Article Text
PDF
Abstract
Aims It is a standard practice to examine multiple levels of needle core biopsies taken for mammographic calcification, but there is almost no evidence on the value of levels in core biopsies taken for other reasons. This study aimed to assess the value of levels for needle core biopsies taken for investigation of lesions other than calcification. A secondary aim was to assess interobserver agreement of diagnosis.
Methods For each of the 375 breast core biopsies with three levels the first level was reviewed and a diagnosis made. Then levels 2 and 3 were reviewed and a final diagnosis was made.
Results The diagnosis after examining three levels was different from that in the initial level in 4 of 272 (1.5%, 95% CI 0.04% to 3%) core biopsies taken for reasons other than calcification and in 13 of 103 (13%, 95% CI 6% to 19%) biopsies taken for investigation of calcification. Interobserver agreement of the original diagnosis at the time of reporting and the final diagnosis at the review for this study was 96% (κ 0.947).
Conclusions This study confirms the value of levels of biopsies taken to investigate mammographic calcification, but suggests that routine levels are of limited value for breast core biopsies taken for other reasons.
Statistics from Altmetric.com
Introduction
In many centres, including Nottingham, needle core biopsy has replaced fine needle aspiration cytology as the preferred method for non-operative diagnosis of breast disease. Needle core biopsy has better sensitivity for diagnosis of malignancy, can distinguish invasive and in situ carcinoma, is better at making definite benign diagnoses and the topographical relation of calcification to the associated pathology can be seen.1 There is evidence that examination of multiple levels is of value in the assessment of core biopsies taken to assess mammographic calcification,2 ,3 and this approach is recommended in guidelines.4 A recent Royal College of Pathologists document states that one level is usually sufficient for core biopsies from mass lesions5 and that ‘In practice most laboratories choose to examine all core biopsies from screen detected lesions at three levels initially’. There is a dearth of evidence on the value of levels in core biopsies taken for reasons other than the investigation of calcification. We have identified only one such study that looked at MRI-guided core biopsies.6 The present study aimed to assess the value of levels in core biopsies taken for reasons other than investigation of calcification. It also gave us the opportunity to assess interobserver and intraobserver variation in diagnosis.
Materials and methods
At the time of this audit all breast needle core biopsies had three levels cut routinely if they were taken for assessment of mammographic calcification or an urgent result was requested. Most of these urgent results were requested for discussion at a multidisciplinary meeting the next day. All biopsies with routine levels over the period of a year were included. One pathologist prospectively examined biopsies as part of routine reporting over a 4-month period. The remaining biopsies were reviewed by one of four specialist breast pathologists blind to the original diagnosis. For each biopsy, the first level was examined and a diagnosis was recorded. Then the other two levels were examined and a final diagnosis was recorded.
The diagnoses were categorised using the National Heath Service Breast Screening Programme system: B1 normal, B2 benign, B3 lesion of uncertain malignant potential, B4 suspicious of malignancy and B5 malignant.4 B5 biopsies were divided into B5a ductal carcinoma in situ (DCIS), B5b invasive malignancy and B5c for biopsies that could not be categorised as B5a or B5b. B3 biopsies were divided into those with (B3b) and without (B3a) epithelial atypia as proposed by Ibrahim et al7 and supported by other studies.8 To assess the interobserver and intraobserver agreement, the final diagnosis after review of all the levels for this study and the original diagnosis at the time of reporting were compared. The study was discussed with the chair of one of the Nottingham research ethics committees; the written opinion was that the ethical approval was not necessary.
Results
Biopsies of lesions other than calcification
Two hundred and seventy-two biopsies were taken for reasons other than calcification (118 symptomatic patients and 154 lesions detected by mammographic screening). A core biopsy was taken in 266 and a vacuum-assisted biopsy in 6. Two hundred and forty biopsies were ultrasound-guided, 6 were under stereotactic guidance, 13 were freehand and the method was not specified for 13. There were four biopsies where levels 2 and 3 resulted in a change in categorisation (1.5%, 95% CI 0.04% to 3%, table 1). In one biopsy the initial level showed an atypical intraductal proliferation (AIDEP, B3b) that was upgraded to suspicious of DCIS in later levels. In one biopsy the initial level was too shallow and showed fragments of normal tissue, but levels showed a fibroadenoma. One biopsy was upgraded from normal to B3b because of a few cells in the stroma on level 3; further levels showed invasive lobular carcinoma. The fourth biopsy showed DCIS on the initial level, but there was focal invasive carcinoma in the third level. The frequency of changes in diagnosis was similar in symptomatic and screening-detected lesions (2 of 118 and 2 of 154 respectively). In 9 of the 272 biopsies (3%, 95% CI 1% to 5%), the final diagnosis was an atypical intraductal lesion: AIDEP (B3b) in 2, suspicious of DCIS in 2 and DCIS in 5.
