Performance of 3 methods for quality control for gynecologic cytology diagnoses.

OBJECTIVE: To evaluate performance and viability of internal quality control (QC) strategies in a public health laboratory of the state of São Paulo. STUDY DESIGN: A retrospective study was performed with 3 QC strategies to improve internal cytologic diagnoses: morphologic guided-list criteria (MGLC), 100% rapid-rescreening (100% RR) of negative slides ("turret" method) and 10% rescreening (10% R) of negative slides. Cases were examined at Adolfo Lutz Institute, São Paulo, Brazil, from 2002 to 2004. Histopathologic results, when available, were considered gold standard; cytologic consensus diagnosis was by 2 pathologists when histologic results were unavailable. RESULTS: MGLC selected 20.7% samples with cytologic atypias, 10% R selected 0.6% and RR selected 2.5%. Cytologic/histologic initial concordance was 57.4%, low-grade squamous intra-epithelial lesion false negative rate was 34.9% and high-grade squamous intraepithelial lesion false negative rate was 12.2%. After diagnosis, consensus concordance was 97.2%. CONCLUSION: The 100% RR and 10% R QC strategies detected more false negative cases in liquid-based cytology than in conventional Pap smears. The 100% RR strategy reduced the false negative results and allowed evaluation of individual staff performance. The 10% R strategy did not offer significant results. We concluded that association of MGLC and 100% RR strategies might improve cytologic diagnostic quality.

How accurate is cytological diagnosis of cervical glandular lesions?

Gynecological cytology has some inaccurate morphological categorization and poor interobserver reproducibility especially for glandular lesions. Liquid-based cytology (LBC) preparations are presumed to reduce artifacts that interfere in diagnosis performance, but its value to correctly identify glandular alterations has not been sufficiently reported. The objective of this study was to compare the diagnostic performance and interobserver agreement of LBC and conventional Pap smear to identify histologically confirmed glandular lesions according to five cytologists. Sensitivity ranged from 55.8 to 73.1% and 32.7 to 48.1% for Pap smear and LBC, respectively. Specificity ranged from 66.1 to 87.1% and 69.4 to 94.4%, respectively. In general, agreement between pairs of cytologists was poor with kappa-values around 0.45. In conclusion, relying entirely on cervical cytology to rule out glandular lesions may be risky. The use of HPV DNA test alone or combined to screening glandular lesions may contribute to minimize the limitations of both conventional and LBC preparations to diagnose glandular abnormalities.

Immunocytochemical expression of p16INK4A and Ki-67 in cytologically negative and equivocal pap smears positive for oncogenic human papillomavirus.

This study was designed to analyze the cross-sectional comparison of the p16 and Ki-67 immunocytochemical expression in negative and equivocal (atypical squamous cells of undetermined significance (ASC-US)) liquid-based cytology (LBC) samples testing positive for high-risk human papillomavirus (HPV) types with HC2 assay or polymerase-chain reaction (PCR). A series of 199 consecutive LBC specimens derived from the same number of women participating in the ongoing Latin American Screening Study at Leonor Mendes de Barros Hospital, Sao Paulo, were analyzed using immunocytochemistry for expression of p16 and Ki-67 in negative and equivocal LBC samples testing positive for high-risk HPV types with hybrid capture II test (HC2) or PCR. All patients with at least one test positive (cytology, PCR, and/or HC2) were followed each 6 months for 3 years. The follow-up procedure consisted of visual examination, colposcopic inspection, cytology, and HC2 assay. Among the negative cytologic samples, 101 were HPV-positive and 55 HPV-negative. Of the HPV-positive group, 59 of 101 cases (58.4%) were positive for both p16 and Ki67 immunostaining, and 17 of 101 (16.8%) were negative for both. The proportion of Ki-67-positivity increased almost in parallel with the increasing grade of p16-positivity (p = 0.0001 for linear trend). In the HPV-negative group, both markers were negative in 41 of 55 cases (74.5%), and no statistical relationship was observed between the two markers (Pearson, p = 0.595). HPV-positive ASC-US samples demonstrated a simultaneous positive immunoreaction for p16 and Ki67 in 11 of 16 cases (68.7%), whereas 3 (18.7%) were concurrently negative. The relationship between the two markers was of borderline significance (Pearson, p = 0.053), but no linear relationship was found between the graded p16 and Ki-67 expression (p = 0.065 for linear trend). In the HPV-negative ASC-US group, there was no statistical association between the graded p16 and Ki-67 positivity (Pearson, p = 0.281). After 36 months of follow-up of the ASC-US patients, 6 women still displayed ASC-US smear, of which 4 of 6 were HPV-positive and expressed both p16 and Ki-67 markers. Two of 43 ASC-US smears had high-grade squamous intraepithelial lesions diagnosed (4.6%), and 1 had low-grade squamous intraepithelial lesion (2.3%). All of those were positive for HPV, p16 and Ki-67. Patients with ASC-US diagnosis and positive high-risk HPV status and positive for p16 Ki67 should be carefully observed to exclude occurrence of a squamous intraepithelial lesion. The combination of these two markers can be a useful implement for management of women with equivocal cytology.

