Information, discrimination and divergence in cytology. II. Total discrimination as a measure of performance

Discrimination is the expectation of log odds ratios. Performance in diagnostic cytology, as compared to the results of histopathology or to a peer-reviewed consensus, can be measured by the total discrimination, which is a well-defined measure of information in information theory and test theory. The total discrimination, as a measure of performance, was calculated for the Gynecologic Cytology Laboratory of the University of Minnesota for the years 1985 through 1987, based on 3,545 sets of single-slide Papanicolaou smears and colposcope-directed biopsies. Similar calculations were made for the performances of physicians and technicians on peer-reviewed "clear-cut" Papanicolaou smears, as reported in a study from the Centers for Disease Control (CDC). For fair comparison of the two different sets of observations, the cytologic categories and corresponding histologic states were merged into five categories and states. The cytologic performance of our laboratory, tested against the histologic diagnoses, and the performances of the physicians and technologists tested against peer-reviewed check samples by the CDC were 0.53, 0.45 and 1.17 decits, respectively. These values generally agree with the conclusions derived by more conventional methods used by the CDC. When sampling errors of cytology specimens are taken into account, the performance of our laboratory, measured by the total discrimination of the cytology-histology confusion matrix, was approximately equal to the performance of the group of technologists studied by the CDC, which was in turn significantly better than that of the group of physicians studied by the CDC. This study demonstrates the value of using the total discrimination for quantifying the performance of a cytology laboratory, a cytopathologist or a cytotechnologist, without the arbitrary means usually used to evaluate such performances.     

Information, discrimination and divergence in cytology. V. General symmetry of total discrimination and total divergence.

Total discrimination and divergence are derived from Baye's theorem and based on backward (a posteriori) probabilities. Total forward (a priori) discrimination and divergence can be computed from a test matrix from which total backward discrimination and divergence were calculated for optimization of the classifications of gynecologic cytology (Papanicolaou smears) and quality control in a laboratory. The total forward discrimination and divergence appears to behave in parallel with the total backward discrimination and divergence, and the discrepancies in the backward and forward discrimination/divergence were the smallest near the optimum classification scheme. Conversely, the discrepancies between backward and forward discrimination and divergence may be helpful in finding the best classification scheme for gynecologic cytology. The general symmetries of the total backward and forward discrimination/divergence may be related to the human cognitive process of preferring symmetry and to the historic process in which cytologic classification followed histologic classification and continuously checked for matches in two directions--i.e., from histology to cytology and vice versa--thus resulting in the preservation of symmetry.     

Information, discrimination and divergence in cytology. I. Theoretical aspects and definitions

The results of medical tests update the probability of diagnosis (diagnosability) of a patient from the population prevalence of a disease. This paper demonstrates a method of combining several mutually dependent tests as the sum of log odds ratios, and of separating the log odd ratios of the test results and those of prevalences. Extending Kullback's discriminant function between two states to m states (m greater than or equal to 2), it is shown that the information of a test can be measured by discrimination and divergence, which are well-defined measures in information theory and test theory. The performance of diagnostic cytology can then be compared to the results of histopathologic diagnoses or to a peer-reviewed consensus without the use of arbitrarily given scores or without unfounded assumptions about underlying continuous variables. They can also provide a way to optimize the categorization (classification) of Papanicolaou smears and a means of quality control for morphologic tests. These methods can be used to evaluate the performances of cytology laboratories that use two classifications consisting of different numbers of categories (classes) and states.     

Information, discrimination and divergence in cytology. III. Optimization of classification of Papanicolaou smears

The performance of a cytology laboratory can be objectively quantitated as the total discrimination, a defined quantity of information. The total discrimination is dependent on the number of categories used in gynecologic cytology and on the corresponding histologic states; over-classification results in a higher rate of misinformation and reduced total discrimination. Total divergence is another measure of the association between cytologic categories and histologic states; in contrast to the total discrimination, the total divergence does not require a one-to-one correspondence between the cytologic categories and the histologic states. Using data from the Gynecologic Cytology Laboratory of the University of Minnesota, the total discrimination was maximized when gynecologic cytology used three categories of diagnosis, consisting of (1) normal, atypical benign or reactive atypia, (2) cervical intraepithelial neoplasia (CIN) and (3) all malignancies. The use of four categories, (1) normal, atypical benign or reactive atypia, (2) mild or moderate dysplasia, (3) severe dysplasia or squamous carcinoma in situ and (4) all malignancies, was almost equally informative. Observations on the total divergences resulted in similar conclusions. These findings generally support the recommendation of the consensus workshop sponsored by the National Cancer Institute (the Bethesda System nomenclature) to group all degrees of CIN into two large categories.     

