Ductal carcinoma in situ (DCIS) represents 20% of screen-detected breast cancers. The likelihood that certain types of DCIS are slow growing and may never progress to invasion suggests that our current standards of treating DCIS could result in overtreatment. The LORIS (LOw RISk DCIS) and LORD (LOw Risk DCIS) trials address these concerns by randomizing patients with low-risk DCIS to either active surveillance or conventional treatment.
To determine the upgrade rate of DCIS diagnosed on core needle biopsy to invasive carcinoma at surgery and to evaluate the safety of managing low-risk DCIS with surveillance alone, by characterizing the pathologic and clinical features of upgraded cases and applying criteria of the LORD and LORIS trials to these cases.
A 10-year retrospective analysis of DCIS on core needle biopsy with subsequent surgery.
We identified 1271 cases of DCIS on core needle biopsy: 200 (16%) low grade, 649 (51%) intermediate grade, and 422 (33%) high grade. Of the 1271 cases, we found an 8% upgrade rate to invasive carcinoma (n = 105). Nineteen of the 105 upgraded cases (18%) had positive lymph nodes. Low-grade DCIS was least likely to upgrade to invasion, comprising 10% (10 of 105) of upgraded cases. Three of the 105 upgraded cases (3%) met criteria for the LORD trial, and all were low-grade DCIS on core needle biopsy with favorable biology on follow-up.
There is a clear risk of upgrade to invasion on follow-up excision; however, applying strict criteria of the LORD trial effectively decreases the likelihood of a missed invasive component or missed aggressive pathologic features.
Ductal carcinoma in situ (DCIS) makes up approximately 20% of all screen-detected breast cancers.1,2 This diagnosis is typically made after a radiologically suspicious area presenting as calcifications or less likely a mass is biopsied. Breast core needle biopsy (CNB), typically using ultrasound or stereotactic guidance, is the standard method of obtaining tissue for pathologic diagnosis.3–5 The current standard of treatment for DCIS diagnosed on CNB includes either breast-conserving surgery, with or without radiotherapy, or mastectomy followed by chemoprevention in patients who do not opt for bilateral mastectomy. Although the goal for treatment of DCIS is to prevent the occurrence of invasive carcinoma, this standard approach has recently come under scrutiny because of evidence showing excellent survival of patients diagnosed with DCIS6 and the likelihood that certain types of DCIS are slow growing and may never progress to invasion. Currently there are clinical trials monitoring disease progression in patients diagnosed with low-risk DCIS.7–10 The LORIS (LOw RISk DCIS) and LORD (LOw Risk DCIS) trials evaluate whether screen-detected non–high-grade (HG) or low-grade (LG) DCIS, respectively, diagnosed on CNB can be safely managed with active surveillance rather than conventional treatment.7 Detailed further in the Materials and Methods, inclusion criteria for the LORIS trial11 include women 46 years of age or older with screen-detected calcifications diagnosed with asymptomatic LG or intermediate-grade (IG) DCIS on CNB. Patients excluded from the trial are those with a mass lesion on screening, clinical presentation of ipsilateral bloody nipple discharge, or a personal history of breast or ovarian carcinoma and those considered high risk based on family history. The LORD trial10 eligibility criteria are similar except that women must be 45 years or older with a diagnosis of pure LG DCIS on CNB.
An obvious risk in omitting surgical excision for a CNB diagnosis of non–HG DCIS is the possibility of a missed invasive component not sampled by CNB. By definition, CNB is a limited sampling of the lesion and may underestimate the presence of invasive carcinoma or higher-grade lesions.12,13 The upgrade rate of DCIS diagnosed on CNB to invasive carcinoma at surgical excision has been reported to range from 13% to 50%.14–20 In addition, many of these studies showed no association between grade of DCIS and upgrade to invasion.14,15,17,21,22
An understanding of the pathologic aggressiveness of the invasive disease identified at surgical excision is pertinent to current discussions questioning the need to surgically excise certain types of DCIS. To our knowledge, ours is the first study to fully characterize the pathologic features of the invasive carcinoma at the time of upgrade on surgical excision. Our findings will provide further insight into the biologic aggressiveness of DCIS diagnosed on CNB when considering the safety of observation alone.
