SEARCH

SEARCH BY CITATION

Keywords:

  • cervical cytology;
  • liquid-based Papanicolaou tests;
  • cervical intraepithelial neoplasia;
  • carcinoma of the uterine cervix

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

BACKGROUND

This study was designed to optimize a liquid-based Papanicolaou (Pap) test by using common cytopathology laboratory equipment and resulted in an inexpensive test that was equivalent at least diagnostically to the conventional Papanicolaou (Pap) smear.

METHODS

Adult women (n = 482) were consented, enrolled, and included in this Institutional Review Board-approved study. After conventional Pap smear slides were obtained, clinicians placed the collection device with residual cells from the uterine cervix in a preservative fluid. In the cytopathology laboratory, a conventional centrifuge device was used to deposit the cells from the liquid onto a glass slide.

RESULTS

Among the conventional Pap smears, 43 were categorized as low-grade squamous intraepithelial lesions (LSIL), and 30 were categorized as high-grade squamous intraepithelial lesions or greater (HSIL+). Among the PapSpin samples, 49 were categorized as LSIL and 24 were categorized as HSIL+. Biopsy confirmation was obtained in 124 patients. There were 23 women diagnosed with LSIL and 27 women diagnosed with HSIL+. Diagnostic agreement between cytologic samples and biopsies is as follows: for conventional Pap smears, there was agreement on 11 of 23 LSIL diagnoses and on 15 of 27 HSIL+ diagnoses; for PapSpin samples, there was agreement on 11 of 23 LSIL diagnoses and on 14 of 27 HSIL+ diagnoses. Exact agreement was achieved between PapSpin and conventional smears in 404 patients (84%). Quality indictors were better in the PapSpin group, except for inadequate endocervical component, which was greater in the PapSpin samples, a difference that was explained by the split-sample study design, which favored the conventional smear.

CONCLUSIONS

The current results indicated that PapSpin is a legitimate, inexpensive alternative to the conventional Pap smear for the detection of cervical intraepithelial neoplasia, resulting in better preservation and improved cell visualization. In addition, the liquid residual allows for reflex human papillomavirus-DNA or polymerase chain reaction testing. Cancer (Cancer Cytopathol) 2006. © 2006 American Cancer Society.

Since the U.S. Food and Drug Administration (FDA) approved the ThinPrep™ test (Cytyc Corporation, Boxborough, MA) in 1996, liquid-based Papanicolaou (Pap) (LBP) tests are fast becoming the industry standard. Only 1 other company, TriPath Imaging (Burlington, NC), has received FDA approval for their gynecologic cytology processing system, SurePath™. Both products perform well, reducing obscuring elements, such as excess blood and inflammation, and revealing diagnostic cells, thereby improving the overall performance of the Pap test compared with the conventional smear (CS). 1–3 However, the cost per liquid-based test is significantly greater than for the CS, making the improved method potentially inaccessible to those underserved women who would benefit most.4

For more than 30 years, the Shandon Cytospin® (ThermoElectron, Pittsburgh, PA), has been a permanent fixture in most cytopathology laboratories. Body fluids, including serous effusions, cerebrospinal fluid, and urine, have been processed successfully by using this inexpensive instrument to produce a 28 mm2 circle of cells on glass slides for microscopy. Recently, the Megafunnel™ was created (ThermoElectron) to capture more cells on a larger (294 mm2) rectangular area (Fig. 1). Of all body fluids, urines benefited most from this new device. Based on the physical similarity between urothelial cells in urine and the cells found in uterine cervical samples, the PapSpin™ (PS) process was developed (ThermoElectron).

thumbnail image

Figure 1. The Megafunnel (right) was developed to capture more cells in a larger rectangular area (294 mm2). The Shandon Cytospin (left) produces a 113.4-mm2 circle of cells.

