The choice of sampling technique for breast disease has been discussed and might depend on the inadequate diagnostic rates for that methodology 1 2 3 . For example, the inadequate and indeterminate rates for fine needle aspiration biopsy (FNAB) and needle biopsy should be <10%, as recommended in the Japanese guidelines 4 . However, some Japanese physicians believe that needle biopsy can better avoid inadequate and/or indeterminate results than can FNAB. Nevertheless, whether this is true or not, some of them perform needle biopsy instead of FNAB. On the other hand, increased breast cancer screening is expected. Therefore, low-cost and simple methods are needed for routine procedures, and cytological examination may fulfill these criteria. As the first-line pathological investigation (i.e., symptomatic lesions, confirming of benign lesions), FNAB could be appropriate for breast cancer screening, with the exception of microcalcifications 1 .

In such circumstances, we recently reported the accuracy of cytological diagnosis compared with histological diagnosis of breast disease in institutes in our area of southern Fukuoka Prefecture, Japan 5 . We found that the accuracy of cytological diagnosis at these institutes was equivalent to that recommended for diagnostic accuracy in Japanese 4 and UK 6 7 guidelines. It was also equivalent to data reported in other countries 8 9 . However, overall performance differed among the institutes, and was related to the characteristics of these institutes (e.g., the number of cases, number of staff, and specialism).

Furthermore, the number of false-positive and false-negative cases should be as low as possible. To confirm the validity of cytological examinations for breast cancer, managing the accuracy of the examinations is essential. Our earlier study also focused on the differences in the performance of FNAB among institutions.

Therefore, in the present study, we evaluated the inadequate, indeterminate, false-negative and false-positive cases to determine the most common histological types and/or features in these settings, and to demonstrate the usefulness and/or limitations of cytological examination for the diagnosis of breast cancer from the viewpoint of histological types.

We conducted a 1-year survey of seven institutes that dealt with many patients, along with a multiple-year survey of institutes with ≤ 100 cases per year. We analyzed data from 2009 and earlier, as previously described 5 .

In this study, we analyzed cytologically inadequate and indeterminate cases, as well as false-negative and false-positive cases, by evaluating the histological features of these cases.

In cases where the sample was rated as inadequate, SIDC was the most common histological type of malignant tumor, while mastopathy was the most frequent benign disease. It was reported that SIDC cases are often classified as inadequate 11 because SIDC usually consists of abundant fibrous elements, which hinders sample collection 11 . Because tumors ≤ 1 cm in size are relatively infrequent among cases with inadequate samples, it seems likely that the histological type or property (i.e., hard tumor or difficult to collect) is cited more frequently than is tumor size as the reason for rating the sample as inadequate. In 1998, it was reported that tumors of ≤ 1 cm tend to be insufficient 12 . However, we consider that techniques for sample collection have improved since that study.

Among the indeterminate cases with benign diseases, the frequencies of mastopathy, fibroadenoma and papilloma were similar. Among indeterminate cases with malignant diseases, the frequency of Papi-tub was high, nearing that of SIDC. These findings suggest that differentiating between epithelial hyperplastic changes and atypical cancer cells may be difficult. In addition, there are two types of Papi-tub. The first is the papillotubular nodular type, including cribriform carcinoma, and the second mainly consists of intraductal carcinoma components associated with invasive lesions. Both types are generally well differentiated with little cell atypia. The size of tumors in cases rated as indeterminate was not always small, similar to that for inadequate samples. The cases were mainly rated as indeterminate because of tissue features rather than tumor size.

The present study revealed that cases with SIDC or mastopathy were likely to be rated as indeterminate or inadequate samples. Cases of SIDC are sometimes classified as inadequate when too few cells are collected for evaluation, especially for tumors rich in fibrous components, as these are difficult to sample 11 . When samples can be extracted, the cells commonly form small clusters with little atypia on cytological specimens, which hinders attempts to distinguish between benign and malignant cases. This often results in the rating of these cases as indeterminate (representative cytological specimens of SIDC as shown in Figure 7). Meanwhile, it may be difficult to obtain samples from cases of mastopathy (fibrocystic disease) and assess hyperplastic changes of the epithelial cells (i.e., ductal hyperplasia). Moreover, some histological types classified as indeterminate differed from those classified as inadequate samples. For example, among benign diseases, papilloma was more frequently classified as indeterminate rather than inadequate. Among malignant diseases, Papi-tub was more frequently classified as indeterminate rather than inadequate.

