Location-guided screening of Pap smears: one lab’s experience
The Department of Pathology at Columbia University Medical Center, New York City, was the first commercial site to use Cytyc Corp.’s ThinPrep Imaging System. Now, five months later, two of its users tell us how the lab got started, what the biggest problem was, and how it works.
Diane Hamele-Bena MD
Ralph M. Richart MD
Teri Wood, CT(ASCP), MPH
The ThinPrep Imaging System is a computerized FDA-approved
slide-scanning system for use in the primary screening of ThinPrep gynecologic
Pap slides. The system combines computer-imaging technology with human microscopic
review and interpretation. It uses a proprietary stain that is essentially stoichiometric
for nuclear DNA content—the principal detection mechanism of the system.
In its pivotal trials, the imager significantly improved diagnostic sensitivity
over conventional manual review for atypical squamous cells of undetermined
significance, or ASC-US, and higher, and it showed an equivalent sensitivity
for low-grade squamous intraepithelial lesions and high-grade squamous intraepithelial
lesions. There was a significant improvement in specificity for HSIL. In addition,
it was reported that cytotechnologists could increase their throughput, with
sensitivity and specificity equal to or better than that achieved in manual
In the Department of Pathology at Columbia University Medical Center, a separate room was dedicated to the ThinPrep Imaging System, or TPIS. An automatic stainer, automatic coverslipper, ThinPrep T2 processors, drying oven, and short-term storage were all placed in a room measuring about 250 square feet situated across the hall from the cytotechnologists’ screening room. Although centralizing this equipment meant moving the processing room to a new location, the proximity of the processing room to the screening room had significant advantages, not only for stringing Imager-related communication cables but also for ease of workflow.
At the time the decision was made to install the TPIS, about 60 percent of the laboratory's gyn Paps were liquid-based; the remainder were conventional Pap smears. Working with the hospital administration, the laboratory was able to take advantage of the Imager installation to convert the entire institution’s gyn Paps to a single preparation type. Now, with rare exceptions, all of the gynecologic specimens received in cytology are taken with the liquid-based ThinPrep method, providing more uniformity in supplies, processing, and staining. A significant additional benefit is the ability to conduct reflex human papillomavirus DNA testing for ASC-US cases using the residual fluid from the ThinPrep samples on all accessioned cases.
When the Imager was first installed, duplicate samples were processed for validation purposes to familiarize the laboratory with the requisite procedures, organize the workflow, and ensure that the processing technicians and cytotechnologists were familiar with the operations before going live. A new automated coverslipper was purchased, as the coverslips must be aligned precisely and the amount of mounting medium must be consistent to prevent the device from rejecting the slides. As we gained more experience with label placement and after we started to use the automated coverslipper, the slide rejection rate decreased. The primary reasons for slide rejection now are the presence of excess blood, inflammation, scant cellularity, air bubbles, and thick preparations. Rejected slides are screened manually on non-Imager microscopes.
The processing of ThinPrep Pap cases for the Imager is the same as for non-Imager cases, but special slides with fiduciary marks must be used if the slides are to be imaged. These fiduciary marks are preprinted on the slides and are used as reference points by the Imager and the review microscope to ensure that the coordinates and Imager-identified field-of-view locations are identical on both instruments. Labels bearing bar codes must be attached to the slides with precision because the Imager will reject the slides if the bar-code labels spill over the edge of slides or over the coverslip. Careful attention to such details was required to achieve a low rejection rate, but once everyone understood what the requirements were and became accustomed to them, they became part of the routine laboratory procedure, and slide rejections fell to a low steady state level (about three percent in our experience).
Once the technical details were addressed, the biggest problem with the TPIS was that the stain was substantially darker than the laboratory's accustomed modified Papanicolaou stain. Although the dark hematoxylin made it easier to identify atypical cells, the stain intensity was so great that it obscured nuclear detail, particularly in high-grade cases and in glandular epithelial nuclei. In fact, the stain was so dark when screening of the TPIS-stained slides was first begun that the professional and technical staff were concerned that there would be a high rate of over-diagnosis and that the ASC-US and AGUS rates would rise to unacceptable levels. The laboratory staff worked with the Cytyc staff for several months to achieve a staining density that would allow the screeners to see nuclear detail but still retain stoichiometry. In the end, after much patient effort at both institutions, the stain was reformulated and refined, and a successful method was implemented as the routine laboratory stain for gynecologic cases. It took a little time to get used to a darker stain, but once the training was completed, the laboratory was left with a consistent, reproducible staining method. After having screened TPIS-stained specimens for several months, the cytotechnologists uniformly preferred this stain to the prior lighter stain with which they had been accustomed.
