Thyroid fine needle aspiration cytology: a review of the National Cancer Institute state of the science symposium.

In October 2007, the National Cancer Institute (NCI) of the United States sponsored a conference reviewing the state of the science of thyroid fine needle aspiration (FNA). Multiple issues were reviewed including pre-FNA requirements, training specifications, criteria for the selection of patients to undergo FNA, diagnostic categories and criteria, ancillary testing and post-FNA follow-up and treatment options. A summation of conclusions covering three of these topics, followed by a question and answer session, was subsequently presented at the 35th European Congress of Cytology (ECC) in Lisbon, Portugal in September of 2009. At the ECC, the findings of the NCI committee proposals regarding the indications for FNA of thyroid nodules, diagnostic categories and criteria, and post-FNA options for follow-up and treatment were discussed. Herein we review the presentations given at that conference.

Improving the yield of circulating tumour cells facilitates molecular characterisation and recognition of discordant HER2 amplification in breast cancer.

BACKGROUND: Circulating tumour cells (CTCs) offer a non-invasive approach to obtain and characterise metastatic tumour cells, but their usefulness has been limited by low CTC yields from conventional isolation methods. METHODS: To improve CTC yields and facilitate their molecular characterisation we compared the Food and Drug Administration-approved CellSearch Epithelial Kit (CEK) to a simplified CTC capture method, CellSearch Profile Kit (CPK), on paired blood samples from patients with metastatic breast (n=75) and lung (n=71) cancer. Molecular markers including Human Epidermal growth factor Receptor 2 (HER2) were evaluated on CTCs by fluorescence in situ hybridisation (FISH) and compared to patients' primary and metastatic cancer. RESULTS: The median cell count from patients with breast cancer using the CPK was 117 vs 4 for CEK (P<0.0001). Lung cancer samples were similar; CPK: 145 cells vs CEK:4 cells (P<0.0001). Recovered CTCs were relatively pure (60-70%) and were evaluable by FISH and immunofluorescence. A total of 10 of 30 (33%) breast cancer patients with HER2-negative primary and metastatic tissue had HER2-amplified CTCs. CONCLUSION: The CPK method provides a high yield of relatively pure CTCs, facilitating their molecular characterisation. Circulating tumour cells obtained using CPK technology demonstrate that significant discordance exists between HER2 amplification of a patient's CTCs and that of the primary and metastatic tumour.

Should atypical squamous cells of undetermined significance (ASCUS) be subcategorized? Accuracy analysis of Papanicolaou smears using receiver operating characteristic curves and implications for the ASCUS/squamous intraepithelial lesion ratio.

We correlated all Papanicolaou test diagnoses over a 6-month period with biopsy results and determined accuracy using receiver operating characteristic curves and biopsy as the "gold standard." Accuracies were calculated using all atypical squamous cells of undetermined significance (ASCUS) cases or by eliminating subsets thereof. Retaining the ASCUS category resulted in significantly greater accuracy for the diagnosis of squamous intraepithelial lesion (SIL) on biopsy compared with eliminating it by diagnosing all such cases as negative. Subcategorization significantly improved the accuracy of the test only when all cases were included. The highest accuracy without subcategorization was achieved when ASCUS, favor reactive, cases were diagnosed as negative, but this threshold was significantly less sensitive than including all ASCUS cases. Increasing or decreasing the estimated ASCUS/SIL ratio from 2.4 without subcategorization significantly reduced accuracy. Similar results were obtained when high-grade SIL on biopsy was used as the gold standard. Use of the ASCUS category significantly improves the accuracy of the Papanicolaou test. Eliminating any subset of ASCUS reduces the ASCUS/SIL ratio but also significantly diminishes the sensitivity of the Papanicolaou test.

An analysis of factors that influence the ASCUS/SIL ratio of pathologists.

