A study of cellular counting to determine minimum thresholds for adequacy for liquid-based cervical cytology using a survey and counting protocol

• Published in: Health technology assessment (Winchester, England) England), 2015-03, Vol.19 (22), p.i-64
• Main Authors: Kitchener, Henry C, Gittins, Matthew, Desai, Mina, Smith, John H F, Cook, Gary, Roberts, Chris, Turnbull, Lesley
• Format: Article
• Language: English
• Subjects: Cell Count - methods; Cell Count - standards; cellular counting; Cervical Intraepithelial Neoplasia - diagnosis; counting protocol; Cytological Techniques - methods; Cytological Techniques - standards; Female; Humans; liquid-based cytology; minimum acceptable cell count; Reference Standards; Reproducibility of Results; surepath; Surveys and Questionnaires; Technology Assessment, Biomedical; thinprep; United Kingdom; Uterine Cervical Neoplasms - diagnosis; Vaginal Smears - methods; Vaginal Smears - standards
• ISSN: 1366-5278; 2046-4924
• DOI: 10.3310/hta19220

Liquid-based cytology (LBC) for cervical screening would benefit from laboratory practice guidelines that define specimen adequacy for reporting of slides. The evidence base required to define cell adequacy should incorporate both ThinPrep™ (TP; Hologic, Inc., Bedford, MA, USA) and SurePath™ (SP; BD Diagnostics, Burlington, NC, USA), the two LBC systems used in the UK cervical screening programmes. The objectives of this study were to determine (1) current practice for reporting LBC in England, Wales and Scotland, (2) a reproducible method for cell counting, (3) the cellularity of slides classified as inadequate, negative or abnormal and (4) the impact of varying cellularity on the likelihood of detecting cytological abnormalities. The study involved four separate arms to pursue each of the four objectives. (1) A questionnaire survey of laboratories was conducted. (2) A standard counting protocol was developed and used by three experienced cytopathologists to determine a reliable and reproducible cell counting method. (3) Slide sets which included a range of cytological abnormalities were each sent to three laboratories for cell counting to study the correlation between cell counts and reported cytological outcomes. (4) Dilution of LBC samples by fluid only (unmixed) or by dilution with a sample containing normal cells (mixed) was performed to study the impact on reporting of reducing either the total cell count or the relative proportion of abnormal to normal cells. The study was conducted within the cervical screening programmes in England, Wales and Scotland, using routinely obtained cervical screening samples, and in 56 participating NHS cervical cytology laboratories. The study involved only routinely obtained cervical screening samples. There was no clinical intervention. The main outcome measures were (1) reliability of counting method, (2) correlation of reported cytology grades with cellularity and (3) levels of detection of abnormal cells in progressively diluted cervical samples. Laboratory practice varied in terms of threshold of cellular adequacy and of morphological markers of adequacy. While SP laboratories generally used a minimum acceptable cell count (MACC) of 15,000, the MACC employed by TP laboratories varied between 5000 and 15,000. The cell counting study showed that a standard protocol achieved moderate to strong inter-rater reproducibility. Analysis of slide reporting from laboratories revealed that a large proportion of the samples