Analysis of US-Guided Core Needle Biopsy of Breast Lesions in a Tertiary Cancer Centre

Abstract

Purpose

Preoperative diagnosis of breast lesions is typically performed using ultrasound (US)-guided core needle biopsy (CNB), and this study analyses factors impacting its performance.

Methods

This study retrospectively analyzed CNBs conducted over 63 months at a tertiary cancer center to assess diagnostic accuracy (DA) and factors influencing biopsy success, including radiologists' experience, needle gauge size, number of cores, and strain elastography.

Results

Of 868 technically successful biopsies, 25 were diagnostically unsuccessful, yielding an overall diagnostic success rate of 97.1%. A statistically significant difference (p < 0.001) was observed between 14G and 18G needles, with diagnostic success improving from 94.4 to 99.8% using 14G needles. Combining Breast Imaging Reporting and Data System (BI-RADS) with elastography improved DA for benign lesions from 93.2 to 95.7% and for suspicious lesions (BI-RADS 4C/5) from 93.5 to 97.4%. In radio-pathologically discordant BI-RADS 4C/5 cases, malignancy was found in 17.5% (n = 7) upon surgical excision or clinical follow-up, highlighting the need for rebiopsy in discordant cases. The false-negative rate was 0.9%.

Conclusion

Overall, US-guided CNB demonstrated high DA, with sensitivity, specificity, and accuracy of 98.4, 100, and 98.8%, respectively—comparable to global standards. Implementing practice-enhancing measures, such as using 14G needles and ensuring that radiologists trained in breast imaging perform biopsies, can further improve technical and diagnostic success.

Keywords

BI-RADS - core needle biopsy - diagnostic accuracy - elastography - VABB

Introduction

Breast cancer is the most common cancer in women worldwide and in India.[1] A steady increase in the incidence of breast cancer has been observed in both the high-income and low- and middle-income countries (LMICs), with almost 45% of breast cancer cases and 55% of breast cancer deaths seen in LMICs.[2] Improvement in awareness, in imaging technology such as full-field digital mammography (MG), high-resolution ultrasonography, and health infrastructure, has led to a shift toward relatively earlier presentation in the stages of breast cancer (stage migration), and higher rates of detection of smaller impalpable masses in clinically asymptomatic women during diagnostic and opportunistic screening. MG and ultrasound (US) of the breast form the mainstay of imaging as primary modalities for initial evaluation of breast lumps. Breast Imaging Reporting and Data System (BI-RADS) stratification is used for assigning a category to the findings detected, indicating the degree of suspicion of malignancy,[3] with robust accompanying recommendations. As technology evolves, 2D MG has encompassed quasi-3D imaging with tomosynthesis and contrast-enhanced MG with improvements in accuracy. Similarly, the US also encompasses advanced imaging such as elastography and contrast-enhanced US. The more widely used elastography estimates the stiffness of a lesion, characterizing it further, and may help in cases of indeterminate lesions, such as a BI-RADS 3 lesion, to be appropriately triaged as BI-RADS 2 or BI-RADS 4, thereby deciding on the need for a biopsy.

Preoperative pathological diagnosis is largely made by percutaneous breast biopsy, either by palpable method or under image guidance. It is best performed by an image-guided approach, particularly for impalpable, small, and complex lesions. Of all breast imaging modalities, US-guided core needle biopsy (CNB) is the preferred method of biopsy for all lesions visible on US, for its ease in technique, option of targeting more suspicious, solid, and stiffer elastography components of the lesion, and feasibility of avoiding traversing vessels, reducing the incidence of hematoma. In the presence of suspicious microcalcifications visible only on MG, stereotactic biopsy or hook wire localization is performed, whereas for CEM or MRI only detected lesions, the respective modality is used for biopsy guidance. Although core biopsies can be performed with these modalities, they are better targeted by the Vacuum-Assisted Breast Biopsy (VABB) technique to procure larger volumes of samples, improving diagnostic accuracy (DA). Our study focuses on the analysis of CNB, which is the mainstay of breast interventions, accounting for more than 95% biopsies performed in a breast unit. The DA of percutaneous US-guided CNB is high, varying in the literature from 84 to 98%.[4] [5] The differences in results depend on the biopsy methods, size of the target lesion and its morphology, gauge of the needle, and number of cores. In this study, we analyzed a large series of US-guided CNB of breast lesions performed over a period of 63 months in a tertiary cancer center to assess the DA and quantifiable parameters that may impact the success of a biopsy, such as the gauge of needle, number of cores, and, role of strain elastography. Nonquantifiable parameters, such as the role of a dedicated breast radiologist performing the procedure and streamlining workflow and appointment schedules, were also assessed to reflect the turnaround time and waiting lists in a busy department.

Methods

A retrospective study was conducted on all consecutive US-guided CNB of breast lesions performed at a tertiary cancer center from January 2017 to March 2022 after clearance from the institutional ethical committee and in accordance with ICMR revised national ethical guidelines for biomedical and health research involving human participants (2017). Waiver of consent was approved as the study was retrospective and involved less than minimal risk with no direct contact between investigators and patients.

Only those patients for whom US and/or MG and histopathology reports (HPR) were available were included in the study. Patients undergoing US-guided VABB were excluded from the study.

The details of the procedure (needle length, needle gauge, number of cores obtained during the biopsy, and procedure complications) were obtained from the radiology procedure report, which is documented as per department protocol. Biopsy cases were documented as (1) biopsy of the index lesion, (2) additional lesion in the same breast, or (3) lesion in the contralateral breast. The indications for biopsy were documented as (1) histopathological diagnosis and (2) procurement of tissue for immunohistochemistry in a known case of malignancy. The lesion characteristics, such as size, morphology, laterality, clock position, elastography scores, and final BI-RADS assessment category, were obtained from radiology images stored on Picture Archiving and Communication System (PACS). The subsequent management, HPRs, and outcome of follow-up were also documented from PACS. A subset of patients who were lost to follow-up following the biopsy was excluded from the analysis of diagnostic success.