In pursuit of physician-specific performance data in cytology, we have been calculating the ASCUS/SIL (atypical squamous cells of undetermined significance/squamous intraepithelial lesion) ratio of cytopathologists (CPs) and providing confidential feedback every 6 months. At the same time, thin-layer technology was introduced as an alternative to conventional smears. Thus we analyzed factors that may influence the ASCUS/SIL ratio, particularly the effect of periodic feedback on outliers (defined by a professional benchmark). For 3 years, the mean ASCUS/SIL ratio for all CPs decreased significantly from 2.92 to 1.87. There was great variability in the mean ASCUS/SIL ratio among 12 CPs (range, 1.11-5.89). Of the 6 CPs who worked continuously during this time, 2 showed a statistically significant decrease in their ASCUS/SIL ratio, including the CP with the highest ratio; 1 showed a significant increase. The mean ASCUS/SIL ratio did not correlate well with years of CP experience or with individual annual case volume. The ASCUS/SIL ratio of some CPs can decrease significantly over time. Whether it was due to feedback or the introduction of thin-layer preparations could not be determined.

Cytologic diagnosis of gastrointestinal stromal tumor with emphasis on the differential diagnosis with leiomyosarcoma.

BACKGROUND: Gastrointestinal stromal tumor (GIST) has only recently been distinguished histologically and immunochemically from morphologically similar neoplasms of the abdomen. METHODS: The authors reviewed 15 cytologic cases of GIST (14 fine-needle aspiration [FNA] specimens and 1 peritoneal fluid specimen) and compared them with 23 cases of leiomyosarcoma (LMS) arising in the abdomen or pelvis (all FNAs). Immunochemistry (IC) was performed on both the cytologic and subsequent tissue specimens if sufficient specimen was available. RESULTS: Cytologic samples of GISTs typically showed irregularly outlined clusters of uniform spindle cells that were spread easily without crush artifact. The cells had wispy cytoplasm with long, delicate, filamentous extensions (13 cases; 87%). A prominent vascular pattern was common (9 cases; 60%); pleomorphism (1 case; 7%) was uncommon. The LMSs showed three-dimensional, tightly cohesive, sharply marginated syncytia of spindle cells, often with nuclear crush artifact. The cytoplasm/stroma had a distinct wiry, refractile appearance (21 cases; 91%); delicate filamentous cytoplasmic extensions (5 cases; 22%) and prominent vessels (3 cases; 13%) were less common. LMSs more commonly exhibited pleomorphism (14 cases; 61%). Epithelioid cytomorphology, mitoses, and necrosis occasionally were observed in both tumor types. IC for c-kit (on cytologic material) was positive in 10 of 10 cases of GIST (usually diffuse and strong) and 2 of 19 cases of LMS (focal). CD34 positivity favored GIST (4 of 9 cases) over LMS (1 of 19 cases). Smooth muscle actin was positive in 20 of 20 LMSs (strong and diffuse) and 6 of 10 GISTs (usually focal). Desmin was positive in 12 of 20 LMSs and was only focally positive in 1 of 11 GISTs. Correlation of IC results was excellent between cytologic and tissue specimens. CONCLUSIONS: Delicate cytoplasmic processes; a prominent vascular pattern; a lack of nuclear pleomorphism; and a c-kit-positive, desmin-negative immunoprofile are characteristic features of GIST and help distinguish these tumors from LMS in cytologic specimens.

Quality assurance in gynecologic cytology. The value of cytotechnologist-cytopathologist discrepancy logs.

