The Role of Colposcopy in Cervical Precancer CHAPTER 1 CHAPTER

CHAPTER 1 CHAPTER

chapter 1. The role of colposcopy in cervical precancer CHAPTER 1 CHAPTER

A positive diagnostic test result squamous intraepithelial lesion and where low rates of default from reveals an abnormality or disease. (HSIL) level (cervical intraepithelial follow-up exist, the threshold for Advice about management is usual- neoplasia grade 2 [CIN2] or greater). treatment may be higher, especially ly accepted willingly. When a wom- However, in many countries with in young women. The management an receives an abnormal cervical established screening programmes of screen-positive women would screening test result, the expecta- tions and fears that she carries are Fig. 1.1. Relative rates of human papillomavirus (HPV) infection, low-grade quite different. Cervical screening squamous intraepithelial lesion (LSIL), and cervical cancer (high-grade tests – whether visual inspection, squamous intraepithelial lesion [HSIL]). cervical cytology, or human papil- 0.5 million cervical cancer cases lomavirus (HPV) tests – do not give a diagnosis; rather, they modify the risk for an individual of developing 10 million HSIL cervical cancer. The progression to precancer and cancer is slow and 60 million is a very uncommon outcome for LSIL/innocent screen-positive women (Fig. 1.1). condylomata The threshold of abnormality at which the risk of cancer outweighs any disadvantage of treatment varies 300 million cases according to patient characteristics of HPV infection and local service considerations. The World Health Organization (WHO) advises treatment at the high-grade

Chapter 1. The role of colposcopy in cervical precancer 1 Fig. 1.2. The ideal, dichotomous screening test. cancer. The problem of imperfect sensitivity and specificity is illustrated in Fig. 1.3. Visual inspection with acetic acid (VIA) is fast becoming the de facto screening method of choice in many regions where cytology and HPV testing are out of reach. A “screen- and-treat” approach is gaining pop- ularity as an efficient method of reaching large numbers of women in difficult circumstances. However, the specificity of VIA is poor, and the difficulty of missing endocervical lesions (whether they are squamous or glandular) is a real problem. Over- treatment of women with false-posi- be much easier if the results of whom there is a risk of precancer. tive VIA test results is perceived as screening tests were diagnostic and Because of this, cytology testing has a necessary trade-off to reduce the dichotomous. An ideal screening test to be performed relatively frequently overall incidence of cervical cancer. would provide two possible answers, (3–5-yearly). The long natural history Thus, screening tests (VIA, cy- and management advice would be of cervical cancer is forgiving of the tology, and HPV testing) are imper- simple (Fig. 1.2). relatively poor sensitivity of cytology. fect, and women with an abnormal But current screening tests for Also, cytology will sometimes primary screening test need further cervical precancer are neither com- recognize cells that are very mildly consideration before reflex referral to pletely sensitive nor absolutely spe- abnormal, or even of a “borderline” colposcopy and/or management. For cific. For example, testing for on- nature: borderline nuclear abnormal- those women in whom a suspicion cogenic (or high-risk) HPV will pick ity, and atypical squamous cells of of CIN2 has been reported, referral up almost all cervical precancerous undetermined significance (ASCUS). to colposcopy is still the appropriate lesions but will also test positive These categories of abnormality cre- advice. (When considering treat- in women who have innocent and ate headaches and frustration for ment, lesser grades of abnormality, transient high-risk HPV infection. In both clinician and patient. They in- or CIN2 with negative p16, the cost– a recent study, 73% of women with clude mostly women who are not at benefit equation is less certain.) The a positive oncogenic HPV test also a high risk of progression, as well as concept of triage is to use other tests had a negative or normal smear (Kat- a minority of women who are. Clear- for women who have an imprecise ki et al., 2011). Cytology, in contrast, ly, the ideal test – not yet available primary screening test report, so is far more specific than HPV testing – would identify only the women who that those women who have a gen- but will miss a number of women in are at a high risk of progression to uine risk of progression to cancer

Fig. 1.3. (a) The ideal screening test. (b) The real screening test. a b

2 CHAPTER 1 CHAPTER may be identified and referred for Fig. 1.4. Age-standardized incidence of invasive cervical cancer and colposcopy and management. Also, coverage of screening in England, showing a decrease in incidence after just as importantly, women who are the introduction of a national call-and-recall screening programme. at a very low risk of progression may be spared the interventions of biopsy and/or treatment. Colposcopy is also important in avoiding overtreatment that may occur with “screen-and- treat” programmes where false-positive rates may be very high (Basu et al., 2015). Finally, colposcopy may recognize invasive cancer not heralded by a screening test.

