Incidence and epidemiology
Due in large part to increased mammographic screening, DCIS incidence increased 7.2 fold in the USA from 1980 to 2001 and LCIS 2.6 fold [
6,
7]. Similar trends were observed in other Western countries [
8]. The median age at diagnosis in most LCIS series varies from 45 to 50 years [
1,
8], i.e. 7–8 years younger than in the DCIS series [
5]. Few data suggest a possible LCIS increase due to hormone replacement therapy (HRT) [
6]. In our series, 57% of 79 menopausal women received HRT. Despite the possible discovery by clinical symptoms, very often due to benign associated lesions (cyst or fibroadenoma), our series confirms that about 65% of LCIS were discovered by clustered microcalcifications, such as reported by other studies [
9,
10]. From a pathological point of view, the distinction between LCIS and ALH is sometimes difficult and subjctive. Both lesions are often associated [
11-
13] and finally represent a temporal continuum in which the extent of lobular involvement varies [
1]. Moreover, the loss of E-cadherin expression characterizes LCIS, facilitating a distinction from ductal lesions.
Significance of LCIS and possible underestimation at core-needle biopsy (CNB)
For many years, LCIS (only representing 1-2% of all breast cancers) has just been considered a marker of risk for subsequent (in situ or invasive) cancer in both breasts [
1]. The discovery by mammography or presence of another benign lesion leads to a complementary excision after CNB only in a variable part of cases, ranging from 40% to 70% in the literature and without clear selection criteria. Some authors proposed surveillance or complementary excision according to several and often different parametres, sometimes including «lobular neoplasia» encompassing both LCIS and ALH. In a large review by Hussain [
2], including 31 studies, 789 out of 1229 (64%) patients diagnosed with LN on CNB had surgical excision: 280 were classified ALH, 241 LCIS, 22 pleomorphic LCIS (PLCIS) and 246 unspecified LN. In this review, the overall underdiagnosis rate.
(DCIS and/or invasive BC at subsequent excision) was 27% (211 out of 789). In a wide series from New York, among 201 out of 285 (71%) excisions performed for LN after CNB, 26 (13%) were upgraded; 7 had DCIS, 9 ILC and 10 IDC [
14]. However, the risk of underestimated malignancy at subsequent excision (DCIS or infiltrating carcinoma) was only 8% (DCIS: 2.7% and infiltrating BC: 5.3%) among 478 lesions in a literature review reported by Murray, but with wide variations (from 3% to 35%) [
15]. Finally, the initial underestimation risk after CNB showing LCIS or ALH is quite difficult to assess, without clear predictive criteria in the literature [
16].
Comparisons between DCIS and LCIS treatments
DCIS now represent 15 to 20% of all breast cancers (63 000 cases in the USA in 2014) and are considered long-term precursors of invasive BC in 40-50% of the cases, even in low-grade lesions [
5]. However, DCIS encompass several heterogeneous lesions due to many biological and molecular features leading to variable progressions to invasion. There are much more DCIS treatment data than for LCIS [
13,
17,
18].
Mastectomy remains performed in 30-42% of DCIS according to the series [
19,
20], due to extension of the lesions, multicentricity and/or patients' preference. The local control rate was about 98% [
4]. On the other hand, very few data are available on mastectomy rate in LCIS [
7,
13]. In our series, multicentricity seemed to be a factor leading to mastectomy.
In a recent American report using the Surveillance, Epidemiology and End Results (SEER) database, among 11 641 patients diagnosed with LCIS from 2000 to 2009, 16% underwent mastectomy, but with large differences based on geographic areas and age (from 12% to 24%) [
7]. The local control rate, such as in our 24 patients, was almost 100%.
Due to more frequent discovevry of «small lesions», a conservative approach was gradually used in DCIS [
17,
18]. The standard treatment includes BCS and whole breast irradiation (WBRT) according to four large randomized trials (combined in a meta-analysis including over 4 500 cases) and retrospective studies showing that WBRT halfed the local recurrence rates (LR), both in situ and invasive [
21-
25]. Consequently, for these lesions, the 10-year LR rates are about 5-8% in the most recent series [
26-
28], reaching only 1% at 7 years in the very selected study on low-risk DCIS by RTOG [
29]. However, in 20-25% of the cases, for various reasons, WBRT is not used in DCIS, leading, even in very selected cases (small lesions, wide margins, low grade), to a 15-20% 10-year LR rate, half of which invasive [
30,
31].