Aspiration biopsy cytology and special stains

Histochemical methods play a vital role in the histology laboratory. The application of histochemistry to aspiration biopsy cytology (ABC) has shown equal promise in the demonstration of distinctive intracytoplasmic and extracytoplasmic substances. In our examination of more than 8,000 fine needle aspirates, we have found that certain histochemical studies may be of special diagnostic importance. These include mucicarmine, periodic acid-Schiff, oil red O, Fontana-Masson argentaffin and Gomori's iron stains. With the exception of the oil red O stain for neutral lipids (alcohol soluble), each may be applied to the destained smear initially prepared according to the method of Papanicolaou. Modifications of these histologic stains have been made to suit the "rapid service" motto of our busy cytology laboratory.     

Information, discrimination and divergence in cytology. VI. Biases and errors of measurement in small samples.

Total discrimination and total divergence have been shown to be useful measures of performance in diagnostic cytology. The sample size, Ns (observed number of cytology-histology pairs, essential components of a confusion matrix), may be small for the comparison of two or more laboratories or periodic quality control using observed values. From actual data from previous reports in this series, the best estimation of the confusion matrix of a population was obtained by fitting a Gaussian-type curve after pseudoscalar transform of ranks (row and column numbers). Small sample confusion matrices were generated by Monte Carlo simulation to 2.5 x 10(-7) resolution. To keep measurement biases within +/- 0.5 decits, we found that 100-200 samples of cytology-histology pairs were required in the best classifications of three, four and five category-states. At these sample sizes, measurement errors (standard errors) were also contained within +/- 0.5 decits. This study also confirmed that previously reported overestimated propagated errors in small samples were in fact overestimation and that their use for testing a null hypothesis was valid. The number of samples with indefinable statistics due to a zero denominator can be as high as 30% when the sample sizes are 500 for three, four and five category-state classifications. Biases due to small samples were positive for most category-states except for the optimum three category-states, in which bias changed to negative ("bias inversion"), and observed errors of discrimination and divergence paradoxically decreased as Ns decreased ("error-sample paradox") for a small sample size (Ns less than 700).(ABSTRACT TRUNCATED AT 250 WORDS)     

Fine needle aspiration cytology in the management of male breast masses. Nineteen years of experience.

OBJECTIVE: To determine the utility and accuracy of fine needle aspiration cytology (FNAC) as well as its sensitivity, specificity and predictive value in the diagnosis of male breast masses. STUDY DESIGN: Data on male breast FNAC done between 1978 and 1997 were retrieved from the records of the cytopathology laboratory. FNAC diagnoses were categorized as positive, negative, inconclusive or unsatisfactory. Cytohistologic correlation was done with data from histopathology records. Sensitivity, specificity, diagnostic accuracy and predictive values of FNAC were calculated using standard statistical methods. RESULTS: Five hundred seven of 13,175 patients undergoing breast FNAC were males. Of them, 393/507 had satisfactory aspirates. Of these, 70 were positive (13.8%), 295 were negative (58%), and 29 were inconclusive (5.7%). A total of 114 FNACs (22.5%) were unsatisfactory. Histopathology was available in 97/507 cases. There were no false positive or false negative diagnoses. FNAC had a sensitivity, specificity and diagnostic accuracy of 100% for male breast lesions. CONCLUSION: This large study shows that FNAC is a very accurate tool for diagnosis of male breast lesions. It is highly sensitive and specific, with good cytohistologic correlation. FNAC should therefore be an integral part of the primary assessment of breast lumps in males.     

