Introduction: Breast cancer is rapidly emerging as the leading cause of cancer in Indian women. Robust cytopathology and histopathology services are required to tackle this growing burden. The use of rapid on-site evaluation (ROSE) and the International Academy of Cytology (IAC) Yokohama System for Reporting Breast Fine-Needle Aspiration Biopsy (FNAB) Cytopathology, which offers structured protocols, are expected to improve breast cytopathology reporting. Methods: We retrieved the cytopathology slides, categorized them by the IAC Yokohama System and histopathology data of all the patients who had been investigated for breast lesions from September 2016 to December 2018, and compared the cytopathology and histopathology. Risk of malignancy (ROM) and performance metrics, like sensitivity, specificity, predictive values, accuracy, and area under the curve were computed. Results: A total of 1,147 FNABs were evaluated, of which 442 (38.5%) underwent ROSE and 624 (54.4%) histopathology. Reported using IAC categories, our cohort recorded 4.9% inadequate, 65.3% benign, 7.8% atypical, 3.3% suspicious for malignancy, and 18.7% malignant lesions. The overall sensitivity and specificity for identifying in situ and malignant lesions were 99.1% and 99.3%, respectively, and were substantially improved by ROSE. ROSE improved the concordance between cytopathology and histopathology from 76.9% to 90.2%, by reducing inadequate (p < 0.001) cases. The ROM increased along a gradient from inadequate to malignant categories, with the gradient being sharpened by ROSE. The false negativity rate was 0.7% and false positivity rate 0%. Conclusion: Incorporating ROSE and the IAC Yokohama System for breast cytopathology reporting improves accurate diagnosis of breast lesions, prevents missed diagnoses, and provides reliable estimates of ROM. These protocols also aid in standardizing a reproducible system for monitoring and auditing of breast pathology services, identify areas that need strengthening, and improve training at pathology centers.

Clinical examination, radiological imaging, and breast cytopathology with or without core needle biopsy (CNB) comprise the 3 key modalities employed for diagnosing breast lumps. These modalities aim to maximize the preoperative identification of malignancy so that early, definitive, one-stage surgery or appropriate treatment can be offered to the patient [1, 2]. As a corollary, these diagnostic procedures also aim to accurately recognize benign breast lesions and avoid unnecessary invasive investigations and anxiety to the patient [2]. For breast biopsy, fine-needle aspiration biopsy (FNAB) and CNB offer complementary advantages, and both can be utilized gainfully in different settings. The recent advances in breast cytopathology with the implementation of rapid on-site evaluation (ROSE) and the development of the International Academy of Cytology (IAC) Yokohama System for Reporting Breast FNAB Cytopathology have structured the reporting of breast cytopathology with the aim of improving cytopathology procedures, training, interpretation, and diagnosis [3, 4].

In recent years, breast cancer has become the most common malignancy in Indian women, surpassing cervical cancer. An incidence of 2,05,424 new cases of breast cancer annually is projected in India, comprising ∼29.9% of all malignancies reported in women, with 87,000 annual deaths [5, 6]. FNAB and CNB are crucial preoperative tests for diagnosing these cases. However, the accuracy of FNAB depends on several factors, including the skill and experience of the operator, the localization method used, the skill of the pathologist, and the proportion of symptomatic and nonpalpable lesions being examined [7, 8]. Since numerous sources of error can confound these procedures, it is important that the performance of breast cytopathology is regularly monitored at a center using a standardized and reproducible system [3, 4, 7]. We undertook this study to incorporate ROSE and the IAC Yokohama System into our breast cytopathology protocols, evaluating their impact on the performance of breast cytopathology versus histopathology in accurately identifying breast malignancies and providing a patient’s risk of malignancy (ROM) under various IAC categories. This afforded us the opportunity to upgrade our breast reporting, create a reproducible monitoring and auditing system, and better identify the areas that need attention for further improvement of our services.

This study was carried out at the Seth GSMC & KEM Hospital, Mumbai, India, which is a tertiary care center catering to a large population in western India. In order to evaluate the impact of ROSE and the IAC System on our breast pathology reporting, we retrieved the cytopathology data of all consecutive patients investigated for breast lesions between September 2016 and December 2018. Ethics clearance was obtained from the Institute’s Ethics Committee for retrieving and analyzing this data [EC/OA-150/2019]. The data was anonymized before undertaking analyses. All the samples had been collected from the patients for routine clinical investigation, after informed consent. An attempt was made to obtain follow-up and any, histopathology report of a biopsy or excision (if performed) in all cases; however, histopathology was available only in 624 cases.

Cytopathology Protocols

Guided and nonguided FNAB samples were collected from enrolled patients using standard procedures. ROSE was carried out using 1% aqueous toluidine blue staining. After provisional assessment by ROSE, the slides were subjected to Papanicolaou staining for standard cytopathology reporting [8]. Air-dried slides were stained using Giemsa. Special stains like Ziehl-Neelsen, Periodic acid Schiff’s, Gomori’s methanamine silver stain, and immunocytochemistry were employed on the cytology slides wherever needed and feasible.

Breast cytopathology reporting with and without ROSE, we used the diagnostic categories recommended by the IAC Yokohama System for Reporting Breast FNAB Cytopathology [1, 3]: insufficient, benign, atypical, and suspicious for malignancy and malignant. The histopathology diagnoses were broadly categorized under nonrepresentative, benign, indeterminate, in situ, and malignant, as has been reported previously [9]. Nonrepresentative histopathology included cases where there was insufficient material to reliably diagnose the lesions. Benign histopathology included breast abscess, fibroadenoma, fibrocystic changes, intraductal papilloma, scar tissue, etc. The indeterminate group included atypical fibroepithelial lesions, including phyllodes tumor, atypical ductal hyperplasia, etc. The in situ group included cases where at least one focuses of in situ pathology in the form of ductal carcinoma in situ or lobular carcinoma in situ was observed. Finally, the malignant histopathology group included cases of invasive carcinoma of no special type, medullary carcinoma, mucinous carcinoma, lobular carcinoma, and non-Hodgkin’s lymphoma.