1.1 Traditional screening: rationale and practice

Systematic high-coverage and quality-assured population screening and recognize morphological changes cancer, but rather the subsequent treatment for precursors to cervi- allows cytologists to report different management of screen-positive cal cancer are highly effective. The degrees of abnormality, classified as women. Screen-positive women conditions for an ideal screening test borderline, low-grade, or high-grade may be stratified according to risk as enunciated by Wilson and Jung- in squamous or glandular cells. and either referred to a colposcopist ner (1968) apply very precisely to High-grade smear abnormalities are or retested some time later. Classi- cervical cancer. The disease has a associated with a higher risk of cer- cally, colposcopic examination facili- long precancerous phase, effective vical cancer development over the tates the recognition and localization screening tests are available and are subsequent decade, and low-grade of genuinely high-grade abnormality easily performed, and the disease and borderline smears are associat- within the “at-risk” epithelium – i.e. is common enough to justify the ex- ed with a dramatically lower risk. the transformation zone (TZ) – and pense of population screening, even Of course, it is not the screen- facilitates either confirmatory biopsy in low- and middle-income countries ing itself that prevents cervical or excision/ablation of the TZ, or both. (LMICs) (Denny and Prendiville, 2015). Finally, there are effective and Fig. 1.5. Projected cervical cancer deaths without any screening (England low-morbidity preventive treatments and Wales). Dashed line represents cervical cancer deaths that would have of proven value for screen-positive happened after 1987 without screening. Solid line shows annual deaths women. In those countries and re- from 1953 to 2002. Arrow indicates start of national screening in 1988. gions that have implemented quality-assured, high-coverage call-and- recall screening programmes for cervical precancer, large reductions have been demonstrated in the rates of both incidence of and mortality from cervical cancer (Figs. 1.4 and 1.5).

1.2 Management of screen-positive women

In the traditional system, a cervical smear test is the usual screening tool (Fig. 1.6). The microscopic examination of cellular material to

Chapter 1. The role of colposcopy in cervical precancer 3 Fig. 1.6. Traditional algorithm for screen-positive cases. routine office or outpatient LLETZ could be learned and performed, compared with laser treatment or cold-knife cone biopsy, meant that the threshold for treatment fell. At about the same time, reporting rates for low-grade and borderline abnormalities varied enormously. In Ireland’s National Cancer Screen- ing Service, for example, at the beginning of its cervical screening programme all smears were sent to a laboratory in the USA. Of these specimens, 14.9% were reported as “not normal”, rates similar to those Established treatment modalities are electrosurgical excision procedure for VIA. Finally, as the very high effective. Through well-organized (LEEP) in the USA, was introduced sensitivity of oncogenic HPV testing screening programmes and the man- as a simple excisional outpatient became apparent, the test became agement of screen-positive women, treatment (Prendiville et al., 1989). used widely, either on its own or in incidence and mortality rates have This technique largely replaced la- combination with cytology. The man- been reduced significantly (Arbyn ser treatment and other destructive agement of screen-positive women et al., 2009; IARC, 2005; Miller, methods because of its simplicity, became more complex. The optimal 1993; Peto et al., 2004; Sasieni et cost–effectiveness, and facility for “next step” for most women with any al., 2003). For many years, this sys- comprehensive histological assess- “not normal” smear is no longer au- tem or model of screen, colposcope, ment of the removed TZ. In select- tomatic referral for colposcopy and and treat selected patients has been ed cases it obviated the need for a treatment. standard and highly successful. preliminary colposcopically directed But the protocol is not perfect. It biopsy. 1.3 Clinical guidelines was a practical and workable system Although it has been known for 30 years ago. For example, in the some time that low-grade abnormal- Against this background, some na- United Kingdom, women were not, ities have a low risk of progression tional societies of colposcopy and at that time, referred for colposcopic (Moscicki et al., 2010), it became cervical pathology have taken on the examination unless the cytologist clear that some cytological low- responsibility of generating clinical considered there to be a significant grade or borderline smears harbour guidelines for physicians and colpos- risk of progression to cancer – in oth- higher-grade lesions. As a result, copists as an aid to management. er words, when there was a smear women with minor-grade abnormali- For example, in the United King- report describing changes suggest- ties were increasingly referred for im- dom, the National Health Service ing severe dysplasia (i.e. a smear mediate colposcopic evaluation, and (NHS) Cervical Screening Pro- report of severe dyskaryosis). At that in time the majority of women with an gramme produced a clinical guide- time, women with minor abnormal- abnormal smear of almost any grade lines document entitled Colposco- ities were followed up with repeat were, in some regions, referred for py and Programme Management: cytology. The threshold for referral colposcopic examination. In some Guidelines for the NHS Cervical to colposcopy was relatively high, parts of Europe and the USA, where Screening Programme (NHS, 2004), and most women who were referred screening was routinely offered an- which encompassed screening, had a high-grade smear report that nually and often to very young wom- management, and follow-up guide- was usually confirmed and managed en, the risk of unnecessary treatment lines for clinicians involved in the at colposcopy. The decision to pro- became commonplace. United Kingdom screening pro- ceed to treatment was uncomplicat- During the 1980s and 1990s, the gramme. It was updated in 2010 ed and had consensus support. Also number of women referred for col- and 2016 (NHS, 2010, 2016) and is in the 1980s, large loop excision of poscopic examination increased ex- a valuable reference document for the transformation zone (LLETZ), ponentially. To add fuel to the fire of anyone involved in cervical precan- which later became known as loop overtreatment, the ease with which cer screening and/or management.