Mucosal cytology in the determination of the pubertal status in girls

AIM: To assess and validate the diagnostic value of buccal and vaginal smears in the ambulatory care of girls with potential disorders of puberty. METHODS: Smears were obtained from 77 girls who presented for assessment of their pubertal status, stained as described by Papanicolaou, examined for signs of estrogenization according to the method of Schmitt, and compared with the clinical status. RESULTS: Vaginal but not buccal smears reflect accurately the changes of sexual maturation, even more sensitive than single serum hormone measurements. CONCLUSIONS: Vaginal smears represent a valid, reproducible, and well-tolerated diagnostic tool in the ambulatory care of peripubertal girls to identify their estrogen status with high sensitivity and specificity. The decision to perform confirmative but invasive diagnostic procedures can be based on auxology, physical examination, bone age determination, and analysis of vaginal smears.     

Cervical cytology quality assurance in Washington State.

The objectives of this study were to establish a profile of cervical cytology laboratories in Washington State, identify quality assurance problems amenable to correction through education or legislation, and describe differences between large and small cytology laboratories. All 43 Washington laboratories that perform cervical cytology were surveyed by mail during 1989. Completed surveys were returned by 37 (86%) of the laboratories. Nearly half (43%) of the respondents reported processing less than 10,000 Papanicolaou smears annually. Only one-third (35%) of the respondents reported participating in relevant proficiency programs. A proportion of smaller cytology laboratories were compensating their cytotechnologists on the basis of the number of slides read and allowing Papanicolaou smears to be read outside the confines of the laboratory. The results of this study suggest that cytotechnologists in some larger Washington laboratories have been exceeding work load limits recommended by professional associations. Recent legislation includes regulations that address cervical cytology quality assurance. However, continued efforts will need to be made to encourage voluntary adoption of quality control measures not addressed by this legislation.     

Liquid-based cytology in breast fine needle aspiration. Comparison with the conventional smear

OBJECTIVE: To compare the various cytologic features on AutoCyte Prep (ACP) (AutoCyte, Inc., Burlington, North Carolina, U.S.A.) and conventional preparation (CP) specimens from breast fine needle aspiration cytology material with a semi-quantitative scoring system. STUDY DESIGN: A total of 100 randomized cases were studied. In each case, 2 passes were performed. One pass was used for CPs (Giemsa and Papanicolaou stain). The other pass produced material for the ACP technique and Papanicolaou stain. Both the conventional and liquid-based preparations were studied independently by two observers and compared for cellularity, obscuring and/or informing background, representative diagnostic material, preservation of cytomorphologic features, presence of monolayer cells and architectural arrangement. RESULTS: Comparing the two preparations, the results were as follows: (1) ACP was superior to CP in 2 features, lack of obscuring background and presence of monolayer arrangement with preservation of cell architecture; (2) ACP was inferior to CP in 1 feature, lack of informing background; and (3) ACP was equal, with small deviations, to CP in the rest of the features evaluated: cellularity, representative diagnostic material, preservation of cell morphology and architectural arrangement. CONCLUSION: The new technology of liquid-based cytology in breast FNA showed a good correlation with CP plus the advantages of: (1) easier and less time consuming evaluation of cell morphology (clear background, no overlapping, smaller area to screen); (2) reproducibility, a factor of great importance to quality control; and (3) possibility of adjunctive investigations (immunocytology, flow cytometry) on the same material.     

Fine needle aspiration cytology of the thyroid: a cytohistologic study with evaluation of discordant cases

OBJECTIVE: To analyze the diagnostic efficacy of fine needle aspiration cytology (FNAC) in the initial evaluation of thyroid nodules, with special emphasis on discrepant cytologic diagnoses. STUDY DESIGN: A total of 192 thyroid fine needle aspirates with subsequent histopathologic follow-up were analyzed. The cytologic diagnoses were divided into 4 categories: positive for malignancy, negative for malignancy, indeterminate for diagnosis and nondiagnostic. The detailed cytologic features were studied along with histopathology sections in all these cases by 2 observers (S.J. and P.D) independently. RESULTS: Cytohistologic correlation was seen in 78.1% of cases and discordance in 21.9%. Indeterminate diagnoses accounted for 15.1% of cases. The majority of these were "follicular neoplasms." The overall sensitivity was 84.44% and specificity 99.11 %. A false positive diagnosis was made in 1 case (0.5%), proven a follicular adenoma on histopathologic examination. A false negative diagnosis was seen in 3.6% of cases. These were cases of papillary microcarcinoma. CONCLUSION: FNAC is a safe, sensitive and specific technique in the initial evaluation of thyroid nodules. A correct cytologic diagnosis can be achieved in a majority of cases, thus obviating the need for a second surgical intervention. A careful and diligent search for various cytologic features and accurate sampling can help in reducing the number of indeterminate, false positive and false negative diagnoses.     

