1. Results of the HTA Adequacy Study JHF Smith Royal Hallamshire Hospital, Sheffield BAC Scientific Meeting 2013

2. Previous publications on cervical LBC adequacy (1) Most cases with HSIL have more than one abnormal cell among 4000 normal cells. Geyer 2000 The Bethesda system requires a minimum of 5,000 well preserved and well visualised squamous cells for a liquid based preparation to be deemed ‘satisfactory’. Bethesda 2001

3. Previous publications on cervical LBC adequacy (2) One study reported a higher detection rate of high grade lesions when cellularity of SurePath [SP] LBC preparations exceeded 20,000 cells. Bolick 2002 SP preparations with < 10000 cells should be designated less than adequate. Bishop 2002 Clear demarcation in sensitivity between SP specimens with 5000 cells. Studeman 2003

4. Previous publications on cervical LBC adequacy (3) Minimum cellularity ThinPrep [TP] samples cannot be defined in practical terms, but value somewhere between 5000 – 10000 cells would give acceptable inadequacy rates whilst maximising detection of dyskaryosis. McQueen & Duvall 2006 TP slides reported as unsatisfactory or SBLB shown to contain < 20000 cells. Siebers 2013

5. LBC Adequacy in UK Most UK laboratories have adopted an interim figure of up to15,000 cells, based on manufacturers’ guidance and the LBC pilot experience

6. Adequacy of LBC samples An adequate liquid-based sample is defined as one that contains the minimum level of squamous epithelial cellularity necessary to ensure a squamous abnormality detection rate equivalent to that offered by conventional smears. Achievable standards, Benchmarks for Reporting, and Criteria for Evaluating Cervical Cytopathology. 3 rd edn. NHSCSP 2012.

7. A multistranded study to determine the minimum cellularity required for the reliable assessment of Liquid Based Cervical (LBC) cytology samples A multistranded study to determine the minimum cellularity required for the reliable assessment of Liquid Based Cervical (LBC) cytology samples Survey current practice in UK laboratories using LBC to establish the cellularity of a large cohort of inadequate, negative and abnormal slides Evaluate the ability of cytotechnologists in a large number of laboratories to detect abnormalities of differing type and relative abundance by preparing sets of slides which vary in their total cellularity and in the total number, type and relative proportion of abnormal cells http://www.hta.ac.uk/project/1600.asp

8. Objectives 1: Survey of Cellularity To assess current standards and practice for the reporting of LBC preparations across England, Scotland and Wales To determine the cellularity of samples deemed inadequate, negative or abnormal by a range of laboratories across the country. To determine the cellularity of samples deemed negative, HPV+ in the ARTISTIC/MCM trial.

9. Objectives 2: Ability to Detect Abnormality To assess the impact of varying the overall cellularity; relative proportion of abnormal cells and the type and presentation of dyskaryotic cells on their likelihood of detection. To determine the threshold of cellularity of LBC preparations which allow the majority of samples containing abnormal cells to be detected by routine screening.

10. Survey of Cellularity Slide Survey 56 (28 TP & 28 SP) laboratories submitted 20 consecutive inadequate, low grade and high grade dyskaryosis plus 50 consecutive negative LBC cases. 56 (28 TP & 28 SP) laboratories submitted 20 consecutive inadequate, low grade and high grade dyskaryosis plus 50 consecutive negative LBC cases. Slides relabelled with an anonymised code, placed in transport boxes in a pre-determined randomised order, so boxes contained a mixture of different diagnostic groupings from different originating laboratories. Slides relabelled with an anonymised code, placed in transport boxes in a pre-determined randomised order, so boxes contained a mixture of different diagnostic groupings from different originating laboratories. Boxes were sent in sets of 5 to each of the participating laboratories via courier. Boxes were sent in sets of 5 to each of the participating laboratories via courier.

11. Survey of Cellularity Cell Counting Nominated primary screening staff Nominated primary screening staff All slides assessed for presence of TZ indicators and had a formal cell count by protocols agreed with the manufacturers All slides assessed for presence of TZ indicators and had a formal cell count by protocols agreed with the manufacturers Similar proportions of slides from each laboratory and of each cytological type sent to each laboratory Similar proportions of slides from each laboratory and of each cytological type sent to each laboratory

12. AA – 4 - 15 Cell Count HTA LBC Adequacy Study ThinPrep MM – 3 - 15 Cell Count HTA LBC Adequacy Study SurePath Total cell count = mean cell count of 10 counts x area of cell deposit x area of ocular Adjacent SP counts starting at edge of cell deposit Alternate TP counts starting at edge of cell deposit, weighted to allow for cell distribution