4 CHAPTER 1 CHAPTER In 2001, a new reporting system influences modify the risk equation. 1.5 Screening and triage for cytology smears was published in Finally, the likelihood of default from options in current practice the USA, and at about the same time follow-up monitoring of untreated pa- the results were published of a large tients needs to be taken into consid- There are several different scenarios study in the USA reporting different eration. This thinking was developed where triage might be useful in the strategies for managing minor cyto- in a clinical opinion paper in 2007 management of cervical precancer. logical abnormalities (ASCUS-LSIL (Castle et al., 2007). The paper put Local circumstances, cost, avail- Triage Study (ALTS) Group, 2003a, forward cogent arguments for using ability of test facilities, and expertise 2003b). To guide physicians in the a quantifiable risk assessment rather will all play a role in determining USA, the American Society for Col- than an individual test result as the which primary screening tool is used poscopy and Cervical Pathology arbiter of management. It extended and which secondary or triage test (ASCCP) implemented a process the principle beyond cytology to col- is used. Examples are illustrated in that developed broad consensus poscopy and histology (Fig. 1.7). Figs. 1.8–1.11. guidelines to aid clinicians in man- When assessing the thresh- Primary screening tests may aging women with abnormal cervical old for referral to colposcopy or for also be used as triage tools (cytolo- cytology. These guidelines became treatment, some researchers have gy, HPV testing, or VIA), and some a defensible and practical aid for a defined thresholds of risk. The 2012 are used in conjunction with others busy gynaecologist to use in every- ASCCP clinical guidelines document to improve the test characteristics of day practice in the USA (Massad et (Massad et al., 2013) includes a risk the primary screening test (co-test- al., 2013; Nayar and Wilbur, 2015; of progression (Table 1.1) that re- ing with cytology and HPV; dual test- Wright et al., 2003), and they apply flects this approach. ing, i.e. p16/Ki-67, with HPV). Finally, to the context of screening principle and practice in the USA. This is not Graphical representation of the risk of cervical precancer at different meant to be a criticism of the process Fig. 1.7. stages and results of screening and clinical management for cervical cancer but should make the reader wary prevention. The risks for each stage and result are approximate risks for about using management algorithms cervical intraepithelial neoplasia grade 3 (CIN3) within a screening interval. outside of the context in which they The axis to the right of the figure represents increasing risk, from nearly were generated. 0% (blue) to 100% (red), of cervical precancer on a logarithmic scale. Each stage of screening and clinical management is represented by a different 1.4 Risk assessment in pattern, and the arrows at the bottom indicate the sequence of the stages. patient management # Less than half of the cases of CIN2 on biopsy are subsequently diagnosed as CIN3 on excisional tissue (precancer). † Within a screening interval. * Test results at the next follow-up visit (≥ 6 months). When to advise the asymptomatic screen-positive woman to have fur- 100% ther intervention requires an assessment of risk. Does the risk of intervention outweigh the risk of cancer † evolving? Many of the triage recommendations contained in management algorithms depend exclusively on the result of the screening test. For example, it may advise an oncogenic HPV test for a woman with a smear report of ASCUS. But there are other factors that modify the relative risk of progression (Mergui et al., 2010). A smear reporting HSIL-mod- erate dyskaryosis carries a very dif- Increasing Risk of Precancer (CIN3) ferent risk of progression to cancer in a woman younger than 30 years compared with a woman older than 40 years. Age, smoking, and other