We describe a simple method for displaying and evaluating the concordance or discordance between cytotechnologists (CTs) and cytopathologists (CPs) on gynecologic cases. The provisional diagnoses made by the CTs and the final diagnoses of the CPs are captured by the laboratory information system; data generated for specified periods are displayed as a 10 x 10 matrix that classifies each possible diagnosis made by the CT and CP into 1 of 10 major categories. Matrices are generated for the entire laboratory and for individual CTs; individual CTs are evaluated based on their deviation from the laboratory average. Three statistical measures are generated: percentage of discordant diagnoses, a kappa statistic, and a weighted measure. During a 2.5-year period, approximately 4,200 cases were referred to a CP for review every 6 months. The median discordance in diagnoses increased during 2 years from 21% to 34%, and the kappa value fell from 0.69 to 0.38. This was attributed primarily to 1 CT, whose performance, as well as that of the entire laboratory, improved after remedial action. Measures of CT-CP diagnostic concordance are a useful and efficient measure of CT performance and can be incorporated into mandatory semiannual performance evaluations.

Extensively keratinized squamous intraepithelial lesions of the cervix are difficult to grade.

We tested the hypothesis that extensively keratinized squamous intraepithelial lesions (SILs) are difficult to grade precisely by identifying 100 Papanicolaou smears with a keratinizing SIL that had been originally judged difficult to grade. Of these, 65 were confirmed as low-grade SIL (LSIL) or high-grade SIL (HSIL) on subsequent biopsy. The 65 smears were reviewed independently by 3 cytopathologists who graded each case as LSIL or HSIL (by Bethesda System criteria). The accuracy of the grade was determined by the subsequent biopsy results; accuracy was compared with that of a historic control group of SILs with biopsy follow-up. In the study group, biopsies showed LSIL in 41 cases and HSIL in 24. The mean accuracy for a smear diagnosis of LSIL was 60% for the study group and 92% for the control group. For a smear diagnosis of HSIL, the accuracy was 60% for the study group and 95% for the control group. The overall kappa value for the study group confirmed poor interobserver agreement. Some keratinizing SILs are difficult if not impossible to grade precisely using standard criteria. For such lesions, the diagnosis "SIL, grade cannot be determined due to extensive keratinization" is justified.

Malignant epithelioid angiomyolipoma ('sarcoma ex angiomyolipoma') of the kidney: a case report and review of the literature.

Malignant epithelioid angiomyolipoma is a recently described rare tumor of the kidney. Its existence has been questioned, however, on the basis of incomplete evidence of malignant behavior, the absence of an associated classic angiomyolipoma component, or the absence of immunoreactivity for HMB-45 in some cases. We describe a case that was HMB-45-positive and arose in association with a classic angiomyolipoma. The patient was treated with a partial nephrectomy. Three years later, she developed rapidly enlarging liver nodules. A fine-needle aspiration of the liver confirmed the presence of pleomorphic epithelioid cells morphologically and immunohistochemically identical to those comprising the primary renal tumor. After two cycles of treatment with doxorubicin, there was a 50% reduction in the size of the tumors with marked improvement in performance status. We believe this case confirms the existence of a malignant epithelioid angiomyolipoma.

Usefulness of ultrasonography in the management of nodular thyroid disease.

BACKGROUND: Fine-needle aspiration biopsy is the standard diagnostic test for evaluating possible malignancy in a thyroid nodule. OBJECTIVE: To evaluate the role of routine ultrasonography in the management of nodular thyroid disease. DESIGN: Retrospective chart review. SETTING: Multidisciplinary thyroid nodule clinic (endocrinology and radiology). PATIENTS: Patients with suspected nodular thyroid disease or suspected recurrent thyroid cancer referred between October 1995 and March 1997. All patients had thyroid ultrasonography and ultrasonography-guided fine-needle aspiration biopsy of nodules at least 1 cm in maximum diameter. MEASUREMENTS: Medical records, ultrasonography findings, cytology reports, and histologic reports were reviewed. Ultrasonography findings were compared with the referring physician's findings on physical examination. RESULTS: 223 patients were seen in the clinic. A total of 209 fine-needle aspiration biopsies were performed on 156 patients. Among 50 of 114 patients referred for a solitary nodule, ultrasonography detected additional nonpalpable nodules at least 1 cm in diameter in 27 and determined that no nodules required aspiration in 23. Of 59 patients referred for a diffuse goiter or a multinodular gland, ultrasonography detected discrete nodules at least 1 cm in diameter that required aspiration in 39 and determined that aspiration was unnecessary in 20. CONCLUSIONS: Ultrasonography altered the clinical management for 63% of the patients (109 of 173) referred to the thyroid nodule clinic after abnormal results on thyroid physical examination.