Parameters for Analysis

Diagnostic success: defined as the outcome of biopsy leading to appropriate management, either by surgical/medical intervention for suspicious lesions or follow-up for benign lesions. Biopsies yielding nonrepresentative, inadequate, or scanty tissue that did not adequately suggest the diagnosis or necessitated a repeat biopsy were considered as diagnostically unsuccessful biopsies.

DA of CNB of suspicious breast lesions: defined as the proportion of biopsies correctly identifying the true status (malignant or benign) of a tissue sample, encompassing both true positive and true negative results, as per the HPR of surgical specimen or biopsy specimen, in order of preference.[6]

Comparison of performance between 18G and 14G needles: based on comparing the diagnostic success and DA in the two cohorts.

Rate of Underestimation: number of cases diagnosed as ductal carcinoma in situ (DCIS) on CNB that turned out to be invasive malignancy on the final excision histology as a percentage of the total number of biopsies diagnosed as DCIS.

Diagnostic performance of Tsukuba strain elastography scores: based on DA, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of predefined Tsukuba scores on a “point-system,” characterizing lesions as benign or malignant with histopathology as gold standard.

Statistical analyses were performed using SPSS (the statistical package for social sciences), IBM Corp. Released 2017. IBM SPSS Statistics for Windows, Version 25.0. Armonk, NY: IBM Corp. Descriptive statistics were used to summarize the dataset. Categorical data were summarized in terms of counts and percentages. Normality of continuous data was assessed using Shapiro–Wilk's or Kolmogorov–Smirnov test. If the data were normally distributed, then mean (SD) was used to summarize the data, whereas if it was nonnormally distributed, then median (IQR) was used to summarize the data. The rate of diagnostic success and DA of biopsies performed using 14G and 18G needles and parameters such as number of cores obtained, incidence of complications, and average lesion size between the two groups were compared using the t-test or Mann–Whitney U test depending upon normality of the variables.

Results

A total of 927 US-guided needle biopsies of breast lesions were performed during the period from January 2017 to March 2022. Of these, 58 were less than 5 mm and were biopsied by VABB and hence excluded. A total of 869 were included in the study as per the eligibility criteria defined earlier ([Fig. 1]). There was a gradual increase in the annual number of CNB performed from 2017 to 2019, with a fall in the numbers in 2020 owing to the COVID-19 pandemic and its associated lockdown. The numbers significantly increased in 2021 after the lifting of the lockdown restrictions.

Fig. 1 Selection of study cohort.

One out of 869 biopsies was technically unsuccessful. The lesion was entirely cystic without a solid component. On insertion of the coaxial needle, the fluid content was aspirated, and no core specimen could be obtained. The fluid was sent for cytological analysis, which yielded scanty histiocytes and inflammatory cells. On 6-month follow-up, the patient had no lesion on palpation or imaging, confirming appropriateness of biopsy converted to cytology.

Of the 868 technically successful biopsies, 861 were females and 7 were males. The mean age of the participants at the time of the breast biopsy was 49.4 years (SD = 11.9; range: 18–86 years), with 62.7% of lesions representing malignancy (544 malignant, 291 benign, 8 normal breast parenchyma, 25 inadequate). Overall, the sensitivity, specificity, and accuracy of CNB were 98.3, 100, and 98.8%, respectively ([Table 1]).

Table 1 Overall sensitivity, specificity, NPV, PPV, and accuracy of US-guided CNB of breast lesions

True positive	True negative	False positive	False negative	Accuracy	Sensitivity	Specificity	NPV	PPV
544	223	0	9	98.8%	98.3%	100%	96.1%	100%

Note: Twenty-five diagnostically unsuccessful biopsies were excluded, and 67 benign biopsies (out of 299) could not be assessed as true negatives as they were lost to follow-up.

Of the 868 technically successful biopsies, 835 were diagnostically successful as the patients were directed toward appropriate treatment as per the histopathologic report. Twenty-five biopsies were diagnostically unsuccessful, reported as scanty tissue, necrosis, or benign parenchyma, rendering the need for repeat biopsy ([Table 2]). Of these, 15 were malignant on repeat biopsy and 10 were benign. Of 868 biopsies, 8 were excluded from the analysis of diagnostic success as patients were lost to follow-up after the biopsy.

Table 2 Case-wise details of all diagnostically unsuccessful biopsies (n = 25)

Sr. no.	Needle gauge	Size (cm)	BI-RADS	HPR on CNB	HPR on follow-up
1	18	3.2	4C	Infarcted hyalinized core	IBC
2	18	1.2	5	Unremarkable fibroadipose tissue	ILC
3	18	2.3	3	Nonrepresentative	Inflammatory
4	18	1.7	4A	Inadequate	Duct ectasia
5	18	1.4	4B	Fibrofatty parenchyma with scanty atypical cells	IBC
6	18	2.1	5	Fragmented parenchyma with scanty cells	IBC
7	18	2.1	4B	Normal breast tissue	Benign breast parenchyma
8	18	3.6	5	Nondiagnostic necrotic tissue	Therapy-related changes
9	18	Diffuse lump	4A	Fibrofatty tissue and clot	Fibrocystic disease of the breast
10	18	1.9	5	Fibromuscular tissue	IBC
11	18	1.4	4C	Fibroadipose tissue	IBC
12	18	5.4	4C	Nonrepresentative	Infarcted IDP
13	18	1.1	4A	Atypical cells are inadequate for diagnosis	Fibroadenoma
14	18	2.3	3	Fibrocollagenous tissue is inadequate for diagnosis	Benign on follow-up
15	18	2.1	5	Fibrocollagenous tissue	IBC
16	18	15	NA	Necrosis	Phyllodes, malignant
17	18	2.6	4C	Scanty fragments of mucinous cancer were seen, inadequate for opinion	Mucinous carcinoma
18	18	3	5	Necrosis	Metaplastic Ca
19	18	4.3	4A	Nondiagnostic	FCC
20	18	3	4C	Scanty atypical cells, inadequate tissue for diagnosis	IBC
21	18	2	5	Nonrepresentative	DCIS
22	18	1.9	4B	Inadequate	Metastatic breast cancer
23	18	2	5	Normal breast tissue	Microinvasive breast cancer
24	18	1.8	4C	Nondiagnostic	DCIS
25	14	0.7	4A	Inadequate tissue for opinion	Benign on follow-up

