David C. Wilbur, MD
William D. Tench, MD
On Dec. 3, 2008, the Food and Drug Administration granted approval for the BD FocalPoint GS Imaging System (BD Diagnostics—TriPath, Burlington, NC). The approval extends the capabilities of the BD FocalPoint Slide Profiler1 to include primary guided screening of cervical cytology specimens.
The BD FocalPoint GS Imaging System is designed to process BD SurePath Pap test slides, initially providing slide ranking and adequacy information to cytotechnologists, followed by relocation on a microscope with automated stage of the 10 microscopic fields of view having the highest probability of containing abnormality. An additional field of view, or FOV, is presented initially to allow for location verification/calibration before the FOV screening process is begun. Following review of the FOVs, if no abnormality or adequacy abnormalities or issues are identified, the case may be signed out as negative for intraepithelial lesion or malignancy, or NILM, if not subject to quality control rescreening. Slides in which adequacy issues or potential abnormal cells or patterns are identified are subjected to a full manual review. The Directed QC Technology selects 15 percent of the highest ranked NILM slides for QC review. Such a nonrandom rescreening process has been shown to be more effective in identifying false-negative cases than is the current standard CLIA ’88-mandated random QC process.2,3 Slides found to be potentially abnormal after primary guided screening are reviewed by pathologists as per the current standard of practice.
The FDA approval was received after results were submitted from a clinical trial of more than 12,000 clinical specimens processed at four clinical sites. The trial results were presented at the 2009 annual meeting of the U.S. and Canadian Academy of Pathology in Boston.4 The clinical study compared the results of a routine manual screening and routine quality control rescreening process with the primary guided screening and directed quality control rescreening process explained in the prior paragraph. Slides were entered into the protocol sequentially from the recent archives of each of the clinical sites, including a small number of “seeded” slides (2.9 percent of the total) that were newly made alongside the daily work from prior abnormal BD SurePath samples. The seeded slides were intended to slightly increase the abnormal prevalence, thus raising the statistical power of the results. The trial was performed in a “masked” fashion, meaning no observer at any point in the protocol had prior knowledge of any other result obtained. All abnormal and unsatisfactory slides, and a subsample of NILM slides, were sent for external adjudication (as is the routine for FDA two-armed clinical trials). The adjudication process provided a reference diagnosis for each slide. The results for each slide in the two study arms were then compared with the reference diagnosis to determine the sensitivity, specificity, and predictive values for abnormal/unsatisfactory detection in each arm. In addition, timing studies compared the screening productivity in each arm of the study.
The trial data show that the GS Imaging System was more robust in the correct classification of LSIL+ (LSIL, HSIL, AIS, and invasive carcinoma) and HSIL+ (HSIL, AIS, and invasive carcinoma), by +9.8 percent and +19.6 percent (both P<.0001), respectively. There was a slight tendency to call more cases LSIL+ and HSIL+ in the GS arm with a concomitant decrease in specificity (-2.6 percent [P<.0001] and -1.9 percent [P=.0032], respectively). However, the negative predictive values for not-HSIL+ and not-LSIL+ were higher in the GS arm than in the manual screening arm, indicating that the overall efficiency of the screening procedure was improved in the GS arm. There was no statistical difference between the two study arms regarding sensitivity and specificity for the detection of ASC-US+ (all abnormal slides), indicating there is no increase in equivocal cases with the use of the GS Imaging System.
The GS system performance in the invasive carcinoma category is interesting. Of 49 total cases of cancer in the trial, the GS arm correctly identified 34 (69 percent) as cancer, while the manual screening arm correctly classified 22 (45 percent). Each arm had only one case of truly missed cancer (other cases in each arm were identified as HSIL, AIS, ASC-H, ASC-US, or AGC), but the GS arm more consistently arrived at the reference diagnosis, indicating that targeting high probability fields of view on each slide may allow cytotechnologists to identify the best examples of diagnostic cells. Slide ranking data for the GS system was robust, with 87 percent of HSIL+ cases ranked in the top 20 percent of slides and 95 percent in the top 40 percent of slides.
Adequacy rates were not statistically different between the two study arms (0.16 percent and 0.20 percent in the manual and GS arms, respectively). However, the GS arm result agreed with the reference diagnosis of unsatisfactory 19 percent more often than did the manual screening arm.
Timing data showed that the GS arm was more efficient, with higher overall throughput of slides. The sites varied in the difference between manual and GS arms slides screened per hour, ranging from a 10 percent to a 57 percent increase. Based on the results obtained, the FDA granted a new workload limitation for slides processed on the GS system of 170 slides per 24-hour period.
Overall, the study concluded that the BD FocalPoint GS Imaging System was robust in terms of slide ranking, presentation of abnormal cells in fields of view to cytotechnologists, and adequacy determination. As a result of the overall process, slides were categorized more precisely into the reference diagnostic categories, with fewer false-negatives and without increases in the number of equivocal cases. The process was more efficient and will allow laboratories to process greater numbers of cervical cytology slides. Devices such as the GS system will be important in the post-HPV vaccine implementation era, when the prevalence of abnormal cases is expected to decline. With declining prevalence, the sensitivity of the screening process will decrease.5 Devices that concentrate abnormal cases by ranking, and abnormal cells by guided screening, would be expected to increase the “functional prevalence” of abnormality in the manually reviewed population. Therefore, the use of such devices is likely to prolong the utility of cervical cancer screening by the Papanicolaou method.
- Patten SF, Lee JSJ, Wilbur DC, et al. The AutoPap 300 QC System multicenter clinical trials for use in quality control rescreening of cervical smears. I: prospective intended use study. Cancer (Cancer Cytopathol). 1997;81:343–347.
- Patten SF, Lee JSJ, Wilbur DC, et al. The AutoPap 300 QC System multicenter clinical trials for use in quality control rescreening of cervical smears. II: prospective and archival sensitivity studies. Cancer (Cancer Cytopathol). 1997;81:337–342.
- Wilbur DC, Prey MU, Miller WM, et al. Detection of high grade squamous intraepithelial lesions and tumors using the AutoPap system: results of a primary screening clinical trial. Cancer (Cancer Cytopathol). 1999;87:354–358.
- Wilbur DC, Black-Schaffer WS, Luff RD, et al. Clinical trials of the FocalPoint GS System show significant improvements in sensitivity for the detection of squamous intraepithelial lesions when compared to manual screening. Mod Pathol. 2009;22(suppl 1):97A.
- Schiffman M. Integration of human papillomavirus vaccination, cytology, and human papillomavirus testing. Cancer. 2007;111:145–153.
Dr. Wilbur, chair of the CAP Cytopathology Committee, is director of cytopathology, Massachusetts General Hospital, Boston. Dr. Tench, a member of the Cytopathology Committee, is with Palomar Medical Center, Escondido, Calif.
Dr. Wilbur was a member of the BD Speakers Bureau in 2008. Dr. Tench served as the BD medical director for the clinical trial.