Download figure to PowerPoint

The main objective of this study was optimization of the procedure recently described in 2 separate studies. 4, 5 The objective of the investigators and ThermoElectron was to develop a procedure that used common laboratory equipment, that produced a cytologic sample of cells from the uterine cervix that improved the final interpretation but depended on established and familiar morphologic criteria, and that was considerably less expensive than the 2 FDA-approved LBP tests. Finally, the sample would be an appropriate platform for reflex testing for the presence of human papillomavirus (HPV) and other sexually transmitted diseases; however, reflex testing was not included in the current study.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

To achieve the study objectives, first, the slide preparation procedure was optimized. Following the manufacturer's instructions, the liquid suspension of cells was processed using the Shandon Cytospin apparatus and the Megafunnel. Variation in cellularity immediately was apparent, occasionally resulting in obscuring layers of cells; other times, there were too few cells for visual evaluation. Working with discarded cellular samples, it was determined that the optimal amount of cell suspension was a volume of 2 mL to produce a satisfactory sample without detracting factors. Once the slide preparations were deemed visually acceptable to the cytotechnologists and cytopathologists, the comparability phase of the study was initiated.

With approval from the Western Institutional Review Board, 538 patients were enrolled from gynecology clinics at Johns Hopkins Bayview Medical Center after informed consent was obtained. Cervical cellular samples were taken from the cervical os by using the Papette™ (Wallach, Orange, CT) brush. A CS was made for routine diagnosis. The brush head with residual material was ejected into a vial of PapSpin Collection Fluid™ preservative. Once it was in the laboratory, the specimen was examined to ensure that it was acceptable macroscopically (i.e., the collection device both was present and was submersed completely in preservative). The specimen vial then was agitated by a vortex mixer for approximately 10 seconds to remove additional cells from the brush head. Immediately, 2 mL of cellular suspension were transferred to the Megafunnel chamber, followed by an equal amount of fresh preservative. The chamber was capped; the hub was sealed with the protective lid and loaded into a Shandon Cytospin 3® cytocentrifuge. Centrifugation at 1250 revolutions per minute for 5 minutes at high acceleration deposited the cellular specimen onto a coated Megafunnel Cytoslide® in an area defined by an etched rectangle (294 mm2). The slide was removed from the Megafunnel assembly, allowed to air dry briefly, and immersed in 95% ethanol for at least 10 minutes. The Megafunnel was discarded. Both PS and CS slides were stained by using the traditional Pap method with Gill hematoxylin, and each slide was protected by a glass coverslip. Slides were screened by experienced cytotechnologists without knowledge of the interpretation of the corresponding pair. For patient safety reasons, pairs periodically were matched to detect significant discrepancies between the 2 samples. Patient reports were amended when the pathologist (P.M.A. or D.L.R.) determined that patient management might possibly be altered with the new information.

Statistical Analysis

For comparisons of diagnostic agreement, pair-matched analyses were used. Percent agreement of the PS and CS was calculated for all 482 patients who were included in the study and excluding patients who had negative results on both tests. The percent agreement of PS and CS with the biopsy diagnosis was calculated for the 124 patients for whom biopsy material was available. The percent of correctly classified diagnoses and the predictive value of the PS and CS diagnostic tests were calculated for biopsy diagnoses of negative/reactive, atypical squamous cells of undetermined significance (ASC-US), low-grade squamous intraepithelial lesion (LSIL), and high-grade squamous intraepithelial lesion (HSIL)/malignant. For the diagnoses of ASC-US, LSIL, and HSIL, these represent the sensitivity and the positive predictive value; for a diagnosis of negative/reactive, these values represent the specificity of the screening tests. The results were compared y using a chi-square test of proportions. For the comparison of quality indicators, proportions of each indicator were calculated for each group, and the group proportions were compared by using chi-square tests. Analyses were performed using SAS software (version 8; SAS Institute Inc., Cary, NC).