Chapter 1. The role of colposcopy in cervical precancer 5 Table 1.1. Referral or follow-up according to risk of CIN3 by primary care staff trained in a

Risk of CIN3 Recommended action relatively short time. Visual inspection provides an immediate result > 5% Immediate referral for colposcopy that can be determined on-site in 2–5% 6–12-month follow-up hospitals, in clinics, or in the field

0.1–2% 3-year follow-up and allows the health-care worker the opportunity to immediate treat < 0.1% 5-year follow-up those TZs that are possibly abnor- CIN3, cervical intraepithelial neoplasia grade 3. mal. The sensitivity and the specificity of visual inspection techniques Fig. 1.8. Generic use of triage after primary screening. are highly variable and are very re- liant on quality-assured training and retraining (Sankaranarayanan et al., 2007; Sauvaget et al., 2011). Of course, these methods only assess the ectocervix and will miss endocervical lesions, with consequent poorer performance in older women. Final- ly, visual inspection is very unlikely to detect glandular intraepithelial lesions or squamous lesions in the canal (some type 2 TZs and most type 3 TZs).

1.5.1.1 Digital VIA Fig. 1.9. Possible triage of low-grade and borderline smears using p16/ Ki-67 dual testing. Given the concerns about subop- timal sensitivity and lack of an efficient quality assurance mechanism for VIA-based screening, modifi- cations of the method have been used in some regions. For example, Parham et al. (2015) have used enhanced magnification of cervical lesions, peer review, quality assurance, continuing medical education, objective recording of screening test results, and access to expert opinion in their screen-and-treat programme in Zambia. Treatment decisions are made primarily on the basis of VIA. However, if there are disagreements between VIA and enhanced mag- repeat cytology has also been used iodine (VILI). They may be per- nified images, then the images are as a triage tool for minor-grade cytol- formed by nurses or other prima- used to make the final decision. ogy reports. ry health-care workers. They use light-illuminated speculum examina- 1.5.2 Cytology 1.5.1 VIA and VILI tion of the cervix after the application of 5% acetic acid (VIA) or Lugol’s io- Cervical cytology smears need to Two naked-eye inspection methods dine (VILI). be examined by properly trained are in widespread use in LMICs: VIA VIA and VILI are inexpensive and quality-assured cytologists. The and visual inspection with Lugol’s and simple, and can be carried out subsequent treatment of women