MATERIALS AND METHODS
With institutional review board approval, a database search was performed at 2 academic centers for a pathologic diagnosis of DCIS on CNB during a 10-year time period from 2006 to 2016. Cases of DCIS with microinvasion diagnosed on CNB were excluded. From those cases, a second search was performed selecting only the cases that had undergone a subsequent surgery with an available pathologic report. Patients with a diagnosis of invasive mammary carcinoma (IMC) in the ipsilateral breast prior to CNB as well as patients who had undergone neoadjuvant treatment prior to follow-up surgery (FUS) were excluded from the study. All data were collected from radiology and pathology reports and included patient age; risk factors for breast cancer; imaging modalities used to evaluate the disease; presence and characteristics of calcifications on imaging; presence of a mass lesion on imaging; size of calcifications and mass lesions; size of needle gauge; pathologic characteristics of DCIS, including nuclear grade, estrogen receptor (ER), and progesterone receptor (PR) on CNB (when available) and FUS; and pathologic characteristics of the invasive component on FUS, including histologic type, Nottingham histologic grade, tumor size, mitotic index (low defined as ≤3 mitoses/mm2, intermediate as 4–7 mitoses/mm2, and high as ≥8 mitoses/mm2), hormone receptor (HR) status (ER and PR), HER2 status, number of lymph nodes positive, and final pathologic stage. Histologic slides were not reviewed.
Both academic centers follow College of American Pathologists–approved criteria23 to determine grade for DCIS. For LG DCIS, nuclei are monotonous and small, 1.5 to 2 times the size of normal breast epithelial cell nuclei, with finely dispersed chromatin and only occasional nucleoli and mitoses and with polarization toward luminal spaces (Figure 1). For HG DCIS, nuclei are markedly pleomorphic and greater than 2.5 times the size of normal breast epithelial cell nuclei, with vesicular chromatin with irregular chromatin distribution, prominent nucleoli, and frequent mitoses, usually not polarized toward luminal spaces (Figure 2). Intermediate-grade DCIS grading criteria are intermediate between LG and HG criteria (Figure 3). Architectural pattern and necrosis were not factors in determining DCIS grade. Quantity of DCIS on CNB was not recorded and was not consistently provided in pathology reports at one institution and never provided in the pathology reports at the second institution. Quantity and size of DCIS on CNB do not impact final pathologic stage and have previously been shown to have poor correlation with radiographic findings.24 In addition, size of DCIS was not included in the eligibility criteria for the LORD and LORIS trials.
Overall upgrade rate of DCIS on CNB to IMC on FUS was calculated. Cases were stratified into 3 groups based on the nuclear grade of DCIS on CNB (LG, IG, and HG) and upgrade rates were determined within each group and compared. The χ2 test was used and a P value of .05 was considered statistically significant.
We applied eligibility criteria of the LORIS and LORD trials to our cohort to determine the number of our cases with upgrade that would have qualified for surveillance alone. Inclusion criteria for the LORIS trial11 include women 46 years or older with screen-detected calcifications associated with histologically confirmed LG or IG DCIS diagnosed either on CNB or on open diagnostic surgical biopsy without clear margins. The LORIS trial allows for bilateral DCIS as long as non–HG DCIS is confirmed in both breasts at the time of diagnosis and allows for previously surgically treated contralateral DCIS of any grade. The trial excludes patients with a previous or current diagnosis of invasive breast carcinoma or previous ipsilateral DCIS, patients with recent onset of ipsilateral bloody nipple discharge without a benign explanation, and patients at high risk for developing breast cancer (familial breast cancer or prior exposure to mantle field radiotherapy). The trial also excludes patients with a mass lesion at the site of calcifications not proven on biopsy to be a specific benign lesion and unequivocal comedo necrosis on histologic examination. The LORD trial10 eligibility criteria include women 45 years or older of any menopausal status with calcification-only lesions detected on screening mammogram associated with pure LG DCIS diagnosed on representative vacuum-assisted CNB and with marker placement at the biopsy site. Any size DCIS is allowed; however, good pathologic and radiologic correlation without suspicion for IG or HG DCIS or invasive breast cancer is required for inclusion. The LORD trial excludes any patient with presence of a mass, increased density around calcifications, or architectural distortion on mammography; symptomatic DCIS (DCIS detected by palpation or nipple discharge); bilateral DCIS; Paget disease; invasive breast cancer; or lobular carcinoma in situ. Patients with a history of invasive breast cancer or DCIS, history of other cancers except cervical carcinoma in situ and basal cell carcinoma of the skin, or family with BRCA 1/2 mutation and patients who are currently pregnant or breastfeeding are excluded.
