Abstract
Aim: To investigate the validity and reproducibility of touch imprint cytology (TIC) of core needle biopsies (CNBs) in the assessment of conspicuous breast findings. Materials and Methods: One hundred and fifty-eight conspicuous breast findings were assessed by CNB during a two-year period in this prospective study. TIC of the CNBs was used for cytological analysis. The performance of this technique was evaluated by comparing the results with the histology of the CNB, and in cases of a subsequent surgery, with the final histology. In addition, inter-observer reproducibility was analyzed. Results: The inter-observer variability was very high (kappa-value of 0.8508; excluding inadequate imprints, 0.9502). Inadequate cellularity for cytological analysis was found in 11.4% (18/158) of findings. Compared with CNB histology, the sensitivity, specificity, positive predictive value, negative predictive value and overall accuracy of TIC was 99%, 100%, 100%, 94%, and 99%, respectively. Conclusion: TIC of CNB specimens of the breast may be a valid option for providing a diagnosis without delay for a histological procedure, assuming good quality of the specimen.
Conspicious findings in the breast are assessed with clinical examination, imaging and finally, if warranted, through invasive measures, removing tissue for microscopical examination. The processing of a tissue specimen in the histopathological laboratory needs a certain amount of time, making a same-day diagnosis practically impossible (1-3). With fine-needle aspiration cytology (FNAC) it is possible to inform the patient about the diagnosis on the same day, avoiding a waiting period many patients find distressing. However, there is strong operator dependence not only in gaining the material, but also in the actual microscopic assessment. In comparison with the histopathological assessment of core needle biopsies (CNB), FNAC is less sensitive and specific (4-11). Therefore, FNAC is not recommended for the preoperative establishment of the diagnosis of breast cancer in German guidelines (12). Whether touch imprint cytology (TIC) taken from CNB specimens is suitable for immediate cytological evaluation has been recently studied (13-16, 4-11, 17-20). This method would provide a good compromise between getting a fast preliminary result to the patient and at the same time allowing a very sensitive diagnosis through the histopathological assessment of every case. The results of these studies were very promising because they showed the high sensitivity and high specificity of imprint cytology (17-19, 11, 8, 20). In Germany, cytological assessment of breast masses is usually not performed, and according to guidelines, histopathological assessment is mandatory (12). However, the concept of imprint cytology of CNB specimens seemed promising enough to warrant a prospective study. The aim of the present study was, therefore, to investigate the validity and reproducibility of TIC of CNBs in the assessment of conspicuous breast findings.
Materials and Methods
In a two-year period between November 2008 and November 2010, 158 conspicuous findings from 143 patients were assessed at the Breast Cancer Centre of the Department of Obstetrics and Gynaecology, University Medical Centre Mannheim, Heidelberg University, Mannheim, Germany. After diagnostic work-up, biopsies for conventional histopathological diagnosis were obtained. These biopsies were taken under ultrasonographic guidance using a disposable, spring-loaded biopsy system with a 14-gauge needle (Achievee; Allegiance, UK). Up to five needle cores were taken, which were all rolled on microscopical glass slides to obtain TIC specimens. The slides were air-dried and then stained using a quick stain (21) (Table I, and Figure 1). For reporting of the results, the five categories of the National Health Service breast screening program guidelines (NHSBSP) were used (22): C1, Inadequate; C2, benign; C3, atypia probably benign; C4, probably malignant; C5, malignant. Imprints categorized as C5 or C4 were considered malignant, while imprints classified as C2 or C3 were scored as benign. The cytological evaluation was carried out separately by two investigators knowing only the age of the patient and the clinically-suspected diagnosis; the same information was provided to the pathologist performing the routine evaluation of all CNBs. One physician evaluated all 158 specimens, whereas the second observer only the first 58 cytological specimens for quality control. Inter-observer reproducibility was assessed by the kappa coefficient providing a measure for agreement beyond chance. The maximum value of 1 shows perfect agreement, whereas a kappa value of 0 indicates that there is no agreement better than that from chance. Furthermore, the cytological results were compared to the histological diagnosis of the CNB and the final histological diagnosis of the resection specimen. For these two reference methods, sensitivity and specificity, as well as kappa values and predictive values, were calculated together with the corresponding confidence intervals (CI). All these measurements were analysed separately for both cytology observers, using all results deemed adequate (C2-C5). To assess the validity of the CNB results compared to the results of the final histology of the resection specimen, sensitivity, specificity and kappa values for this comparison were also calculated. Additionally, the demographic data and histological characteristics were evaluated. SAS release 9.3 (SAS Institute Inc.; Cary, NC, USA) was used for statistical evaluation.
Quick stain process used.
