Fax: (011) 852 28725197
Liquid-based cytology and conventional cervical smears
A comparison study in an Asian screening population
Article first published online: 7 OCT 2003
Copyright © 2003 American Cancer Society
Volume 99, Issue 6, pages 331–335, 25 December 2003
How to Cite
Cheung, A. N. Y., Szeto, E. F., Leung, B. S. Y., Khoo, U.-S. and Ng, A. W. Y. (2003), Liquid-based cytology and conventional cervical smears. Cancer, 99: 331–335. doi: 10.1002/cncr.11786
- Issue published online: 12 DEC 2003
- Article first published online: 7 OCT 2003
- Manuscript Accepted: 11 AUG 2003
- Manuscript Revised: 6 AUG 2003
- Manuscript Received: 30 MAY 2003
- ThinPrep Papanicolaou test;
- conventional smear;
- Asian screening population
This study compared the findings of cervical cytology screening by ThinPrep Papanicoloau (Pap) tests (TP) with the findings of screening by conventional cervical smears (CS) in a screening population involving what to the authors' knowledge isthe largest sample of ThinPrep Pap tests published to date.
Data from 191,581 CS that were screened in the period from March 1, 1998 to February 28, 2000 were compared with data from 190,667 TP performed from March 1,, 2000 to February 28, 2002 and that were obtained from the same sources.
With TP, the unsatisfactory rate was reduced from 0.48% to 0.32%. Fewer cases were considered to be suboptimal (19.12% vs. 12.97%). The detection rates of squamous cell carinomas, adenocarcinomas, and high-grade squamous intraepithelial lesions (HSIL) were essentially unchanged in the TP group (0.005%, 0.003%, and 0.25%, respectively) compared with the rates for the CS group (0.01%, 0.006%, and 0.25%, respectively). There was an increase in the detection of atypical squamous cells of undetermined significance (ASCUS; 3.74% vs. 3.19%) and low-grade SIL (LSIL; 1.67% vs. 1.01%) with a decrease in the ASCUS-to-LSIL ratio from 3.15 for CS to 2.33 for TP. The detection of atypical glandular cells of undetermined significance (AGUS) increased slightly from 0.07% to 0.09%. The proportion of cells reported to have reactive atypia dropped from 2.71% to 1.48%. Limited biopsy correlation (range, 73.2–76.2%) confirmed the increased sensitivity of TP. More actinomyces (1.07% vs. 0.52%) were detected in TP samples despite of a similar portion of intrauterine-device users. The average primary screening and rapid rescreening time of each slide were reduced from 8 minutes to 4 minutes and from 2 minutes to less than 1 minute, respectively.
The preliminary experience of the authors of the current studys appears to support the use of the ThinPrep Pap test to enhance the efficiency of cervical cytology screening. Cancer (Cancer Cytopathol) 2003;99:331–5. © 2003 American Cancer Society.
The ThinPrep Papanicoloau (Pap) Test (TP) (Cytyc Corporation, Boxborough, MA) was approved as a replacement for the conventional cervical smear (CS) in May 1996 and is considered to besignificantly more effective for the detection of low-grade squamous intraepithelial lesions (LSIL) and more severe lesions when compared with the conventional Pap smear in a variety of patient populations.1, 2 Although a number of studies have reported increased sensitivity and improved specimen adequacy with TP, many of these were split-sample studies.3–17 To our knowledge, there have been few publications based on a screening laboratory in which TP was implemented for routine use.18–27
This study presents the data from a laboratory serving an Asian screening population with full-scale conversion and routine application of TP and compares it with data obtained from CS screening during the previous years. Data obtained over a span of 4 consecutive years (2 years served by CS and 2 years by TP) were compared to minimize the effects of seasonal changes.28–30 To our knowledge, this is the largest TP series reported to date, on the basis of interpretation of nearly 200,000 routine TP screenings.
MATERIALS AND METHODS
Full-scale conversion to and application of the ThinPrep Pap test took place on March 1, 2000, by the Cervical Cytology Laboratory, Department of Pathology at the University of Hong Kong. Until that time, CS was the mode for screening. This test was adopted after the successful completion of a trial of 5000 cases and after all reporting cytopathologists and cytotechnologists had attended a certifiable training program organized by Cytyc Corporation.
From March 1, 1998 to February 28, 2000, and from March 1, 2000 to February 28, 2002, 194,281 CS and 190,667 TP samples were received from the clinics of the Family Planning Association of Hong Kong. Women attending these clinics were considered representative of the screening population in Hong Kong, since the prevalence of LSILs and more severe SIL was < 5%.9, 31 The age distribution of these 2 groups of patient populations were similar (CS: mean age of 36.2 years, standard deviation [SD] = 9.1; TP: mean age of 36.8 years, SD = 9.5). From March 1, 1998 to February 28, 2000, 16,359 (8.5%) intrauterine device (IUD) users were screened, and 18,409 (9.7%) IUD users were screened from March 1,s 2000 to February 28, 2002.