Abbreviations: BBP, benign breast parenchyma; DCIS, ductal carcinoma in situ; FCC, fibrocystic change; IBC, invasive breast cancer; IDP, intraductal papilloma; ILC, invasive lobular cancer.

Factors Affecting the Outcomes

1. Needle gauge: biopsies performed from 2017 to December 2020 used an 18-G needle of either 9 or 10 cm length, depending on the manufacturer. In December 2020, there was a change in breast biopsy practice in our department with the adoption of the internationally recommended standard of 14-G needles for breast biopsy. Our dataset included 433 biopsies performed using 18G needles and 435 biopsies performed using 14G needles. There was a statistically significant (p < 0.001) difference in the diagnostic success rates of biopsies performed with 14G and 18G needles, with higher diagnostic success seen with 14G needles, increasing from 94.4 to 99.8%. Additionally, the percentage of lesions requiring re-biopsy was statistically significantly greater for 18G needles (24/433 for 18 G vs. 1/435 for 14 G needles, p-value < 0.001; [Table 2]). No significant complication was observed in any of the patients.

2. Size of lesions: of 868 lesions, 17 had no definite size mentioned in the radiological report, nor could it be extrapolated from the available imaging (this may be attributed to large lesions partially included in the imaging frame or images acquired without callipers). Of the lesions for which sizes were available (n = 851), the mean size was 2.49 cm. A subset analysis was performed for the subcentimeter-sized lesions (n = 103), of which 51 were benign and 52 were malignant on final histopathology. Overall, the diagnostic success rate amongst the subcentimeter-sized lesions was 99%. There was no statistical significance between the diagnostic success rates of subcentimeter-sized and > 1 cm-sized lesions (p < 0.18). Interestingly, except for one lesion, the other 24 of 25 lesions that required a re-biopsy were above 1 cm in size, pointing toward a greater precision in smaller lesions.

3. Tsukuba elastography score: of the 868 technically successful biopsies, Tsukuba strain elastography scores were available for 342 lesions ([Table 3]). A lesion scoring from 1 to 3 points was categorized as benign, while a lesion scoring at 4 or 5 points was categorized as malignant.[7] The mean elasticity score for malignant lesions was 3.76 ± 0.7, and for benign lesions was 3.1 ± 0.7. The sensitivity, specificity, PPV, NPV, and DA of the elastography score categorized as benign or malignant were 66.6, 70.5, 80, 53, and 67.2%, respectively.

Table 3 Elastography assessment and HPR diagnosis cross-tabulation

		HPR
		Malignant	Benign	Total
Elastography assessment	Malignant (score: 4–5)	144	36	180
	Benign (score: 1–3)	76	86	162
	Total	220	122	342

A false negative benign elastography assessment was in 76 cases, wherein 53 were IBC on final histopathology. On analysis of these 53 cases, B-mode features suggested necrosis in 25 cases, thereby rendering them suspicious. Two cases (BIRADS 4a) showed internal necrosis; the rest 74 cases were given a final BI-RADS assessment of 4B or above, with 87% of these categorized as BI-RADS 4C or 5 ([Fig. 2]).

Fig. 2 Grayscale US image (A) shows a taller-than-wider hypoechoic lesion with perilesional echogenic halo (arrows). On strain elastography (B), it shows a Tsukuba score of 2, favoring benignity. In view of the grayscale morphology, it was assigned BI-RADS 4B with CNB suggestive of papillary carcinoma. In this case, elastography was a false negative.

False positive elastography assessment was seen in 36 cases, of which the majority, that is, 56% (n = 20), were fibroadenomas, with predominance in 31% (n = 11) hyalinized and sclerosed subtypes. These are known to be hard on elastography, and interpretation of these lesions should be done in conjunction with the rest of the BI-RADS descriptors to maintain higher accuracy[8] ([Fig. 3]).

Fig. 3 Grayscale US image (A) shows an oval hypoechoic lesion in parallel orientation. On strain elastography (B), it shows a Tsukuba score of 4, favoring malignancy. In view of the grayscale morphology, it was assigned BI-RADS 4A with CNB suggestive of hyalinized fibroadenoma. In this case, elastography was a false positive.

Subset Analysis of Performance of the Tsukuba Score according to BI-RADS Category

The combined DA of BI-RADS and elastography to predict benign lesions improves from 93.2 to 95.7% as opposed to using BI-RADS alone. In case of suspicious lesions, that is, BI-RADS 4C and 5, there was an improvement in DA from 93.5 to 97.4% when combined with the elastography assessment ([Table 4]).

