Fax: (011) 39-0471-908175
ThinPrep versus conventional Papanicolaou smear in the cytologic follow-up of women with equivocal cervical smears
Version of Record online: 22 OCT 2003
Copyright © 2003 American Cancer Society
Volume 99, Issue 6, pages 342–345, 25 December 2003
How to Cite
Negri, G., Menia, E., Egarter-Vigl, E., Vittadello, F. and Mian, C. (2003), ThinPrep versus conventional Papanicolaou smear in the cytologic follow-up of women with equivocal cervical smears. Cancer, 99: 342–345. doi: 10.1002/cncr.11856
- Issue online: 12 DEC 2003
- Version of Record online: 22 OCT 2003
- Manuscript Accepted: 22 SEP 2003
- Manuscript Received: 15 AUG 2003
- liquid-based cytology;
- atypical squamous cells of undetermined significance (ASCUS);
- atypical squamous cells;
- cannot exclude a high-grade lesion (ASC-H)
The purpose of the current study was to compare the efficacy of liquid-based cytology and conventional smears in the cytologic follow-up of cases with “atypical squamous cells, cannot exclude a high-grade lesion” (ASC-H) or “atypical glandular cells” (AGC).
Cytologic follow-up was performed on 214 cases with ASC-H/AGC diagnosis an conventional smears using either ThinPrep (n = 100) or conventional Papanicoloau (Pap) tests (n = 114). Results were then compared with further histologic and/or cytologic follow-up.
Repetition on conventional smears enabled a definite diagnosis (within normal limits [WNL], squamous intraepithelial lesion [SIL] or carcinoma) in 58 cases (50.9%). ASC/AGC was confirmed in 50 cases (43.9%), and 6 of the smears (5.3%) were inadequate. WNL, SIL, or carcinoma was diagnosed in 82 (82.0%) cases by following the patients with ThinPrep cytology, whereas ASC or AGC was confirmed in 18 cases (18.0%). No inadequate specimens were found. A diagnosis of SIL or greater (SIL +) was confirmed histologically in 11 of 11 (100.0%) conventional smears and in 31 of 34 (91.2%) ThinPrep specimens. Of the 87 WNL specimens, 9 (8 conventional smears and 1 ThinPrep specimen) developed a histologically confirmed SIL during further follow-up. Specimen adequacy was significantly better in the ThinPrep specimens compared with conventional smears.
Because of better specimen adequacy, ThinPrep cervical cytology appears to significantly reduce the occurrence of ASC/AGC compared with conventional Pap smears. Cancer (Cancer Cytopathol) 2003;99:342–5. © 2003 American Cancer Society.
Since its introduction in 1943 by Papanicolaou, Papanicolaou (Pap) smear screening has been a widely accepted and cost-effective method for the prevention of cervical carcinoma. Nevertheless, a high ratio of false-negative smears and of atypical squamous cells (ASC) diagnoses have led to the development of new diagnostic techniques.1, 2, 3 The ThinPrep Pap test (TP) was proposed as an alternative technique that should guarantee a more reliable diagnosis compared with the conventional Pap smear (CS). Although a number of studies have reported an increase in sensitivity and specimen adequacy with a better ASC:squamous intraepithelial lesion (SIL) ratio some recent studies and reviews of the literature have not been able to confirm any significant difference in sensitivity, specificity, or occurrence of ASC between TP and CS.4–10 In a recent study, CS performed even better than TP.11 However, to our knowledge to date no study has compared liquid-based cytology (LBC) and CS in the follow-up of equivocal smears. Solomon et al. reported on the follow-up of a large number of atypical squamous cells of undetermined significance (ASCUS) cases with TP as part of a triage including also colposcopy and human papilloma virus (HPV) typing.3 The report pointed out the utility of HPV typing in the management of ASCUS. However, high grade cervical intraepithelial neoplasia (CIN3) cases with negative colposcopy are not rare,12 and because of the low specificity of HPV typing, some cases may actually profit more from an effective cytologic follow-up.
The aim of the current study was to compare the efficacy of direct-to-vial TP and CS in the cytologic follow-up of selected cases with a primary diagnosis of atypical squamous cells that cannot exclude a high-grade lesion (ASC-H) or atypical glandular cells (AGC) on CS.
MATERIALS AND METHODS
From our database, 195 women (mean age, 46.7 years; range, 25–90 years) with a diagnosis of abnormal cervical epithelia of unknown significance on conventional smears (high-grade dysplasia or carcinoma cannot be excluded or definitely diagnosed) were selected. All cases were primarily diagnosed during the same period (2000–2002). The original diagnosis, according to the Muenchener Nomenklatur II (MNII), was Pap III (not clear), which is nearly equivalent to the ASC-H/AGC category of the Bethesda System. At the time of review of the cases, the equivalent in the Bethesda system was used (177 ASC-H and 37 AGC). Pap III cases that did not satisfy the criteria of ASC-H or AGC were not included in the study.