Comparison of the diagnosis made on level 1 and on all three levels of biopsies taken for reasons other than calcification
Biopsies taken for assessment of calcification
One hundred and three biopsies were taken for mammographic calcification (91 screen-detected and 12 symptomatic). A core biopsy was taken in 40 and a vacuum-assisted biopsy in 63. Twenty-four biopsies were ultrasound-guided, 75 were under stereotactic guidance, 1 was freehand and the method was not specified for 3. In 13 of 103 (13%, 95% CI 6% to 19%) specimens, levels 2 and 3 showed features resulting in a change in the diagnostic category compared with level 1 (table 2). The proportion with a change in diagnosis was similar for core biopsies (5 of 40, 12%) and for vacuum-assisted biopsies (8 of 63, 13%). In 49 biopsies (48%, 95% CI 38% to 57%), the final diagnosis was an atypical intraductal lesion: AIDEP (B3b) in 12, suspicious of DCIS in 3 and DCIS in 34.
Comparison of the diagnosis made on level 1 and on all three levels of biopsies taken for calcification
In 11 biopsies there was either no or minimal calcification and a non-malignant diagnosis in the three initial levels, so further levels were cut. In six biopsies the further levels resulted in a change of diagnosis and calcification was seen (one B1 to B5a, one B1 to B3b, one B1 to B2, one B2 to B3b and two B3b to B5a). In two biopsies there was a change of diagnosis although no calcification was seen (B1 to B3b and B1 to B2). In three biopsies calcification was seen in the further levels, but the diagnosis was not changed (two B2 and one B4).
Axillary nodal core biopsies
Forty patients had ultrasound-guided core biopsies of axillary nodes. For 39 there was no difference between the diagnosis on level 1 and the diagnosis on all three levels (normal 1, reactive 13, metastatic carcinoma 23 and lymphoma 2). The diagnosis changed in one biopsy: the initial level showed small fragments of fibrofatty tissue, but the other levels showed a nodal metastasis.
Interobserver and intraobserver agreement
The intraobserver agreement (95 of 102–93%, 95% CI 88% to 98%) and interobserver agreement (156 of 162 (96%, 95% CI 93% to 99%)) were similar (see tables 3 and 4). The κ statistics were 0.910 and 0.947, respectively, using the seven categories in tables 3 and 4. The discrepancies were: normal versus fibrocystic change or sclerosed fibroadenoma (six biopsies), AIDEP (B3b) versus suspicious of DCIS (2), suspicious of DCIS versus definite DCIS (3), radial scar or papilloma without epithelial atypia versus with atypia (2) and atypical fragments separate from the core categorised as B3b by one observer and B4 by another. The frequency of discrepancies was similar in symptomatic (6 of 104, 6%) and screening-detected lesions (7 of 160, 4%).
Comparison of the diagnosis made at original reporting and on review for this study with the same observer
Comparison of the diagnosis made at original reporting and on review for this study with different observers
Discussion
This study found that two additional levels were of value in 13% of biopsies taken to assess mammographic calcification, but in only 1.5% of biopsies taken for other reasons.
A number of studies have investigated the value of levels of core biopsies taken to assess calcification. It is important to identify calcification histologically in a pattern comparable to that seen in the specimen radiograph unless a malignant diagnosis is made. Grimes et al2 found that calcification was seen in the first three levels in 112 of 168 core biopsies. Additional levels in 50 biopsies showed calcification in 21 with a change in diagnosis in 11 (22%) including 2 that changed from atypical ductal hyperplasia to DCIS. Dahlstrom et al9 also identified DCIS that was found only in further levels as calcification was not seen in the initial sections. Kumaraswamy and Carder3 examined nine levels. An accurate diagnosis was made in 89% on three levels and in 97% on six levels. In only one case was there a change from a normal or benign diagnosis to an atypical or malignant diagnosis. In the present study, levels 2 and 3 resulted in a change in diagnosis in 13% of biopsies. Eight of the 13 changes involved the spectrum from AIDEP to DCIS. Furthermore, there were eight additional changes in diagnosis seen in further levels that were requested as the calcification was not adequately sampled in the initial three levels.