6 CHAPTER 1 CHAPTER Fig. 1.10. Possible triage of positive HPV test using cytology. predictive value of HPV testing is very high. A large number of studies have investigated how best to use this information. HPV testing for all known oncogenic types has been available and approved for many years. There are essentially three realms where oncogenic HPV testing is of proven clinical utility: • as a screening tool in women older than 30 years; • as a triage tool for women with low- grade cytological abnormalities; and • as a follow-up tool for women who have been treated for squamous or with high-grade CIN prevents the to collect, process, and report cy- glandular cervical precancer. development of cancer (Miller, 1993). tology are simply not available. It is However, in LMICs, cytology has not highly unlikely that health systems in 1.5.3.1 Oncogenic HPV testing been as successful, for several rea- LMICs will wish to establish cytology as the primary screening tool sons. Unless strict and continuing as the primary screening tool for cer- quality control programmes are in vical cancer prevention. Sankaranarayanan et al. (2009) first place, cytology may perform poor- demonstrated in a large cluster ran- ly in terms of recognizing abnormal 1.5.3 HPV testing domized controlled trial (RCT) that lesions. It has good specificity but a single round of HPV testing was lacks sensitivity, and in some series HPV DNA testing will probably re- superior to cytology or VIA or no may be associated with sensitivity place or complement cytology as screening in reducing the incidence rates of only 50% (Arbyn et al., 2009; the primary screening tool in many of advanced cervical cancer and Nanda et al., 2000). Also, the test re- developed countries for women old- in reducing mortality from cervical quirements are expensive, and the er than 30 years (Arbyn et al., 2013). cancer. screening interval needs to be rela- Because of the absolute relationship Four RCTs of HPV screening tively frequent (3–5 years). Finally, in between oncogenic HPV infection versus routine cytological screening many regions the facilities necessary and cervical cancer, the negative have been undertaken in Europe. In these studies together, 176 464 women aged 20–64 years were randomly Possible use of HPV testing for triage of positive visual inspection Fig. 1.11. assigned to HPV-based (experimen- with acetic acid (VIA) test. tal arm) or cytology-based (control arm) screening in England, Italy, the Netherlands, and Sweden. In all four studies, the incidence of CIN3 was lower in those women who had initially been screened by HPV than in those initially screened by cytology, and similar rates of reduction were recorded in each study. This reduced rate was true across all studies despite differenc- es in screening protocols between the studies. Because none of the individual studies was sufficiently large to show a reduction in cancer

Chapter 1. The role of colposcopy in cervical precancer 7 incidence, a recent overview of the with low-grade cytology reports, HPV testing is more cost-effective pooled data has been undertaken by even as high as 30% (Kinney et al., than cytology in the context of a Ronco et al. (2014). 1998). The search has continued for European national screening pro- The 176 464 women were followed a triage tool to discriminate women gramme (Coupé et al., 2007; Legood up for at least two rounds of cervi- at genuine risk of having or develop- et al., 2012). cal screening, equating to a median ing CIN3. of 6.5 years. A total of 107 invasive However, the management of 1.6 Colposcopy cervical carcinomas were found. The women with minor-grade lesions re- authors concluded in their summary: mains controversial, and follow-up 1.6.1 What is colposcopy? “HPV-based screening provides 60– recommendations for women with 70% greater protection against inva- ASCUS and low-grade squamous in- Colposcopy is low-powered micro- sive cervical carcinomas compared traepithelial lesion (LSIL) have varied scopic and light-illuminated exami- with cytology. Data of large-scale from conservative management (i.e. nation of the lower genital tract ep- randomised trials support initiation repeat cytology) to immediate refer- ithelium. The first reported use of a of HPV-based screening from age ral for colposcopy and biopsy (Sout- colposcope was in Hamburg, Ger- 30 years and extension of screen- ter et al., 2004). Because of the cru- many, in the early 1920s as a result ing intervals to at least 5 years.” cial role that HPV infection plays in of a collaboration between the Uni- The results of this overview are very the genesis of cervical cancer, HPV versity of Hamburg and the German convincing. testing has been investigated as an microscope manufacturer Leitz. alternative to repeat cytological test- The early work published in the 1.5.3.2 Oncogenic HPV testing ing, in a large number of disparate 1930s from Hamburg described the as a triage tool for women studies. Several formal reviews of origins of cervical cancer being in with low-grade cytological these studies have been performed. a sheet of epithelium, i.e. intraepi- abnormalities The most recent Cochrane review thelial, as opposed to arising from a (Arbyn et al., 2013) advised that for single focal lesion. During the 1930s As the threshold for referral for in- triage of women with LSIL, the Hy- and 1940s, colposcopy practice vestigation of smear abnormalities brid Capture 2 (HC2) oncogenic spread and evolved throughout Eu- fell over the past two decades, the HPV test yields a significantly higher rope. It was not until the 1960s and number of women attending colpos- sensitivity, but a significantly lower 1970s that colposcopy became more copy increased. As a general princi- specificity, compared with repeat widely established, through individu- ple, women with any degree of cy- cytology. al experts. tological abnormality require either follow-up or treatment, depending 1.5.3.3 Oncogenic HPV testing 1.6.2 What can colposcopy be on the risk of progression to cancer. as a follow-up tool for women used for? There is consensus that most wom- who have been treated for en with high-grade lesions should squamous or glandular cervical Colposcopy may be used to ex- be referred for colposcopy and precancer amine any epithelial surface of the managed according to colposcopic lower genital tract. Some of the indi- assessment and/or biopsy results. Because residual disease and re- cations for colposcopy are given in However, for low-grade abnormali- current disease can occur up to Table 1.2. It may be used as a pri- ties, no such consensus exists (Cox, 20 years after treatment, it is impor- mary screening tool and as a way of 2005; Sawaya, 2005; Soutter, 1994). tant to implement a follow-up proto- facilitating different treatment modal- The reason is that it is not possible col wherever possible (Soutter et al., ities. Colposcopy does not perform to predict the natural history of mi- 2006). A number of RCTs and sever- well as a primary screening tool nor-grade lesions on the basis of cy- al meta-analyses have demonstrat- (Leeson et al., 2014). It is also used tology alone. Low-grade abnormali- ed that HPV testing is the best test to examine the vulva, the anus, the ties often regress (Melnikow et al., of cure (Arbyn et al., 2005). It has vagina, and more recently the oro- 1998; Ostör, 1993), and the potential replaced cytology and colposcopy pharynx as well as the penile epithe- for overtreatment is obvious. Howev- in several national clinical guidelines lium, because each of these sites is er, some studies reported relatively documents, although many still ad- prone to developing colposcopically high rates of high-grade or moder- vise co-testing with cytology. Finally, recognizable precancerous lesions. ate-grade abnormalities in women several reviews have concluded that Colposcopy has also been used in

