Ductal carcinoma in situ is a preinvasive tumour of the breast originating from the cells that line the mammary ducts. A broad range of diseases is grouped under this term, ranging from low-grade indolent lesions to high-grade aggressive precursor of invasive tumours. From an architectural point of view, several subtypes are described: solid, cribriform, micropapillary or papillary, with or without necrosis (comedocarcinoma), with different behaviours, different potential of local relapse and different association with microinvasive or invasive carcinoma [1]. Its incidence is increasing due to the adoption of screening programmes, though without a decline in invasive cancer incidence. In addition, there is a strong consensus that DCIS treatment in most patients has no clear effect on mortality reduction. This suggests an overdiagnosis and an overtreatment particularly of low-grade in situ lesions that amount for a 20% of all DCIS [2].

Historically, survival for these patients is 97–100% with death presumed to be linked to the spread of unrecognized small invasive component: the risk of cancer-related death in DCIS patients is estimated at 1.9% within 10 years [3].

The treatment of the axilla in this pattern of patients represents a strong example of a debated matter and overtreatment.

Given that DCIS is characterized by neoplastic cell proliferation within the mammary ductal system, with no evidence of invasion into the surrounding stroma, it has theoretically 0% potentiality for axillary or distant metastases by definition, and consequently there wouldn’t be any role for axillary evaluation even in case of high-grade DCIS [4–7].

In fact the preoperative diagnosis of DCIS by core needle is upstaged after the final pathologic report in as many as 10–40% of cases (increasing in relation to the method of biopsy and the number/size of samples done, being more unlikely with the VAB). As a consequence, these patients require to be subsequently submitted to a separate surgical axillary procedure, with the technical difficulties related to the tracer migration and the lower detection rate at a second operation [5, 8–10].

In any case, there is currently no validated method to predict which DCIS-diagnosed women will have invasive cancer at the final pathological report, but this is not sufficient to justify performing SLNB in all DCIS diagnosis at the preoperative biopsy.

One thing is for sure: until the late 1990s, axillary dissection was indicated also in cases of DCIS; later, according to the pioneering experience of Silverstein that questioned the need for routine ALND in DCIS and recommended that it must be abandoned, the standard became SLNB.

Now the issue in these patients is “to do or to avoid sentinel node biopsy?” [11]. The ALND has no more any role in DCIS; moreover, there are precise indicators (e.g. from EUSOMA, the European Society of Breast Units) of good quality of the procedure, among which there is “Proportion of patients with DCIS who do not undergo axillary clearance” [12].

We reached this awareness also thanks to important protocols such as National Surgical Adjuvant Breast and Bowel Project NSABP B17 and B24 (and others) and their subsequent reviews where the risk of axillary recurrence in DCIS was found to be less than 1% [13–17].

Therefore, the real topic should be “how to be sure that the preoperative diagnosis of DCIS will effectively relate with a final diagnosis of pure DCIS?” [18, 19]

The centre of the matter is effectively that the reliability of pure DCIS diagnosis can only be subsequent to surgical excision, as an underestimation of the presence of microinvasiveness and invasiveness has been described in even up to 42% of patients, when preoperative DCIS diagnosis is performed by core-needle biopsy on an area of microcalcifications [20–23].

At the same time, the rate of sentinel node positivity in DCIS is described as 1–13%; it is also reported that axillary metastases are found in about 1–2% in case of pure DCIS patients treated with axillary dissection. In a meta-analysis of more than 3000 patients, a 7.4% of sentinel node positivity has been found with a 3.7% in patients with a definitive diagnosis of pure DCIS. ASCO reports 0.9% of pN1 and 1.5% of pN1mic for patients proved to have pure DCIS on final resection [24].

For this reason the vast majority of authors traditionally believe that there is a subset of DCIS at high risk for microinvasive or invasive tumour and a little subgroup of pure DCIS anyway with positive axilla, consequently with the indication for sentinel node biopsy.