Diagnostic performance of United States Air Force cytology laboratories. An eight-year experience in central monitoring

Based on experience obtained through review of historical case material, the cytology program of the United States Air Force was regionalized into ten cytocenters, which annually process approximately 375,000 gynecologic cases. In order to assure uniformity and quality of diagnosis to multiple supplying clinics, standardized diagnoses were developed with required central reporting of all abnormal recovery rates. Initial results showed a sign-out abnormal rate (class II or worse) ranging from 1% in two cytocenters to nearly 6% in two others--over a fivefold variation on a demographically similar population. Variation by diagnostic class and follow-up tissue biopsies are reported to validate the higher abnormal rates. Through an applied program over the last seven years of formal education of pathologists and cytotechnologists, consultative visits and sharing of comparative referred statistics, the subsequent abnormal rates were substantialy altered to a range of 3.5% to 7.0%. More recently, required reporting of abnormal rates per technologist appears to be another monitor of laboratory quality, in that a given cytocenter may have a total abnormal rate of 4.5%, but a range among cytotechnolgists of 1.3% to 8.7%. These and other presented statistics may prove valuable for external monitoring (including assessment of quality performance) of individual cytology laboratories.     

Imprint cytology of core needle biopsy specimens of breast lesions. A rapid approach to detecting malignancies, with comparison of cytologic and histopathologic analyses of 173 cases

OBJECTIVE: To investigate whether imprint cytology of core needle biopsy (CNB) specimens from breast lesions is a useful method of rapidly obtaining additional diagnostic information and potentially can be used to reduce the number of biopsies needed. STUDY DESIGN: Cytologic analysis was performed on 173 breast lesions and compared with their histopathologic diagnoses (143 malignant and 30 benign). For imprint cytology, one CNB specimen was rolled between two slides and stained with Diff-Quik and Papanicolaou stain. RESULTS: The diagnostic overall accuracy of Diff-Quik stain (Papanicolaou stain) was 95.4% (95.9%), with a sensitivity of 96.5% (97.2%), specificity of 90% (90%), positive predictive value of 97.8% (97.8%) and negative predictive value of 84.3% (87.0%). There was no statistically significant difference between the stains. Histopathologic analysis had an overall accuracy of 97.7%, with a sensitivity of 97.2%, specificity and positive predictive value of 100% and a negative predictive value of 88.2%. CONCLUSION: Imprint cytology of CNBs is a sensitive method of detecting malignancies in breast tumors. Diff-Quik is a rapid and reliable approach that can reduce the number of biopsies. Inadequate and suspicious cases should be evaluated based on complementary diagnostic procedures for breast lesions.     

Interobserver variability in cytologic subclassification of squamous intraepithelial lesions--the Bethesda System vs. World Health Organization classification