Statistical Methods

Descriptive statistics were used to describe the patient cohort. Confusion matrices and heatmaps were computed to compare how well cytopathology (with and without ROSE) predicted the diagnostic categories of breast lesions against those assigned by histopathology. χ2 test was used to compare the performance of cytopathology with and without ROSE, across different IAC categories. Performance metrics, including sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, and area under the receiver operating characteristic curve (AUC) were computed to compare the performance of cytopathology against histopathology. The PPV of various IAC categories for diagnosing in situ or frank malignancy was computed to estimate the ROM of each category. A p value of <0.05 was taken as significant. All statistical analyses were performed in R v3.6.3 and SPSS v23.0.

A total of 1,147 breasts FNAB were examined in this study. A majority of these specimens were received from female patients (96.8%, median age: 34.5 years, interquartile range [IQR]: 23–45 years), with remainder of male patients (3.2%, median age: 45.0 years, interquartile range: 28–59 years). Breast lesions were equitably distributed with 45.2% arising in the right breast, 51.2% in the left, and 3.5% were bilateral. Precise quadrant-wise information was available in 59.6% of cases and their distribution is depicted in Figure 1. Of the total 1,147 cases, 442 (38.5%) underwent ROSE. Of the total 1,147 cases, 624 (54.4%) had histopathology follow-up.

Fig. 1.

Distribution of breast lesions observed in this study. All figures in percentages.

Fig. 1.

Distribution of breast lesions observed in this study. All figures in percentages.

Close modal

Overall, the 1,147 cases were subclassified as 4.9% inadequate, 65.3% benign, 7.8% atypical, 3.3% suspicious for malignancy, and 18.7% malignant. The most common breast lesion under the benign category was fibroadenoma (309, 41.3%). Other entities included inflammatory lesions (47, 6.3%), granulomatous mastitis (26, 3.5%) (Fig. 2a, b), gynecomastia (23, 3.1%), galactocele (8, 1.1%) (Fig. 2c), and miscellaneous entities, like keratinous cyst, fat necrosis (Fig. 2d), and filariasis (Fig. 3a, b). The atypical category included lesions with atypical cells (44, 48.9%) and fibroepithelial lesions (e.g., fibroadenoma vs. low-grade phyllodes) (46, 51.1%). The suspicious for malignancy category included lesions suspicious for invasive carcinoma no special type (30, 79%) and suspicious for papillary carcinoma (9, 23.6%), and low-grade lymphoma (1, 2.6%). Ductal carcinoma was the most common lesion in the malignant category accounting for 203 (94.9%) cases followed by mucinous carcinoma (2, 0.9%) and malignant phyllodes tumor (1, 0.5%). The ROM was also assessed for the different categories and was found to be 16% for the inadequate, 0.7% for benign, 23.3% for atypical, 94.1% for suspicious for malignancy, and 100% for the malignant category.

Fig. 2.

a, b Granulomatous mastitis (IAC II): epithelioid cell granuloma with necrosis and nuclear debris (Toluidine blue, ×200; Papanicolaou, ×400). c Galactocele (IAC II): bluish granular proteinaceous background with scattered histiocytes (Giemsa, ×100). d Fat necrosis (IAC II):granular necrotic fat fragments and multi nucleated histiocytes (Papanicolaou, ×400). IAC, International Academy of Cytology.

Fig. 2.

a, b Granulomatous mastitis (IAC II): epithelioid cell granuloma with necrosis and nuclear debris (Toluidine blue, ×200; Papanicolaou, ×400). c Galactocele (IAC II): bluish granular proteinaceous background with scattered histiocytes (Giemsa, ×100). d Fat necrosis (IAC II):granular necrotic fat fragments and multi nucleated histiocytes (Papanicolaou, ×400). IAC, International Academy of Cytology.

Close modal
Fig. 3.

a, b Filariasis breast (IAC II): numerous unfertilised eggs of filariasis along with wet mount preparation showing the adult filaria (Papanicolaou, ×200; Wet mount, ×200). IAC, International Academy of Cytology.

Fig. 3.

a, b Filariasis breast (IAC II): numerous unfertilised eggs of filariasis along with wet mount preparation showing the adult filaria (Papanicolaou, ×200; Wet mount, ×200). IAC, International Academy of Cytology.

Close modal

In 705 (61.5%) cases where ROSE was not implemented the categories were as follows: inadequate (48, 6.8%), benign (443, 62.8%), atypical (53, 7.5%), suspicious for malignancy (31, 4.4%), and malignant (130, 18.4%). Of these cases without ROSE, histopathology follow-up was undertaken in 390 (55.3%) cases. Among these, 22 cases inadequate on cytopathology, were found on histopathology to be suspicious/malignant in 5 cases and benign (fibroadenoma, galactocele, lipoma, keratinous cyst, etc.) in the remaining 17 cases. A total of 173 benign cases (39.1%) had histopathology and only 3 cases showed a discrepancy in classification. Among the atypical, 44 (83.0%) underwent histopathology follow-up and were found to be indeterminate, in situ or malignant lesions. Of the suspicious of malignancy, 30 (96.8%) cases were followed up with histopathology and 28 (93.3%) of these revealed in situ or invasive malignancy. One case of suspected papillary carcinoma reported on cytopathology was later categorized to be indeterminate as the CNB was reported as a papillary neoplasm with atypia. Similarly, the malignant cytopathology group had histopathology follow-up in 121 (93.1%) cases, with malignancy confirmed in 98.3%, and in 2 cases (1.7%) the CNB yielded a nondiagnostic sample. The corresponding figures for samples that were subjected to ROSE are described in detail in the following paragraphs.

