False-negative core needle biopsies of the breast

An analysis of clinical, radiologic, and pathologic findings in 27 consecutive cases of missed breast cancer

Authors


Abstract

BACKGROUND

A benign diagnosis in a core needle biopsy (CNBx) of the breast performed for a clinically and/or radiologically suspicious abnormality is often due to a nonrepresentative sample. However, the discordance may not be recognized, resulting in a logistic delay in the diagnosis.

METHODS

Twenty-seven false-negative CNBxs were identified in 952 consecutive CNBxs of the breast (653 benign, 266 malignant, and 33 atypical) performed during a 1-year period. Biopsies were analyzed with respect to clinical and radiologic findings, biopsy type, type of malignancy, and interval between the original CNBx and final diagnosis. Four hundred thirty-eight (67%) of the patients with a benign CNBx diagnosis either underwent excision or had a minimum of 1-year follow-up (mean, 35.6 months; median, 36 months).

RESULTS

The cancers missed on CNBx included 6 ductal carcinomas in situ, 17 invasive ductal carcinomas, 3 invasive lobular carcinomas, and 1 non-Hodgkin lymphoma. The overall false-negative rate was 9.1%. For palpable lesions, ultrasound-guided CNBx had a lower rate of missed cancer (3.6%) compared with CNBx without image guidance (13.3%). The false-negative rate for vacuum assisted CNBx biopsy was 7.6% (3.3% for the 11-gauge needle, 22.2% for the 14-gauge needle; 5.6% for nonpalpable mass lesions, 8.2% for microcalcifications). In all seven false-negative CNBxs performed by radiologists, the discordance between the radiologic and pathologic findings was promptly recognized due to their standard follow-up protocol. The discordance between the degree of clinical suspicion, radiologic impression, and the pathologic findings was not immediately recognized in 5 of 20 false-negative CNBxs performed by surgeons (4 without radiologic guidance and 1 with ultrasound guidance), resulting in a delay in the diagnosis ranging from 112–336 days.

CONCLUSIONS

A false-negative diagnosis of breast carcinoma was found to be more common in CNBx performed without image guidance but occurred to a lesser degree in image-guided biopsies. A delay in diagnosis can be avoided by establishing a standard post-CNBx follow-up protocol. Cancer 2003;97:1824–31. © 2003 American Cancer Society.

DOI 10.1002/cncr.11278

In recent years there has been a marked increase in percutaneous core needle biopsy (CNBx) performed for palpable and nonpalpable breast lesions. Fine-needle aspiration cytology (FNAC) had been the principal sampling method of obtaining a tissue diagnosis for palpable lesions. 1–5 However, CNBx has become more popular with the introduction of the spring-loaded automated “gun,” which is easier to use and has a higher sampling success rate, 6 particularly when performed under radiologic guidance. 7 Because FNAC is not a very sensitive method for nonpalpable abnormalities (sensitivity rate of 44–92%), 8 image-guided large-core needle biopsy (sensitivity rate of 85–99%) 9–15 has gained widespread acceptance in this clinical situation. The recent introduction of the directional, vacuum-assisted biopsy instrument with a 14-gauge or 11-gauge needle (Mammotome, Biopsy Medical, Irvine, CA) has allowed harvesting of a larger tissue sample with a single needle insertion compared with the stereotactic automated 14-gauge needle. 16

Sampling error is generally recognized to be a major factor with FNAC, and several studies have addressed this problem. 17–21 Despite advances in biopsy devices and techniques, CNBx of breast also is associated with a sampling error. This could result in a false-negative diagnosis, which may delay treatment if there is no established protocol to determine the clinical, radiologic, and pathologic concordance. A protocol for radiologists requires the correlation of histopathologic findings of CNBx with the imaging findings, and if the findings are discordant, prompt repeat biopsy is indicated. 22 However, to the best of our knowledge, no such protocol has been proposed for nonimage-guided CNBxs. Few studies have addressed the frequency and clinical implications of false-negative image- 23–28 and nonimage-guided CNBxs. 29 In this study, we have analyzed 27 consecutive cases of breast carcinoma missed by initial CNBx, paying particular attention to the clinical and radiologic findings, final pathologic diagnosis, impact of radiologic guidance, and technical variables related to the procedure such as the needle gauge.