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Of the 538 patients enrolled, samples from 5 patients were rejected because of incomplete identification, and 1 patient had only a conventional Pap smear collected. Fifty samples were excluded because they were fixed in a developmental preservative that was introduced toward the end of patient enrollment and were identified as unsatisfactory by the cytologists involved in this study. This left 482 patients to form the basis for the current data analysis. Distribution of diagnostic categories, using The Bethesda Terminology, among the CS slides were as follows: 296 within normal limits (WNL), 107 ASC-US, 4 atypical glandular cells of undetermined significance (AGUS), 43 LSIL, 28 HSIL, 2 cancer cases, and 2 unsatisfactory cases. The PS results were similar: 327 WNL, 80 ASC-US, 1 AGUS, 49 LSIL, 21 HSIL, 2 cancer cases, and 1 unsatisfactory case. The percent agreement for the 2 preparations is shown in Table 1. The PS performance approximates the 2 FDA-approved methods (ie, indeterminate results [ASC-US/AGUS] are decreased, and LSIL numbers are increased). The ASC-squamous intraepithelial lesion (SIL) ratio for the PS method was lower (1.1) than for the CS method (1.5). Four reports (0.8% of total patients) were amended because cells that most likely reflected SILs were detected on PS samples but not on CS samples. Quality indicators, which previously were categorized as “satisfactory but limited by” (Table 2) differed significantly. A total of 449 PS samples and 331 CS samples were considered satisfactory (P <.0001); an endocervical component was absent in 144 of 481 PS samples (30%) and in 43 of 481 CS samples (9%) (P <.0001), a finding that was explained by the split-sample study design that favored the CS method. However, the PS samples were much improved by a reduction in obscured inflammation (17 of 482 PS samples [4%] vs. 70 of 482 CS samples [15%]; P <.0001) and fewer thick cellular clumps (10 PS samples vs. 20 CS samples; P = .06) (Fig. 2A,B ). Poor preservation of cellular material was present in 4 PS samples and in 55 CS samples. Obscuring blood was a problem in 26 CS samples but was nonexistent in the PS samples.

Table 1. Agreement of Diagnoses for the PapSpin and Conventional Methods*
PapSpin DiagnosisConventional DiagnosisPapSpin Totals
NegativeASC-USAGUSLSILHSIL/MalignantUnsatisfactory
  • ASC-US: atypical squamous cells of undetermined significance; AGUS: atypical glandular cells of undetermined significance; LSIL: low-grade squamous intraepithelial lesion; HSIL: high-grade intraepithelial lesion.

  • *

    The percent agreement for all diagnoses was 83.8%. Excluding the 286 negative/reactive diagnoses in agreement, the percent agreement was 60.2%.

Negative286363101327
ASC-US858086080
AGUS0010001
LSIL2130340049
HSIL/malignant000024024
Unsatisfactory0000011
Conventional totals296107443302482
Table 2. Comparison of Quality Indicators
CategoryPapSpin (% of Total)Conventional Smear (% of Total)P Value
  1. EC indicates endocervical; NS, not significant.

Satisfactory449 (93)331 (69)<.0001
No EC component144 (30)43 (9)<.0001
Inflammation17 (4)70 (15)<.0001
Poor preservation5 (1)55 (11)<.0001
Thick smear10 (2)20 (4).06
Blood0 (0)26 (5)<.0001
Scant cellularity1 (0.2)1 (0.2)NS
Unsatisfactory1 (0.2)2 (0.4)NS
thumbnail image

Figure 2. (A) This conventional smear specimen demonstrates marked cellular obscuring by inflammation. (B) This split-sample PapSpin specimen demonstrates reduced cellular obscuring by neutrophils and a delicate cluster of endometrial cells. Papanicolaou [Pap] stain; original magnification, × 200 (A, B).

