Abstract
Aim: Determination of the prevalence, of the radiological and clinical characteristics, and outcome of atypical hyperplasia (AH) of the breast within a population subjected to routine screening (double-view mammography with double reading, performed every two years between 50 and 75 years of age). Patients and Methods: The clinical and radiological records and histological findings of percutaneous and surgical biopsy specimens of sixty-eight patients presenting with AH were reviewed together with patient follow-up data after percutaneous and surgical biopsy. Results: AH incidence in the population was 0.19‰ with the following distribution of lesions: atypical epithelial hyperplasia (AEH, 53%), columnar cell metaplasia with atypia (CCMA, 32%), and lobular intraepithelial neoplasia (LIN, 8%). The mean patient age was 58 years and 24% of patients were receiving hormone replacement therapy. The main radiological finding was the presence of microcalcifications for AEH and CCMA lesions in particular, and the mammograms were valid (correlation between American College of Radiology score and risk of lesion, only 3% of lesions were recognized on the second reading). A total of 13.7% of AH cases were underestimated and a real risk of AH progression was observed, regardless of whether or not surgical biopsy had been performed. Conclusion: The clinical and radiological characteristics of AH observed in a population subjected to routine breast cancer screening are identical to those for patients with the same lesions referred to specialist centers. Surgical biopsy remains the recommended option due to the risk of underestimation of lesions by percutaneous biopsy and the risk of progression justifies the need for continued close monitoring.
- Mammogram
- screening
- lobular intraepithelial neoplasia
- atypical epithelial hyperplasia
- columnar cell metaplasia
Widespread routine screening for breast cancer combined with the development of imaging techniques [ultrasound, mammography and magnetic resonance imaging (MRI)] and percutaneous biopsies has increased diagnosis of atypical hyperplastic breast lesions. Accounting for just 3.6% of cases in 1985 (1), prevalence had risen to 23% in 2007 (2). Atypical hyperplasia of the breast (AH) is currently subdivided into three categories according to the World Health Organization classification of borderline lesions of the breast (Table I) (3); A typical epithelial hyperplasia (AEH), lobular intraepithelial neoplasia type 3 (LIN3), and columnar cell metaplasia with atypia (CCMA). These lesions raise issues that are left unresolved: their clinical significance remains controversial. They are either linked to risk for breast cancer (with the patients concerned having to undergo closer monitoring), or considered a true precancerous condition (in which case excision is recommended) (4, 5); percutaneous biopsy used to establish diagnosis leads to a risk of underestimation, which in most cases today ends in surgical excision to enable full lesion analysis (6-8).
To our knowledge, few teams (9) have focused on the characteristics and outcome of borderline lesions diagnosed in low-risk patients having undergone systematic breast cancer screening. The aim of this study was to assess the prevalence of borderline lesions in unselected populations taking part in screening programs, together with their characteristics and outcome, and to ascertain the diagnostic accuracy of percutaneous biopsy techniques in a population-based study.
Patients and Methods
National guidelines govern routine breast cancer screening in France (10). The target population includes patients aged from 50 to 74 years inclusive. Two views of each breast are taken (craniocaudal and oblique views) every two years preceded by a clinical examination by the attending radiologist. Mammographic findings initially classified as normal by the first radiologist are referred for a second independent reading, while abnormal findings may lead to further investigation at the discretion of the radiologist, including X-ray, ultrasound or MRI scan and puncture or percutaneous biopsy (ultrasound-guided (14 to 16 G), stereotactic-guided (11 G) or surgical biopsy). Patients excluded from the national screening program are either those with an apparent clinical sign, having already had breast cancer, already monitored for an abnormal X-ray image, presenting with high-risk factors (known borderline lesion, strong family history of breast or ovarian cancer), or BRCA1 and - 2 mutation carriers. Between October 1, 1995 and August 31, 2009, 115 files of patients for whom a diagnosis of AH had been established following routine breast cancer screening were reviewed. Twenty-eight patients were excluded as either cancer or a benign lesion was found on the biopsy and 19 files were not admissible (missing data). The following data were analyzed in the remaining 68 files: clinical data [age, clinical examination, hormone replacement therapy (HRT) use], radiological data [American College of Radiology (ACR) grade on first and second reading, symptomatology of abnormal findings], biopsy technique, histological findings, therapeutic management and follow-up.
World Health Organization classification of borderline lesions (3).
The statistical analysis was carried out using the SAS® software (SAS France, Evry-Grégy-sur-Yerres, France). Chi2 and Fisher's exact tests were used when the number of patients was less than five for qualitative variable analysis. The test was considered positive if the p-value was less than or equal to 0.05.