Results
All patients were females between 17 and 89 years old at the time of biopsy (average=58.6±16.5 years). Of the 158 lesions investigated by CNB, 109 (69.0%) were malignant and 49 (31.0%) benign, on histological examination. In 109 cases, surgical resection was performed, providing a final histopathological result based on the resection specimen. In 94 cases with subsequent surgery, the final histology report showed a malignant result (86.2%), and in 15 cases, a benign result (13.8%). Further histological characteristics are provided in Table II. The TIC results of the first and the second investigator in comparison to these from CNB are shown in Table III. As investigator 2 reviewed considerably fewer cases, the estimated measurements are less precise, resulting in broader CIs. The inter-observer variability for the categories “malignant”, “benign” and “inadequate” gave a kappa-value of 0.8508 (95%CI=0.7303-0.9714). If one excludes the category “inadequate” from the calculation, then the kappa-value is 0.9502 (95%CI=0.8536-1.0000). Table IV shows the comparison of the histology result of the CNB and of the surgical specimen, respectively, with the cytology result of both investigators. The CNB accurately predicted the final histology in approximately 99% of instances. Interestingly, one of the CNBs that failed to predict the final histological result (an invasive ductal carcinoma with apocrine differentiation categorized as benign on CNB) was classified as malignant (C5) on the corresponding TIC. Cytology was deemed as inadequate in 18 out of 158 cases by investigator 1 (11.4%) and in 11 out of 58 cases (19.0%) by investigator 2 (Table III). A higher percentage of benign cases were rated as inadequate. In the 18 cases of investigator 1, seven were malignant (ductal, mixed ductal and lobular or lobular carcinoma) and 11 benign (fibrosis, fat necrosis, fibroadenoma, fibro-cystic change, mastitis, ductal papilloma) as classified by CNB. In the 11 cases of investigator 2, four were malignant (ductal, mixed ductal and lobular or lobular carcinoma) and seven benign (fibroadenomas, fibrosis, fat necrosis, fibrocystic change). The two false-positive cytology results of investigator 1 were fibrosis of the breast, and mastitis and fibroadenoma. One case was rated malignant on cytology by investigator 1, but regarded as benign fibrosis on CNB; the final histological result confirmed the malignancy as invasive ductal carcinoma. Investigator 2 also rated this case as malignant on cytology. False-negative cases exhibited a very low number of cells on cytology; on histology, these cases were diagnosed as lobular or invasive carcinomas, or ductal in situ carcinoma. For investigator 2, no false-positive or false-negative results were found, since the case being false-positive in comparison with CNB, and was proven to be truly malignant on definitive histology.
Histopathological characteristics.
Touch imprint cytology of a malignant lesion (invasive ductal carcinoma). The black arrow shows a solitary mitosis.
Discussion
Cytological assessment of conspicuous breast findings is not in widespread use in Germany. But the possibility of a rapid cytology-based test on cells obtained from CNB with subsequent histology combines the advantages of both methods. In the present study, the reproducibility and the validity of this approach was tested. The proportion of malignant findings is very high in the study cohort because of the characteristics of patients referred to a breast cancer centre of a university hospital. In the final histological diagnoses, the percentage of malignant cases is even higher because many cases with a benign result on CNB will not go on to surgery. In line with palliative care, some patients with malignant findings will forego an operation, so that a final histological result is missing in some cases. In the present study we used a quick stain method established in an earlier study on touch imprint specimens of breast resections (21), whereas in most other studies, staining was carried out using Diff-Quick (17-19, 11, 8, 20). Our results showed an excellent sensitivity of 91.9% for the detection of a malignant tumors at a somewhat lower specificity (85.7%) on the basis of the final histology of the surgical specimen. Only a few benign findings, e.g. fibroadenomas, were difficult to interpret on cytology due the large number of cells. After adding Mayer's haematoxylin to the staining solution it was easier to discriminate between benign and malignant cases due to a better demonstration of myoepithelial cells. The percentage of results deemed inadequate in this study was 11.4% for investigator 1 and 19.0% for investigator 2. These values are comparable to those of other studies showing an overall inadequacy rate for TIC of 0 to 38% (mean=13%) (6, 7, 23, 8, 11). Inadequate samples are found both for benign and malignant cases, but more often for benign lesions (24, 25, 15, 26). In accordance with other trials, the number of samples rated as inadequate in this study was higher in benign cases. The high percentage of those for investigator 2 may be due to the fact that only the first 58 samples were evaluated; during the course of the study, the quality of the samples improved considerably. The correct rolling of the cores on the slide, good staining quality and a sufficient number of cores were important for good results (10, 8). The sensitivity for malignancy in this study was high, which is comparable to similar studies (8, 24, 14, 16, 15, 26, 20). For certain types of histology, e.g. invasive lobular carcinoma, false-negative results were more frequent in our study; this was also found in other studies (11). For benign results, especially fibroadenomas in our study, the evaluation can be difficult (8, 24, 16). The cytological appearance of a rapidly-dividing fibroadenoma and papillomatosis are known to very closely resemble malignant cytology. While investigator 2 had no false-positive results, investigator 1 reported three false-positives. The comparison to investigator 2 shows that in the case of low quality preparation/staining and a low amount of material, one should refrain from the evaluation of the specimen. In earlier studies, the percentage of false-positive results ranged between 0 and 32% (8, 24, 14, 16, 15, 26, 6, 7, 23), showing the utmost importance of the need for high quality of specimens. The inter-observer variability of the specimens deemed adequate for evaluation was very high (kappa-value of 0.9502). In conjunction with clinical and radiological findings as part of triple assessment, TIC is an option very quickly providing a diagnosis to the patient. Critical for its implementation into routine practice may be the lack of cytopathologists trained in this method. Future telemedicine applications may provide new possibilities, but this has to be investigated in further studies.
Conclusion
TIC of CNB specimens may be a realistic option for providing a diagnosis without the need for delay for the histological work-up, assuming a good quality of the specimen. The implementation into routine practice and the limitations have to be evaluated in future studies.
Touch imprint cytology (TIC) results of both investigators in comparison to core needle biopsy (CNB) histology. The numbers are presented together with relative frequencies in parentheses; measurements such as sensitivity are given with the corresponding 95% confidence intervals.
Comparison of touch imprint cytology (TIC) and core needle biopsy (CNB) histology and the final histology of the surgical specimen. The numbers are presented together with relative frequencies in parentheses; measurements such as sensitivity are given with the corresponding 95%confidence intervals.
- Received October 31, 2013.
- Revision received November 14, 2013.
- Accepted November 18, 2013.
- Copyright© 2014 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.