The TP samples are direct-to-vial samples, not split samples. The samples were collected via Papette, a broom-type sampling device (Wallach Surgical Devices, Inc., Orange, CT), then rinsed directly into the PreservCyt vial (Cytyc Corporation). The labeled samples were then processed using TP 2000 Processors (Cytyc Corporation). There were 9466 cases (4.96%) of TP samples that needed to be prewashed with CytoLyt/acetic acid solution to lyse red blood cells, dissolve mucus, and reduce protein precipitation. Both the CS and TP slides were stained using the Pap staining technique and the Autostainer XL (Leica, Nussloch, Germany) and then coverslipped in the same fashion. The specimen adequacy and the cytologic features in these slides were then evaluated by the cytotechnologists, using the Bethesda system.32 Slides containing abnormal cells or clinical information or those classified as inadequate were referred to pathologists.
The cytologic abnormalities detected during the two screening periods were compared. Analysis was performed with the Fishers exact test. The chi-square test was utilized and P values were derived. The screening and rapid rescreening times were also compared.
Patients with LSIL or more severe lesions were referred for colposcopic examination. To correlate cytology findings with cervical biopsy results, referral data of these patients were collected from major hospitals in Hong Kong.
The unsatisfactory rate was reduced from 0.48% to 0.32% with TP. Similarly, fewer cases were considered to be suboptimal (19.12% vs. 12.97%) (Table 1).
|No. of CS (%)||921 (0.48)||15,824 (16.5)||470 (0.49)||1167 (1.22)||876 (0.91)|
|No of TP (%)||608 (0.32)||23,725 (12.44)||1006 (0.53)|
There were differences in the incidence of abnormal cells being detected (Table 2). The detection rates of squamous cell carcinomas, adenocarcinomas, and high-grade squamous intraepithelial lesions (HSIL were essentially unchanged with TP (0.005%, 0.003%, and 0.25%, respectively) compared with rates of CS in the previous years (0.01%, 0.006%, and 0.25%, respectively). There was a slight increase in atypical squamous cells of undetermined significance (ASCUS) (3.74% vs. 3.19%) and LSIL (1.67% vs. 1.01%). The ASCUS-to-LSIL ratio was reduced from 3.15 for CS to 2.33 for TP. The incidence of atypical glandular cells of undetermined significance (AGUS) slightly increased from 0.07% to 0.09%. The proportion of cells reported to have reactive atypia dropped from 2.71% to 1.48%. Biopsy correlation was available in 73.2–76.2% of referral cases. Limited biopsy results confirmed the increased sensitivity of TP.
|Unsat||Neg||React||ASCUS||AGUS||LSIL||HSIL||SCC||Adeno||Total no. of cases|
|No. of CS (%)||921 (0.48)||176,774 (92.27)||5188 (2.71)||6117 (3.19)||134 (0.07)||1939 (1.01)||476 (0.25)||19 (0.01)||13 (0.006)||191,581|
|No. of TP (%)||608 (0.32)||176,285 (92.4)||2819 (1.48)||7124 (3.74)||189 (0.09)||3183 (1.67)||453 (0.24)||10 (0.005)||6 (0.003)||190,667|
|Monilia||Trichomonas||Actinomyces||HSV||Total no. of cases|
|No of CS. (%)||14,952 (7.7)||886 (0.46)||1003 (0.52)||62 (0.03)||194,281|
|No of TP. (%)||15,159 (7.97)||830 (0.44)||2048 (1.07)||68 (0.03)||190,667|
|Author||Year||Samples||No. of cases detected or screened|
|Hutchinson et al.3||1992||Split||2655||2655|
|Awen et al.4||1994||Split||1000||1000|
|Wilbur et al.5||1994||Split||3218||3218|
|Bur et al.6||1995||Split||114||114|
|Laverty et al.7||1995||Split||2026||2026|
|Tezuka et al.8||1996||Split||251||251|
|Lee et al.9||1997||Split||7360||7360|
|Robert et al.10||1997||Split||35,560||35,560|
|Sherman et al.11||1998||Split||1780||1780|
|Hutchinson et al.12||1999||Split||8000||8000|
|Shield et al.13||1999||Split||300||300|
|Wang et al.14||1999||Split||972||972|
|Park et al.15||2001||Split||483||483|
|Cockill et al.18||1997||Direct-to-vial||7000||499|
|Bolick et al.19||1998||Direct-to-vial||39,408||10,694|
|Papillo et al.20||1998||Direct-to-vial||8,574||16,314|
|Diaz-Rosario et al.22||1999||Direct-to-vial||74,756||56,339|
|Carpenter et al.21||1999||Direct-to-vial||5000||2727|
|Ashfaz et al.23||1999||Direct-to-vial||43,289||25,783|
|Weintraub et al.24||2000||Direct-to-vial||130,381||39,864|
|Obwegeser et al.25||2001||Direct-to-vial||1002||997|
|Limaye et al.27||2003||Direct-to-vial||1,985,350||166,619|
The average primary screening and rapid rescreening time of each slide were reduced from 8 minutes to 4 minutes and from 2 minutes to less than 1 minute, respectively.
Performance of the ThinPrep Pap test has been evaluated in several studies. Results have varied considerably and appear to be influenced by the collection device and collection protocols.