13. ThinprepField12345678910 High Grade N=572 Mean37.536.936.534.233.132.330.329.328.125.9 SD31.929.730.634.027.829.625.525.624.521.5 Median3130 27.52726 242321 Min0000000000 Max267208220450200400203200 121 Inad N=623 Mean11.08.79.08.17.37.68.37.27.47.0 SD17.716.019.116.914.714.831.917.023.917.6 Median4333232222 Min0000000000 Max199127207158125165700250500250 Mild Dysk N=562 Mean44.446.443.440.536.435.834.834.333.032.4 SD35.737.234.136.629.029.131.329.628.740.8 Median37383633312927 2625 Min0000000000 Max230320249485232200350226164750 Neg N=1418 Mean43.841.941.038.937.135.834.233.531.931.1 SD44.634.835.135.040.034.137.534.131.832.3 Median33 3230 29262524 Min0000000000 Max820300341350999482800597504534 Total N=3175 Mean36.335.334.332.330.429.628.527.726.625.7 SD39.034.334.034.634.331.534.930.930.231.5 Median26272524232220 1917 Min0000000000 Max820320341485999482800597504750 Weights* F20231.43280.53248.47216.41184.35152.29120.2388.1756.1125.05 F22209.43251.51219.45187.38155.32123.2591.1959.1227.062.00 Summary Statistics for each field (Thinprep Summary Statistics for each field (Thinprep)

14. SurepathField12345678910 High Grade N=559 Mean79.389.289.794.391.995.595.796.294.594.1 SD56.262.946.963.252.052.467.265.854.253.0 Median71828584 8987 8588 Min3458173240 Max999 369999584569999 750775 Inad N=587 Mean21.722.023.023.122.024.926.025.824.625.0 SD30.827.933.229.125.139.750.237.235.949.2 Median15 1615 16 1516 Min0000000000 Max506318550310235559999350500999 Mild Dysk N=561 Mean76.084.793.188.692.3 92.491.191.693.1 SD38.240.969.244.450.245.647.643.943.245.1 Median70818485868786 87 Min0050000010 Max361376955610560318404310339380 Neg N=1402 Mean84.592.596.599.3100.0101.2100.6104.1103.4103.9 SD60.960.666.565.160.566.464.572.472.275.8 Median71.5808286.5878988 89 Min0206527230 Max850650900999508999800999 Total N=3109 Mean70.277.280.882.182.484.1 85.684.885.3 SD56.959.465.263.059.763.166.167.865.769.2 Median62707475 78 7978 Min0000000000 Max999 955999584999 Weights* F16125.76117.78109.79101.8193.82100.81 F20100.3092.2784.2576.2368.2051.15 F2290.1483.1375.1267.1159.0937.06 Summary Statistics for each field (Surepath)

15. Weighted v Unweighted Specimen Cellularity (TP & SP)

16. Weighted vs Un-weighted Specimen Cellularity split by Slide Diagnosis (TP)

17. Weighted vs Un-weighted Specimen Cellularity split by Slide Diagnosis (SP)

18. Relative Frequency Histograms of Specimen Cellularity Score (TP & SP)

19. Relative Frequency Histograms split by Slide Diagnosis (TP)

20. Relative Frequency Histograms split by Slide Diagnosis (SP)

21. Ability to Detect Abnormality 180 Routine TP & SP samples LSIL & HSIL Scanty, numerous, pale & HCCG Serial dilutions 5-10K, 10-15K, 15-20K, 20-25K, 25-30K, 35-45K, 45-55K, 55K+ Mixed dilution with known negative cases <25, 25-49, 50-99, 100-149, 150-199, 200-399, 400-799, 800-1600, 1600+ 50%

22. Slide evaluation Total cell count and total abnormal cell count determined for each slide by two members of a four member expert cytology group HSIL or LSIL observed in all but one SP and all but two TP slides Slides randomised into batches of 308 (88 serial dilution, 88 mixed dilution, 66 negative, 66 inadequate) Each batch circulated to 3 of 24 laboratories using either SP or TP giving three independent assessments for each slide

23. Percentage of assessments in each cytological category by cellularity and LBC system SurePathThinPrep Estimated cellularity <15000>=15000<5000>=5000 HSIL46.8 82.1 41.3 89.1 50.0 86.0 49.7 93.3 LSIL35.347.836.042.6 Inadequate14.41.010.81.6 Negative3.59.83.26.1 No of slides 6763462642

24. Percentage of assessments in each cytological category by cellularity and LBC system SurePathThinPrep Estimated cellularity <15000>=15000<5000>=5000 HSIL46.8 82.1 41.3 89.1 50.0 86.0 49.7 93.3 LSIL35.347.836.042.6 Inadequate14.41.010.81.6 Negative3.59.83.26.1 No of slides 6763462642 OR=0.56 95% CI 0.34,0.94, p=0.019 OR=0.51 95% CI 0.30,0.88, p=0.016 p<0.001 p<0.017p<0.195

25. Conclusions Low cellularity reduces the likelihood that dyskaryotic cells will be detected by about 7% in both the SP and TP LBC systems. The Bethesda system requirement for a minimum of 5000 cells is appropriate only for the ThinPrep system. A minimum cellularity of 15,000 cells is appropriate for the SP system.

26. Acknowledgement This study was funded by the National Institute for Health Research Health Technology Assessment Programme. http://www.hta.ac.uk/project/1600.asp

27. Acknowledgement G Cook G Cook MS Desai M Gittins HC Kitchener C Roberts P Sasieni LS Turnbull

28. Any Questions ?