Diagnosing lymphoproliferative disorders involving the cerebrospinal fluid: increased sensitivity using flow cytometric analysis.

Flow cytometric immunophenotypic analysis (FCA) can be performed to evaluate lymphoid cells in cerebrospinal fluid (CSF). We compared this method with conventional cytologic diagnosis to determine its utility. A retrospective comparison of 35 consecutive CSF flow cytometry results with the corresponding cytologic diagnoses was undertaken. Twenty-five of 35 CSFs (71%) were successfully analyzed by flow cytometry. The 10 samples which could not be analyzed were either too old (greater than 3 days) or had an insufficient number of cells. A total of 9 lymphomas was detected: 4 by both flow cytometry and cytology; 2 by cytology alone; and 3 by flow cytometry alone. This represents a 50% increase in the detection of lymphoproliferative disorders in CSF by a combination of flow cytometry and cytology vs. cytology alone. Furthermore, in 3 cases with follow-up where the cytologic diagnosis was "atypical cells of undetermined significance" and the flow cytometric findings were negative for malignancy, the clinical course confirmed a benign pleocytosis in all three. We conclude that flow cytometric analysis markedly improves sensitivity when used in combination with cytology in the evaluation of lymphoid cells in CSF.

Cytologic detection of Toxoplasma gondii tachyzoites in cerebrospinal fluid.

We reviewed our case records to see how often Toxoplasma gondii organisms were identified by cytologic evaluation of cerebrospinal fluid (CSF). During a 12-year period, 6,090 CSF specimens were examined, and 2 cases (0.03%) showed tachyzoites. Both patients were immunocompromised. One patient underwent lumbar and ventricular taps, and the other underwent only ventricular tap. Organisms were identified in the ventricular specimens but not in the lumbar sample. Both patients were treated, and subsequent ventricular CSF samples were negative. Toxoplasma gondii can be identified by cytologic examination of CSF. Our results confirm prior observations that in patients with obstructive hydrocephalus, tachyzoites are more likely to be found in ventricular rather than lumbar specimens.

Preclinical feasibility study of NMP179, a nuclear matrix protein marker for cervical dysplasia.

OBJECTIVE: To evaluate, in a preclinical feasibility study, the efficacy of NMP179, a monoclonal antibody recognizing a cervical tumor-associated nuclear matrix antigen, for the early detection of high and low grade cervical intraepithelial neoplasia. STUDY DESIGN: In a blind study involving two clinical sites, NMP179 immunocytochemical staining data from 261 cervicovaginal Thin-Prep specimens were evaluated. Assay sensitivity and specificity were calculated based upon a positive threshold of > 10 immunostained cells per case, using cytologic diagnosis as an end point. RESULTS: Based upon the examination of squamous epithelial cells, NMP179 detected 96.7% of cases with cytologically diagnosed high grade squamous intraepithelial lesions (HSIL) and 70.5% of low grade squamous intraepithelial lesions. The antibody also reacted with 29.6% of normal (within normal limits or benign cellular changes) smears. CONCLUSION: The NMP179 assay detected HSIL with very high accuracy (96.7%). The assay was 79.3% sensitive for the detection of low and high grade cervical intraepithelial neoplasia (grades 1-3), with a specificity of 70.4%. NMP179 may be an effective marker for the early detection of preneoplastic squamous intraepithelial lesions of the cervix and may be useful as an adjunctive tool for better management of cervical intraepithelial neoplasia.

LSIL biopsies after HSIL smears. Correlation with high-risk HPV and greater risk of HSIL on follow-up.