A total of 1271 cases of DCIS diagnosed between 2006 and 2016 that fit criteria were identified at 2 academic institutions. Of these cases, 200 (16%) were LG, 649 (51%) were IG, and 422 (33%) were HG DCIS on CNB (Table 1). On FUS, we found an 8% upgrade rate to IMC (105 of 1271 cases). Of the 105 upgraded cases, 10 (10%) were previously LG DCIS, 43 (41%) were IG DCIS, and 52 (49%) HG DCIS on CNB (Table 1). Nuclear grade of DCIS significantly correlated with upgrade to invasion on FUS (P < .001). Upgrade rate based on nuclear grade of DCIS was 5% (10 of 200) for LG DCIS, 7% (43 of 649) for IG DCIS, and 12% (52 of 422) for HG DCIS (Table 2). Patient age ranged between 27 and 86 years, with an average age of 56 years. Size of IMC ranged from microinvasion to pT3 tumors, with the majority of cases, 54% (57 of 105), staged as pT1a or less. Only 7% (7 of 105) of tumors were pT2 or pT3 (Figure 4).
Lymph node status was available for 87 cases with upgrade on FUS. Nineteen of 87 cases (18%) had positive lymph nodes, constituting 1% (19 of 1271) of all cases diagnosed as DCIS on CNB. Of the 19 lymph node–positive cases, 12 (63%) were diagnosed as HG DCIS, 6 (32%) as IG DCIS, and 1 (5%) as LG DCIS on CNB (Figure 4). Patient age for these 19 cases ranged from 34 to 65 years, with an average age of 50 years. Eleven of these 19 patients (58%) had available information on risk factors, with 6 of 11 patients (55%) having a first-degree relative with breast cancer and 5 of 11 patients (45%) with no significant risk factors for breast cancer. Sixteen of the 19 lymph node–positive cases (84%) had available imaging results. Of these 16 cases, 12 (75%) presented as calcifications, 3 (19%) presented with a mass lesion, and 1 (6%) presented as a mass with calcifications (Table 3). Size of the radiologic lesion was provided in 8 of the 16 cases (50%), with a mean size of 30 mm. Five of these 8 cases (63%) were calcification-only lesions, ranging from a 4-mm cluster of calcifications to a 75-mm span of calcifications. Two of the 8 cases (25%) were mass lesions measuring 10 mm and 50 mm by imaging, and 1 of the 8 cases (12%) was a mass with calcifications measuring 20 mm. Of the 16 lymph node–positive cases with radiologic information, 13 (81%) were calcifications sampled by stereotactic-guided CNB, 3 (19%) were mass lesions sampled by ultrasound-guided CNB, and 10 (63%) had information on size of the needle gauge used for CNB. Seven of these 10 cases (70%) presented as calcifications and were sampled with needles that ranged from 7 to 12 gauge, with a median needle gauge of 9. Three of these 10 cases (30%) were mass lesions. Two of these 3 (67%) were sampled with a 14-gauge needle and 1 (33%) was sampled with a 12-gauge needle. Grade of IMC was available in 18 of the 19 lymph node–positive cases (95%). Of these, 8 (44%) were poorly differentiated, 9 (50%) were moderately differentiated, and 1 (6%) was well differentiated. Statuses for ER, PR, and HER2 were available in 17 of the 19 lymph node–positive cases (89%). Of these, 8 (47%) were HR positive/HER2 negative, 6 (35%) were HR negative/HER2 positive, 1 (6%) was HR positive/HER2 positive, and 2 (12%) were HR negative/HER2 negative.
