Giorgio Gherardi, M.D., Cristina Marveggio, B.D., Stefania Rossi, M.D.
Between January 1994 and December 1996, 1250 patients were referred to this Institution for FNA evaluation of breast masses. Of these, 65 cases showed cytologic changes consistent with breast borderline lesions. Immediate evaluation of the harvest from first passes allowed preparation of an extra slide to be processed for DNA image analysis in 47 cases. All these cases underwent subsequent surgical excision of the lesion which was histologically evaluated in this Institution. The age ranged between 33 and 81 (mean age 50). Thirty-one patients were premenopausal. In all cases X-Ray mammograms and ultrasound imaging data were available. Most lesions were easily palpable. Their size ranged between 0.4 cm and 1.6 cm. Follow-up information of at least 12 months after surgical excision of the lesion was available in all cases.
Sample preparation and processing
FNA biopsy was performed by the pathologist in all cases by means of 23 and/or 25 Gauge needles attached to a 20 mL suction syringe. Smeared cell samples were either fixed in 95% ethanol, for Papanicolaou stain, or air-dried, for Romanowsky stain. Immediate evaluation of an ethanol fixed slide after staining with a quick Papanicolaou staining method allowed evaluation of the adequacy of the harvest and, in the cases under study, prompted performance of an additional aspiration sampling in order to obtain extra slides to be used for DNA-image analysis and cytologic scoring. These latter were prepared with a "pull-apart" smearing technique which provides two slides with comparable content ( 20 ); the smearing slide was fixed in ethanol 95% for Papanicolaou staining while the stationary slide was processed for DNA-image analysis. The former was used to assess specimen adequacy and for subsequent cytologic score evaluation. The latter was air-dried for 30 minutes, then fixed in 10% neutral buffered formalin for 30 minutes, washed in phosphate buffered saline and stained according to the Feulgens method (CAS DNA staining kit, Cell Analysis System, Elmhurst, IL, U.S.A.). Touch preparation slides were obtained from surgical excisional biopsy specimen of each case and were processed for DNA-image analysis according to the same procedure.
Cytologic grading system and histologic evaluation
The cytologic score evaluation was performed on the Papanicolaou stained smearing slide obtained with the "pull-apart" procedure ( 20 ). The system we used for scoring the aspirates was introduced by Masood et al ( 6 ) and is based on evaluation of six cytologic parameters, each being scored separately. They consist of: cellular arrangement, cellular pleomorphism, myoepithelial cells, anisonucleosis, nucleoli, chromatin clumping. Each component is scored from 1 to 4 and are added up to give a total cytologic score. On surgical pathologic specimens histologic distinction of florid hyperplasia without atypia (FH), atypical ductal hyperplasia (ADH), and ductal carcinoma in situ (DCIS) was based on the Guidelines indicated by Page and Rogers ( 21 ).
DNA image analysis
For quantitation of the DNA staining, the CAS 200 Image Analysis System was used. The principles of this system as well the accuracy and reproducibility of results have been described ( 22,23 ). Quantitation of DNA relies on the Feulgen´s staining reaction to specifically and quantitatively stain DNA. Cytological samples offer the best environment for cytophotometric ploidy analysis since there is no nuclear sectioning artifact and overlap. The instrument was calibrated by measuring 20-50 rat hepatocytes with every stained batch of slides. Slides were stained with azure A according to the method reccomended by Becton-Dickinson ( 24 ). Based on the known ratio of DNA content between rat and human epithelial cells, the computer calculates the expected sum of the optical density of normal human cells. In this video-based interactive image cytometer the operator can scrutinize and classify the nuclei according to their morphologic characteristics. This allowed evaluation of nuclear DNA content of stromal cells and lymphocytes that served as internal diploid control from each slide. Histograms were classified as diploid if more than 70% of cells had the DNA index (DI) in the range of 0.9 to 1.1 compared with the diploid standard. The percentage of cells in the S+G2/M area of the histogram was calculated for diploid cases. Tetraploidy was diagnosed if at least 20% of the cells had a DI in the range of 1.9 to 2.1. Aneuploidy was diagnosed for cellular proliferations having other major peaks containing >=20% of the cells with DI in the range between 1.11 to 1.89, or >10% of the cells with a DI <0.9 or>2.1.
The negative and the positive predictive value (NPV and PPV) of some parameters was calculated by using the following formulae:
NPV = true negatives / true negatives + false positives
PPV = true positives / true positives + false positives