The aim of the study was to compare interobserver variability for The Bethesda System (TBS) and World Health Organization (WHO) classification of cervical squamous intraepithelial lesions. A total of 1,000 conventional Papanicolaou smears (156 positive and 884 negative) were examined "blindly" by three cytologists and one cytotechnician. The degree of observer agreement was expressed by kappa statistics using a program for the calculation of interobserver variation and association "Agree" (Svanholm and Jergensen, 1989). Kappa (kappa) was determined for each cytologic diagnosis within a particular classification and total for either classification. The association with and separation from other diagnoses was determined for each cytologic diagnosis in the form of conditional probability (P(j)). In WHO classification, the diagnoses of dysplasia media and dysplasia gravis showed poor reproducibility (kappa = 0.114 and kappa = 0.259, respectively), the diagnosis of dysplasia levis good reproducibility (kappa = 0.639), and the diagnosis of carcinoma in situ excellent reproducibility (kappa = 0.762). WHO classification yielded pool kappa of 0.741. In TBS classification, the diagnosis of LSIL showed good, and HSIL excellent reproducibility (kappa = 0.542 and kappa = 0.763, respectively). TBS classification yielded pool kappa of 0.699. Dysplasia media (P(j) = 0.121) and dysplasia gravis (P(j) = 0.274) were found to be morphologically poorly defined, and carcinoma in situ (P(j) = 0.777) and dysplasia levis (P(j) = 0.651) well defined diagnoses. LSIL was morphologically moderately defined (P(j) = 0.587) and HSIL well defined (P(j) = 0.789) diagnosis. Accordingly, TBS does not substantially improve diagnostic reproducibility of the cytologic diagnoses of squamous intraepithelial lesions, while providing considerably less information to the clinician than the four-grade dysplasia/CIS terminology, thus eliminating the opportunity of choosing a different procedure for the diagnosis of dysplasia media, which is of utmost importance in the population of young nulliparae.     

Value of image-guided needle aspiration cytology in the assessment of pelvic and retroperitoneal masses. A study of 112 cases

OBJECTIVE: To evaluate the diagnostic value of the noninvasive method of image-guided needle aspiration cytology (NAC) in the assessment of radiologically detected pelvic and retroperitoneal space-occupying lesions (excluding the pancreas, kidney and adrenal). STUDY DESIGN: NAC was performed under computed tomographic or ultrasound guidance on 112 patients suspected of having a pelvic or retroperitoneal mass. Cytologic examination was performed on site after staining smears with the Papanicolaou method. In addition, air-dried smears, fixed smears, filter preparations from needle washings and cell blocks were studied. The NAC diagnosis was supported by examining cell blocks; further support was obtained with a tissue biopsy in some cases. Additionally, pertinent immunoperoxidase and/or histochemical studies were done. RESULTS: Eighteen cases were diagnosed as inflammatory lesions, 17 cases consisted of normal cellular elements, and 12 cases showed scanty material and were considered unsatisfactory/inadequate for a diagnosis. Five cases were suspicious for malignancy, and in 39 cases metastatic tumors were diagnosed from a previously known primary. Thirteen cases were diagnosed as lymphoma, and in 8 cases a diagnosis of soft tissue sarcoma was made. There were no false positive diagnoses of malignancy. Cell block preparations and immunohistochemistry were helpful with tumor typing, although lymphoma subtyping and soft tissue tumor typing generally required open biopsy. CONCLUSION: NAC, as the first-line investigation, is not only useful in the diagnosis of space-occupying lesions of the pelvic and retroperitoneal region but can also help in choosing appropriate management. The technique is most useful in diagnosing metastases but is also helpful in excluding malignancy in some cases and in suggesting diagnoses of lymphomas and soft tissue tumors.     

Utility and diagnostic accuracy of fallopian tube touch imprint cytology

OBJECTIVE: To determine the diagnostic accuracy of cytology smears in distinguishing between tube and non-tube structures. METHODS: One hundred cytology smears of fallopian tube and non-tube structures (vessels, round and ovarian ligaments) were prepared from surgically removed uterus and fallopian tube specimens and stained by the Papanicolaou method. The slides were reviewed blindly by pathologists and interpreted as tube or non-tube structures. The results were compared to the histological examination of the same specimens. FINDINGS: Results indicated an overall accuracy of 97% with a specificity of 98% and sensitivity of 96% for cytology smears, taking histology as the gold standard. Positive and negative predictive values were 96.1% and 97.9%, respectively. CONCLUSION: Cytology smears are a convenient and cost effective tool for laboratory confirmation of tubal sterilization. This method can reduce the costs of laboratory examination, especially in developing countries, where tubal sterilizations are done in large cohorts. However, histological slides remain the gold standard in cases of medicolegal problems.     

Quantitative cytology of the positive region in flow sorted vaginal smears.