Impact of ROSE

ROSE was carried out in 442 (38.5%) samples yielding a preliminary diagnosis of inadequate (8, 1.8%), benign (316, 71.5%), atypical (23, 5.2%), suspicious for malignancy (14, 3.2%), and malignant (81, 18.3%), respectively. These preliminary diagnostic categories were compared against those assigned by final cytopathology and histopathology (Fig. 4). The comparison of ROSE categories against the final cytopathology categories showed 96.2% concordance. ROSE categories were compared against the final categories assigned by histopathology (Fig. 4a, b). Benign and malignant conditions showed good levels of congruence with histopathology.

Fig. 4.

Comparison of diagnostic categories assigned by ROSE versus those assigned by (a) final cytopathology and (b) histopathology. ROSE, rapid on-site evaluation.

Fig. 4.

Comparison of diagnostic categories assigned by ROSE versus those assigned by (a) final cytopathology and (b) histopathology. ROSE, rapid on-site evaluation.

Close modal

We further evaluated whether performing ROSE improved the final IAC results. For this, we compared the frequency with which various IAC categories were assigned to specimens which had undergone ROSE (442, 38.5%) versus those which had not undergone ROSE (705, 61.5%) (Table 1). The most significant impact of ROSE was seen in the reduction of inadequate (p < 0.001) and suspicious for malignancy (p = 0.010) categories. ROSE enabled better classification of benign lesions (p = 0.027) but showed little change in the frequency with which atypical and malignant specimens were categorized.

Table 1.

Comparison of FNAB cases distributed across various cytology categories when evaluated with and without ROSE

 Comparison of FNAB cases distributed across various cytology categories when evaluated with and without ROSE
 Comparison of FNAB cases distributed across various cytology categories when evaluated with and without ROSE

Performance of Breast Cytopathology versus Histopathology

We evaluated the performance of breast cytopathology in our cases against histopathology. This was important to identify the areas that were lagging behind in our breast pathology services, and also to establish a reproducible system incorporating ROSE, the IAC Reporting System and histopathology to monitor and improve various components of our services in the future.

We first compared how well cytopathology categorized cases against broad histopathology categories (Fig. 5). We stratified this by including ROSE in the cytopathology workup (Fig. 5a–c). As is evident from the heatmaps, ROSE reduced the number of outliers’ diagnoses substantially and assigned benign and malignant categories more accurately. Samples that were subjected to ROSE showed only 1 benign, 6 atypical, and 1 suspicious of malignancy case that had to be reclassified on histopathology (Fig. 5a). The single benign case that had to be reclassified on histopathology showed fat necrosis on ROSE. It was subjected to CNB due to high clinical suspicion and was found to be a malignant lesion. FNAB most likely had sampled the fat necrosis surrounding the tumor. Six cases categorized as atypical by ROSE were fibroadenomas on histopathology. The young age of the patients and the marked stromal cellularity had flagged the mismatch, prompting us to confirm the diagnosis by histopathology. Three cases with benign, suspicious for malignancy, and malignant diagnoses on ROSE, could not be confirmed on histopathology due to nonrepresentative samples and lack of patient follow-up.

Fig. 5.

Comparison of diagnoses assigned by cytopathology and histopathology. The first 3 heatmaps compare diagnostic categories assigned by final cytopathology with ROSE (a), without ROSE (b), and overall (c), versus the diagnostic categories assigned by histopathology. d Compares the predominant histopathology diagnoses observed in the study versus the diagnostic categories they were assigned by final cytopathology. DCIS, ductal carcinoma in situ; FA, fibroadenoma; ROSE, rapid on-site evaluation.

Fig. 5.

Comparison of diagnoses assigned by cytopathology and histopathology. The first 3 heatmaps compare diagnostic categories assigned by final cytopathology with ROSE (a), without ROSE (b), and overall (c), versus the diagnostic categories assigned by histopathology. d Compares the predominant histopathology diagnoses observed in the study versus the diagnostic categories they were assigned by final cytopathology. DCIS, ductal carcinoma in situ; FA, fibroadenoma; ROSE, rapid on-site evaluation.

Close modal

In contrast to the above analysis, specimens which did not undergo ROSE showed a larger number of recategorizations on histopathology. These included 5 inadequate, 4 benign, 16 atypical, 3 suspicious for malignancy, and 2 malignant cases which had to be recategorized (Fig. 5b). The 16 atypical cases showed lactational atypia (Fig. 6a, b), infarcted fibroadenoma, benign breast changes, and fat necrosis on histopathology. The 5 inadequate cases all were malignant, and underscore the importance of ROSE in assessing adequacy on-site minimizing missed and delayed diagnoses.

Fig. 6.

a, b Lactational change with mild atypia (IAC III)-Mild to moderate nuclear atypia in the ductal epithelial cells with dissociation and milky background (Giemsa, ×400; Papanicolaou, ×400). IAC, International Academy of Cytology.

Fig. 6.

a, b Lactational change with mild atypia (IAC III)-Mild to moderate nuclear atypia in the ductal epithelial cells with dissociation and milky background (Giemsa, ×400; Papanicolaou, ×400). IAC, International Academy of Cytology.

Close modal

While the above analysis drew comparisons across broad categories of cytopathology and histopathology, we also wanted to assess how specific lesions had fared between the 2 techniques: how accurate was cytopathology (both with and without ROSE) in identifying the most common lesions (Fig. 5d). We found that a large majority of benign conditions were correctly classified by cytopathology, with 96.2% of fibroadenomas and 100% of acute and granulomatous mastitis identified accurately. Only 2.7% of fibroadenomas and 9.8% of benign breast lesions were labeled atypical on cytopathology. Half (50.0%) of the fibrocystic lesions and papillomas was classified as benign or atypical. Other less common benign entities included lipoma, fat necrosis, galactocele, filariasis, benign fibroepithelial polyps, gynecomastia, adenolipoma, and duct ectasia. These were largely (76.3%) identified correctly by cytopathology (Fig. 5d).