Smear adequacy was always recorded. Although the category “satisfactory but limited by” was eliminated from the Bethesda System, this category was maintained according to the MNII to point out troublesome smears.
A cytologic control with either CS (n = 114) or TP (n = 100) was performed after a 2- to 4-month period. After cytologic follow-up, a biopsy to confirm the cytologic diagnosis was taken in 90 cases (31 ASC/AGC cases, 45 cases of SIL, and 14 WNL cases). In addition, 99 cases had a further cytologic follow-up, and 24 had no follow-up.
Cytologic sampling was performed by 81 different gynecologists. Sampling for CS was performed with heterogeneous collecting methods (swab, spatula, and brush), whereas all TP specimens were sampled using a cervex brush. TP specimens were processed using the TP 2000 or TP 3000 processor (Cytyc Corporation, Boxborough, MA). Erythrolytic pretreatment with acetic acid was performed where necessary. Both CS and TP slides were stained according to the routine Pap method.
All slides were screened by experienced cytotechnologists. Abnormal slides were referred to a supervisor or pathologist.
Statistical differences in the proportions of unclear cytologies between the two analyzed groups were evaluated using the z-statistics. A P value < 0.05 was considered statistically significant.
At the time of the initial diagnosis of ASC-H/AGC on CS, 96 smears (44.9%) were classified as satisfactory. All other cases were judged as satisfactory but limited by (SBLB) inflammatory exudate, cell overlapping, presence of blood, absent endocervical or metaplastic cells, poor fixation, low cellularity, or other limiting factors (Table 1).
|Initial cytology (n = 214)||Follow-up on conventional smears (n = 114)||Follow-up on ThinPrep (n = 100)|
|Satisfactory||96 (44.9%)||50 (43.9%)||82 (82.0%)|
|SBLB||118 (55.1%)||58 (50.9%)||18 (18.0%)|
|Unsatisfactory||0 (0.0%)||6 (5.2%)||0 (0.0%)|
At follow-up on conventional cytology, 50 smears (43.9%) were classified as satisfactory, 58 (50.9%) as SBLB, and 6 (5.3%) as unsatisfactory. In 58 CS (50.9%), a clear diagnosis was possible (Table 2). Of these 58 clear cases, 50% were satisfactory and 50% were SBLB. In 56 CS (49.1%), an unclear cytologic diagnosis (ASC, AGC, or unsatisfactory) was confirmed. In addition, 37.5% of these smears were satisfactory, 51.8% were SBLB, and 10.7% were unsatisfactory.
|Conventional smear (n = 114)||Satisfactory CS (%)||ThinPrep (n = 100)||Satisfactory TP (%)|
|All definite diagnoses||58 (50.9%)||50.0||82 (82.0%)||84.2|
|All unclear diagnoses||56 (49.1%)||37.5||18 (18.0%)||72.2|
On repetition of cytology with TP, 82 smears (82.0%) were classified as satisfactory, 18 (18.0%) were classified as SBLB, and no case was considered unsatisfactory. A definite diagnosis was possible in 82 TP specimens (82.0%). Of these 82 cases, 84.2% were satisfactory and 15.8% were SBLB. The diagnosis remained unclear (ASC–AGC) in only 18 TP specimens (18.0%). In addition, 72.2% of these cases were classified as satisfactory, and 27.8% were classified as SBLB (Table 2).
Histologic follow-up confirmed a dysplastic or neoplastic lesion in 11 of 11 CS (100.0%) and in 31 of 34 TP with a cytologic diagnosis of SIL + (91.2%) (Table 3). Two of the three negative histologic specimens were only partially evaluable because of marked artifacts. Histologic follow-up of ASC/AGC was positive for dysplasia or more severe diagnosis in 17 of 24 CS (70.8%) and 5 of 7 TP specimens (71.4%). Of the 87 WNL, 14 had a histologic follow-up. In nine cases (eight CS and one TP specimen), a CIN was diagnosed. For 60 WNL cases, at least 1 additional cytologic follow-up had occurred, and 13 cases had no follow-up. In 3 of 29 CS (10.3%) with a previous diagnosis of WNL, an abnormal cytology (2 ASC/AGC and 1 SIL) was diagnosed. None of the 31 TP cases with a previous WNL diagnosis demonstrated an abnormal cytology.
|Cytologic diagnosis||Positive histology (CIN/AIS +)||Negative histology|
|SIL/AIS +||11/11 (100%)||0/0|
|ASC/AGC||17/24 (70.8%)||7/24 (29.2%)|
|WNL||8/10 (80.0%)||2/10 (20.0%)|
|SIL/AIS +||31/34 (91.2%)||3/34 (8.8%)|
|ASC/AGC||5/7 (71.4%)||2/7 (28.6%)|
|WNL||1/4 (25.0%)||3/4 (75.0%)|
Statistical analysis showed significant differences between the proportions of unclear cytologies in the CS group (49.1%) and the TP group (18.0%) (P < 0.001).