Table 4 Elastography assessment according to BI-RADS category

Category	Total number of cases	Benign on HPR	Accuracy
• BI-RADS 4A	74	69	93.2%
• Combined BI-RADS + elastography assessment as soft	47	45	95.7%
• BI-RADS 4B	50	26	52%
• Combined BI-RADS 4B + elastography assessment as soft	27	19	70%
Category	Total number of cases	Malignant on HPR	Accuracy
• BI-RADS 4B	50	24	48%
• Combined BI-RADS 4B + elastography assessment as hard	23	16	69.5%
• BI-RADS 4C and BI-RADS 5	186	174	93.5%
• Combined BI-RADS 4C/5 + elastography assessment as hard	117	114	97.4%

For BI-RADS 4B lesions, the DA to detect malignancy was 48% when using BI-RADS alone; however, in combination with elastography assessment, the DA improved to 69.5%. On the other hand, to detect benign lesions, a BI-RADS assessment of 4B, with elastography assessment as soft, improved the accuracy from 52 to 70%. This implies that elastography can be used with B-mode US to stratify 4B lesions, with hard lesions (Elastography score: 4–5) warranting an upgrade to BI-RADS 4C and soft lesions (Elastography score: 1–3) warranting a downgrade to BI-RADS 4A. This highlights the importance of the supplemental value of elastography in the assessment of breast lesions, emphasizing its reliability not as an independent feature, but in conjunction with B-mode US features. Higher Tsukuba scores increase the likelihood of malignancy, and when interpreted with the rest of the imaging features on B-mode, improve the DA. This has a bearing on the evaluation of the lesions in the event of radiological-pathological discordance.

DA and Radiological-pathological discordance: [Table 5] summarizes the DA as per the BI-RADS categories.

Table 5 Diagnostic accuracy of breast lesions as per BI-RADS assessment

BI-RADS category	Number of benign results	Number of malignant results	% Malignant	Reference
BI-RADS 3	20	0	100%	≤2%
BI-RADS 4A	143	14	8.9%	>2% but ≤ 10%
BI-RADS 4B	66	45	40.5%	>10% to ≤ 50%
BI-RADS 4C	28	283	90.9%	>50% to <95%
BI-RADS 5	5	178	97.3%	≥95%

Traditionally, radiological and pathological discordance is considered in two contexts—for a BI-RADS category 4C or 5 when reported as benign on histopathology warranting a rebiopsy, and for a BI-RADS category 2 or 3, where follow-up is recommended over biopsy, but its histopathology is reported as malignancy, in the eventuality where biopsy was performed for a clinical suspicion, physician's recommendation or patient's preference.

Of the 22 lesions assigned BI-RADS category 2 or 3, none were malignant. Forty-two of 311 BIRADS 4C lesions, and 11 of 183 BIRADS 5 lesions were benign on CNB. Among the benign discordant biopsies of BIRADS 4C, 11 were lost to follow-up, 28 remained benign on final HPR or follow-up, and 3 were malignant on final HPR. Among the benign discordant biopsies of BIRADS 5 lesions, two were lost to follow-up, five remained benign on final HPR (diagnoses of intraductal papilloma in two cases, granulomatous mastitis, inflammatory etiology, and duct ectasia), and four were malignant on final HPR. Together, repeat histological evaluation of the discordant BI-RADS 4C/5 lesions on surgical excision or clinico-radiologic follow-up showed malignancy in 17.5% (n = 7) cases, emphasizing the importance of re-biopsy in discordant cases.

Malignancies with features of benignity: for the purpose of the study, we additionally analyzed BI-RADS 4A lesions (with only a 2–10% chance of malignancy), which turned out to be malignant on histopathology. This was to identify the imaging morphology of certain malignancies, falsely masking the lesions to fall into a category of low suspicion. Twenty-one percent (n = 3) were triple-negative cancers, which are prone to being misdiagnosed as benign on imaging as they often appear oval or round masses, more likely to have circumscribed margins, and to show posterior attenuation.[9] [10] 36% (n = 5) were diagnosed as invasive lobular carcinoma (ILC), papillary carcinoma, mucinous carcinoma, and metaplastic carcinoma ([Table 6]).

Table 6 BI-RADS 4A lesions with malignant histopathology report

Case no.	Vascularity	Size (cm)	HPR
1	Yes	1.2	IBC
2	No	2.2	TNBC
3	Yes	3.5	TNBC
4	NA	2.1	ILC
5	NA	1.5	IBC
6	No	1.6	IBC
7	Yes	2.4	IBC
8	Yes	1.6	TNBC
9	Yes	2.1	Metaplastic Ca
10	Yes	3.1	Mucinous carcinoma
11	NA	1.5	Papillary Ca
12	Yes	2.4	IBC
13	No	0.5	IBC
14	No	0.8	Metastases

Underestimation of the invasive component of DCIS: of the 868 lesions, 26 were found to have DCIS on HPR of the CNB. Of these, four were excluded as they were lost to follow-up and no surgical histopathology was available. Fifteen of the remaining 22 showed invasive disease on surgical HPR, giving an underestimation rate of 68.1%. Of these 15, 5 were microinvasive on final surgical HPR. The underestimation rate for frank invasive breast cancer was 45%. Of the 15 cases with underestimation, 73.4% (n = 11) had ultrasonographic imaging morphology of mass-forming disease, favoring invasive malignancy ([Fig. 4]). The mean size of the lesions in this subset was 2.15 cm, with a range of 0.6 to 4.7 cm, and there was a trend of increasing underestimation with an increase in the size of the lesion.

Fig. 4 Grayscale US image shows a taller-than-wide hypoechoic lesion favoring a malignancy. CNB showed DCIS; however, final surgical HPR was suggestive of invasive breast carcinoma. The presence of a mass on US is more in favor of invasive malignancy.

