Of the 962 lesions, 603 (63%) were malignant, two (0.2%) high-risk, 353 (37%) benign, and four (0.4%) insufficient for diagnosis at LCNB specimen. Of the 603 malignant specimens, 584 were infiltrating carcinoma and 19 DCIS; of the high-risk lesions, one was a LCIS, the other an ADH. In 659 cases, surgical follow-up by means of an excision biopsy was performed where the lesion was totally removed. Surgical excision yielded 615 malignancies (609 infiltrating carcinoma, seven DCIS) and 44 benign lesions (). Of the 609 infiltrating carcinoma, 11 have been assigned to be DCIS (DCIS underestimates) and two high-risk lesions (high-risk underestimates) at LCNB specimen. Three cases also came out to be invasive, where LCNB did not yield sufficient material for diagnosis. Overall histological agreement between the LCNB specimen and finding at excision biopsy was highly significant (P
=0.001). Two patients refused excision biopsy, where LCNB specimen showed DCIS or insufficient tissue for reliable tumour classification. The observed high cancer rate in this series reflects the fact that only lesions were biopsied where breast imaging yielded a lesion of BI-RADS III or higher. The average size was 2.5
cm (median 2.2
cm; range 0.2–11
cm). The sizes between benign and malignant lesions were statistically not different. In all, 236 lesions were palpable, 726 were not, without a difference between benign and malignant.
Figure 2 Follow-up outcome of all breast lesions sampled by LCNB (n=962). *One patient refused excision biopsy despite LCNB-based DCIS diagnosis. **One patient refused excision biopsy despite inconclusive LNCB diagnosis. *** (more ...)
For the diagnostic assessment of 962 breast lesions, 2.044 3D-US-guided core specimens were obtained in a period of 28 months. In most (730, 75.9%) cases, two biopsies per lesion were sufficient to obtain a clear and reliable histological result. Only one core was taken in 31 (3.2%), three in 165 (17.2%), four in 12 (1.2%), and five in two (0.2%) of the sonographically detectable lesions. In few (22, 2.3%) cases, the number of biopsies was not available. The sensitivity of LCNB under 3D-US guidance to identify infiltrating breast lesions was 96.1% (24/616), the specificity 100%. The average of follow-up was 22.2 months (median 21; range 8–36 months). However, not following our recommendation to come in 3-month intervals for follow-up examinations, not all patients had returned for re-evaluation despite their remaining risk of misdiagnosed lesions. In all, 34% (121/353) patients with benign lesions have not returned for follow-up. For these patients, cross-reference with other institutions was performed, and no case of breast cancer was found. However, since cancer databases might not be 100% reliable, those 121 cases were excluded from statistical analysis. This did not alter the results. Three patients who underwent re-examination in our hospital for follow-up showed an increasing lesion size and subsequently underwent excision biopsy that yielded a benign diagnosis.
Among the 962 analysed lesions, almost all (603 of 616; 97.9%) malignancies were identified by LCNB and immediately surgically excised as a consequence. Of the 55 lesions that have been classified by LCNB to be benign, 29 lesions were determined to be BI-RADS IV and V lesions by breast imaging. Therefore, excision biopsy was carried out and yielded eight malignancies (seven infiltrating carcinoma, one DCIS) and 21 benign lesions. Of 23 breast tumours where LCNB and breast imaging yielded a benign diagnosis, and which have been removed, three came out to be malignant in the surgical specimen (two infiltrating carcinoma, one DCIS).
Analysis of data from surgical follow-up, performed in 659 of the lesions, revealed complete agreement between histological findings in LCNB and subsequent open surgical biopsies for 635 (96.2%) lesions. Partial agreement was found in 13 (2%) and clear disagreement only in 11 (1.8%) of the 659 excised lesions (). Those 11 malignant lesions that were initially misdiagnosed by LCNB consisted of nine infiltrating carcinoma and two DCIS ().
Characteristics of single cases of core needle biopsy misses (complete disagreement)
All 11 LCNB misses occurred in lesions that were sonographically classified as masses and central hits of the needle. Eight of these lesions appeared as BI-RADS IV or V lesions in breast imaging and underwent subsequent excision biopsy (confirmed seven invasive tumours and one DCIS) because of disagreement between sonographic and histological finding in LCNB specimen. Only three lesions classified as benign by means of breast imaging and LCNB came up to be malignant (two invasive cancers, one DCIS) at excision biopsy. None of the recorded parameters like lesion size, palpability or number of biopsies was significantly different in the group of malignant lesions not recognised by LCNB-based diagnosis.
Half of all masses (51.4%) were larger than 20
mm, 36.8% were between 10 and 20
mm, and 11.8% smaller than 10
Clinically significant complications (that required additional medical intervention as a consequence of the biopsy) occurred in one (0.1%) of all biopsied patients: an infection required surgical drainage and antibiotic treatment. Cases of minor interstitial haemorrhage, ecchymosis, or self-limiting inflammation were not considered to be significant complications.