MATERIALS AND METHODS

During a 1-year period between November 1997 and October 1998, percutaneous CNBx was performed on 952 breast lesions at Henry Ford Hospital, Detroit, Michigan. Of these, 342 CNBxs were performed without image guidance, 241 with ultrasound (US) guidance, and 369 using stereotactic imaging and a directional vacuum-assisted biopsy device (Mammotome).

All the nonimage-guided CNBxs of palpable mass lesions were performed by surgeons using a spring-loaded automated Biopty gun with a 14- or 18-gauge needle having a 20-mm throw (Microvasive Boston Scientific Corporation, Watertown, MA) until June 1998. A 14- or 18-gauge needle having a 22-mm throw (CR Bard, Inc., Covington, GA) was used after June 1998. One hundred seventy-six US-guided CNBxs (73%) were performed by radiologists using a 12-megahertz (MHz) linear array transducer (GE Logic, Milwaukee, WI) or a 7.0-MHz transducer (Acuson, Mountain View, CA) with a disposable 14-gauge Monopty (CR Bard, Inc.) or Achieve (Allegiance Health Care Corp., McGaw Park, IL) needle. An average of five core samples were obtained, with the number of cores taken from each patient determined by the radiologist at the time of the procedure. The surgeons performed 65 US-guided CNBxs (27%) using a 5/8-MHz linear array transducer (B & K Medical Products, Grand Rapids, MI). Of the 369 CNBXs performed using stereotactic imaging and a Mammotome, an 11-gauge needle was used in 311 cases (84.3%) and a 14-gauge needle was used in 58 cases (15.7%). After early experience with the 14-gauge Mammotome needle, exclusive use of the 11-gauge needle became standard practice in our institution. The standard procedure was to obtain six CNBx specimens. The need for more specimens was determined by the radiologist at the time of biopsy. For CNBXs of calcifications, additional samples were obtained until calcium could be documented on the specimen radiograph.

The pathologic diagnoses of the 952 CNBxs and the outcome on subsequent excision or follow-up are summarized in Figure 1.

Figure 1.

Data analysis of core needle biopsies.

In 33 cases, the CNBx diagnosis had been atypical hyperplasia (4 atypical lobular hyperplasia cases and 29 atypical ductal hyperplasia cases). In 5 of these cases, the final pathologic diagnosis after excision of the lesion was carcinoma (4 ductal carcinoma in situ cases and 1 invasive lobular case). A diagnosis of atypical hyperplasia in a needle biopsy generally would prompt an excision biopsy, hence these cases were not considered to be false-negative.

Of the 653 cases with a benign diagnosis on CNBx, 188 cases (27.7%) did not undergo subsequent excision or have a follow-up period of at least 1 year either by mammography or clinical breast examination by a surgeon. In 389 of the cases that were not confirmed by excision, the lesion in question either remained stable or regressed during a follow-up period of at least 1 year (mean, 35.6 months; median, 36 months). The remaining 76 cases with benign CNBx diagnosis underwent repeaet CNBx /excision biopsy; benign results were found in 49 cases, whereas in the remaining 27 cases a final diagnosis of malignancy was rendered. The benign CNBx diagnosis in these 27 cases was confirmed by review of the original slides. These cases were analyzed with respect to clinical and radiologic findings, biopsy type, number of needle cores taken, and the time interval between the initial CNBx and final pathologic diagnosis of malignancy. The required information was acquired by review of medical records including radiology and pathology reports.

RESULTS

Of the 952 patients who underwent CNBx, 298 (31.3%) had a final malignant pathologic diagnosis. Of these 298 patients, 27 had been diagnosed as benign on the original CNBx and 5 as atypical ductal hyperplasia.