Results
Eighty-five radiologists, working in 40 centers were involved in the screening program in our area of France (Ille-et-Vilaine). Approximately 37,000 mammograms were performed per year with a participation rate of 60%, 9.1% of which were positive with a biopsy rate of 1.2%. Three percent of these biopsies revealed borderline lesions, accounting for 0.19‰ of all mammograms performed over the entire study period. The percentage of borderline lesions diagnosed with routine breast cancer screening rose over time (Table II).
The mean age of patients was 58±6.6 years (range=50-74 years). Sixteen women (29%) were receiving hormone replacement therapy. Five out of the 68 patients presented with a clinical abnormality in the form of a nodule or with an ambiguous clinical lump, which on biopsy revealed two cases of AEH (one of which associated with CCMA) and three of LIN3. The various histological types were: AEH=36/68, i.e. 53%, CCMA=22, i.e. 32%, LIN3=6, i.e. 8%, with a mean age of 62.1, 56.5 and 56.3 years, respectively.
We reviewed 63 mammograms (the images of five patients, including four cases of AEH and one of LIN, were unavailable for review). The clinical and radiological findings are summarized in Table II. The presence of microcalcifications is the major radiological sign indicative of AEH (p=0.03) and CCMA (p=0.02) as these lesions are not characterized by opacity. LIN3 lesions had no particular radiological characteristics (Table III). ACR classification proved valid for diagnosing the lesions. Out of the 68 cases, 57 were diagnosed according to the ACR classification, and an ACR score of 3 or more was significantly associated with the risk of cancer in borderline lesions observed in the surgical biopsy specimen (p=0.03).
The second reading led to re-classification in two cases (3%), scored as ACR 2 on the first reading and subsequently re-classified as ACR 3 and ACR 4, leading to a diagnosis of AEH and LIN3, respectively.
The histological diagnosis was initially confirmed by surgical biopsy in 10 cases (14.5%) and by percutaneous biopsy in 58 cases. The biopsy technique depended on the radiological sign. In the presence of areas of opacity, microbiopsies (p=0.006), guided by ultrasound (p=0.0002) were used, whereas microcalcifications were investigated by macrobiopsy (p=0.003) under stereotactic guidance (p=0.002). Among the 58 patients who underwent percutaneous biopsy, 7 were monitored (12%) and 51 underwent surgical biopsy (88%). Seven cases of cancer were found on the surgical specimens, indicating an underestimation rate of 13.7% for percutaneous biopsy (17%, 15% and 0%, respectively for AEH, CCMA and LIN3 lesions (Figure 1). No cancer or residual borderline lesions were found for 21 surgical specimens (surgical biopsy was therefore unnecessary for 41% of women) and 23 patients had pure borderline lesions.
During follow-up, three cases of cancer were observed in 12 patients presenting with AEH, including two cancers occurring after surgical biopsy. On a final note, the percentage of borderline lesions diagnosed over time increased slightly with the screening programs, albeit with no significant differences between the percentages (Table IV).
Discussion
The incidence of AH breast lesions in an unexposed population benefiting from routine screening is not precisely known. However, although low (0.19‰ of screening mammograms in our study), it cannot be overlooked owing to the widespread introduction of breast cancer screening and the controversy surrounding it which is mainly due to overdiagnosis of cancer (11, 12) and overtreatment of lesions of low-grade carcinoma, in situ in particular (13). We noticed a rise in incidence over time, in our region, which is most plausibly explained by the adoption of new screening practices starting in the year 2000 (change to two readings, screening every two years instead of three), a growing interest in borderline lesions by radiologists and the availability of more efficient digital mammographic facilities, especially for infraclinical lesions such as microcalcifications (14).
Clinical and radiological characteristics of borderline lesions in our cohort.
Analysis of radiological signs.
The mean age of diagnosis in this population is comparable to that found in French series analyzing LIN3 (7) and CCMA (6), concerning patients not undergoing systematic screening, whereas in our study the mean age for AEH is higher (62 years) than that found in another recent French series analyzing a population referred to a specialist center [aged 52.2 years (15)]. This may be explained by current knowledge of the natural history of AEH which is today considered a precursor of low-grade carcinoma (16), whereas LIN3 (except for the pleomorphic subtype) and CCMA are understood to be risk markers for cancer with no potential for progression per se (17, 18): (a) the frequency of AEH increases with age, reaching a peak in the post-menopausal years and then falling after the age of 60 (1, 4), (b) the risk of AEH progression persists for at least 25 years (1, 4), indicating the persistence of these lesions. However, the oldest patients are less at risk than those presenting with AEH at a young age [relative risk of 6.69 in patients aged under 45 years and of 3.37 for over 55-year-olds in the Mayo Clinic cohort (4), as also found in the Nurse Health Study with an odds ratio of 7.3 during the pre-menopausal years vs. 3.1 in the post-menopausal years (19)].