Download figure to PowerPoint

Follow-up biopsies from 6 months to 18 months after the Pap smear were available for cytohistologic correlation in 124 patients (26%). Most tissue studies were triggered initially by the CS; however, after the 4 revised reports, 4 more women were examined by colposcopy, and 2 biopsies were obtained (ASC-US PS result, negative biopsy; ASC-cannot rule out HSIL [ASC-H] PS result, biopsy interpreted as atypia short of dysplasia). The proportions of correctly classified diagnoses and the predictive value of the PS and CS methods relative to the biopsy diagnoses are reported and compared in Table 3. There were no significant differences noted between the 2 methods, although there was a suggestion (P <.07) that the PS method correctly classified a greater proportion of negative diagnoses. Although the cytologic diagnoses in the 2 groups were not identical, the numbers of lesions identified on biopsy were similar (PS samples, 49%; CS samples, 44%), thereby fulfilling the purpose of the screening test with the 2 methods (Table 4). Similar to other studies that used LBP testing, the relatively poor cytohistologic correlation of the LSIL category can be explained by the rendered diagnosis based on only a few characteristic koilocytes, which are the hallmark of HPV infection (Fig. 3). These cells are visible clearly on the LBP, but they may originate in a very early LSIL lesion, which is less obvious on colposcopic examination and is defined poorly on histologic sections. 2, 6

Table 3. Correctly Classified Diagnoses and Predictive Value of the PapSpin and Conventional Smear Methods for 124 Patients with Biopsies
Biopsy DiagnosisCorrectly Classified (%)*Predictive Value (%)
PapSpinConventionalP ValuePapSpinConventionalP Value
  • ASC-US: atypical squamous cells of undetermined significance; LSIL: low-grade squamous intraepithelial lesion; HSIL: high-grade intraepithelial lesion.

  • *

    Correctly classified percentages were calculated as the percent of patients who had a biopsy-confirmed diagnosis that was classified correctly for that diagnosis on the screening test.

  • The predictive value was calculated as the percent of patients whose diagnosis on screening was confirmed on biopsy.

Negative/reactive58.341.7.0688083.77
ASC-US9.136.4.13411.64
LSIL47.847.8.104146.78
HSIL/malignant51.955.6.787868.72
Table 4. Surgical Pathology with Both PapSpin and Conventional Smear Correlation Diagnoses*
Cytologic DiagnosisBiopsy DiagnosisTotal Cytologies per Diagnosis
Negative/ReactiveASC-USAGUSLSILASC-HHSIL/MalignantUnsatisfactory
PSCPSCPSCPSCPSCPSCPSCPSC
  • ASC-US: atypical squamous cells of undetermined significance; AGUS: atypical glandular cells of undetermined significance; LSIL: low-grade squamous intraepithelial lesion; ASC-H: atypical squamous cells, cannot rule out high-grade squamous intraepithelial lesion; HSIL: high-grade intraepithelial lesion; PS: PapSpin; C: conventional smear.

  • *

    PapSpin study Patients 1 through 488 (includes follow-up biopsy).

  • In total, 124 patients underwent biopsies.

Negative/reactive352532  212121  4430
ASC-US142214  671213  2338
AGUS12            12
LSIL8532  1111  56  2724
ASC-H1332  21  52  118
HSIL/malignant1311  23  1415  1822
Unsatisfactory              00
Total biopsies per diagnosis60110233270124
thumbnail image

Figure 3. Characteristic koilocytes are present on this PapSpin specimen (Papanicolaou stain; original magnification, × 600).

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The Pap smear for detection of precursor lesions to cervical cancer generally is acclaimed as the most successful screening test for cancer detection. Epidemiologists consider it among the top 3 contributors to public health, along with the vaccines to eradicate polio and small pox. An effective screening test must be minimally invasive, inexpensive, and reliable. The LBP tests were a response to the need to improve the sensitivity of the test without sacrificing specificity, which always is a trade-off for any screening test. The subjectivity of Pap smear interpretation negatively influences these 2 parameters. The experience of the clinicians who conduct the test and those who examine the slides makes the test less than reliable on a single pass. Only the frequency of sampling (traditionally annually) has provided the test with any chance of success. Most lesions take from 8 years to 10 years from first HPV infection to a diagnosis of HSIL. Therefore, the natural history of cervical cancer enables the detection of most lesions at an early stage, even after 1 or 2 missed opportunities because of under interpreted Pap smears.