The clinical, radiologic, and pathologic features of the 27 cases of breast cancer missed by CNBx are summarized in Table 1. In 22 (81.5%) of these cases, the breast lesion was either a palpable or radiologically detected mass lesion. In the remaining 5 cases, the radiologic abnormality was suspicious microcalcifications. Eighteen of the 27 patients had undergone CNBx without any radiologic guidance, 3 had US-guided CNBxs (2 performed by surgeons and 1 performed by a radiologist), and 6 underwent Mammotome biopsies (Table 2).

Table 1. Clinical, Radiologic, and Pathologic Findings of False-Negative Cases
Case no.Clinical findingsRadiologic findingsCNBx type (needle gauge), [performer]No. of coresPathologic diagnosis on CNBxFinal pathologic diagnosis (size of tumor)Interval to final diagnosis (days)
  1. CNBx: core needle biopsy; mam: mammogram; A: core needle biopsy without radiologic guidance; S: surgeon; FCC: fibrocystic change; IDC: invasive ductal carcinoma; U: unknown; ILC: invasive lobular carcinoma; Ca++: calcifications; DCIS: ductal carcinoma in situ; NHL: non-Hodgkin lymphoma; B: ultrasound-guided biopsy; RS: radial scar; R: radiologist; C: Mammotome biopsy with an 11-gauge needles; D: Mammotome biopsy with a 14-gauge needle.

1MassMass (mam)A(14)[S]4FCCIDC (3 cm)168
2MassMass (mam)A(U)[S]3Benign breast tissueIDC (3.5 cm)336
3MassMass (mam)A(U)[S]3Fatty tissueIDC (1.8 cm)112
4MassNot doneA(20)[S]2Acute hemorrhageIDC (3.5 cm)168
5MassAsymmetric density (mam)A(U)[S]5Fat necrosisILC (4.5 cm)28
6MassMass (mam)A(U)[S]5Fat onlyIDC (1.5 cm)14
7MassMass and Ca++ (mam)A(U)[S]4NonspecificDCIS (1 cm)28
8MassMass (mam)A(U)[S]5FibrosisNHL (3 cm)21
9MassMass (mam)A(U)[S]4NonspecificILC (2 cm)7
10MassMass (mam)A(U)[S]1NonspecificIDC (3 cm)14
11MassMass (mam)A(U)[S]6BenignIDC (1 cm)7
12MassMass (mam)A(14)[S]4Perilobular hemangiomaIDC (2 cm)42
13MassMass (mam)A(15)[S]2BenignTubular carcinoma (1.2 cm)14
14MassMass (mam)A(U)[S]5BenignILC (1.8 cm)7
15Vague massAsymmetric densityA(U)[S]4Fat onlyIDC (2 cm)7
16MassMass and Ca++ (mam)A(U)[S]2Fat onlyDCIS7
17MassMass and Ca++ (mam)A(U)[S]4FibrosisDCIS (3.5 cm) and IDC (0.3 & 0.1 cm)21
18MassMass (mam)A(14)[S]2FCCIDC (4 cm)56
19MassMass (ultrasound)B(14)[S]4BenignIDC (4 cm)112
20NoneMass (mam and ultrasound)B(14)[S]3BenignRS + DCIS (1.2 cm)14
21MassMass (mam and ultrasound)B(14)[R]4BenignIDC (3 cm)42
22NoneCa++ (mam)C(11)[R]7UnsatisfactoryDCIS (4 mm)56
23NoneCa++ (mam)C(11)[R]10UnsatisfactoryDCIS (5 mm)21
24NoneMass (mam)D(14)[R]15NonspecificIDC (1 cm)56
25NoneCa++ (mam)D(14)[R]14FCC, no Ca++IDC (1.2 cm)56
26NoneCa++ (mam)D(14)[R]> 10Benign, no Ca++DCIS (0.5 cm)14
27NoneCa++ (mam)D(14)[R]12Benign with Ca++IDC (0.8 cm)21
Table 2. Details of False-Negative Core Needle Biopsies Performed with Different Techniques
Type of biopsyFalse-negative cases (%)aInterval to excision (no. of cases)Final pathology (no. of cases)
≤ 56 days> 56 days
  • CNBx: core needle biopsy; IC: invasive carcinoma; DCIS: ductal carcinoma in situ; NHL: non-Hodgkin lymphoma; RS: radial scar.