Can the risk associated with a high-grade cervical smear be disregarded when followed by a low-grade biopsy? We examined the distribution of human papillomavirus (HPV) types in such cases to see whether they segregated preferentially with low-risk or high-risk viruses and compared the distribution with that reported in the literature for women with high-grade squamous intraepithelial lesions (HSILs) and low-grade squamous intraepithelial lesions (LSILs). We identified 48 cases of HSIL smears with corresponding LSIL biopsy specimens. Biopsy specimens were tested and typed for HPV by polymerase chain reaction amplification with consensus primers followed by restriction fragment length polymorphism analysis, and HPVs were scored as low-risk or high-risk types. Thirty-seven cases scored positive for HPV DNA: 2 for low-risk HPV types, 17 for high-risk types, and 18 for types of unknown oncogenicity. The prevalence of high-risk HPV was significantly higher than that of low-risk HPV. There was a higher rate of high-risk HPV than that seen in historic unselected LSIL cases. Cases of HSIL cytology/LSIL histology represent a group distinct from unselected LSILs by virtue of their higher prevalence of high-risk HPV types and, therefore, warrant closer clinical follow-up.

Cytologic characteristics of abnormal cells in prior "normal" cervical/vaginal Papanicolaou smears from women with a high grade squamous intraepithelial lesion.

BACKGROUND: Some cervical/vaginal Papanicolaou (Pap) smears previously diagnosed as normal in women with a high grade squamous intraepithelial lesion (HSIL) are found to contain abnormal cells on retrospective review. This study characterized and quantitated such cells in 100 Pap smears. METHODS: One hundred prior negative Pap smears from 49 women with a newly diagnosed HSIL were rescreened, as were 107 negative Pap smears from 100 women with normal subsequent smears. Cases were assorted randomly so that the rescreener was unaware of the subsequent follow-up. RESULTS: All 12 Pap smears found to contain atypical mature and immature metaplastic cells belonged to the group with a subsequent HSIL (P = 0.001). In addition, 18 of the 100 previously negative Pap smears (18%) had immature metaplastic cells without nuclear atypia compared with 4 of 107 Pap smears (4%) in the control group (P = 0.0007). CONCLUSIONS: This study supports the observations of other authors that atypical metaplastic cells, especially those of the immature type, are associated with HSIL. These cells most likely are HSIL lesional cells, which are not easily recognizable as such. Immature metaplastic cells without atypia also were shown to be associated significantly with HSIL in this study. These cells may be unrecognizable lesional cells or a marker of increased risk for HSIL and deserve further study.

Performance characteristics of rapid (30-second) prescreening. Implications for calculating the false-negative rate and comparison with other quality assurance techniques.

Rapid (30-second) prescreening of cervicovaginal smears can be used to detect false-negative cases and determine the false-negative rate of primary screening, but the performance characteristics have not been evaluated fully. A test set of 242 cases including 80 originally false-negative cases were rapidly screened by 4 different cytotechnologists on 2 occasions. Intraobserver and interobserver reproducibility were good. Median specificity for each round of observations was 89% (range, 30%-96%). Median sensitivity for all true-positive cases was 78% (range, 63%-97%); for all false-negative cases it was 59% (range, 38%-89%). The relative sensitivity of rapid screening for true-positive and false-negative cases varied with the diagnosis. Rapid screening detected almost the same percentage of false-negative cases of atypical squamous cells of uncertain significance (ASCUS) as true-positive ASCUS cases (median ratio, 1.12; range, 0.72-1.52). The median ratio of false-negative to true-positive ASCUS cases was significantly different than the ratio for low-grade plus high-grade squamous intraepithelial lesions (0.68; range, 0.50-0.96). Although performance varies between individuals, in this test population the reproducibility, specificity, and sensitivity were good. Because it detects more false-negative cases at a lower cost per case than routine rescreening, rapid prescreening should be considered as an alternative to current quality control measures.