Among breast lesions with indeterminate histopathology, low-grade phyllodes tumor was the commonest and 32 (94.1%) of these were classified as atypical on cytopathology and 2 (5.9%) as benign fibroadenomas. The review of these 2 cases showed increased stromal cellularity, altered stromal-to-epithelial ratio and plump spindle cells in the background (Fig. 7a–d). Benign phyllodes tumor can be difficult to distinguish from fibroadenoma on cytopathology [10]. Other less common indeterminate entities included atypical or suspicious of malignancy lesions (Fig. 5d). Our cohort had very few cases of in situ carcinomas, and 3 (75.0%) of these were flagged as atypical, suspicious of malignancy, or malignant by cytopathology, with 1 high grade DCIS being misclassified as a benign lesion on cytopathology (Fig. 5d), which on review was due to inadequate sampling of a deep-seated lesion. Among invasive carcinomas, 187 (82.0%) were correctly categorized as malignant by cytopathology (Fig. 8a, b), while the remainder were categorized as atypical 12 (5.3%) or suspicious for malignancy 24 (10.5%). Other less common malignant lesions, like papillary carcinoma (Fig. 8c, d), lobular carcinoma, mucinous carcinoma, non-Hodgkin’s lymphoma (Fig. 9a, b), metaplastic carcinoma and malignant phyllodes tumor (Fig. 9c, d) were labeled as malignant (11, 50.0%), or suspicious for malignancy (7, 31.8%) with 4 (18.2%) being classified as atypical (Fig. 5d). The 16 atypical cases that had been typed as atypical on FNAB had revealed limited cytological features, which could be confirmed as malignant on histopathology (Fig. 5d).

Fig. 7.

a–d Phyllodes tumor (IAC III)-Large cellular stromal fragments with numerous spindle-shaped nuclei. Histopathology of the same case revealed cellular stroma with atypia and occasional mitotic figures (Papanicolaou, ×100, ×200 & ×1,000; H&E, ×200). IAC, International Academy of Cytology.

Fig. 7.

a–d Phyllodes tumor (IAC III)-Large cellular stromal fragments with numerous spindle-shaped nuclei. Histopathology of the same case revealed cellular stroma with atypia and occasional mitotic figures (Papanicolaou, ×100, ×200 & ×1,000; H&E, ×200). IAC, International Academy of Cytology.

Close modal
Fig. 8.

a, b Invasive carcinoma of no special type (IAC V): loosely cohesive bizarre ductal epithelial cells with high N/C ratio, hyperchromatic nuclei and prominent nucleoli (Toluidine blue, ×400; Papanicolaou, ×400). c, d. Papillary carcinoma (IAC V): cellular smears comprising of jigsaw patterned tissue fragments with crowded epithelial cells having enlarged nuclei and high N/C ratio (Papanicolaou, ×100 & ×400). IAC, International Academy of Cytology.

Fig. 8.

a, b Invasive carcinoma of no special type (IAC V): loosely cohesive bizarre ductal epithelial cells with high N/C ratio, hyperchromatic nuclei and prominent nucleoli (Toluidine blue, ×400; Papanicolaou, ×400). c, d. Papillary carcinoma (IAC V): cellular smears comprising of jigsaw patterned tissue fragments with crowded epithelial cells having enlarged nuclei and high N/C ratio (Papanicolaou, ×100 & ×400). IAC, International Academy of Cytology.

Close modal
Fig. 9.

a, b Non-Hodgkin lymphoma (IAC V): dissociated large atypical lymphoid cells on a background of lymphoglandular bodies (Giemsa ×40; Papanicolaou, ×400). c, d Malignant phyllodes tumor (IAC V): atypical plump spindle cells with marked atypia and tumor giant cells (Papanicolaou, ×400). IAC, International Academy of Cytology

Fig. 9.

a, b Non-Hodgkin lymphoma (IAC V): dissociated large atypical lymphoid cells on a background of lymphoglandular bodies (Giemsa ×40; Papanicolaou, ×400). c, d Malignant phyllodes tumor (IAC V): atypical plump spindle cells with marked atypia and tumor giant cells (Papanicolaou, ×400). IAC, International Academy of Cytology

Close modal

The analysis of breast FNAB against histopathology excluding the cases excluded final inadequate and atypical cases (Table 2). We stratified this analysis by 2 scenarios. First, how well did cytopathology identify the cases that were eventually flagged in situ or malignant by histopathology, and how did incorporating ROSE affect this performance. FNAB picked up malignancies with an overall 86.1% sensitivity and 99.6% specificity, and excluded malignancies with a NPV of 89.9%. Importantly, all of these performance metrics were substantially improved by using ROSE (Table 2). Second, how well did the IAC categories of suspicious for malignancy and malignant identify in situ and malignant histopathology? As would be expected, by widening the net with both suspicious and malignant cases, the performance of FNAB went substantially up with 99.1% sensitivity and 99.3% specificity. Incorporating suspicious for malignancy cases in the comparison, the impact of ROSE was minimal (Table 2).

Table 2.

Performance of FNAB in detecting in situ and malignant histopathology, with “inadequate” and “atypical” cytopathology removed

 Performance of FNAB in detecting in situ and malignant histopathology, with “inadequate” and “atypical” cytopathology removed
 Performance of FNAB in detecting in situ and malignant histopathology, with “inadequate” and “atypical” cytopathology removed

The performance of FNAB against histopathology was also analyzed with atypical cytology cases included (Table 3). We re-analyzed the above scenarios and found that cytopathology identified malignancies with an overall 80.5% sensitivity and 99.7% specificity, and NPV of 91.7%, and was substantially improved by ROSE (Table 3). When we expanded the definition and compared the performance of suspicious for malignancy and malignant cases in correctly identifying in situ and malignant histopathology, we noted a substantial increase in sensitivity (92.4%), specificity (99.4%), and NPV (94.6%). Cases subjected to ROSE again performed better on all metrics; however, the gain was smaller due to the inclusion of atypical and suspicious for malignancy cases in the analysis (Table 3).