  • a

    Number of false-negative cases/no. of malignant cases that underwent that type of biopsy.

Nonimage-guided CNBx18/135 (13.3)144IC (15), DCIS (2), NHL (1)
Ultrasound-guided, performed by surgeon2/37 (5.4)11IC (1), RS + DCIS (1)
Ultrasound-guided, performed by radiologist1/47 (2.1)10IC (1)
Mammotome (11-gauge)2/61 (3.3)20DCIS (2)
Mammotome (14-gauge)4/18 (22.2)40IC (3), DCIS (1)

Of the 20 CNBxs of palpable masses performed by surgeons (2 with ultrasound guidance), the discordance between the radiologic and/or clinical findings and histology was promptly recognized in 15 (75%) cases and a repeat CNBx or surgical excision was performed. The time interval from the original CNBx to the final diagnosis of malignancy in the 15 cases in which the discordance was promptly recognized ranged from 7 to 56 days (mean, 19.1 days; median, 14 days). In five cases (four CNBx obtained without radiologic guidance and one CNBx obtained under US guidance), the discordance was not promptly recognized by the surgeon (Fig. 2). A repeat CNBx excision biopsy was subsequently performed because of the persistence and/or increase in size of the lesion. The delay in the final cancer diagnosis in these cases ranged from 112 to 336 days (mean, 179 days; median, 168 days). In four of the five cases, the patient had presented with a suspicious breast mass clinically with a corresponding soft-tissue density by imaging, but this information had not been clearly stated on the histopathology request forms. In two of the cases, the delay in recognition of the discordance was partly the result of unusual clinical presentation. One patient (Table 1, No. 4) presented with a painful mass, preventing mammography initially and the patient was treated for inflammation. The other patient (Table 1, No. 19) was a young woman with a palpable, circumscribed mass with a clinical impression of a fibroadenoma.

Figure 2.

(A) Mammogram showing a relatively well-circumscribed suspicious mass lesion. (B) Core needle biopsy (performed without image guidance) showing benign breast tissue with no discrete mass lesion. (C) Excision biopsy showing a Grade 3 invasive duct carcinoma (Case no. 2, Table 1).

The single missed case of cancer in a US-guided CNBx performed by a radiologist had immediate recommendation for excision followed by immediate excision and a definitive diagnosis. In five of the six false-negative Mammotome CNBXs, biopsy procedures were considered either suboptimal or unsatisfactory at the time of procedure and the discordance was expected. In one case in which the Mammotome CNBx was performed for microcalcifications, CNBx revealed microcalcifications in association with a benign lesion. However, excision biopsy was performed within 21 days due to high radiologic suspicion.

DISCUSSION

Evaluation of a breast lesion by CNBx often involves three different specialties (surgery, radiology, and pathology) with a final diagnosis of cancer requiring pathologic confirmation. CNBx may be performed without radiologic guidance by a surgeon or under US guidance by either a surgeon or a radiologist. The availability of US guidance to the surgeons as an office procedure has increased the precision of the CNBx considerably, decreasing the false-negative rate as noted in the current study. The standard method for the diagnosis of mammographically detected nonpalpable masses or microcalcifications used to be excisional biopsy with wire localization. However, with the advent of minimally invasive image-guided CNBx techniques, the preoperative diagnosis of cancer has become routine even for nonpalpable lesions. Any of the CNBx procedures may fail to sample a cancer and result in a benign, often nonspecific histologic diagnosis. This problem of sampling error has been well documented with FNAC (sensitivity rate, 66–100%), 30 which in some studies was more sensitive than the older (Trucut, Travenol Laboratories, Deerfield, IL) needle biopsy CNBx with a sensitivity rate of 67–89%. 31, 32 It has been well documented in studies from Denmark, Italy, Israel, Norway, and the U.S. that the diagnostic accuracy of malignancy improved considerably if FNAC results were correlated with physical examination and mammography findings. 2, 33–39 This triple approach to interpretation of FNAC results has now been incorporated into the National Cancer Institute (NCI) guidelines. 40 Similarly, for CNBx, if the discordance between the histologic appearances and the clinical/radiologic findings is promptly recognized, the procedure could be repeated or the lesion excised. Delay in diagnosis occurs when benign histologic findings are mistakenly regarded as representative of the lesion in question.