8 CHAPTER 1 CHAPTER Table 1.2. Common indications for colposcopy Federation of Cervical Pathol- What is colposcopy best used for? ogy and Colposcopy (IFCPC) • A suspicious-looking cervix nomenclature or terminology (see Annex 3). • Symptoms suggestive of cervical cancer, e.g. persistent postcoital bleeding, persistent intermenstrual bleeding 5. Compile a Swede score (see An- • Cervical leukoplakia nex 4). 6. Where possible, take a video or a • A cytological abnormality number of pictures of the exam- • A positive VIA or VILI screening test ination findings so as to record: • A positive high-risk HPV test in the presence of a low-grade or borderline smear a. the TZ type and size abnormality or other screening test abnormality b. the site(s) of greatest abnor- HPV, human papillomavirus; VIA, visual inspection with acetic acid; VILI, visual inspection with Lugol’s iodine. mality c. the site of any biopsy clinics investigating lower genital tract gen before colposcopic d. the treatment, if performed. infection. However, the great majority assessment? Misunderstanding the role of the of colposcopic examinations are of b. Determine whether there is colposcopic examination is common, the cervix with suspected precancer. inflammation. to the extent that some authors con- The performance of colposcopy i. Is infection (viral, fungal, sider the major role of colposcopy as a purely diagnostic tool is known bacterial) present, and is to be guiding the diagnostic biopsy to be influenced by the result of the investigation and treat- (Jeronimo and Schiffman, 2006; screening test, and there are sever- ment prudent before col- Wentzensen et al., 2015). Some al studies where colposcopy has poscopic assessment? aspects of colposcopic evaluation not performed well (Jeronimo and c. Confirm full visibility of the en- are contextual, and some are not. Schiffman, 2006; Pretorius et al., tire cervix and upper vagina Diagnostic acumen and recognition 2011). However, in those countries under colposcopic view. of high-grade abnormality will vary where colposcopy is part of a prop- d. Determine whether there is according to the prevalence of high- erly constructed, quality-assured evidence of previous treat- grade abnormality in the clinic refer- programme, it is associated with a ment, or any degree of epi- ral population. Other aspects of col- very high negative predictive value thelial fibrosis. poscopic evaluation are independent (Cruickshank et al., 2015; Kelly et Once these assessments have of case characteristics – for example, al., 2012; Ricci et al., 2015). Also, been made, it will be possible to de- TZ type, adequacy of examination, colposcopy is not just a diagnostic termine whether a complete colpo- hormonal status, and infection state. tool; indeed, that is not even its most scopic examination can be undertak- Every colposcopy should assess valuable role. en. If so, the following steps should the degree of abnormality as reflect- The colposcopic examination be performed. ed in a simple scoring system, for ex- should undertake and document 2. Determine the type and size of the ample the Swede score (Strander et the following: TZ: al., 2005) (see Annex 4). Sometimes 1. Assess the state of the cervix at a. TZ type (see nomenclature in it will be appropriate to take a biop- the time of examination, and de- Chapter 7, and Annex 1) sy, and sometimes not. Sometimes it termine whether it is possible to b. TZ size (small or large). will be appropriate to treat at the first/ undertake an adequate examina- 3. Recognize epithelial abnormality assessment visit, and sometimes tion (see Chapter 6). (i.e. is disease present?). not. Sometimes it will be appropri- a. Assess the hormonal status. a. Cervical precancer non-inva- ate to take a sample for cytology, for i. Is the epithelium well-es- sive or intraepithelial abnor- HPV testing, for endocervical brush trogenized? mality classified as: cytology, or for other biomarkers of ii. Are pregnancy changes i. CIN1 or LSIL cervical cancer progression. If the present? ii. CIN2 or HSIL-CIN2 TZ is not fully visible, the examina- iii. In postmenopausal wom- iii. CIN3 or HSIL-CIN3. tion will be incomplete. In that case, en, is the degree of atro- 4. Document the above examination the decision about management will phic epithelial change findings in a standard and au- depend on other case characteristics sufficient to consider ditable format (see Annex 2) us- and whether to excise the TZ by way prescribing topical estro- ing the most recent International of a type 2 or type 3 excision. These