Radiologic data were available for 95 of the 105 upgraded cases (90%). Of the 95 cases, 58 (61%) presented as calcifications, 21 (22%) presented as mass lesions or asymmetry, and 16 (17%) showed a mass lesion with calcifications. Of the 37 cases that presented as mass lesions with or without calcifications, the radiologic size was available for 29 cases (78%), ranging between 7 and 60 mm (mean = 18 mm; median = 12 mm). Of the 58 cases that presented as calcification-only lesions, the size of the calcifications was reported in 27 cases (47%) and ranged between 3 and 120 mm (mean and median = 34 mm). Information on needle gauge used for CNB was available for 35 of 58 calcification-only lesions (60%), with a median needle gauge of 11, and in 34 of 37 cases of mass lesions (92%), with a median needle gauge of 14. Five of the 37 cases (14%) presenting as mass lesions were considered concordant by radiology, with 3 of the 5 cases (60%) diagnosed as DCIS involving a papilloma, 1 (20%) diagnosed as papillary carcinoma in situ, and 1 (20%) diagnosed as DCIS associated with a mucocele-like lesion. The remaining 32 cases (86%) diagnosed as purely DCIS were not concordant with a mass lesion. All 58 cases of DCIS presenting as calcification-only lesions were considered concordant by radiology. When comparing the radiologic features with the tumor stage on FUS, we found that there was no significant correlation between the two (P = .93) (Table 3).
Nine of the 10 upgraded LG DCIS cases had available imaging. Of these 9 cases, 4 (44%) presented as calcifications, 4 (44%) presented as a mass, and 1 (11%) presented as a mass with calcifications. Hormone receptor status of DCIS on CNB was available in 3 of 10 upgraded cases (30%), and all were ER and PR positive. In 5 of the 10 upgraded cases (50%), the patients underwent mastectomy, and the rest underwent lumpectomy as surgical treatment. On FUS, Nottingham histologic grade of IMC was available in 9 of the 10 cases (90%). Of these 9 cases, 4 (44%) were well differentiated, 5 (56%) were moderately differentiated, 8 (89%) demonstrated a low mitotic index, and 1 (11%) displayed an intermediate mitotic index. Tumor size ranged from less than 1 mm (n = 1) to 15 mm, with an average size of 8.6 mm. All cases were pathologic stage pT1c or lower (Figure 4). In 7 of the 10 cases (70%) for which HR results of IMC were available, all were ER and PR positive. HER2 receptor status was available for 8 of the 10 upgraded cases (80%) and none were positive. Of these 8 cases, 5 (63%) were negative by HER2 fluorescence in situ hybridization, 2 (25%) were negative (0) by immunohistochemistry, and 1 (12%) was equivocal (2+) by immunohistochemistry, without available fluorescence in situ hybridization results (Table 4). The case of LG DCIS on CNB that did not have imaging results available involved a 44-year-old patient who on FUS was diagnosed with IMC, no special type. The IMC was moderately differentiated and HR positive/HER2 negative, with a final pathologic stage of pT1c N1a. This was the only LG DCIS case with positive lymph nodes. Based on age, this patient would not have qualified for either the LORD or LORIS trial.
The remainder and the majority of DCIS cases that were upgraded to IMC on FUS were IG (43 of 105; 41%) or HG (52 of 105; 49%). These cases accounted for all HER2-positive IMCs (28 of 105; 27%) (Table 4), and all were staged as pT2 or pT3 (7 of 105; 7%) (Figure 4).
Eligibility criteria for the LORD and LORIS trial are very similar with the exception of nuclear grade. The LORD trial includes only pure LG DCIS, whereas the LORIS trial includes both LG and IG DCIS. Of the 105 upgraded cases, 3 (3%) met criteria for both trials because these cases were LG DCIS. An additional 17 of the 105 upgraded cases (16%) met criteria for the LORIS trial but not the LORD trial because these were IG DCIS, resulting in a total of 20 of 105 upgraded cases (19%) meeting LORIS trial criteria (Table 5).
The 3 LG DCIS cases that met criteria for both the LORD and LORIS trials represent a small fraction (2%) of the 200 CNB cases diagnosed as LG DCIS. Pathologic characteristics of these 3 LG DCIS cases on FUS included histologic type of IMC, no special type, in 2 cases (67%) and invasive mucinous carcinoma in 1 case (33%); histologic grade of well differentiated in all cases (100%); and stage pT1aN0 in 2 cases (67%) and pT1bN0 in 1 case (33%). The remaining 7 LG DCIS cases would have been excluded from the LORD and LORIS trials because of 5 cases (71%) presenting with a mass lesion, 1 case (14%) in which the patient had a strong family history of breast cancer, and 1 case (14%) in which the patient was 44 years old at diagnosis.