Accuracy and complication rate of brush cytology from bile duct versus pancreatic duct.

BACKGROUND: The accuracy and complication rates of brush cytology obtained from pancreaticobiliary strictures have not been fully defined. In this study we compared the accuracy and complications of brush cytology obtained from bile versus pancreatic ducts. METHODS: We identified 148 consecutive patients for whom brush cytology was done during an ERCP from a database with prospectively collected data. We compared cytology results with the final diagnosis as determined by surgical pathologic examination or long-term clinical follow-up. We followed all patients and recorded ERCP-related complications. RESULTS: Forty-two pancreatic brush cytology samples and 101 biliary brush cytology samples were obtained. The accuracy rate of biliary cytology was 65 of 101 (64.3%) and the accuracy rate of pancreatic cytology was 30 of 42 (71.4%). Overall sensitivity was 50% for biliary cytology and 58.3% for pancreatic cytology. Of 67 patients with pancreatic adenocarcinoma, sensitivity for biliary cytology was 50% versus 66% for pancreatic cytology. Concurrent pancreatic and biliary cytology during the same procedure increased the sensitivity in only 1 of 10 (10%) patients. Pancreatitis occurred in 11 (11%) patients (9 mild cases, 2 moderate cases) after biliary cytology and in 9 (21%) patients (6 mild cases, 3 moderate cases) after pancreatic cytology (p = 0.22). In 10 patients who had pancreatic brush cytology, a pancreatic stent was placed. None of these patients developed pancreatitis versus 9 of 32 (28%) patients in whom a stent was not placed (p = 0.08). Pancreatic cytology samples obtained from the head of the pancreas were correct in 13 of 18 (72%) cases, from the genu in 7 of 7 (100%) cases, from the body in 5 of 9 (55%) cases, and from the tail in 4 of 7 (57%) cases. CONCLUSION: The accuracy of biliary brush cytology is similar to the accuracy of pancreatic brush cytology. The yield of the latter for pancreatic adenocarcinoma is similar to that of the former. Complication rates for pancreatic cytology are not significantly higher than the rates for biliary cytology. The placement of a pancreatic stent after pancreatic brushing appears to reduce the risk of postprocedure pancreatitis.

Accuracy and reproducibility of estimating the adequacy of the squamous component of cervicovaginal smears.

In the Bethesda System for reporting cervicovaginal cytology results, 1 criterion for smear adequacy is an adequate squamous component. The accuracy of a cytologist's estimate that 10% of the slide is covered by squamous cells, the adequacy threshold, has not been determined. The percentage of the surface of a glass slide covered by squamous cells was independently estimated by 4 cytologists on 2 occasions by microscopic examination of 83 buccal smears prepared to display estimated coverage of 1% to 20% of the slide surface. The accuracy of visual estimates was compared with measurements by the TracCell System. Each observer made a third set of estimates after receiving 5 slides with known coverage. Median coverage by visual estimation ranged from 4% to 25%, but as measured by the TracCell system was 2%. Median estimated coverage was significantly different for 2 of 4 observers between first and second viewings and between all but 1 pair of observers. For all observers, it was significantly higher than the true coverage. A visual estimate of 10% coverage corresponded to a true median coverage of 3%. When provided with a physical standard, the median estimated coverage by 3 of 4 observers was not statistically different from the true coverage, and interobserver kappa values improved. Unaided visual estimation of the adequacy of squamous cell coverage is neither reproducible nor accurate. What most cytologists consider "adequate" coverage represents only 3% coverage. The availability of a physical standard dramatically increases reproducibility and accuracy.

Nontyrosine crystalloids in fine-needle aspiration specimens of the parotid gland: a report of two cases and review of the literature.