False negative rate: false negative rate is defined as the number of false negative results expressed as a percentage of the total number of malignancies.[4] Of the 868 technically successful biopsies, 299 yielded a benign/normal histopathological report. Sixty-seven of these had no imaging or histopathological follow-up as they were radiologically concordant. Of the remaining 232 with follow-up, 9 were false-negative benign diagnoses, with subsequent surgical histopathology or clinical follow-up suggestive of malignant disease. However, of these, three were diagnosed as phyllodes on CNB, which are known to have heterogeneous stromal properties. Studies have shown that the rate of concordant diagnosis between CNB and surgical excision in the case of phyllodes is approximately 60% and hence, diagnosis as well as grading of these tumors based on the histological findings in CNBs has limitations.[11] Differentiation of phyllodes from fibroadenomas on histology includes assessment of the capsule, where fibroadenomas have a true, smooth capsule and phyllodes have a nodular capsule with perforations and finger-like projections.[12] [13] Inclusion of the capsule in the core biopsy specimen may be beneficial in improving the DA of CNB in case of suspected phyllodes. A diagnosis of phyllodes, thus anyways mandates complete histopathology evaluation. Excluding the cases of phyllodes (n = 3), the false negative rate was 0.9%, with a significant difference between the two needle gauges (1.5% with 18G vs. 0.4% with 14G; [Table 7]).

Table 7 Cases one to six, which were false negatives on CNB

Case no.	Needle gauge	Size (cm)	Morphology	BI-RADS	HPR on CNB	Follow-up diagnosis
1	18	1.3	Solid	4B	Hyalinized stroma	DCIS (after 1 y)
2	18	2.3	Solid	4C	Fibrosis and inflammation	IBC
3	18	2.5	Solid	5	Fibroadipose tissue	IBC
4	18	2.9	Solid-cystic	4B	IDP	Papilloma/DCIS-intermediate grade on final
5	18	1.7	Solid	4C	FCC	IBC
6	14	4.9	Solid-cystic	4C	BBP	DCIS

Abbreviations: BBP, benign breast parenchyma; DCIS, ductal carcinoma in situ; FCC, fibrocystic change; FCC, fibrocystic change; IBC, invasive breast cancer; IDP, intraductal papilloma.

Note: Three cases of phyllodes were excluded here.

Radiological causes of false-negative results of core needle biopsies in the analyzed data were divided into two groups:

1. Inadequate sampling of the material: the causes include the selection of the wrong biopsy system. Intraductal lesions with calcifications carry a high likelihood of DCIS and may be better targeted using VABB. Other causes can be poor visualization of the lesion or needle and nonconfirmation of in situ positioning of the needle by verifying the needle tip-mass relation in both orthogonal planes. Selection of a smaller gauge, acquisition of fewer cores, and lesser expertise are factors that are operator-dependent. Of the nine false negative biopsies, only one was performed by a radiologist specialized in breast imaging (over 10 years of experience), while others were performed by nonspecialist/general radiologists (around 4 years of experience). Further, on analysis of the cohort of lesions biopsied by nonspecialist/general radiologists with an 18G needle versus those performed by a breast imaging radiologist with a 14G needle, there was a statistically significant improvement in the DA from 94.4 to 99.8% (p < 0.001).

2. Histopathological nonhomogeneity of the lesion: one of the cases, diagnosed as complex papilloma with atypical ductal hyperplasia on CNB, eventually upgraded to DCIS on surgical HPR ([Fig. 5]). Underestimation rates of malignancy in the case of papillary lesions can be as high as 25%, with a higher underestimation rate in lesions more than 1 cm.[14] Therefore, a diagnosis of atypical papilloma warrants surgical excision to rule out malignancy, which may be seen in up to 67% of cases.[15] In our dataset, the lesion was palpable, measured 2.9 cm, and showed atypical features on CNB; hence, surgical excision was justified. Although this was a discordance, the atypical features suggested an alternative diagnosis and resulted in appropriate management.

Fig. 5 Grayscale US image (A) shows an irregular isoechoic to hypoechoic mass with angular margins, assigned as BI-RADS 4B. CNB image (B) shows appropriate targeting of the lesion, with the needle within the lesion (arrow).

Discussion

Since its introduction in the 1980s, CNB has gained popularity as a tool for assessing both palpable and nonpalpable/radiologically detected breast abnormalities, which has transformed the practice of preoperative diagnoses of breast lesions in both symptomatic and screen-detected patients.

In our study, we retrospectively analyzed 868 technically successful CNBs of breast lesions done in our department under US guidance. Overall sensitivity, specificity, NPV, PPV, and accuracy to detect malignancy were 98.3, 100, 95.4, 100, and 98.8%, respectively. A study by El-Sayed et al analyzed over 20,001 cases over a period of 10 years (1997–2007), and found sensitivity to be 96.4% and specificity to be 99.7%. Further, they found a steady increase in accuracy over the years, from 90.5 to 99.3% corresponding to the improvement in biopsy and imaging techniques, which was also seen in our analysis, particularly with the usage of 14G needles and performance of biopsies by radiologists specialized in breast imaging.[4] Another study by Hari et al showed image-guided biopsy to have a sensitivity and specificity of 96.3 and 100%, respectively, as opposed to 46.7 and 100% in palpation-guided biopsy.[16]

A study by Radhakrishna et al analyzed the concordance of radiological interpretation and histopathology findings in 467BI-RADS category 3 to 5 lesions following a core biopsy, which showed 93.25% PPV for BI-RADS 5 lesions for malignancy and 98.4% NPV of BI-RADS category 3 lesions for cancer.[17] Our data showed PPV for cancer in BI-RADS 5 lesions to be 97.2% and NPV of BI-RADS category 3 lesions to be 100%, which is concordant with the reference standard established by the ACR BI-RADS Atlas Fifth Edition.[3]

The false negative rate in our study was 1.5%. This result falls in the range reported in literature from 0.8 to 6%.[18] [19]

In our study, the malignancy rates of BI-RADS 4A, 4B, and 4C lesions were 8.9, 40.5, and 90.9%, respectively, which are consistent with the reference standard.[3] We further analyzed the cases where there was radiological-pathological discordance, particularly the BI-RADS 4C and 5 lesions, which were benign on CNB. Seventeen point five percent of these cases showed malignancy on repeat sampling, which highlights the importance of re-biopsy in discordant cases.