Chapter 1. The role of colposcopy in cervical precancer 9 Fig. 1.12. (a) Colposcopic image of a normal transformation zone (TZ). (b) Colposcopic image of a TZ exhibiting low-grade changes. (c) Histological section of a normal squamocolumnar junction (40× magnification).

a b c

characteristics include: 1.6.3 Diagnostic performance or equivocal abnormalities. There is • the patient’s age and fertility of colposcopy no gold standard. A systematic and aspirations; adequate colposcopic examination • the reliability of the referral smear The diagnostic accuracy of col- by a properly trained colposcopist or other screening test; poscopic examination, like that of will nearly always recognize HSIL • the risk of default from follow-up; any subjective test, will vary ac- when cytology has heralded it and • the grade of suspected abnormal- cording to the training and expertise the report is known (Fig. 1.13). It will ity; and of the colposcopist as well as the sometimes recognize microinvasive • the availability of ancillary inves- prevalence of the disease. Also, it disease (Fig. 1.14), but not always tigations (e.g. endocervical brush is influenced by knowledge of the (Howe and Vincenti, 1991). This is cytology, HPV testing, other bio- referral screening test report. Final- not an issue if the TZ has been ex- marker tests). ly, it performs better at the extremes cised in its entirety. Discovering high-grade abnor- of abnormality: normal/low-grade When the cytology report sug- mality when the smear reports a (Fig. 1.12) and definite high-grade gests a low-grade or borderline low-grade or borderline abnormal- (Fig. 1.13). The weakest diagnostic abnormality and an adequate col- ity is more difficult, and it will often performance is with the middle or poscopic examination by a trained be prudent to take one or even two equivocal grade of abnormality. This colposcopist reveals low-grade or biopsies, but in a quality-assured is also true for cytology and pathol- normal appearance (Fig. 1.12), the colposcopy service the negative ogy. The subjective error inherent risk of HSIL occurring in the next predictive value of a negative/nor- in colposcopic assessment is very 4 years or more is very low. In this mal colposcopic examination is very similar to the range of disparity that situation, the negative predictive val- high, even without a colposcopically exists among cytologists and pathol- ue of a quality-assured colposcopic directed biopsy. ogists when assessing middle-grade examination, even with a positive

Fig. 1.13. (a) Colposcopic image of high-grade cervical intraepithelial neoplasia (CIN); coarse punctation. (b) Low- power colposcopic image of high-grade CIN; coarse mosaic pattern. (c) Low-power colposcopic image of high- grade CIN; note the atypical vessels and sharp margin at the 5 o’clock position.

