thought to be involved, such as smoking-related carcinogens, and dietary and hormonal factors.

Prevention of cervical premalignancy

The recognition that risk factors for cervical neoplasia were sexually related suggested the potential for a number of primary preventive strategies.

Primary prevention using behavioural changes

An obvious strategy was modification of sexual behaviour. Options included the use of sexual heath education, particularly advocating the use of barrier methods of contraception which are associated with a reduced incidence of CIN. Smoking is another consideration. Reducing or quitting smoking is associated with improvements in CIN. Nonetheless, the potential of these approaches is limited by the reality of only limited success in persistently changing human behaviour.

Primary prevention using immunization

Recognition of the central role of HPV in the pathogenesis of cervical neoplasia suggested that primary prevention was possible through the development of prophylactic HPV vaccines. These have been shown to be highly immunogenic, to generate high levels of neutralizing immunoglobulin G antibodies, and to persist for at least 5 years. Antibody responses are higher around the time of puberty (9–15 years), indicating that this should be the target population. What is not known is how long the duration of protection lasts, and whether or not booster immunization is needed.

Several phase III trials (Ault 2007) have shown that more than 90% of persistent HPV 16/18 infections can be prevented for up to 5 years after vaccination, and that more than 90% of precancerous lesions can be prevented in subjects who were HPV negative prior to vaccination. The long-term effects on cervical cancer incidence will require another 10–20 years of follow-up.

The optimal target age for prophylactic vaccination is just before the start of sexual activity (i.e. 9–14 years), as HPV infection may occur soon afterwards. The UK HPV vaccination programme began in 2007. Girls in Year 8 at school (aged 12–13 years) are offered a bivalent HPV vaccine (Cervarix) which protects against infection with HPV 16/18 types. The programme involves three injections over 6 months.

Women will still develop cervical cancer despite HPV vaccination. It is estimated that 70% of cervical cancers might be prevented by HPV vaccination, so primary cervical screening will still be needed, albeit in a modified version.

Detection of cervical premalignancy

Population-based screening should be conducted using well-organized and high-quality programmes with high coverage, as well as providing adequate treatment for detected lesions.

In the developed world, cervical cytology has formed the basis of screening programmes, but the resources and infrastructure required have precluded its use in poorer countries. Current debate concerns the use of the HPV detection assay as a screening tool, whereas in poorer countries, attention has focused on cheaper screening techniques that involve visually assessing the cervix, such as cervicography and visual inspection with acetic acid.

Cervical cytology

The recognition that cervical cytology could be used to detect precancerous change led to the introduction of cervical cytology as a screening test (Figure 38.5). Early detection and treatment can prevent the development of 75% of cancers. Whilst cytology is used to detect women at risk of having cervical premalignancy, most abnormalities are not precancerous. Only a small proportion of women with abnormal smears would develop cancer, although these women are high risk compared with the normal population. There is therefore huge potential for overtreatment unless one can accurately select which lesions require treatment.

Figure 38.5 (A) Dyskaryotic squamous cells showing mild and moderate dyskaryosis. There is cytoplasmic clearing resulting in a clear area around the nucleus indicative of koilocytosis, which is a papilloma-virus-related cytopathic change (Sure Path, original magnification ×60). (B) Glandular cells show disturbance in the normal honeycomb architecture. There is nuclear hyperchromasia, crowding of cells and ‘feathering’ of outline. These are features of glandular dyskaryosis (Thin Prep, original magnification ×60).

Images courtesy of Dr C.A. Waddell, Birmingham Cytology Training Centre.

In the UK, the incidence of cervical cancers has halved since the National Health Service’s (NHS) cervical screening programme was introduced in 1988. The NHS cervical screening programme is highly organized. In the UK, women aged 25–65 years are invited for screening every 3 or 5 years. It is thought that screening under the age of 25 years may do more harm than good as cervical cancer is rare in this age group (Sasieni and Adams 1999). There are clear service guidelines, effective data collection systems using a number of mandatory returns from cytological laboratories, and internal and external quality assurance systems. Target population coverage is the key to success. The programme aims for coverage of over 80% of the target population, but there has been a worrying fall in levels in recent years, falling as low as 66% in women aged 25–30 years (Figure 38.6).

Figure 38.6 Cervical screening: coverage by age, England, 2003 and 2008, showing the fall in screening uptake in younger women.

Source: Health and Social Care Information Centre, 2009. Cervical Screening Programme 2007/2008. The Health and Social Care Information Centre, Sheffield.

In spite of the success achieved by cervical cytology, it is not without its shortcomings. The assessment and definition of cytological abnormality are subjective with considerable interobserver variation. The process is laborious and tiring with considerable scope for operator error, especially when the workload is high. Furthermore, cytology screening has little effect on the incidence of adenocarcinoma of the cervix.

False-negative results have been variously estimated as being between 2.4% and 26% in various types of study. False-negative results can occur because of inadequate sampling, incorrect laboratory processing, or detection and interpretative errors of the cell samples. This type of error is potentially serious, as a falsely reassuring result may result in no further investigation for 3–5 years, but it is relatively uncommon. There is a need for less labour-intensive and more reliable screening methods.