Table 3.

Performance of FNAB in detecting in situ and malignant histopathology, with “inadequate” cytopathology removed

 Performance of FNAB in detecting in situ and malignant histopathology, with “inadequate” cytopathology removed
 Performance of FNAB in detecting in situ and malignant histopathology, with “inadequate” cytopathology removed

Predicting the ROM

A crucial goal of breast cytology and histopathology is to recognize malignancy in a timely and accurate fashion: the PPV of the IAC categories yielding a diagnosis of in situ or invasive carcinoma in histopathology was calculated and stratified by the application of ROSE (Table 4). There was a distinct rise in ROM across IAC categories from inadequate to malignant, and ROSE enhanced this gradient by minimizing potentially malignant cases in the inadequate (0%), benign (0.9%), and atypical (4.3%) IAC categories. In contrast, the specimens which did not undergo ROSE showed a significantly higher ROM in inadequate (18.2%) and atypical (34.9%) categories (p < 0.001).

Table 4.

ROM observed under various cytopathological categories for detecting in situ and malignant histopathology

 ROM observed under various cytopathological categories for detecting in situ and malignant histopathology
 ROM observed under various cytopathological categories for detecting in situ and malignant histopathology

Quality Control Outcomes

A key metric in quality control is the false negative (FN) rate and is due to operator failure due to poor localization and poor FNAB technique of the sample or inability of the pathologist to detect subtle cytopathological features of malignancy. International guidelines recommend that FN rates should be <5% [2, 3]. Our pathology services performed well on this metric with a 0.71% (2/282) FN rate, with only 2 cases benign by ROSE and/or final cytopathology and malignant on histopathology (Fig. 5c). These 2 cases had revealed fat necrosis from deep-seated breast lumps. A high index of clinical-radiological suspicion ensured that we confirmed these cases with histopathology.

However, out of the total 749 cases that were categorized as benign on FNAB, only 285 were followed up with a CNB or excision biopsy based on their clinical-radiological suspicions. Of these 285 cases, 282 could be confirmed as benign on CNB, but 3 cases were indeterminate and lost to follow-up. In the remaining 464 cases the FNAB and clinical-radiological findings were unequivocally diagnostic of benign lesions and were not biopsied and not followed up for potential FN lesions and late development of malignancy in the same quadrant of the breast. Hence, the FN rate of 0.71% reported in this study represents only those benign lesions that underwent further histopathology examination. An equally important measure is the false positivity (FP) rate, representing over-diagnosis and mandate careful auditing. Ideally, FP rates should be <1% [2, 3], with our cases showing no malignant categorization on cytopathology to be benign on histopathology (Fig. 5c).

There are considerable variations in the literature regarding insufficient/inadequate rates, which can be due to the inexperience of FNAB operators, the varying mix of patients, use of mammographic screening and type of lesions (palpable or impalpable) vary from 0.7 to 47% [1, 7, 9]. We noted that 4.9% of the total 1,147 FNAB cases in this study were inadequate and less (1.8%) with ROSE (Fig. 4). Furthermore, 4.0% of FNAB that ultimately underwent histopathology had been deemed inadequate (Fig. 5c). We found 0.64% (4) cases with inadequate cytopathology to be malignant on histopathology, and 3.3% (38) were suspicious for malignancy out of the total 1,147 cases. Of the cases which had histopathology, 5.9% had been flagged as suspicious for malignancy (Fig. 5c). Of the specimens that underwent ROSE, 3.2% were suspicious for malignancy (Fig. 4) The ROM for atypical and suspicious of malignancy cases were 23.2% and 94.1%, respectively (Table 5), and are comparable to those previously reported [2].

Table 5.

A comparison of breast cytopathology performance guided by the IAC Yokohama System across recent studies

 A comparison of breast cytopathology performance guided by the IAC Yokohama System across recent studies
 A comparison of breast cytopathology performance guided by the IAC Yokohama System across recent studies

Breast lumps are common in women of all ages and breast cancer is the leading cause of cancer in women worldwide. Breast cancer comprises 23% of cancer burden worldwide, with 2.1 million new cases and 6,27,000 deaths annually [11‒14]. FNAB and CNB, along with mammography and clinical evaluation, form crucial parts of the “triple test” for diagnosing breast lumps [9]. FNAB is sensitive and specific for diagnosing both benign and malignant breast lesions [9, 10]. It can also reliably pick up radiologically detectable nonpalpable lesions, and can achieve sensitivities of up to 90–95% [1, 8, 15‒17]. But FNAB has its limitations and CNB is required to confirm lesions, such as carcinoma in situ, calcifications, and proliferative tumors [18‒20]. However, CNB is not only more invasive but also remains prohibitively expensive and inaccessible in low-resource countries, like India [10, 21‒23]. FNAB and CNB can complement each other depending on the type of resources available at a health-care facility, the economic constraints, the nature of a lesion, the skill of the operator sampling the lesion, and the skill of the pathologist examining the specimens [24, 25].

The introduction of ROSE and the IAC Yokohama System for Reporting Breast FNAB Cytopathology have structured and significantly improved the utility of FNAB [1]. We studied the impact of these protocols in our breast FNAB reporting and observed distinct improvements. ROSE substantially improved the final diagnoses (Fig. 4) and reduced the discordance between cytopathology and histopathology (Fig. 5). ROSE significantly reduced inadequate sampling and the suspicious for malignancy category (Table 1). ROSE reduced the financial burden and inconvenience of patients from having to revisit. Overall, ROSE was significantly helpful in ruling out malignant lesions, but showed no significant improvement in ruling in malignancy (Table 1). This reflects the spectrum of breast cases which are predominantly benign with a smaller proportion of malignancies. It also reflects a tendency to err on the side of caution in our cytology reporting. Despite its better performance, a few cases did show a mismatch between ROSE and the final cytopathology and/or histopathology diagnosis. These were predominantly atypical and suspicious lesions, including 10 cases of low-grade/benign phyllodes tumor which required reassignment at final reporting. Another 6 cases suspicious for malignancy at ROSE were eventually found malignant by final cytopathology and histopathology.