Fifty-one gynecologic specimens were collected from three women's hospitals and mailed in a prefixed status to our laboratory. The specimens were classified into a negative, a suspicious, a postradiation, and a positive group. After single cell dispersion the samples were stained for DNA and protein, analyzed, and sorted in the dual laser equipped Heidelberg flow analyzer sorter (HEIFAS). Particles with elevated DNA values (beyond 3.5 ploidy) and with intermediate protein values were sorted as the positive fraction directly on microscopic slides. After restaining according to Papanicolaou, they were re-evaluated cytologically and identified as tumor cells, dysplastic cells and false alarms. The latter consist of doublets and aggregates of more than two cells, binucleated cells, sperm aggregates and epithelial cells contaminated with bacteria. The different groups showed significant differences regarding the total rate of aggregates to single cells. In general, false alarms were very frequent in the positive region and impeded the statistical classification of the sample. The reduction of false alarms is a prerequisite for prescreening with flow instrumentation.     

Biliary tract brush cytology

OBJECTIVE: To prospectively review brush smears obtained during endoscopic retrograde cholangiopancreatography (ERCP) primarily from the biliary tree. STUDY DESIGN: A total of 175 specimens from 147 patients were included in the study. The smears, prepared directly from the endoscopic brush, were stained by the Papanicolaou technique and analyzed for standard cytologic features. RESULTS: The smears were categorized into benign/reactive, significant atypia and suspicious/positive. The consistent features seen in suspicious or positive smears were tightly cohesive, small, three-dimensional cell clusters that formed cell balls. The cells in the clusters displayed features of malignant cells. CONCLUSION: ERCP-guided brushing is a safe diagnostic procedure for the evaluation of biliary tree lesions. Small, three-dimensional epithelial clusters with marked atypia signify malignancy and warrant the diagnosis of a malignant neoplasm even when only one or two such clusters are seen in the smears. Single cells, cytoplasmic vacuoles and prominent nucleoli are not essential for a diagnosis of malignancy.     

Large-volume cytocentrifuge for processing alcohol-fixed cytologic specimens. Application in urinary cytology

The use of a large-volume cytocentrifuge for the routine processing of urine specimens was investigated. Most specimens arrived in the laboratory mixed with 50% ethanol. Polyethylene glycol was added to the fixative supplied to the clinicians or to the cell suspension received in the laboratory. The slides used in the cytocentrifuge were coated with gelatin or poly-L-lysine to minimize cell loss. The resulting monolayers of cells, with occasional true tissue fragments, were stained by the Papanicolaou method. When indicated, immunocytochemical and histochemical stains were used. The technique gave good morphologic details, good cell yield and consistent staining quality. This method has also been applied to other cytologic specimens, such as serous fluids and fine needle aspirates, with good results. The method saves the time of cytotechnologists, and slides prepared by this technique are suitable for automatic staining.     

Haemorrhagic cytology samples – how to get the best diagnostic results

OBJECTIVE: To describe and evaluate the value of a simple filtration technique for use in processing haemorrhagic samples for cytomorphological evaluation and immunocytochemistry. METHODS: One hundred and sixty haemorrhagic cytological samples (133 FNAs, 27 effusions) received in our laboratory from August 2002 to September 2005 were included in this study. After preparing two smears for diagnostic evaluation, the residual sample was suspended in 2 ml of a cell medium prepared in our laboratory. These primary haemorrhagic suspensions were filtered through disposable nylon filter devices and the cells deposited on the upper membrane surface were transferred into 2 ml of fresh cell medium. From all three fractions - primary cell suspension, filter deposit and filtrate - cytospins were prepared and stained by Giemsa or Papanicolaou methods. Cytospins were examined under the microscope for the presence of diagnostic cells, red blood cells (RBCs) and debris. Additional cytospins for immunocytochemistry were prepared at the cytopathologist's request. RESULTS: RBCs and debris were successfully removed in 142 out of 160 haemorrhagic samples (88%) by using this new filtration technique. In all these cases the tumour cells were well presented and allowed substantially improved cytomorphological evaluation. Immunocytochemical staining was performed on 112 filtered samples with three different markers per case on average. Filtration did not improve the quality of cytospins in 18/160 haemorrhagic samples, mostly attributable to insufficient numbers of diagnostic cells in the original samples. CONCLUSION: The presented filtration technique is very simple and quick. It substantially improves the quality of haemorrhagic samples for cytomorphological evaluation; moreover, the samples are well suited for multiple immunocytochemical stainings.