100% rapid rescreening for quality assurance in a quality control program in a public health cytologic laboratory.

OBJECTIVE: To verij5 the efficacy of the quality control (QC) program in a cytologic laboratwy with a rapid rescreening (RR) protocol. STUDY DESIGN: RR, according to the Turret RR method, of all samples initially screened as negative at the Laboratory of Cytology, Adolfo Lutz Institute, was performed. The slides were reviewed for 60 seconds. Suspect smears were fully checked by 2 reviewers to determine the final diagnoses. A total of 2954 sequential cytologic results were considered in this study. Of the 2954, 2568 (86.9%) were considered initially negative according to our internal QC, and these cases underwent RR. Also, 10% were randomly selected from these negative cases for full reviewing. The internal QC in our laboratory includes review of cases selected according to clinical and cytomorphologic criteria. RESULTS: Among the 2954 total cases, QC detected 386 (13%) atypias with final diagnoses reported according to The Bethesda System 2001 as follows: 82 (2.18%) low grade squamous intraepithelial lesions (LSILs), 35 (1.18%) high grade squamous intraepithelial lesions (HSILs), 2 (0.06%) squamous cell carcinomas, 105 (3.5%) atypical cells of undetermined significance (ASC-US), 4 (0.12%) atypical endocervical cells (AECs) and 158 (5.3%) unsatisfactory samples. RR of 2568 smears initially considered negative selected 194 (7.5%) slides. Of the 194, 146 (75.3%) were negative, 28 (14.4%) ASC-US, 5 (2.6%) AEC, 1 (0.5%) LSIL and 14 (7.2%) unsatisfactory. Full review of a 10% random fraction of the 2568 cases interpreted as negative did not detect lesions but did detect 5 (1.95%) unsatisfactory samples. CONCLUSION: Internal QC used in our laboratory based on clinical and cytomorphologic criteria to select cases for review proved to be an efficient method of detecting HSIL and cervical cancer. The consensus basis of this program strongly limits the false positive and false negative rates and also provides subjects with continuing education. One hundred percent RR is more efficient than 10% full reviewing in detecting cervical abnormalities.

Cervical biopsy-based comparison of a new liquid-based thin-layer preparation with conventional Pap smears.

The objective of this study is to compare the diagnostic efficacy of universal collection medium (UCM) liquid-based cytology (LBC) (Digene Corp., MD) and the conventional Pap smear in a comparative study, using histologic results as the gold standard. This was a cross-sectional study. Conventional Pap smears and UCM LBC specimens, obtained from women in a low socioeconomic outpatient population referred to a tertiary center for gynecologic care, were compared. For the purpose of this study, when cervical specimens were collected for cytology, all women underwent colposcopy and biopsy was done if a cervical abnormality was observed. Cytologic evaluation of UCM LBC and conventional Pap smears were carried out separately, masked to the results of the other method. Agreement beyond chance between the two cytologic methods was ascertained by means of the unweighted kappa statistic. Sensitivity, specificity, and predictive values with 95% confidence intervals were calculated for both methods. McNemar's test was used to determine the level of association between the two cytology procedures. A total of 800 women were evaluated. Assessment of the overall agreement between the two cytologic methods yielded a kappa of 0.777 (P < 0.0001). After adjustment for histologic diagnosis, the computed kappa in each stratum was as follows: normal = 0.733; CIN 1 = 0.631; CIN 2/3 = 0.735; cancer = 0.652. The sensitivity and specificity of UCM LBC for detection of cervical intraepithelial lesions and cancer were 75.3% and 86.4%, respectively, not statistically different from the 81.8% and 85.2% seen with the conventional method. This study demonstrates that the UCM LBC method is as accurate as the conventional Pap smear cytology in detecting cervical intraepithelial lesions and cancer even so the UCM samples were systematically prepared from a second sampling of the cervix.

Atypical glandular cells of undetermined significance. Cytologic predictive value for glandular involvement in high grade squamous intraepithelial lesions.

OBJECTIVE: To verify the cytologic predictive value of a diagnosis of atypical glandular cells of undetermined significance (AGUS) in high grade squamous intraepithelial lesions (HSIL) cases. STUDY DESIGN: In a retrospective study, 98 cases of HSIL were reviewed. All patients were referred for colposcopy and directed biopsy to confirm the cytologic diagnoses. Loop excision of the transformation zone was performed to treat clinical lesions. Sensitivity, specificity, positive and negative predictive value were evaluated. Kappa statistics and logistic regression analysis were used to evaluate the findings statistically. RESULTS: By logistic regression analysis, we found that the chance of finding squamous intraepithelial lesions involving glands in AGUS smears was 5.32 times higher than in those with no AGUS. It was 5.74 times higher in cervical intraepithelial neoplasia (CIN) 3 lesions than in CIN 2. CONCLUSION: The cytologic predictive value for HSIL involving glands is statistically significant when specific and objective criteria are used for the AGUS diagnosis.