Of the 20 cases that met LORIS trial eligibility criteria (Table 5), 8 (40%) demonstrated aggressive pathologic features on FUS with HG carcinoma and/or positive lymph nodes. Of these 8 cases with aggressive pathologic features on FUS, all were previously diagnosed with IG DCIS on CNB, 4 (50%) had background HG DCIS on FUS, and 3 (38%) were poorly differentiated on FUS; 1 (12%) was staged as pT2N0, 1 (12%) was pT1cN1a, and 1 (12%) was pT1miN1mi.
Upgrade rate on excision after a CNB diagnosis of DCIS has been the subject of a number of studies revealing the inherent sampling error associated with CNB. On review of the literature, we found 9 studies that reported on upgrade rate based on nuclear grade of DCIS.* As summarized in Table 2, study size ranged from 86 to 608 cases of DCIS on CNB, with overall upgrade rate ranging from 13% to 50%. Of note, none of these studies had exclusion criteria beyond excluding cases that did not have FUS information available. Similar to our findings, the minority of cases in these 9 studies were LG DCIS and most showed that LG DCIS was least likely to upgrade to IMC on FUS. Only 2 of the 9 studies provided information on pathologic characteristics of the IMC on FUS, and that information was limited. Go et al22 reported whether the IMC was microinvasion (≤1 mm) or frank invasion (>1 mm), showing that 5 of the 15 upgraded LG DCIS cases were microinvasion. Wahedna et al15 provided the grade of IMC and reported that all 3 LG DCIS cases were upgraded to well-differentiated IMC. Four of these 9 studies found that DCIS grade significantly correlated with upgrade to IMC on FUS,18,20,25,26 similar to our findings.
With recent Surveillance, Epidemiology and End Results data showing excellent prognosis for women diagnosed with DCIS,6 overtreatment of DCIS is of concern. Two recent trials assessing the safety of surveillance in place of surgery and radiation for certain types of DCIS diagnosed on CNB, the LORIS and LORD trials, will answer important questions concerning overtreatment. Although both trials have similar eligibility criteria, including screen-detected calcifications, no mass lesion or architectural distortion on imaging, no symptomatic DCIS, no personal history of breast cancer, and no high-risk features for breast cancer, the LORIS trial allows for both LG and IG DCIS on CNB, whereas the LORD trial allows only pure LG DCIS on CNB. Based on our cohort of upgraded cases, and applying eligibility criteria of these 2 trials, the inclusion of both LG and IG DCIS would have resulted in 20 eligible patients with missed IMC on CNB, with 8 of these 20 also missing aggressive pathologic features, including HG DCIS, poorly differentiated IMC, and positive lymph nodes. In contrast, restricting inclusion to only pure LG DCIS would result in 3 eligible patients with missed IMC on CNB with none of these cases associated with aggressive pathologic features.
Although we and others have shown a clear risk of upgrade to IMC on FUS after a CNB diagnosis of DCIS, our data suggest that applying the most strict criteria implemented in the LORD trial would effectively decrease the likelihood of a missed invasive component or missed aggressive pathologic features. Close surveillance, including additional imaging modalities such as magnetic resonance imaging or tomosynthesis, would be essential in detecting those few cases that meet criteria of the LORD trial but have an invasive component not sampled on CNB. The 3 cases that fit eligibility criteria for the LORD trial were upgraded to small invasive carcinoma with pathologic features compatible with biologically favorable disease. It is possible that these invasive carcinomas would be slow growing without much consequence of delaying surgery until detection through close surveillance.
In conclusion, our data show that the subset of patients defined by the eligibility criteria of the LORD trial could safely avoid surgery after a CNB diagnosis of LG DCIS for screen-detected calcification-only lesions and instead opt for close surveillance. Definitive guidance concerning changes to standard therapy for DCIS awaits results of these trials.
The authors have no relevant financial interest in the products or companies described in this article.