Breast cytopathology at our center showed overall high sensitivity (86.1%) and specificity (99.6%) at diagnosing malignancies. Including ROSE in the workup, substantially enhanced the sensitivity (92.9%) and specificity (100%) of cytopathology (Table 2). Expanding the net by including both suspicious for malignancy and malignant cases raised the sensitivity (99.1%) of FNAB further, with a specificity of 99.3%. However, with the inclusion of suspicious for malignancy cases, ROSE contributed negligible improvement in picking up malignant cases. This indicates that ROSE helps weed out suspicious and ambiguous cases early in the workflow (Table 2).

The IAC Yokohama System also helps monitor the ROM in patients classified under different IAC categories, enabling the pathologist to better flag the cases that would need further histopathology confirmation, and also offer more robust recommendations to the treating surgeons [1]. We found the ROM increase along a gradient from inadequate to malignant IAC categories (Table 4). However, without ROSE, the ROM among inadequate cases jumped to 18.2%, while with ROSE it was zero. Furthermore, ROSE shifted the ROM gradient sharply toward the suspicious for malignancy and malignant end of the IAC spectrum, which otherwise spilled onto the atypical and inadequate categories without ROSE (Table 4). This underscores the advantage of combining ROSE and the IAC Yokohama System for monitoring and predicting the ROM.

The major performance metrics were FN rate (0.71%), FP rate (0.0%), inadequate rate (4.9%), and suspicious for malignancy rate (3.3%), and showed further improvements in each of these parameters with ROSE. FNAB is preferred to CNB partly because of financial and resource constraints. FNAB is less reliable in the small number of patients with deep-seated breast lumps surrounded by excessive fatty tissue, which are at risk of delayed and missed diagnoses, and after thorough clinical-radiological evaluation should be offered FNAB by an experienced operator under ultrasound-guidance or immediate CNB [26].

A majority of our patients, including those financially constrained, often travel several miles to access pathology services. Even the few cases that get diagnosed as inadequate or inconclusive, are missed opportunities because most of these patients fail to return for follow-up until much later when the disease has advanced and/or has become unmanageable. Such missed opportunities highlight the importance of ROSE, regular monitoring of IAC ROM rates, and early ultrasound-guided FNAB where CNB is unaffordable.

Moving forward, harnessing long-term follow-up data will be a goal at our center to bring out the true FN rates in our patients and further strengthen our services. Besides the above challenges that are unique to low-resource countries, we also noticed the need to improve supervision and training of less experienced pathologists. Low-grade phyllodes tumors are difficult to diagnose on cytopathology, and all suspect cases should be reviewed and confirmed by an experienced pathologist [10, 27]. Such measures along with inclusion of ROSE can prevent delayed and missed diagnoses.

If a malignant FNAB diagnosis is in accordance with clinical-radiological findings in the triple test, and material is available in the cell block for prognostic markers, neoadjuvant chemotherapy and surgery can be commenced [3, 28]. If the FNAB diagnosis does not correlate with the clinical and imaging findings, then CNB or simple excision becomes mandatory [2, 23]. When lymph nodes are palpable or suspicious on ultrasound, FNAB is recommended to stage the patient [2, 29, 30].

We compared our findings with other studies and found largely similar and a few divergent trends (Table 5). Several studies have reported similar rates of inadequate samples (1.4–5.7%) as reported in our study (overall 4.9%; 1.8% with ROSE) (Table 5) [4, 7, 15, 17, 31‒33]. However, a few studies had recorded much higher rates of inadequate samples (11.0–15.5%) [9, 34, 35]. The IAC Reporting System recommends that operators should aim for an inadequate sample rate of <5.0%, which should be even lower if ROSE is implemented [3, 7, 9, 35, 36]. The proportion of benign cases (65.3%) in our cohort was also similar to that observed in various studies (41.8–82.6%) (Table 5) [3, 7, 9, 11, 13, 37]. Among benign cases, fibroadenomas comprised 58.1% of total cases, which was similar to that noted by previous studies, 48.8–67.7% [38‒40]. Follow-up histopathology was not required for most benign cases, especially if the patient was negative on clinical-radiological evaluation for >6–12 months [3, 41]. The proportion of atypical (7.8%) and suspicious (3.3%) cases in our study were similar to that observed in other studies [4, 7, 9, 15, 17, 37], except a few studies where such cases were recorded in higher proportions, 10.4–13.7% atypical cases and 9.3–11.0% suspicious for malignancy cases [4, 34, 42, 43]. The proportion of malignant lesions varied across studies, possibly reflecting the type of patients being cared for by the respective centers (Table 5). The overall performance of cytopathology in accurately diagnosing in situ and malignant breast lesions in our study was similar [4, 9, 15, 17] to better [7, 37] than that observed in contemporary studies (Table 5).

Our study shows that incorporating the IAC Yokohama System for Reporting Breast FNAB Cytopathology with ROSE can improve early and accurate diagnosis of breast lesions, prevent missed diagnoses, and provide reliable estimates of ROM in a given patient population. This is especially important for low- and middle-income countries where CNB is an expensive option. Our results also demonstrate that these protocols can help create a standardized and reproducible system for monitoring and auditing of breast pathology services, which can help identify the areas that need strengthening and improve the training being delivered at pathology centers.

Written informed consent was taken from all patients for the procedure of fine-needle aspiration biopsy. The retrospective analysis was undertaken after approval from the Institutional Ethics Committee on human research (EC/OA -150/2019).