We also analyzed the BI-RADS 4A lesions, which were malignant on HPR. Of these, 21% (n = 3) were triple-negative cancers, which are prone to being misdiagnosed as benign on imaging as they often appear oval or round masses, more likely to have circumscribed margins, and to show posterior.[10] ILC, papillary carcinoma, mucinous carcinoma, and metaplastic carcinoma, were the other encountered histopathologies, all of which tend to show atypical imaging features often mimicking benign etiologies, and hence appropriately categorized as BI-RADS 4A.[20] [21] [22] [23] [24]

Further, on analysis of the six cases with false negative CNB results (excluding the three cases of phyllodes), all were categorized as BI-RADS 4B and above, with suspicious morphology. Based on radiological-pathological discordance, a repeat biopsy or surgical excision was performed.

Radiological-pathological discordant cases should be prompted by the reporting pathologist, re-evaluated by the radiologist and surgeon in conjunction, and a biopsy should be repeated to avoid missing a malignancy. In some cases, despite a concordance, a repeat biopsy may be warranted, driven by the inherent pathology of the finding. This includes the entity of B3 pathologies, where inherent heterogeneity in the tumor milieu demands wider and larger sampling, either by an extended vacuum-assisted biopsy or image-guided surgical excision biopsy.[25]

Our data was collected over a span of 63 months from January 2017 to March 2022. Biopsies performed from January 2017 to December 2020 used an 18-G needle of either 9 or 10 cm length, depending on the manufacturer. In December 2020, there was a change in breast biopsy practice in our department with the adoption of 14-G needles for breast biopsy. Our dataset includes 433 biopsies performed using 18G needles and 435 biopsies performed using 14G needles. A study by Uematsu et al studied the performance of 18G US-guided CNB in breast lesions, and found a sensitivity of 96% in the diagnosis of malignancy.[26] A more recent study by Lai et al comparing 14G needles to 16G needles showed no significant difference in sensitivity and specificity between the two groups. However, better overall accuracy (p = 0.02), less underestimation (p < 0.001), and lower false-negative (p = 0.02) rates were found for the 14-gauge CNB.[27] Our study demonstrated higher diagnostic success rates (p < 0.001) with the 14G needle (99.8%) as compared with the 18G needle (94.4%). There were more cases with inadequate tissue for diagnosis, scanty cells, and nonrepresentative sampling with an 18-G needle. Underestimation rates in our study were not statistically different between the two needle gauge CNB (p = 0.35); however, overall, lesser underestimation was observed with 14G needles. The false negative rate was also not statistically different between the two needle gauge CNB (p = 0.38), but was lower with 14 G needles.

Prior studies have shown a minimum of two to four cores to be sufficient for the establishment of a diagnosis of malignancy.[28] [29] Hsin-Ni and Chen have shown that adequate tissue for immunohistochemical biomarkers can usually be obtained through one core US-guided biopsy, except under certain circumstances, such as small size of lesion (≤1 cm), heterogeneous tumor components, and fibrotic texture. Further, additional cores are necessary to account for the loss of specimen due to crush artifacts, which are seen more often with large bore CNB.[30]

Gruber et al demonstrated that a minimum of two cores were required with a 14G needle and five cores with a 16G needle for higher DA.[18] In our study, the mean number of cores obtained was higher in the 14G group (5.82) as opposed to the 18G group (5.15). This may factor in influencing the higher diagnostic success with 14G needles. Further, the use of 14G needles in our department coincides with the initiation of revised breast intervention protocols, with biopsies being performed by dedicated breast radiologists, which might have further led to better selection of lesions and appropriate sites to be biopsied. A similar analysis by Brenner et al on the effect of operator experience and number of samples on DA of stereotactic biopsy showed an increase in the DA with an increase in the number of core biopsy samples obtained for any given lesion seen on mammograms and with increased experience in performing the procedure.[31]

Underestimation of the invasive component of DCIS ranges in the literature from 16.2 to 47.8%.[4] [32] [33] [34] Our study showed a higher underestimation rate of 68.1%. The underestimation rate was higher with the 18G needle (83.3%). An analysis of underestimation rates reveals that various factors, such as lesion size, higher BI-RADS assessment, gauge of needle, and technique of biopsy, affect the rate of underestimation. Alejandro et al showed the highest underestimation rates in lesions with an initial size > 2 cm compared with those with a size <2 cm.[32] Other studies have shown that the presence of a palpable lump and a nodular mass on US negatively affected underestimation rates.[33] [35] [36] Schulz et al showed that with a palpable mass, which remained the only independent predictor of invasion after multivariate adjustment, and the presence of at least three of the remaining five risk factors (a mass lesion on US, the presence of a mammographically detectable mass/architectural distortion/density, a BI-RADS score of 5, a lesion diameter ≥ 50 mm, and ≥ 50% of histologically affected ducts), the predicted probability of invasion was 56.0%.[36] Our data showed a mean lesion size of 2.1 cm, with 73.4% of cases showing a mass lesion on US, which explains the higher rate of underestimation. A higher underestimation rate would be expected in LMICs such as India, where the average size at presentation is often larger than 2 cm, and patients tend to present with clinically and radiologically apparent masses. Further, the subset of DCIS diagnoses in our data provided a small sample size for analysis. Although an accurate target of the lesions yielded a diagnostic CNB result, underestimation of the invasive component remains a concern. Retrospective analysis of our cases showed that of the 15 cases with underestimation, at least four cases had imaging appearance of an aggregate of ducts, and six cases had a target lesion size equal to or less than 9 mm; such lesions would be better targeted via VABB to improve the underestimation rates.