To our knowledge, several studies to date have already reported a better sensitivity and adequacy for TP compared with CS smears.4, 5, 6, 7 Recently, however, statistical reviews of the literature have concluded that there is no evidence that LBC is more or less sensitive or specific than CS.8, 9 Major critical points to the studies already published were the inconsistent histologic follow-up for high-grade lesions and the use of brush sampling for both CS and LBC in “split sample” protocols, which may reduce the performance of CS. Furthermore, Moseley et al. claim that some studies compare LBC with suboptimal CS.9 This is, however, the daily routine in many cytologic laboratories dealing with a large number of clinicians with heterogeneous sampling, and it may be a reason for choosing LBC. In the current direct-to-vial study, contributors were not selected. The objective of the current study was to compare the efficacy of TP and CS specimens in the cytologic follow-up of cases in which the primary diagnosis on CS was of abnormal squamous or glandular epithelia of unknown significance and a high-grade squamous or glandular lesion could not be excluded. In the study laboratory these diagnoses account for less than 0.2% of all smears. Although these patients are usually referred for colposcopy and biopsy, the clinicians in these cases simply repeated the cytology without performing biopsy.
Because of better fixation, the reduction of artifacts, and better sampling of the squamocolumnar junction, a definite diagnosis was easier to formulate in TP than in CS. The results demonstrated a significant increase in clear diagnoses (WNL, SIL +, and adenocarcinoma in situ [AIS] +) and a decrease in ASC/AGC on TP specimens compared with cases in which cytology was repeated on CS (P < 0.0001). The poor performance of repeated CS can be explained by the persistence of factors limiting the interpretation of the smears.
Although the category “satisfactory but limited by” was eliminated from TBS, in the current study we maintained the category to point out troublesome smears. In fact, our analysis showed a significant association between ASC/AGC and poor smear adequacy. Fully satisfactory smears were recorded at follow-up in 43.9% of CS cases and 82.0% of TP specimens. Overall at follow-up, 45.9% of the ASC/AGC specimens and 64.3% of the smears with a clear diagnosis were judged to be completely satisfactory for evaluation. Smears were considered unsatisfactory or “satisfactory but limited by” mostly as a result of granulocytosis, cell overlapping, presence of blood, cell degeneration, poor fixation of the specimen, scanty cellularity, or absence of endocervical cells. Despite the elimination of the category SBLB, the problem of troublesome smears still exists. Even if not unsatisfactory for evaluation, these smears can be difficult to interpret and may lead to unnecessary ASC diagnoses or false-negative or false-positive reports.
Obwegeser et al. consider the greatest advantage of LBC to be that it forces the clinician to use an adequate collection device.10 This would also explain the better detection rates of SIL. In the study by Obwegeser et al. the comparison of results between CS and LBC did not reveal any statistically significant differences in any diagnostic categories. Sampling was in fact performed by a small number of well trained clinicians, and specimen collection for CS was homogeneously performed using a Szalay spatula under colposcopic guidance. This may be unrealistic in large laboratories with heterogeneous experience of the clinicians taking smears. Well trained clinicians using reliable sampling tools may achieve only minor benefits from TP. Adequate training of a large number of gynecologists may be more cost-effective than LBC, although this may be difficult in practice. Furthermore, correctly obtained CS with marked inflammatory exudates or hematic smears may still be difficult to interpret. Also, in the study by Coste et al., only skilled physicians were selected, whereas cytopathologists were experienced in CS but not in LBC.11 This fact may explain the better results of CS compared with LBC in that study.
Overall, in the current study the detection rate of severe lesions was significantly higher with TP than with CS (P < 0.0001). This significant difference is reflected by the fact that only challenging ASC-H/ACG cases, for which the advantages of LBC were more evident, were analyzed. Histologic diagnoses confirmed a dysplastic lesion in more than 90% of the TP cases. The follow-up of WNL cases also demonstrated a higher rate of false-negative CS results compared with TP, therefore indicating the better sensitivity of the latter.
The ThinPrep Pap test appears to perform significantly better than CS in the follow-up of cases with previously unclear cytologic diagnoses. Colposcopy and biopsy are obviously the management of choice of ASC-H and AGC smears.13 However, cases with negative colposcopy in which a clinically unapparent lesion cannot be excluded may profit from an effective cytologic follow-up. The current study data indicate an overall decrease in unclear diagnoses because of the better specimen adequacy of TP. Although good results might also be attained by training the clinicians in specimen sampling, it may be difficult to achieve a homogeneous high quality of CS in laboratories with a large number of contributors. TP might therefore serve as a method for improving the diagnostic reliability of cervical cytology.
The authors thank Hubert Dorfmann for his support.
- 5Utility 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.