This is a crucial problem, because now several studies confirm the unfavourable impact of invasive LR in patients treated for DCIS with a 12-18% long-term of metastatic evolution [
21,
22,
32]. For LCIS, there are no clear therapeutic guidelines, both after CNB and surgical excision [
1,
2,
9]. The previously SEER-quoted study [
7] showed that 10% of the patients had biopsy alone (precise criteria for the choice not specified) and 73% excision alone, whereas only 1% (116 patients) underwent complementary WBRT, without specific choice criteria or data on local control.
Table
3 shows the risk of subsequent invasive or in situ carcinoma after treatment of LCIS by excision alone in the main literature series [
33-
36]. This risk is quite similar in several DCIS series with the same treatment, i.e. San Francisco or Ontario series [
30,
31], as well as a very selected study by ECOG [
37], with approximately 50% of invasive local recurrences.
Table 3
Risk of subsequent invasive or in situ carcinoma after treatment of LCIS by excision alone
Salvadori (1991) | | 78 | 58 | 5 | 6,4 | - | - |
Ottesen (1993) | | 69 | 61 | 8 | 11,6 | 4 | 5,8 |
Fisher E. (2004) | | 180 | 144 | 9 | 5 | 17 | 9,4 |
Chuba (2005) | | 3141* | 120 | NS | 8,8 | NS | NS |
In comparison with DCIS, the slope of the LR curve for LCIS seems quite less sharp and more spread out in time (see Figure
1), suggesting that the overall cancerization process was slower.The increase of local recurrence rates due to young age and incomplete excision (close and/or positive margins) is not clearly documented in LCIS compared to DCIS. However, few LCIS subtypes, e.g. pleomorphic LCIS, very extensive LCIS (with over ten involved acini) or LCIS with a great amount of necroses, seem to be clearly more aggressive lesions [
3,
9] and should be treated such as DCIS, as suggested by 2009 French Breast Carcinoma in Situ guidelines (
www.e-cancer.fr).
Therefore, in these aggressive LCIS, WBRT addition should be discussed in the multidisciplinary team, in order to minimize the LR risk. In our previous report, we observed only one invasive LR out of 25 cases, with a 153-month follow-up [
38]. In the present study, the LR is slightly higher (3/17), but the number of cases is lower and all LR occurred beyond 10 years.
These results are quite similar to DCIS treated by lumpectomy and WBRT in the series with a more than 120-month follow-up, both in retrospective studies and randomized trials [
5,
21-
23]. However, it remains difficult to better identify the «aggressive» LCIS subtypes. A recent study by the Curie Insitute showed that KI67 should be used as a «discriminant» parameter to select more aggressive LCIS. Indeed, among 43 patients, the 5-year LR rates were 1/34 (3%) for low Ki 67 (≤10%) and 3/9 (33%) for high Ki 67 (>10%) (p = 0.0054) [
39].
Another possible way of reducing the long-term risk of LR is the use of Tamoxifen. Indeed, the positivity rate of ER and/or PgR in LCIS is higher than in DCIS [
9], and the results of the former NSABP P-1 prevention study showed that the most effective reduction of invasive BC by tamoxifen was observed among women with LCIS and ALH, suggesting a possible increased impact of anti-estrogens in the first steps of breast cancerization [
40]. Similar results were observed in the NSABP P-2 prevention trial (STAR) using Raloxifen [
41]. On the other hand, it should be remembered that the CBC rate in LCIS is quite higher than in DCIS, reaching 14% at 10 years in a large study from Connecticut tumor registry [
42]. Consequently, the use of tamoxifen should be considered, especially in pre-menopausal women with a lower incidence of thrombo-embolic accidents and uterine cancers.
In short, LCIS is not always an indolent disease. Table
4 shows several features in comparison with DCIS. In fact, many data are quite similar with the almost same long-term LR rates, both after BCS alone and BCS + WBRT.The scarcity of this lesion urgently requires further larger studies, at least prospective long-term records, to clearly assess the LR and CBC rates according to radiological, clinical, histopathological and molecular features, in order to evaluate the potential benefit in selected cases of WBRT and tamoxifen [
43].
Table 4
Comparison between DCIS and LCIS (data adapted from references 1, 2, 5, 9, 13, 17, 36, 37)
Frequency | 15-20 | 1-3 |
Calcifications at diagnosis | 90 | 60-70 |
Excision rate after CNB | 100 | 50-80 |
Hormone receptor positivity | 70-80 | 90 |
Treatments: | | |
Mastectomy | 30-40 | 10-15 |
BCS | 20-25 | 80-85 |
BCS + WBRT | 60-70 | 1-3 |
CBC | 5-8 | 12-15 |
10-year LR (after BCS) | 15-30 | 10-15 |
Invasive LR | 50-60 | 30-40 |