a b c

10 CHAPTER 1 CHAPTER Fig. 1.14. (a) Microinvasive squamous carcinoma. (b) Colposcopic image of unscreened population or women microinvasive disease. referred because of a screening test with low specificity. a b In deciding whether to take a biopsy, the colposcopist should consider whether the biopsy will alter management. At each end of the spectrum of suspected abnormality, a biopsy will not usually affect management. It is where uncertainty pre- vails that a colposcopically directed biopsy is valuable. Figs. 1.15 and 1.16 illustrate simplified approaches to managing suspected low-grade HPV test, is very high (Cruickshank by infection, previous treatment, or and high-grade lesions, respective- et al., 2015; Kelly et al., 2012; Ricci atrophy. ly, in the context of a cytologically et al., 2015). Completely normal cer- It is possible to miss a high-grade screened population. vical epithelium in a fully visible TZ lesion when performing colposco- Attempts to improve colposcopic is, again, usually very clear. Lesser py, particularly if the lesion is small diagnostic accuracy vary in their ap- grades of CIN are more difficult to and possibly transient or when the proach. One way is to improve col- discriminate from normal epithelium, colposcopic examination is compro- poscopic quality control. The Italian but low-grade disease carries an ex- mised by inflammation, bleeding, region of Emilia-Romagna has intro- ceedingly low risk of progressing to or hormonal changes (atrophy or duced a voluntary quality assurance cancer. pregnancy) such that the examina- system. The programme recently Colposcopically directed biop- tion should be recognized as being reported an Internet-based quality sies are sometimes necessary and inadequate. Other reasons why col- assurance programme. Of 65 col- sometimes not. For most women, a poscopy might underperform at a poscopists in the region, 59 partici- colposcopic impression of CIN3 in diagnostic level are that the colpos- pated in a review of 50 selected col- the presence of a high-grade smear copist is inadequately trained or that po-photographs and classified them warrants excision of the TZ rather the women being examined are an according to colposcopic visibility of than a directed biopsy. Indeed, when a quality-assured laboratory reports Fig. 1.15. Algorithm of management where smear report is low-grade a smear as CIN3, colposcopy will squamous intraepithelial lesion (LSIL)/atypical squamous cells of undeter- reveal a high-grade lesion in the mined significance (ASCUS), given an adequate colposcopic examination great majority of cases. In this situa- in a type 1 transformation zone (TZ) by a properly trained colposcopist and tion, when colposcopy does not find with a cytology report from a quality-assured laboratory. evidence of CIN3 the colposcopist should consider a colposcopically directed biopsy and, even more importantly, consultation with the referring laboratory and review of the referral smear, before deciding management. It is far better to com- petently perform a colposcopy than to rely on random biopsies. When an adequate colposcopic examination is normal, a random biopsy will very rarely find high-grade CIN (Song et al., 2015; Wentzensen et al., 2015). These comments, of course, pertain to the colposcopic examination that is adequate and is not compromised

Chapter 1. The role of colposcopy in cervical precancer 11 Fig. 1.16. Algorithm of management where smear report is high-grade the squamocolumnar junction (i.e. squamous intraepithelial lesion (HSIL), given an adequate colposcopic TZ type), degree of abnormality, and examination in a type 1 transformation zone (TZ) by a properly trained need for biopsy. In that study, a bi- colposcopist and with a cytology report from a quality-assured laboratory. opsy was advised in 99% of women who ultimately had histologically proven CIN3. The authors (Bucchi et al., 2013) acknowledged that assessment of still images is not as good as video or in vivo assessment, but the colposcopists in that programme performed exceptionally well. Another approach has been to advocate multiple random biopsies (Pretorius et al., 2004). The Chi- nese group that originally reported the advantage of random biopsy at colposcopy has recently reviewed its database (Song et al., 2015) and concluded that “random biopsy is not effective in the negative quad- rant in women with positive colposcopy”. Wentzensen and colleagues (Wentzensen et al., 2015) have come to a similar conclusion. Key points • Colposcopy is an assessment and diagnostic tool and offers the best way to manage women with suspected cervical precancer.

• A colposcopic examination should be systematic and structured and should always record the adequacy of the examination, the transformation zone type and size, and the degree of abnormality as reflected in an objective diagnostic scoring system, for example the Swede score.

• When quality-assured, colposcopic examination has a high negative predictive value.

• Excisional therapy for cervical precancer should always be performed under colposcopic vision.