We found that the most frequent histological type of tumor among the false-negative cases was SIDC. Indeed, cases of SIDC are often reported as false-negatives 11 , a finding confirmed by our data. SIDC usually shows with thick fibrosis and, when samples can be extracted, the cells commonly form small clusters with little atypia; this is also why cases with SIDC are often rated as inadequate and indeterminate. Moreover, the atypical cells were relatively small and were similar in appearance to benign or hyperplastic cells.

False-positive cases included cases with papilloma or fibroadenoma, which involve marked hyperplasia of epithelial cells 12 13 . Therefore, particular care is needed during cytology of cases in which a papillary structure or a cribriform pattern is seen. Fibroadenomas are divided into subtypes based on epithelial morphology. For example, the mastopathic type is often mistaken as a malignant tumor based on clinical signs, and the presence of epithelial hyperplasia within the tumor at the time of cytology and needle biopsy 14 15 16 . In the present study, one fibroadenoma was incorrectly classified as malignant because epithelial hyperplasia within the tumor was incorrectly identified. In this case, however, the fibroadenoma was of a type undergoing marked stromal myxomatous changes (i.e., myxomatous fibroadenoma). We previously reported that the interstitial components of fibroadenoma should receive particular attention during histological analysis 17 . This type of fibroadenoma can be mistaken for a malignant tumor based on US findings, particularly marked growth of the stroma. Therefore, evaluation of the surrounding stroma (e.g., by checking for metachromasia on Giemsa staining) in addition to epithelial cells will help avoid misdiagnosis 17 18 .

With the increasing implementation of screening programs in Japan, cytological examination of breast tissue is of increasing importance. Therefore, improving the accuracy of FNAB is an important issue, even though we showed that the accuracy of FNAB is as high as that recommended under the relevant clinical guidelines 5 . We believe that there is room for improvement. For example, it is important to acknowledge that certain histological types, such as SIDC for malignant breast diseases, make tissue aspiration and diagnosis difficult. Indeed, SIDC tended to be diagnosed in cases classified as inadequate, indeterminate and false-negative using aspirated samples in the present study. SIDC was reported to be easy to recognize using appropriate imaging modalities (i.e., mammography and/or US) 11 . Therefore, the present findings should encourage the use of needle biopsy rather than aspiration when this histological type is suspected based on imaging. Furthermore, triple assessment (i.e., clinical, pathological and radiological assessment) should be considered in some cases 5 19 . It is also important that staff receive appropriate pathological and clinical training, and that there is good communication between clinicians and pathological staff 5 . Finally, because there are institutional differences, we recommend that a management system is established to facilitate accurate cytological examination and diagnosis, as we have already reported 5 .

The inadequate, indeterminate, false-negative and false-positive cases showed similar histological types, notably SIDC for malignant tumors, and mastopathy, fibroadenoma and papilloma for benign cases. We need to pay particular attention to the collection and assessment of aspirates. Diagnosis should be reached while considering possible errors in diagnosis for these histological types of breast disease. In particular, several inadequate, indeterminate and false-negative cases with samples collected by aspiration were diagnosed as SIDC. These findings should encourage the use of needle biopsy rather than aspiration when this histological type is identified on imaging.

FNAB: Fine needle aspiration biopsy; SIDC: Scirrhous growing type, invasive ductal carcinoma; Papi-tub: Papillotubular growing type, Invasive ductal carcinoma; Solid-tub: Solid tubular growing type, invasive ductal carcinoma; IDC: Invasive ductal carcinoma; NOS: Not otherwise specified; DCIS: Ductal carcinoma in situ.

The authors declare that they have no competing interests.

RY is the corresponding author and wrote the manuscript. SiT and TK designed the study. TY (cytologist, facility D), KM (cytologist, facility G), YN (cytologist, facility F), SI (cytologist, facility E), EH (cytologist, facility B), MN (cytologist, facility A), JW (pathologist, facility B), KH (pathologist, facility E), and MK (pathologist, facility D) performed the study at each facility. HY provided administrative support and interpreted the data. All authors provided important contributions to the conception and design of the study, reviewed the analytical results, and read and approved the final manuscript.