Progression of borderline lesion incidence over time.
Distribution of borderline lesions according to type and outcome. AEH: Atypical epithelial hyperplasia; CCMA: columnar cell metaplasia with atypia; LIN: lobular intraepithelial neoplasia; DCIS: ductal carcinoma in situ; IDC: invasive ductal carcinoma; ILC: invasive lobular carcinoma.
No significant relationship was found between the existence of clinical abnormality and the type of borderline lesion. We did not find a correlation between the use of HRT during the menopause and the risk of discovering a high-risk lesion during screening mammography but our study lacks sufficient power to draw firm conclusions. However, there are several arguments linking HRT and AH lesions: tissue hormone receptivity increases parallel with lesion severity (20), there is a higher incidence of AEH in HRT users (21) and, conversely, there has been a fall in the incidence of these lesions in the United States since the significant drop in the number of hormone prescriptions (22).
Borderline lesions are unevenly distributed, with a predominance of AEH (53%), as found by Flegg et al. (9), which reflects the relative frequency of the respective borderline lesions. We did not observe any symptomatological differences between the ages covered by screening and younger patients. Clinical examination is still valid, as within screening programs that theoretically target asymptomatic patients, 5 out of 68 patients with unspecific clinical abnormalities were also diagnosed with borderline lesions. As expected, the most important mammographic symptoms are microcalcifications which are a major sign of AEH and CCMA. As mammograms of older women are easier to read (23) the ACR classification provides an efficient scoring system. This is borne-out by the fact that on the one hand, the ACR 3+ lesions correlate with the highest histological grade, and on the other, a low percentage of cases were re-classified on the second review [two cases, i.e. 3% of diagnoses, whereas a percentage of 13% has been observed (24)]. A diagnosis of LIN3 is difficult to achieve in this population due to the absence of conventional radiological signs. However, although breast MRI is superior to mammography for diagnosing LIN3, it appears unrealistic to use it for screening given its inadequate specificity (25).
A standard diagnostic strategy is generally adopted, preferably with an ultrasound-guided biopsy technique for areas of opacity and mammographically-guided biopsy for microcalcifications. The diagnostic performance of these ultrasound-guided biopsies remains unchanged in populations undergoing screening with a mean lesion underestimation rate of 13.7% comparable to that observed for patients who underwent non-systematic screening (6, 15). However, our figures differ significantly from the figures reported by Flegg et al. (9). Their study reveals an underestimation rate of 24% (with a majority of non-invasive lesions diagnosed at surgical biopsy), whereas the surgical specimens showed no residual lesion in 20% of cases and persistent borderline lesions in 55% of cases (vs. 41% and 45% respectively in our experience). A plausible explanation is related to the variation in percutaneous biopsy techniques (choice of several probe rotations with stereotactic vacuum-assisted large-core biopsy, and preference for larger needles for ultrasound-guided biopsies) which greatly influence false-positive and residual lesion rates (26). Consequently, a surgical biopsy procedure is recommended after discovery of fibrocystic breast disease associated with risk factors in order to carry out a full lesion analysis, included as part of a systematic routine screening program (27, 28).
Follow-up of patients with AH is still required, including those who have undergone a surgical biopsy. Three cases of breast cancer were operated on out of 12 patients presenting with AH in our series (two of whom after surgical biopsy). These figures were comparable to those reported in the Australian series (9).
Conclusion
Borderline breast lesions are the “black hole” of breast screening. Although mammography is an efficient screening method and percutaneous biopsy has reduced the number of surgical biopsy procedures, to date there are still two unresolved issues: (a) the lack of consensus on criteria for deferring a surgical biopsy after diagnosis by percutaneous biopsy (8) (41% of patients in our series were over-treated), and (b) the risk of progression of these little-known lesions ranging from 4.3% (29) to 20% at 14 years, as seen in the Mayo Clinic cohort (5). A national observatory could be set up to record and follow-up a cohort of patients and help address clinical issues related to these borderline breast lesions.
Acknowledgements
Acknowledgments are due to Mrs. Odile Audrain and Tracey Westcott for their assistance in the preparation of the manuscript.
Footnotes
-
Conflicts of Interest
None.
- Received August 27, 2012.
- Revision received October 21, 2012.
- Accepted October 22, 2012.
- Copyright© 2012 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.