Schaefer et al showed a mean elasticity score of 4.1 ± 0.9 for malignant lesions, and 2.1 ± 1.0 for benign lesions (p < 0.001). With a best cut-off point between elasticity scores 3 and 4, sensitivity was 96.9%, and specificity was 76%.[37] Sensitivity of strain elastography ranges between 60 and 86%.[38] [39] [40] Studies have shown the supplemental value of elastography to B-mode US in improving overall accuracy.[37] [41] Our study showed overall sensitivity, specificity, and accuracy of elastography to be 63.6, 70.5, and 66%, respectively.

Conclusion

Incorporating practice-changing steps, such as utilizing 14G needles and the performance of biopsies by a radiologist with training in breast imaging, can improve technical and diagnostic success. Statistically significant improvement in diagnostic success is seen with 14G compared with 18G needles, with statistically nonsignificant improvement in underestimation rate, rate of complications, and false negative rate. DA of US-guided CNB for breast lesions is high, with sensitivity, specificity, and accuracy of biopsies being 98.3, 100, and 98.8%, comparable to global standards. Supplemental utility of strain wave elastography can improve diagnostic performance. US-guided biopsies offer high diagnostic success with minimal complications, making them essential for diagnosis and treatment.

Conflict of Interest

None declared.

• References

• 1 Express TF. Breast cancer now more common than lung, WHO says. The Financial Express; 2021
• 2 Tfayli A, Temraz S, Abou Mrad R, Shamseddine A. Breast cancer in low- and middle-income countries: an emerging and challenging epidemic. J Oncol 2010; 2010: 490631
• 3 D'Orsi CJ, Sickles EA, Mendelson EB, Morris EA. (2013) ACR BI-RADS Atlas Breast Imaging Re-porting and Data System. American College of Radiology, Reston, VA, USA. - References - Scientific Research Publishing;
• 4 El-Sayed ME, Rakha EA, Reed J, Lee AH, Evans AJ, Ellis IO. Audit of performance of needle core biopsy diagnoses of screen detected breast lesions. Eur J Cancer 2008; 44 (17) 2580-2586
• 5 Brancato B, Crocetti E, Bianchi S. et al. Accuracy of needle biopsy of breast lesions visible on ultrasound: audit of fine needle versus core needle biopsy in 3233 consecutive samplings with ascertained outcomes. Breast 2012; 21 (04) 449-454
• 6 Abalı H, Şimşek VN, Uslu B, Tokgöz AF, Tural ÖS. Factors Influencing Diagnostic Success of Computed Tomography-guided Transthoracic Needle Biopsy in Intrathoracic Lesions: An Experience of a Reference Chest Disease Hospital. Istanbul Med J 2023; 24 (01) 14-21
• 7 Farrokh A, Wojcinski S, Degenhardt F. [Diagnostic value of strain ratio measurement in the differentiation of malignant and benign breast lesions]. Ultraschall Med 2011; 32 (04) 400-405
• 8 Balleyguier C, Ciolovan L, Ammari S. et al. Breast elastography: the technical process and its applications. Diagn Interv Imaging 2013; 94 (05) 503-513
• 9 Du HY, Lin BR, Huang DP. Ultrasonographic findings of triple-negative breast cancer. Int J Clin Exp Med 2015; 8 (06) 10040-10043
• 10 Alluri P, Newman LA. Basal-like and triple-negative breast cancers: searching for positives among many negatives. Surg Oncol Clin N Am 2014; 23 (03) 567-577
• 11 Choi J, Koo JS. Comparative study of histological features between core needle biopsy and surgical excision in phyllodes tumor. Pathol Int 2012; 62 (02) 120-126
• 12 Oprić S, Oprić D, Gugić D, Granić M. Phyllodes tumors and fibroadenoma common beginning and different ending. Coll Antropol 2012; 36 (01) 235-241
• 13 Zhou ZR, Wang CC, Sun XJ, Yang ZZ, Yu XL, Guo XM. Diagnostic performance of core needle biopsy in identifying breast phyllodes tumors. J Thorac Dis 2016; 8 (11) 3139-3151
• 14 Boufelli G, Giannotti MA, Ruiz CA. et al. Papillomas of the breast: factors associated with underestimation. Eur J Cancer Prev 2018; 27 (04) 310-314
• 15 Sydnor MK, Wilson JD, Hijaz TA, Massey HD, Shaw de Paredes ES. Underestimation of the presence of breast carcinoma in papillary lesions initially diagnosed at core-needle biopsy. Radiology 2007; 242 (01) 58-62
• 16 Hari S, Kumari S, Srivastava A, Thulkar S, Mathur S, Veedu PT. Image guided versus palpation guided core needle biopsy of palpable breast masses: a prospective study. Indian J Med Res 2016; 143 (05) 597-604
• 17 Radhakrishna S, Gayathri A, Chegu D. Needle core biopsy for breast lesions: an audit of 467 needle core biopsies. Indian J Med Paediatr Oncol 2013; 34 (04) 252-256
• 18 Gruber I, Oberlechner E, Heck K. et al. Percutaneous ultrasound-guided core needle biopsy: comparison of 16-gauge versus 14-gauge needle and the effect of coaxial guidance in 1065 breast biopsies - a prospective randomized clinical noninferiority trial. Ultraschall Med 2020; 41 (05) 534-543