The current paradigm for patient management relies on an inexpensive test. With the introduction of the LBP tests, costs at least doubled, and new processing devices required the training of laboratory personnel and changes in laboratory space allocation. However, the decrease in indeterminate results (ASC-US/AGUS), increases in the detection of neoplasia, 1–3 the and ability to perform reflex HPV testing on the residual material6, 7 has convinced many laboratories to convert in part or completely to the LBP test.

However, cytopathologists and cytotechnologists encountered new cytomorphologic criteria, some daunting, depending on the preparation. This required days of training and from weeks to months of a potentially dangerous learning curve. The PS method meets the need for an inexpensive processor with familiar cellular criteria. Almost every cytopathology laboratory in North America has at least 1 Shandon Cytospin® machine. Some laboratories already are using the Megafunnel™ device for body cavity fluids. Little adjustment is needed to incorporate this procedure into the laboratory menu. Specimens are processed in batches of 12, taking only 15 minutes per batch before routine staining. Multiple batches may be processed if a laboratory has more than 1 machine. In a moderate volume laboratory (15-30,000 annual gynecologic samples), the PS is more appropriate than larger multisample processors and is more efficient than a single test device. According to Cytyc marketing materials, 25 ThinPrep samples can be processed in 1 hour using the T2000 instrument, which is the single-sample processing device. According to the experience reported by Weynand et al., 4 48 PS samples can be processed in 1 hour by using batch processing, as described earlier.

Cost comparison places the PS higher than the CS but still less expensive than the 2 FDA-approved procedures (usually >$8 per patient). Although the PS kit is not sold in the U.S., the current price in Europe is 4.5 Euros per kit (approximately U.S. $5.50). Each kit includes collection vial with fluid, collection brush, Megafunnel with cap, and slide. All components are disposable except for the slide.

Visual perception of the cellular material on the PS is comfortably traditional but much improved over the often obscured or air-dried CS. The layering of cells on the PS is not as thin as other LBP tests but is transparent enough to allow visibility of cells with minimal through focusing. Inflammatory exudate floats above the cells and is present for diagnosis but without interfering with cell visibility (Fig. 4). Whether such preparations are suitable for machine screening has yet to be determined. Infectious organisms are recognized easily (Fig. 5A,B ). The rare invasive cancer is obvious on the PS (Fig. 6A ). The counterpart CS is a mess of thick blood and entrapped tissue fragments that have no discernible cellular criteria (Fig. 6B).

thumbnail image

Figure 4. This PapSpin specimen exhibits the even distribution of inflammation and elevation of inflammatory cells above the focal plane of the epithelial cellular component, allowing clear observation of squamous cells (Papanicolaou stain; original magnification, × 200).

Download figure to PowerPoint

thumbnail image

Figure 5. (A) Trichomonas and (B) Candida are obvious on PapSpin specimens. Papanicolaou [Pap] stain; original magnification × 600 (A); × 200 (B).

Download figure to PowerPoint

thumbnail image

Figure 6. (A) Cytologic features of invasive squamous carcinoma are identified easily on a PapSpin preparation (B) The counterpart conventional smear exhibits rare malignant cellular fragments largely obscured by blood and compromised by air-drying Papanicolaou [Pap] stain; original magnification, × 400 (A, B).

