Giorgio Gherardi, M.D., Cristina Marveggio, B.D., Stefania Rossi, M.D.
In the current study we performed DNA-image analysis of FNA cytologic samples illustrating the features of borderline breast lesions and tried to correlate ploidy and cell kinetics data with cytologic score in order to assess the role, if any, of this ancillary technique as a diagnostic tool in equivocal cases. The semiquantitative measurement of the six cytologic parameters ( 6 ) served as a basis to grade cytomorphological changes of borderline breast lesions into three groups of samples with different grades of atypia. This categorization correlated well with the final histologic diagnosis in that in the samples with minor atypia (score value <15) no malignancy was detected and, conversely, no benign lesion fell in the group of samples with marked atypia (score value>16). Premalignant lesions were scored between 14 and 17 and prevailed in the group of cases with intermediate cytologic atypia. Based on these observations, in our hands cytologic score values <15 and>16 would be strongly indicative of, respectively, a benign and a malignant (or premalignant) lesion. Thus, the cytologic score system per se would help delineate the above two categories of low-grade and high-grade lesions thus reducing the number of indeterminate cases which correspond to the group of aspirates with intermediate cytologic atypia (score values 15 and 16).
In this latter category a significant overlap of cytomorphologic features among FH, AH, DCIS and IDC was detected thus confirming that in a relevant number of aspirates of breast lesions diagnosed as "suspicious" or "atypical" even the strict adherence to a reproducible and standardized score system, such as that proposed by Masood et al ( 6 ) for evaluating selected cytomorphologic features, fails to provide a reliable distinction of benign florid vs. premalignant and malignant breast proliferations. This finding is well known in the literature: in fact, Sidawy et al. ( 15 ) evaluated the applicability of the published cytologic criteria in the categorization of proliferative breast lesions with and without atypia and DCIS in FNAs by assessing the diagnostic accuracy and interobserver reproducibility of a panel of experts. They demonstrated a low agreement among the raters as well as a discouraging cytohistologic correlation in FNA diagnoses and concluded that the cytologic criteria of proliferative breast disease need to be further defined and assessed. Since a precise cytohistologic correlation is difficult to achieve, the amalgamation of different conditions in two large categories, i.e. "low grade" lesions (including non proliferative and proliferative breast lesions without atypia) and "high grade" lesions (including AH, DCIS, and well differentiated IDC) would certainly have a better impact on patient management since lesions falling in the former category may be followed clinically within a 6-month interval, and those in the latter category should be pursued by excisional biopsy.
Our DNA ploidy analysis data confirmed previous observations ( 11-18 ) showing that ductal hyperplasia without atypia in the breast is practically always characterized by a diploid DNA pattern of cellular proliferation, while aneuploidy appears in about less than half the cases of AH (range 0% to 40%; 37.5% in this series) and in a consistent number of cases of DCIS (range 30% to 84%; 81.8% in this series). By a combined evaluation of ploidy and cell kinetics in diploid proliferations we, moreover, demonstrated that a SPD pattern is strongly correlated to a benign lesion, while a RPD proliferation is at high risk of being malignant, or premalignant, thus confirming previous observations obtained by Martelli et al. ( 19 ) by flow cytometric DNA-analysis of FNAs of benign and malignant breast tumors. The expansion of the proliferative compartment in diploid cell populations and/or the occurrence of an aneuploid stem-line in the hyperdiploid range of both AH and DCIS cases stand for a genomic instability in these lesions that leads to an increased DNA synthesis with clonal expansion in the course of tumor progression towards invasive carcinoma ( 13,14,26,27 ).
DNA image analysis correlated well with cytologic grading in that no aneuploid samples were present in the group with minor atypia and no SPD histograms were detected in the group of aspirates with marked atypia. Moreover, in the group of samples with intermediate atypia, the SPD and aneuploid patterns were strongly associated with a benign and a malignant (or premalignant) lesion, respectively, or, in other words, with aspirates of "low-risk" and "high-risk" lesions. According to these results, DNA image analysis is a valuable diagnostic aid in subclassifying breast borderline lesions since, irrespective of the score value of cytologic atypia of the aspirate, the detection of a SPD pattern is indicative of a benign proliferation and the detection of an aneuploid pattern is indicative of a malignant (or premalignant) lesion. In fact, if malignant and premalignant lesions are considered together, the NPV of a SPD pattern was 95% and the PPV of an aneuploid pattern was 100% irrespective of the score value. The NPV of the SPD pattern raised to 100% in cases scoring <15. The RPD pattern was detected in aspirates falling in all three categories and it was shared by both benign and malignant (or premalignant) lesions at least in the groups with minor and intermediate cytologic atypia. In these latter the PPV of a RPD pattern for malignant and/or premalignant lesions was 63% thus such a ploidy profile even in cases with a low to intermediate cytologic score should be considered with caution and used as an indication to pursue the case by excisional biopsy to establish the real nature of the lesion.
In summary, in our series the score system introduced by Masood et al. ( 6 ) to evaluate semiquantitatively cytomorphologic changes in aspirates of borderline breast lesions definitely delineated the majority of high risk lesions which were characterized by a high (>16) score value. Coupling cytologic score evaluation to DNA-image analysis was, however, important in cases scoring <16 in that in the categories with minor and intermediate grades of atypia the PPV of an aneuploid pattern was 100% and that of a RPD pattern was 63% while the SPD pattern showed a NPV of 95%.
Further studies are needed to assess the reproducibility of this double approach in subclassifying borderline breast lesions and to establish its efficiency in improving patients management in this challenging field of FNA cytology.