The authors declare no conflicts of interest.

This study received no funding from any source.

N.A. and K.K. conceptualized and designed the study, performed literature search, and carried out the clinical study. S.T. carried out data collection, clinical study, preliminary data analysis, and wrote the manuscript. P.S. carried out formal data analysis, visualization, study design, and wrote the manuscript. M.A. and V.S. assisted in study design and literature search. All the authors contributed to the final review and editing of the manuscript.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.

1.
Field
AS
.
Breast FNA biopsy cytology: current problems and the International Academy of Cytology Yokohama standardized reporting system
.
Cancer Cytopathol
.
2017
;
125
(
4
):
229
30
.
2.
Singh
N
,
Wells
CA
.
Assessment of accuracy in breast cytology
.
Cytopathology
.
2001
;
12
(
4
):
211
8
.
3.
Field
AS
,
Raymond
WA
,
Schmitt
F
, editors.
The International Academy of Cytology Yokohama system for reporting breast fine needle aspiration biopsy cytopathology
.
Switzerland
:
Spinger Nature
;
2020
.
4.
Panwar
H
,
Ingle
P
,
Santosh
T
,
Singh
V
,
Bugalia
A
,
Hussain
N
.
FNAC of breast lesions with special reference to IAC standardized reporting and comparative study of cytohistological grading of breast carcinoma
.
J Cytol
.
2020
;
37
(
1
):
34
9
.
5.
Mathur
P
,
Sathishkumar
K
,
Chaturvedi
M
,
Das
P
,
Sudarshan
KL
,
Santhappan
S
;
ICMR-NCDIR-NCRP Investigator Group
,
.
Cancer statistics, 2020: report from National Cancer Registry Programme, India
.
JCO Glob Oncol
.
2020
;
6
:
1063
75
.
6.
Globocan 2018: India factsheet – India Against Cancer. Available from: cancerindia.org.in.
7.
Montezuma
D
,
Malheiros
D
,
Schmitt
FC
.
Breast Fine needle aspiration biopsy cytology using the newly proposed IAC Yokohama system for reporting breast cytopathology: the experience of a single institution
.
Acta Cytol
.
2019
;
63
:
1
6
.
8.
Kothari
K
,
Tummidi
S
,
Agnihotri
M
,
Sathe
P
,
Naik
L
.
This ‘rose’ has no thorns-diagnostic utility of ‘rapid on-site evaluation’ (rose) in fine needle aspiration cytology
.
Indian J Surg Oncol
.
2019
;
10
(
4
):
688
98
.
9.
Wong
S
,
Rickard
M
,
Earls
P
,
Arnold
L
,
Bako
B
,
Field
AS
.
The International Academy of Cytology Yokohama System for reporting breast fine needle aspiration biopsy cytopathology: a single institutional retrospective study of the application of the system categories and the impact of rapid onsite evaluation
.
Acta Cytol
.
2019
;
63
(
4
):
280
91
.
10.
Tummidi
S
,
Kothari
K
,
Agnihotri
M
,
Nail
L
,
Sood
P
.
Fibroadenoma versus phyllodes tumor: a vexing problem revisited
.
BMC Cancer
.
2020
;
20
(
1
):
648
60
.
11.
Colditz
G
,
Chia
KS
.
Invasive breast carcinoma: introduction and general features
. In:
Lakhani
SR
,
Ellis
IO
,
Schnitt
SJ
,
Tan
PH
,
van de Vijver
MH
, editors.
WHO classification of tumours of the breast
. 4th ed.
Lyon
:
IARC
. p.
14
31
.
13.
Gupta
S
.
Breast cancer: Indian experience, data, and evidence
.
South Asian J Cancer
.
2016
;
5
(
3
):
85
6
.
14.
Malvia
S
,
Bagadi
SA
,
Dubey
US
,
Saxena
S
.
Epidemiology of breast cancer in Indian women
.
Asia Pac J Clin Oncol
.
2017
;
13
(
4
):
289
95
.
15.
Arul
P
,
Masilamani
S
.
Application of National Cancer Institute recommended terminology in breast cytology
.
J Cancer Res Ther
.
2017
;
13
(
1
):
91
6
.
16.
Chakrabarti
I
.
FNAC versus CNB: who wins the match in breast lesions
.
J Cytol
.
2018 JulSep
;
35
(
3
):
176
8
.
17.
Kamatar
PV
,
Athanikar
VS
,
Dinesh
U
.
Breast fine needle aspiration biopsy cytology reporting using international academy of cytology yokohama system-two year retrospective study in tertiary care centre in Southern India
.
Njlm
.
2019
;
8
(
4
):
PO01
3
.
18.
Saha
A
,
Mukhopadhyay
M
,
Das
C
,
Sarkar
K
,
Saha
AK
,
Sarkar
DK
.
FNAC versus core needle biopsy: a comparative study in evaluation of palpable breast lump
.
J Clin Diagn Res
.
2016
;
10
(
2
):
EC05
8
.
19.
Mitra
S
,
Dey
P
.
Fine-needle aspiration and core biopsy in the diagnosis of breast lesions: a comparison and review of the literature
.
Cytojournal
.
2016
;
13
:
18
.
20.
Moschetta
M
,
Telegrafo
M
,
Carluccio
DA
,
Jablonska
JP
,
Rella
L
,
Serio
G
,
.
Comparison between fine needle aspiration cytology (FNAC) and core needle biopsy (CNB) in the diagnosis of breast lesions
.
G Chir
.
2014
;
35
(
7–8
):
171
6
.
21.
Tikku
G
,
Umap
P
.
Comparative study of core needle biopsy and fine needle aspiration cytology in palpable breast lumps: scenario in developing nations