Download figure to PowerPoint

Two larger studies of the effectiveness of the PS method describe experiences similar to ours. The initial feasibility study, which was published by Khalbuss et al., 5 enrolled 791 patients at high risk for uterine cervical neoplasia. The prevalence of abnormal cytologic findings was 27% for the CS method and 25% for the PS method (then called SpinThin). Of the 647 samples biopsied, 31% revealed cervical intraepithelial neoplasia. Those authors concluded that the PS method was comparable to the CS method. Weynand et al.4 examined 3000 cervical samples that were processed first for CS with the residual used for PS preparations. Epithelial cell abnormalities in each group were identical (3.8%). The quality of the PS method was superior to the CS method, except for the presence of endocervical component (5.3% absent from PS preparations; 2.2% absent from CS preparations), a reflection of the split-sample design of the study. PS improved the detection rate of HSIL lesions while decreasing the number of ASC-US diagnoses. The detection of LSIL lesions was diminished slightly because of the absence of diagnostic cells from the residual material used to prepare the PS. That study also examined the cost of the PS compared with the AutoCyte Prep and ThinPrep processes. The cost of disposables was identical to the AutoCyte, with ThinPrep slightly more expensive. The cost of the Shandon Cytospin instrument is magnitudes less expensive than competitor systems.

The objective of the developers of the PS process was to develop an inexpensive, liquid-based replacement for the conventional Pap smear. The specifications were as follows: 1) diagnostic accuracy better than the CS method; 2) cytologic criteria comparable to the standard; 3) instrumentation readily available in virtually every cytopathology laboratory in North America and, potentially, internationally; 4) minimal maintenance of the equipment that is performed easily by laboratory personnel; 5) disposable supplies that are significantly less expensive than the other LBP tests on the market; 6) detection of HPV is possible by testing the residual cell suspension using current methodology 8; and 7) in practice, the procedure is appropriate for a cytopathology laboratory that has a moderate volume.

Based on our early experience with the Shandon PS process and resultant slides, we believe that the company has met all of its objectives except for better diagnostic accuracy. Predictably, when direct-to-vial studies are permitted, endocervical components will increase, and representation of lesions will be greater than what we observed in the current study, which was performed on split samples. A prospective clinical trial will then be appropriate to demonstrate improved performance over the conventional Pap smear.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The authors thank the Johns Hopkins Bayview nurses, who added additional time to their heavy schedules to consent patients to participate in this study. The responsiveness of the ThermoElectron staff was impressive. They also thank Jeff Brant, Vice President of Sales and Marketing at ThermoElectron Company, for his perpetual availability.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  • 1
    Bishop JW, Bigner SH, Colgan TJ, et al. Multicenter masked evaluation of AutoCyte PREP thin layers with matched conventional smears, including initial biopsy results. Acta Cytol. 1998; 42: 189-197.
  • 2
    Limaye A, Connor A, Huang X, Luff R. Comparative analysis of conventional Papanicolaou tests and a fluid-based thin-layer method. Arch Pathol Lab Med. 2003; 127: 200-204.
  • 3
    Tench W. Preliminary assessment of the AutoCyte PREP Direct-to-vial performance. J Reprod Med. 2000; 45: 912-916.
  • 4
    Weynand B, Berliere M, Haumont E, et al. A new, liquid-based cytology technique. Acta Cytol. 2003; 47: 149-153.
  • 5
    Khalbuss WE, Rudomina D, Dauff ND, Chuang L. Melamed MR. SpinThin, a simple, inexpensive technique for preparation of thin-layer cervical cytology from liquid-based specimens. Cancer (Cancer Cytopathol). 2000; 90: 135-142.
  • 6
    Levi AW, Levi AW, Kelly DP, Rosenthal DL, Ronnett BM. Atypical squamous cells of undetermined significance (ASC-US) in liquid-based cytologic specimens: results of reflex HPV testing and histologic follow-up in routine practice with comparison of interpretive and probabilistic reporting methods. Cancer (Cancer Cytopathol). 2003; 99: 191-197.
  • 7
    Bolick DR, Bolick RE, Coates F, et al. Laboratory implementation of human papillomavirus testing. Arch Pathol Lab Med. 2003; 127: 984990.
  • 8
    Ramael M, Van Steelandt H. Detection and subtyping of HPV in PapSpin medium collected cervical cells: Free Communication (abstract FC1–7). Presented at the Fifth International Multidisciplinary Congress EUROGIN 2003 Program, Paris, France, April 13-16, 2003.