Section 3 -

The Future of Cervical Screening Annabelle Farnsworth Douglass Hanly Moir Pathology Sydney , Australia

Attempts have been made to automate cervical cytology for over forty years. The Pap smear, a technique originally describe by George Papanicolaou in the 1940's is one of the most commonly performed pathology tests and yet unlike other high volume tests, is still performed manually in much of the world. The technique persists even though the Pap test has a relatively low sensitivity for significant disease and an acknowledged false negative rate [1]. Where it is used, however, as the screening test especially in an organised screening program it has had a significant effect on the reduction in the incidence of and the mortality from cervical cancer [2]. The impetus for the most recent developments in automation in this area began almost twenty years ago when there was recognition of the failings of Pap smears combined with the rapid development of computer and video technology. Even so, the introduction of automation has taken much longer than would have been first thought. Recent developments however, may offer a partially automated alternative, which should eventually replace the conventional Pap smear. Review of the technology In the late 1980's and early 1990's four new technologies emerged into the arena of cervical screening. These were: Papnet (Neuromedical Systems Inc) AutoPap (Neopath Inc) SurePath (Autocyte, TriPath Imaging) ThinPrep (Cytyc Corp) 1 & 2. Papnet and AutoPap These two technologies will be discussed together as they were initially developed to be automated readers of conventional Pap smears. Papnet was introduced as a pre-screening method of conventional Pap smears. The slides were read though the Papnet imager which generated images of interest. These were then reviewed on a computer screen by a scientist. Papnet used neural net computer technology to analyse the complex nature of the conventional Pap smear. AutoPap was designed to also pre-screen conventional slides. Conventional computer technology was used. Images were not presented for review but rather slides with the most severe abnormalities were grouped together and these slides were then completely screened by scientists. The percentage to be completely screened could be altered but whichever component was not re-screened was archived and was not ever reviewed by a scientist or pathologist. Both of the technologies achieved FDA approval and were used predominantly as quality control measures. Papnet was a commercial failure. This was for a number of reasons including marketing strategies and the workflow design which required the conventional smears to be shipped to a central laboratory for imager screening rather than having the machines situated within the laboratory. Both the Papnet and the AutoPap technology were purchased in some part by a third group, SurePath-Autocyte (see below). They have not been commercially available since then. 3 & 4. SurePath & ThinPrep Both of these technologies rather than developing imagers to read conventional smears concentrated on developing imagers, which would read an especially prepared liquid based sample (LBC). To improve the ability of a machine to read the collected sample, both of these processes depended on placing the sample into a liquid preservative from which a monolayer of cells could be prepared. The sample was cleared of as much extraneous material (blood and inflammatory cells) as possible; SurePath by sedimentation and ThinPrep by filtration. The development of these liquid based samples was a great success; just as they were easier of the machines to read so they were easier for the human eye. Both of these technologies have had considerable success replacing conventional smears in some countries. It is estimated in the USA that 80% of cervical cytology is liquid based. Both of these technologies now have image analyzers available to read their monolayer samples. SurePath has developed FocalPoint, which achieved FDA approval as a primary screener of both conventional and SurePath slides, based on the AutoPap technology but is still developing its location guided system. The ThinPrep image analyzer (TPI) received FDA approval in 2004. The Imager identifies specific fields of interest within the monolayer sample which are then examined by a cytologist on an automated light microscope. Although the TPI has had good uptake in the United States, skepticism still remains in other parts of the world as to both the value of liquid based cytology and the use of automation in this process. An Australian ThinPrep Image Analyzer study Australia remains the only country in the world where LBC is done as a split sample technique. This situation exists because two substantive government funded reviews of the technology failed to show a benefit in an already successful cervical screening program [3]. Government funding was not forthcoming for LBC, as the increased expenditure could not be justified. LBC, predominantly ThinPrep remains relatively widespread in Australia however, funded entirely by the patients, the demand being driven by the manufacturers themselves. Questions have also been raised over the validity of LBC generally [4]. Douglass Hanly Moir is a large laboratory situated in Sydney , Australia , which has been using ThinPrep (TP) in the split sample role for over nine years. In 2004 the ThinPrep Image Analyzer (TPI) became available in Australia and it was decided at that time to use it for reading the TP samples. This was on the basis of both the FDA approval and a preliminary verification study. It was also decided that if the image analyzer was to be introduced this may offer an opportunity to assess its efficacy compared to conventional cytology. The accuracy of the TPI and conventional cytology (CC) was compared during normal laboratory operation where the TP sample had been prepared after taking a conventional pap smear. The TPI and CC reading were done without knowledge of the result of the other reading. The most severe abnormality was reported to the referring practitioner. Histology results for all cases in which the two cytology reading showed more than minimal disagreement were sought from the New South Wales Pap test register. 