Three fine-needle aspiration biopsies from 2 patients with parotid masses yielded large numbers of nontyrosine crystalloids. One patient proved to have a benign cyst that was resected because it did not subside following two fine-needle aspirates. In a second patient, the swelling shrank after a course of antibiotics and fine-needle aspiration/drainage of the cyst. Recognition that crystalloids are associated with benign disease is important and should be considered in the management of these patients. The outcome of the patients studied here and of those previously reported has been benign. Conservative management of patients with parotid masses that contain nontyrosine crystalloids is indicated.

Negative predictive value of imaging-guided abdominal biopsy results: cytologic classification and implications for patient management.

OBJECTIVE: Our purpose was to assess the negative predictive value of imaging-guided abdominal biopsy results and correlate it with cytology classification, lesion size, needle gauge, and cancer history. MATERIALS AND METHODS: A retrospective study was performed of 100 patients with proven diagnoses who had undergone imaging-guided abdominal biopsies showing no cells that were malignant or suspicious for malignancy. Specimens were classified as normal or benign, nondiagnostic, or atypical. Negative predictive value was calculated for each cytologic category, lesion size, needle gauge, and cancer history. Logistic regression analysis was performed to allow us to identify predictors of false-negative results. RESULTS: Overall negative predictive value was 67%. Other negative predictive values were normal or benign result, 78%; nondiagnostic result, 66%; and atypical result, 29%. Negative predictive value was greater when the lesion was large (> or =3 cm) (p = .031). Logistic regression analysis allowed us to predict a 9.3% chance of a false-negative result for a specimen of normal or benign cytology that was taken from a large lesion in a patient with no cancer history but an 87% chance of a false-negative when a specimen of atypical cytology was taken from a small (<3 cm) lesion in a patient with a cancer history. CONCLUSION: Imaging-guided abdominal biopsy specimens containing atypical cells should be viewed with caution. In patients without cancer, if a lesion is large and the specimen contains normal target organ or benign cells, the likelihood of a false-negative result may be low enough that imaging surveillance at appropriate intervals may be sufficient.

Qualifying the cytologic diagnosis of "atypical squamous cells of undetermined significance" affects the predictive value of a squamous intraepithelial lesion on subsequent biopsy.

OBJECTIVE: To evaluate whether different qualifications of a cytologic diagnosis of "atypical squamous cells of undetermined significance" predict a greater or lesser likelihood of cervical pathology. DESIGN: Comparison of different cytologic qualifications of atypical squamous cells of undetermined significance with the frequency of significant cervical disease as documented by tissue biopsy. PARTICIPANTS AND SETTING: Four hundred, fifty-two consecutive Papanicolaou smears showing atypical squamous cells of undetermined significance (diagnosed by nine cytopathologists) in women who had undergone cervical biopsy within the previous 90 days at Brigham and Women's Hospital, Boston, Mass (January-June 1995). MAIN OUTCOME MEASURE: The histopathologic diagnosis of squamous intraepithelial lesion of the cervix. RESULTS: The 452 smears were qualified as "favor reactive" (22%), "not otherwise specified" (42%), "favor squamous intraepithelial lesion" (29%), and "favor high-grade squamous intraepithelial lesion" (6%). High-grade squamous intraepithelial lesions and total squamous intraepithelial lesions were pathologically confirmed by cervical biopsy in 3.6% and 6% of the favor reactive smears, in 11% and 21% of the not otherwise specified smears, in 12% and 30% of the favor squamous intraepithelial lesion smears, and in 53% and 59% of the favor high-grade squamous intraepithelial lesion smears. Significant associations were seen between a favor reactive smear and a benign finding on cervical biopsy (94%, P = .04) and between a favor high-grade squamous intraepithelial lesion smear and a biopsy that showed a high-grade squamous intraepithelial lesion (53%, P = .00001). CONCLUSIONS: Qualifying atypical squamous cells of undetermined significance stratifies women into different risk groups for squamous intraepithelial lesion. It is reasonable for physicians to make patient management decisions based, at least in part, on such qualifications.