In the current study, the overall false-negative rate for malignancy was 9.1% (considering atypical hyperplasia as a positive diagnosis), which is within the 1–15% range reported in the literature. 28, 41–46 The false-negative rate in this and other studies may be an underestimate due to limited or no follow-up of a substantial number of CNBx with benign diagnosis. However, although in our study 188 patients did not have either a mammogram or a physical breast examination by a surgeon, only 43 of the patients had left the Henry Ford Health system within 1 year, whereas the others had no record of breast carcinoma in 1–4 years of follow-up in the system for other problems. Palpable lesions biopsied under US guidance had a lower rate of missed cancer (3.6%; 5.4% for surgeons and 2.1% for radiologists) compared with nonimage-guided CNBxs performed by surgeons (13.3%). The false-negative rate for Mammotome biopsies was 7.6% (3.3% for the 11-gauge needle, 22.2% for the 14-gauge needle, 5.6% for nonpalpable mass lesions, and 8.2% for microcalcifications), which again is within the reported range (0.3–8.2%). 9–15 The false-negative rate decreased significantly after the introduction of the 11-gauge needle as opposed to 14-gauge needle for Mammotome CNBxs.

In our study, all the cases of breast cancer missed by the initial CNBx had a radiologic abnormality. The most common scenario for a false-negative result was a biopsy of a suspicious mass in which the histologic appearances did not provide a satisfactory explanation for a mass lesion. In the remaining cases, the radiologic abnormality was suspicious microcalcifications, but no microcalcifications were noted in any of these CNBxs except one case in which histologically evident microcalcifications were associated with a benign lesion.

There are guidelines requiring pathology-radiology correlation for image-guided CNBxs performed by radiologists, 47, 48 and it is only appropriate that similar principles be applied to CNBxs performed in surgeons' offices. The ideal conditions of triple diagnosis clinics described for FNACs 33, 39 do not exist in most hospitals in the U.S. As noted by Berg et al., 49 a thorough understanding by the pathologists of the limitations of sampling is critical, as is understanding what constitutes an acceptable histopathologic result given the clinical and/or radiologic impression of the lesion in question. However, the pathologists are seldom provided adequate clinical information and mammographic findings are rarely stated. 50 A specific diagnosis such as “fibroadenoma, papilloma, or invasive carcinoma” usually fits a discrete mass with a corresponding mammographic soft-tissue density. A micropapilloma, hyperplasia, or mild degrees of cystic changes are microscopic findings only and usually do not account for a discrete mass. Therefore, overemphasizing a minor microscopic finding as a main diagnosis may lead to false reassurance of adequate sampling. If the CNBx is for mammographic calcifications, pathologists should be familiar with the lesions typically associated with calcifications. Rare, tiny microcalcifications < 80–100 μm common in breast biopsies are not likely to be seen on conventional mammograms and are not to be considered representative of mammographically detected microcalcifications.

Radiologic-pathologic correlations and post-CNBx follow-up protocols are routine and mandatory for image-guided CNBxs performed by radiologists. A similar alertness to clinical and/or radiologic and pathologic discordance may be accomplished by instituting standard post-CNBx follow-up protocols performed by surgeons. Reviewing clinically and radiologically suspicious cases with a benign CNBx finding at multidisciplinary tumor board meetings will help all three specialties to come together to clarify any issues in the pathology report that could have communicated a false sense of security and to ensure prompt repeat CNBx, preferably with image guidance or excision of the lesion.

In the current study, we have highlighted the need for immediate recognition of any discrepancy between the pathologic diagnosis and the clinical/radiologic findings, which should trigger a repeat CNBx or excision biopsy by alerting the surgeon or the radiologist to the possibility of a sampling error.

Ancillary