After the slides are processed, stained, coverslipped, and labeled, they must be dried thoroughly before being imaged. The slides are then placed in cassettes that hold 25 slides, and up to 10 racks are placed in the processor for automated screening. For ease of workflow, we process slides in the morning, allow them to dry in the afternoon, and load them in the Imager for overnight scanning. Slides processed in the afternoon can be loaded on the Imager the next morning. With this routine the lab can easily process about 50,000 slides per year without a weekend shift.
The processor images each slide using an optical cellular collection algorithm by which it measures the integrated optical density of the cell nuclei. The system is PC-based, uses Windows software, and provides a display that tracks the progress of the slides through the Imager slide-by-slide and rack-by-rack. From the approximately 120 fields of view (FOVs) present on a single slide, the Imager selects 22 FOVs for review by the cytotechnologist at the review microscope. It is important to note that every single slide will have 22 FOVs presented to the screener, even if it is completely negative. It is tempting, early in the learning curve, to assume that the FOVs will contain abnormal cells because they have been selected by the Imaging device, but, after a little experience, the cytotechnologists adapt to the fact that the Imager selects the fields but that it is their job to use conventional morphology to decide which, if any, cells in those selected fields are abnormal and which are within normal limits. The coordinates of the selected FOVs and other information for each slide are stored in the computer for later retrieval on the review microscope. The Imager can be interrupted at any time and individual slides or one or more racks of slides can be removed for the cytotechnologist's interpretation. This feature is useful in scheduling the workflow.
After the imaging is completed, the racks are presented to the cytotechnologists
who manually retrieve individual slides from the rack and place them on the
review microscope stage one at a time. When the slide is placed on the stage
of the review scope, the specimen is automatically identified when the unique
bar code is scanned, and the scope retrieves the data for that slide. The review
scope then automatically guides the cytotechnologist to each of the 22 selected
FOVs in geographic order (not hierarchical order, to ensure that the cytotechnologist
views all of the 22 selected FOVs). The cytotechnologist steps through the 22
FOVs using a control pod, switching objectives (10x and 40x) as needed and electronically
marking the relevant cells.
If the cytotechnologist interprets the cells in the 22 FOVs as being within normal limits, the case can be signed out. If, however, the cytotechnologist interprets any of the cells as abnormal, they engage the Autoscan Mode, whereby the entire slide is screened with user-defined preferences, including scanning speed and direction. During the Autoscan process, the cytotechnologist electronically marks cells of interest. After the entire slide has been screened, the cytotechnologist chooses the most significant areas for marking by an integrated marking pen system that physically places an “L” on the areas of the slide selected for the pathologist's review. Because the device does not always display the most abnormal cells in the FOVs, the cytotechnologists use the Autoscan Mode in all cases in which any cellular abnormalities are detected in the 22 FOVs. Following imaging and screening on the review microscope, the usual quality control procedures are performed using non-Imager microscopes.
The Imager has been a useful addition to the laboratory. It is an excellent blending of the best of computer and human attributes and allows the cytotechnologist to focus on interpretive, rather than cell locator, skills.
Although New York State has not yet changed the allowable cytotechnologist screening limits, it is obvious that using the Imager will permit each cytotechnologist to screen a greater number of slides than they do now with equal or greater sensitivity. Greater throughput will make it possible for laboratories to increase the number of cases they process or reduce the number of cytotechnologists for their caseload. This should ameliorate the cytotechnologist shortage that exists in some areas of the country and help make up for the decreasing number of schools of cytotechnology and their graduates. Location-driven screening may also lead to the further consolidation of smaller cytology laboratories into larger ones and to greater efficiency, cost-effectiveness, and laboratory quality.
Dr. Hamele-Bena is assistant professor of pathology and director of the cytology laboratory, Dr. Richart is vice chairman of anatomic pathology, and Teri Wood is manager of the cytology laboratory at Columbia University Medical Center, New York, NY.