We have identified only one study investigating the value of levels of core biopsies taken for reasons other than calcification. Cornea et al5 found that four levels resulted in a change in diagnosis in only 1 of 505 MRI-guided core biopsies. In the present study of 272 core biopsies taken for reasons other than calcification, there were only four changes of diagnosis with two further levels. Two changes were of limited clinical significance: B1 versus B2 and B3b versus B4. The significance of upgrading one biopsy from B5a to B5b is that sentinel node biopsy would be offered. The most important discrepancy was the change from B1 to B3b because of a few stromal cells that on further levels were shown to be invasive lobular carcinoma. However, this patient had a clinically malignant mass, so would have had further biopsy after a B1 result. It is difficult to say what an acceptable rate of discordance would be, but even if numerous levels were examined the rate of discordance would not be zero. It is essential that any result is interpreted in combination with the clinical and radiological findings (the triple approach). The low rate of change in diagnosis compared with the biopsies taken for calcification may be a reflection of the different ranges of lesions biopsied in these two clinical scenarios. The final diagnosis of core biopsies taken for reasons other than calcification were rarely in the spectrum from AIDEP to DCIS (3%), but such diagnoses were frequent in core biopsies taken for calcification (48%).
In the present study, 57% of the biopsies taken for reasons other than calcification were of screening-detected lesions, compared with 23% of core biopsies overall in this period. This bias was because many of the urgent cases were taken at a screening assessment clinic that was the day before a multidisciplinary meeting. Nevertheless, the frequency of changes in diagnosis was similar in symptomatic and screening-detected lesions, suggesting that the low rate of change in diagnosis of 1.5% in the biopsies taken for reasons other than calcification is likely to be representative. There was one change in diagnosis in 40 core biopsies of axillary lymph nodes.
In two cases the diagnosis was not made on the initial level because the biopsy was not cut into properly. It is important that the laboratory staff are trained to embed the biopsies and cut into them properly when taking sections. Also the pathologist must request levels if the initial section is not adequate, for example if there are holes in the section or the amount of tissue is less than one would expect.
The intraobserver (κ 0.91) and interobserver agreements (0.95) were both excellent and higher than two previous studies. Verkooijen et al10 obtained a κ statistic of 0.83 comparing assessment by general and specialist pathologists of 14 gauge biopsies of impalpable lesions using five categories (normal, benign, borderline, DCIS and invasive carcinoma). Stang et al11 obtained a κ statistic of 0.87 comparing interpretation by two experienced breast pathologists of biopsies using five categories (B1 to B5) with biases towards mammographically detected lesions and vacuum-assisted biopsies. The high level of agreement in the present study may be because all the pathologists specialise in breast pathology and regularly show each other biopsies. Most of the discrepancies were of limited clinical importance (B1 vs B2 and B3b vs B4). It was surprising that there were only two biopsies with discordance over whether atypia was present or not as there is an element of subjectivity in this distinction and previous studies have shown difficulty in this area.10 ,11
In conclusion, the present study supports the recommendation of examining multiple levels of core biopsies taken for mammographic calcification, but suggests that routine levels may not be necessary for breast core biopsies taken for other reasons. If only one level is to be examined it is important that the block is properly cut into when taking the section.
Take-home messages
-
This study confirms the value of examining levels of core biopsies taken for assessment of mammographic calcification.
-
Levels were of value in only 1% of biopsies taken for investigation of lesions other than calcification.
Footnotes
-
Contributors AHSL conceived the study. All authors contributed to the design of the study. AHSL, ZH, EAR and IOE all reviewed sections for the study. AHSL analysed the data. All authors contributed to the interpretation of the data, writing of the manuscript and approved the final draft. AHSL is the guarantor.
-
Competing interests None.
-
Ethics approval Not required.
-
Provenance and peer review Not commissioned; externally peer reviewed.