55,164 split sample pairs were examined over a nine-month period. 3.1% of the CC and 1.8% of the TPI were unsatisfactory. There were 1758 women for whom there was more than minimal discrepancy between the TPI and CC results. The TPI gave the more severe result in 1193 of the 1758 cases. In cases where only one of each pair of discrepant cytology results was CIN 1 or higher grade TPI detected 133 cases of high-grade histology among 380 biopsies (35%), whereas the conventional cytology detected 62 cases among the 210 biopsies (29.5%). A repeat reading of the histology where the pathologist was blinded to the initial cervical cytology result showed similar findings. There were only 6 histologically confirmed high-grade glandular lesions in our study population, all of which had a glandular lesion predicted on TPI. One of the TPI results was reported as a low grade glandular but simultaneously predicted a high-grade squamous lesion. CC and TPI's ability to detect high-grade glandular lesions in this study was equivalent. Reading times were measured over five months for both the ThinPrep Imager and conventional cytology for 20 cytologists who read both types of smears. On average they read 13.3 ThinPrep imager slides per hour and 6.1 conventional cytology slides per hour. This study provides substantive evidence that cervical cytology read using the TPI detects more histological high grade squamous disease than conventional cytology, with significantly reduced reading times compared to conventional cytology. The ability of TP to detect glandular lesions remains controversial [5, 6]. Although the numbers of glandular lesions is small in this study, their detection rate was the same. Cervical cytology in the future Although there are numerous other studies showing the efficacy of the ThinPrep imager [7] this large study should provide high quality evidence to allow consideration of its introduction into cervical screening programmes. The cost of automation is not inconsiderable but the imager will provide both labour savings and improved accuracy potentially leading to increased screening intervals. These will offer cost savings which may offset the increased costs of automation. Although this may not be as significant where direct to vial liquid based cytology is already in use, in countries such as Australia, where the technique is split sample this should have considerable impact on staffing levels and efficiency of laboratories. The data from the Australian TPI study will be used for a formal cost benefit analysis. The preparation of a sample in liquid fixative, potentially also allows its use for different testing techniques. Testing for human papillomavirus is already commonplace but in the future this sample may also be useful for biological markers. Testing for markers of disease such as P16 may also be performed using a liquid based sample with image analysis. It is also timely that cervical cytology becomes more efficient and more accurate with the development of anti human papillomavirus vaccines. One potential impact of even partial population vaccination against HPV is an overall reduction in the number of abnormalities presented to the scientist even if there is no reduction in cervical cytology samples. The use therefore of a concentrated sample and pre-screening technique such as is available with image analyzers will help in assessing cervical disease in this new environment. Conclusion All these changes do not herald the end of cervical cytology. Rather, one technique will be replaced by another more efficient, more accurate and potentially more useful technology. This is indeed timely given the rapid changes in molecular techniques and the increasing need for more specific investigations and will indeed bring cervical cytology into the new millennium. References Nanda K, McCrory DC, Myers E R, et al. Accuracy of the Papanicolaou test in screening for and follow-up of cervical cytologic abnormalities: a systemic review. A.. Intern Med 2000; 132:810-819 Farnsworth A, Mitchell H S. Prevention of cervical cancer. MJA 2003; 178:653-654 Medical Services Advisory Committee of Australia Liquid based cytology for Cervical screening MSA ref 12a Assessment repot August 2002; http://www.msac.gov.au/pdfs/reports/msacref12a.pdf Davey et al. Effect of study design and quality on unsatisfactory rates cytology classifications and accuracy in liquid based versus conventional cervical cytology: a systematic review; Lancet 2006;367:122-32 Roberts et al. Comparison of ThinPrep and Pap Smear in relation to prediction of adenocarcinoma in situ; ACTA Cytologica 1999 Jan-Feb; 43(1): 74-80 Bai H, Sung C J, Steinhoff M M. ThinPrep Pap test promotes detection of glandular lesions of the Endocervix. Diagn Cytopathol 2000 July; 23(1):19-22 Biscotti et al. Assisted primary screening using the automated ThinPrep imaging system. Am J Clin Pathol. 2005;123:281-287