• 19 Sauer G, Deissler H, Strunz K. et al. Ultrasound-guided large-core needle biopsies of breast lesions: analysis of 962 cases to determine the number of samples for reliable tumour classification. Br J Cancer 2005; 92 (02) 231-235
• 20 Paramagul CP, Helvie MA, Adler DD. Invasive lobular carcinoma: sonographic appearance and role of sonography in improving diagnostic sensitivity. Radiology 1995; 195 (01) 231-234
• 21 Renshaw AA, Derhagopian RP, Tizol-Blanco DM, Gould EW. Papillomas and atypical papillomas in breast core needle biopsy specimens: risk of carcinoma in subsequent excision. Am J Clin Pathol 2004; 122 (02) 217-221
• 22 Eiada R, Chong J, Kulkarni S, Goldberg F, Muradali D. Papillary lesions of the breast: MRI, ultrasound, and mammographic appearances. AJR Am J Roentgenol 2012; 198 (02) 264-271
• 23 Korhonen KE, Zuckerman SP, Weinstein SP. et al. Breast MRI: false-negative results and missed opportunities. Radiographics 2021; 41 (03) 645-664
• 24 Andreou S, Soule E, Long D, Jasra B, Sharma S. When something seems amiss: radiology-pathology correlation of metaplastic breast cancer. Cureus 2020; 12 (05) e8239
• 25 Sanderink WBG, Camps-Herrero J, Athanasiou A. et al. Image-guided biopsy of breast lesions: when to use what biopsy technique. Insights Imaging 2025; 16 (01) 208
• 26 Uematsu T, Kasami M, Uchida Y. et al. Ultrasonographically guided 18-gauge automated core needle breast biopsy with post-fire needle position verification (PNPV). Breast Cancer 2007; 14 (02) 219-228
• 27 Lai HW, Wu HK, Kuo SJ. et al. Differences in accuracy and underestimation rates for 14- versus 16-gauge core needle biopsies in ultrasound-detectable breast lesions. Asian J Surg 2013; 36 (02) 83-88
• 28 Fishman JE, Milikowski C, Ramsinghani R, Velasquez MV, Aviram G. US-guided core-needle biopsy of the breast: how many specimens are necessary?. Radiology 2003; 226 (03) 779-782
• 29 Bolívar AV, Alonso-Bartolomé P, García EO, Ayensa FG. Ultrasound-guided core needle biopsy of non-palpable breast lesions: a prospective analysis in 204 cases. Acta Radiol 2005; 46 (07) 690-695
• 30 Li HN, Chen CH. Ultrasound-guided core needle biopsies of breast invasive carcinoma: when one core is sufficient for pathologic diagnosis and assessment of hormone receptor and HER2 status. Diagnostics (Basel) 2019; 9 (02) 54
• 31 Brenner RJ, Fajardo L, Fisher PR. et al. Percutaneous core biopsy of the breast: effect of operator experience and number of samples on diagnostic accuracy. AJR Am J Roentgenol 1996; 166 (02) 341-346
• 32 Bouzón Alejandro A, Iglesias López Á, Acea Nebril B, García Jiménez ML, Díaz Carballada CC, Varela Romero JR. Underestimation of invasive breast carcinoma in patients with initial diagnosis of ductal carcinoma in situ: Size matters. Cir Esp (Engl Ed) 2021; 99 (09) 655-659
• 33 Meurs CJC, van Rosmalen J, Menke-Pluijmers MBE. et al. A prediction model for underestimation of invasive breast cancer after a biopsy diagnosis of ductal carcinoma in situ: based on 2892 biopsies and 589 invasive cancers. Br J Cancer 2018; 119 (09) 1155-1162
• 34 Suh YJ, Kim MJ, Kim EK. et al. Comparison of the underestimation rate in cases with ductal carcinoma in situ at ultrasound-guided core biopsy: 14-gauge automated core-needle biopsy vs 8- or 11-gauge vacuum-assisted biopsy. Br J Radiol 2012; 85 (1016) e349-e356
• 35 Sá RDS, Logullo AF, Elias S, Facina G, Sanvido VM, Nazário ACP. Ductal carcinoma in situ: underestimation of percutaneous biopsy and positivity of sentinel lymph node biopsy in a Brazilian Public Hospital. Breast Cancer (Dove Med Press) 2021; 13: 409-417
• 36 Schulz S, Sinn P, Golatta M. et al. Prediction of underestimated invasiveness in patients with ductal carcinoma in situ of the breast on percutaneous biopsy as rationale for recommending concurrent sentinel lymph node biopsy. Breast 2013; 22 (04) 537-542
• 37 Schaefer FK, Heer I, Schaefer PJ. et al. Breast ultrasound elastography–results of 193 breast lesions in a prospective study with histopathologic correlation. Eur J Radiol 2011; 77 (03) 450-456
• 38 Wojcinski S, Boehme E, Farrokh A, Soergel P, Degenhardt F, Hillemanns P. Ultrasound real-time elastography can predict malignancy in BI-RADS-US 3 lesions. BMC Cancer 2013; 13: 159
• 39 Itoh A, Ueno E, Tohno E. et al. Breast disease: clinical application of US elastography for diagnosis. Radiology 2006; 239 (02) 341-350
• 40 Kumm TR, Szabunio MM. Elastography for the characterization of breast lesions: initial clinical experience. Cancer Control 2010; 17 (03) 156-161
• 41 Georgieva M, Prantl L, Utpatel K. et al. Diagnostic performance of ultrasound strain elastography for differentiation of malignant breast lesions. Clin Hemorheol Microcirc 2019; 71 (02) 237-247

Publication History

Article published online:
09 January 2026