.
Turk Patoloji Derg
.
2016
;
32
(
1
):
1
7
.
22.
Mitra
SK
,
Rajesh
R
,
Mishra
RK
,
Rai
P
,
Vahikar
S
,
Singhal
P
.
Comparative evaluation of FNAC, core needle biopsy and excisional biopsy in subtyping of breast lesions
.
J Path Micro
.
2016
;
2
(
1
):
9
15
.
23.
Joshi
M
,
Reddy
SJ
,
Nanavidekar
M
,
Russo
JP
,
Russo
AV
,
Pathak
R
.
Core biopsies of the breast: diagnostic pitfalls
.
Indian J Pathol Microbiol
.
2011
;
54
(
4
):
671
82
.
24.
Nassar
A
.
Core needle biopsy versus fine needle aspiration biopsy in breast: a historical perspective and opportunities in the modern era
.
Diagn Cytopathol
.
2011
;
39
(
5
):
380
8
.
25.
Ljung
BM
,
Drejet
A
,
Chiampi
N
,
Jeffrey
J
,
Goodson
WH
3rd
,
Chew
K
,
.
Diagnostic accuracy of fine-needle aspiration biopsy is determined by physician training in sampling technique
.
Cancer
.
2001
;
93
(
4
):
263
8
.
26.
Farras Roca
JA
,
Tardivon
A
,
Thibault
F
,
El Khoury
C
,
Alran
S
,
Fourchotte
V
,
.
Diagnostic performance of ultrasound-guided fine-needle aspiration of nonpalpable breast lesions in a multidisciplinary setting: the institute curie’s experience
.
Am J Clin Pathol
.
2017
;
147
(
6
):
571
9
.
27.
Field
AS
,
Schmitt
F
,
Vielh
P
.
IAC standardized reporting of breast fine-needle aspiration biopsy cytology
.
Acta Cytol
.
2017
;
61
(
1
):
3
6
.
28.
Schmitt
F
,
Vielh
P
.
Fine-needle aspiration cytology samples: a good source of material for evaluating biomarkers in breast cancer
.
Histopathology
.
2015
;
66
(
2
):
314
5
.
29.
Gibbons
CE
,
Quinn
CM
,
Gibbons
D
.
Fine-needle aspiration biopsy management of the axilla in primary breast carcinoma
.
Acta Cytol
.
2019
;
63
(
4
):
314
8
.
30.
Boughey
JC
,
Moriarty
JP
,
Degnim
AC
,
Gregg
MS
,
Egginton
JS
,
Long
KH
.
Cost modeling of preoperative axillary ultrasound and fine-needle aspiration to guide surgery for invasive breast cancer
.
Ann Surg Oncol
.
2010
;
17
(
4
):
953
8
.
31.
Modi
P
,
Haren
O
,
Jignasa
B
.
FNAC as preoperative diagnostic tool for neoplastic and nonneoplastic breast lesions: a teaching hospital experience
.
Indian J Med Res
.
2014
;
4
:
274
8
.
32.
Haobam
S
,
Thiyam
U
,
Sharma
DC
.
Cytomorphologic Al study of breast lesions with sonomammographic correlation
.
J Evol Med Dent
.
2015
;
4
(
81
):
14137
42
.
33.
Nguansangiam
S
,
Jesdapatarakul
S
,
Tangjitgamol
S
.
Accuracy of fine needle aspiration cytology from breast masses in Thailand
.
Asian Pac J Cancer Prev
.
2009
;
10
(
4
):
623
6
.
34.
Bajwa
R
,
Tariq
Z
.
Association of fine needle aspiration cytology with tumor size in palpable breast lesions
.
Biomedica
.
2010
;
26
:
124
9
.
35.
Hoda
R
,
Brachtel
E
.
International Academy of Cytology-Yokohama System for reporting breast fine-needle aspiration biopsy cytopathology: a review of predictive values and risks of malignancy
.
Acta Cytol
.
2019
;
63
(
4
):
292
301
.
36.
Field
AS
,
Zarka
MA
.
Breast
. In:
Field
AS
,
Zarka
MA
, editors.
Practical cytopathology: a diagnostic approach to fine needle aspiration biopsy
.
Amsterdam
:
Elsevier
;
2017
Chap. 5.
37.
Madubogwu
CI
,
Ukah
CO
,
Anyanwu
S
,
Chianakwana
GU
,
Onyiaorah
IV
,
Anyiam
D
.
Sub-classification of breast masses by fine needle aspiration cytology
.
Eur J Breast Health
.
2017
;
13
(
4
):
194
9
.
38.
Singh
A
,
Haritwal
A
,
Murali
B
.
Pattern of breast lumps and diagnostic accuracy of fine needle aspiration cytology; a hospital-based study from Pondicherry, India
.
Internet J Pathol
.
2010
;
11
(
2
):
1
6
.
39.
Muddegowda
PH
,
Lingegowda
JB
,
Kurpad
R
,
Konapur
P
,
Shivarudrappa
A
,
Subramaniam
P
.
The value of systematic pattern analysis in FNAC of breast lesions: 225 cases with cytohistological correlation
.
J Cytol
.
2011
;
28
(
1
):
13
9
.
40.
Challa
VR
,
Yale Guru
BG
,
Rangappa
P
,
Deshmane
V
,
Gayathri
DM
.
Cytological and pathological correlation of fnac in assessing breast lumps and axillary lymph node swellings in a public sector hospital in India
.
Patholog Res Int
.
2013
;
2013
:
695024
.
41.
Singh
N
,
Wells
CA
.
Invited review: assessment of accuracy in breast cytology
.
Cytopathology
.
2001
;
12
(
4
):
211
8
.
42.
Sneige
N
.
Utility of cytologic specimens in the evaluation of prognostic and predictive factors of breast cancer: current issues and future directions
.
Diagn Cytopathol
.
2004
;
30
(
3
):
158
65
.
43.
al-Kaisi
N
.
The spectrum of the “gray zone” in breast cytology. A review of 186 cases of atypical and suspicious cytology
.
Acta Cytol
.
1994
;
38
(
6
):
898
908
.