The majority of studies adopted a split-sample protocol.3–17 In split-sample studies, the collected material was first applied to the CS and then to the liquid-based medium. Another approach was to compare the diagnostic results of TP with CS from patients from similar sources. The CS data were either collected historically from previous years or collected concurrently.18–27 We support the suggestion that comparative studies be designed so that all cellular material obtained from the cervix is placed in the cell suspension (i.e. direct-to-vial approach), and the assessments be performed in a routine laboratory environment by cytotechnologists working in a routine cervical cytopathology service.33
We also believe that comparison of diagnostic results may be more reliable when data regarding CS and TP from a complete year are compared side by side. This is because clear seasonal rhythms in the detection of infections including monilia, trichomonas, Actinomyces, human papilloma virus (HPV), chlamydia, and herpes,28, 30 as well as squamous carcinoma and its precursors have been demonstrated.29, 30 The current study data are also unique because they provide an analysis of what to our knowledge is the largest hitherto reported collection of TP data from a screening population.
In general, the majority of studies comparing TP and CS found that the quality of slides improved in TP cases.6, 10, 13, 14, 16, 18–20, 22, 24, 33–35 TP significantly enhanced specimen adequacy by reducing the number of cases classified as “unsatisfactory” or “satisfactory but limited by …” due to obscuring inflammation, obscuring blood, and poor fixation. The current study found that the incidence of unsatisfactory cervical samples dropped by approximately one-third. However, our laboratory had a relatively low unsatisfactory rate, even with CS. The use of TP in communities with high rates of unsatisfactory smears may prove more cost-effective.
It has been reported in earlier studies that endocervical cells were detected less frequently with TP than with CS.6, 7, 18 This may be related to the choice of sampling devices. In the current series, the percentage of samples lacking endocervical cells with the TP was actually reduced by 24%. Although some TP slides were still considered to have inadequate squamous cells for assessment and thus suboptimal, the percentage was also significantly reduced.
A few reports have indicated that the TP method of sample preparation has greater diagnostic sensitivity than the conventional direct Pap smear.4, 18 This may be attributed in part to the nonhomogeneous cell sampling during transfer from the sampling device to the microscope slide with the direct smear whereas with the TP method, equivalent diagnostic material is achieved on each of the replicate slides prepared per specimen.36 In the majority of these cases, there was an increase in the detection of LSIL or more severe diagnoses.3, 5, 8, 11, 12, 14, 17, 19–24, 27, 34, 37 In particular, the TP method was considered capable of providing an increase in both sensitivity and specificity for glandular lesions.23
In the current series, the TP technique was associated with an increased detection of LSIL lesions, but no increase in high-grade dysplasias and carcinomas was found. This observation concurs with the findings of some earlier studies.21 There was also a reported increase in the percentage of ASCUS and AGUS. However, the ratio of ASCUS to LSIL substantially decreased in the TP population when compared with the CS population. Moreover, when the data concering ASCUS were broken down into monthly statistics, the number of ASCUS increased in the first 3 months of reporting TP but decreased in subsequent months. This may be related to the learning curve of cytotechnologists and cytopathologists, who required considerable retraining before reaching competence in screening monolayers.33
The incidence of HSIL or more severe lesions did not appear to significantly improve with TP in the current analysis. The incidence of HSIL reported in the population served by the study laboratory has been consistently low (< 0.3% between 1993–2003). This finding may reflect a genuine low incidence of HSIL in this screening population in Hong Kong.
The identification of infectious agents was found to be increased in previous studies.14, 20 Nevertheless, to our knowledge, the significant increase in the detection of Actinomyces has not been reported previously. This may be related to the high number of IUD users in our screening population.
It has been documented in earlier studies that TP smears were easier to screen, and screening times were significantly shorter for TP than for CS.6, 7, 16–18, 33, 36, 38, 39 Our technicians could finish the assessment in approximately half the time required for CS.
The results of the current study demonstrate that the TP reduces the possibility of an unsatisfactory smear and improves diagnostic efficiency as indicated by an increased LSIL-to-ASCUS ratio as well as a reduced number of cells reported to have reactive atypia. The former is due to better demonstration of koilocytes. More cases of Actinomyces were identified by TP. The time needed for screening and rescreening was markedly reduced, leading to improved efficiency in the laboratory screening procedures. Our preliminary experience suggests that the use of TP can enhance the efficiency of cervical cytology screening.
The authors thank Dr. Susan Fan, Dr. R. J. Collins, Professor H. K. Ng, Dr. John Chan, Dr. Andrew Choy, Dr. Elaine Gwi, Dr. K. C. Lee, Dr. W. F. Ng, Dr. S. W. Pang, Dr. Michael Suen, and Dr. M. C. Tang for providing the follow-up data.
- 12Utility of liquid-based cytology for cervical carcinoma screening: results of a population-based study conducted in a region of Costa Rica with a high incidence of cervical carcinoma. Cancer. 1999; 87: 48–55., , , et al.
- 32The Bethesda System for reporting cervical/vaginal cytologic diagnoses: definitions, criteria and explanatory notes for terminology and specimen adequacy. New York: Springer-Verlag, 1994., .