In the attempt to define the subgroup of DCIS where the SLNB could be done for the high probability of an invasive component at the final diagnosis, the most important and updated guidelines (see “An overview on the main international guidelines”) indicate SLNB in these subsets of patients:

The inability to obtain clear margins, multicentric disease, large tumour size and contraindications to radiotherapy are the indications for mastectomy in case of DCIS, but they are also risk factors for the presence of invasive cancer and consequently the presence of a possible metastatic sentinel lymph node. In a series by Tan et al. [31], 33% of patients who underwent mastectomy for DCIS had an occult form of invasive cancer revealed by final pathology; among these patients, more than 10% were found to be positive at SLNB [31, 32].

Guillot et al. [26] made an interesting research on 241 patients with pure extensive DCIS in preoperative assessment submitted to mastectomy (from 2000 to 2009, treated at Institute Curie in Paris) followed by an axillary staging in 221 cases (109 SLNB, 93 ALND , 19 mixed procedures in the period of learning curve for SLNB). This work revealed:

The analysis showed that palpable tumours, opacity on mammography and preoperative high-grade DCIS were significant predictors of invasiveness; BMI >25, palpable tumours and opacity on mammography were statistically significant predictors of ALN metastases. In their paper, Guillot et al. furthermore made an overview on microinvasion, invasion and axillary involvement in patients with preoperative diagnosis of pure DCIS, evaluating 11 studies published from 2005 and 2011 for a total of 2319 patients. They found that at the final pathology report, the confirmed pure DCIS were in 1312 cases (75.7%), with 6.6% of DCIS plus microinvasive carcinoma and 22% of DCIS associated with invasive carcinoma (11.3% positive for axillary lymph node metastasis). At this regard, however, we should consider the high impact of SLNB positivity detected by IHC in those works [26].

To make everything more complex, most of the SLN metastases of DCIS consist of micro-metastases and ITCs, and sentinel node is the only positive node usually even in case of macro-metastasis [17, 18, 33].

A new era in the management of the axilla was introduced by the revolutionary studies published by A. Giuliano et al. [34] and by V. Galimberti et al. [35]: considering patients with invasive cancer, they stated the omission of axillary clearance in case of micro-metastases or in patients treated with a conservative approach with 1–2 macro-metastatic lymph nodes. Considering this trend towards minimizing axillary surgery for invasive cancer, trials are ongoing investigating whether a SLND can be avoided at all in clinically node-negative patients with invasive carcinoma [36].

We should therefore question ourselves about the meaning of performing the SLNB in patients with a diagnosis of DCIS.

To date, there are weak evidences in the literature about the significance of SLNB in pure DCIS: it has been shown that even a positive sentinel node in DCIS does not affect survival and is not associated with a high risk of local or distant recurrence [37–39].

Moreover, the presence of cancer cells in the SLN of DCIS patients is a debated issue with a still unclear significance. Some authors mentioned that the cells might represent the result of a mechanical displacement, a sort of micro-embolism of breast cells that have been dislodged by a sampling procedure through the lymphatic system, particularly in case of vacuum-assisted stereotactic biopsy with its multiple sampling [18, 40–42].

Among many retrospective studies, C. Tunon-de-Lara et al. [46] published in 2015 an original paper reporting a prospective multicentric French experience. They examined the relevance of using the SLNB upfront for patients with:

The study was effectively designed for establishing the rate of needless ALND avoided (in the case of a final histology of microinvasive/invasive carcinoma). Other endpoints were the underestimation of invasion by VAB itself, the rates of SLND and the positive sentinel nodes. The results showed:

Another original clinical trial by L.M. van Roozendaal [39] in 2016 included 910 DCIS patients treated in the Netherlands between 2004 and 2013, with the following results:

The authors evaluated the results as consistent with other two recent studies [48, 49] that showed 2.9% pN1mi and 2.4% pN1 with 21% invasive breast cancer and 0%, 0.5% and 30%, respectively [39].

Coromilas et al. [50] published in 2015 on JAMA Oncology an interesting retrospective research about the influence of hospital- and surgeon-related factors on performing any kind of axillary lymph node surgery in DCIS. Looking at 35,591 women who were diagnosed with DCIS and consequently had lumpectomy or mastectomy between 2006 and 2012, the authors found: