Abstract
Background
Patients with triple-negative breast cancer (TNBC) and a pathological complete response (pCR) after neoadjuvant chemotherapy may be suitable for non-surgical management. The goal of this study was to identify baseline clinicopathological variables that are associated with residual disease, and to evaluate the effect of neoadjuvant chemotherapy on both the invasive and ductal carcinoma in situ (DCIS) components in TNBC.
Methods
Patients with TNBC treated with neoadjuvant chemotherapy followed by surgical resection were identified. Patients with a pCR were compared with those who had residual disease in the breast and/or lymph nodes. Clinicopathological variables were analysed to determine their association with residual disease.
Results
Of the 328 patients, 36·9 per cent had no residual disease and 9·1 per cent had residual DCIS only. Patients with residual disease were more likely to have malignant microcalcifications (P = 0·023) and DCIS on the initial core needle biopsy (CNB) (P = 0·030). Variables independently associated with residual disease included: DCIS on CNB (odds ratio (OR) 2·46; P = 0·022), T2 disease (OR 2·40; P = 0·029), N1 status (OR 2·03; P = 0·030) and low Ki-67 (OR 2·41; P = 0·083). Imaging after neoadjuvant chemotherapy had an accuracy of 71·7 (95 per cent c.i. 66·3 to 76·6) per cent and a negative predictive value of 76·9 (60·7 to 88·9) per cent for identifying residual disease in the breast and lymph nodes. Neoadjuvant chemotherapy did not eradicate the DCIS component in 55 per cent of patients.
Conclusion
The presence of microcalcifications on imaging and DCIS on initial CNB are associated with residual disease after neoadjuvant chemotherapy in TNBC. These variables can aid in identifying patients with TNBC suitable for inclusion in trials evaluating non-surgical management after neoadjuvant chemotherapy.
Ductal carcinoma in situ and microcalcification important
Introduction
Triple-negative breast cancer (TNBC) is a highly aggressive molecular subtype that accounts for 15–20 per cent of all breast cancers. Gene expression analysis has revealed that TNBC comprises a heterogeneous group of tumours. Multiple studies1–11 have shown that TNBC is more sensitive to neoadjuvant chemotherapy (NACT) than other molecular subtypes, with pathological complete response (pCR) rates of up to 50 per cent. Although TNBC is associated with poor prognosis, with high rates of metastasis and short relapse-free survival, patients who achieve a pCR on NACT have improved survival and fewer locoregional recurrences than patients with residual disease12.
High pCR rates increase the potential for non-surgical management after NACT. It is important, however, to accurately identify patients who have a pCR; these patients could be included in clinical trials testing the safety of omitting surgery. The predictive value of different imaging modalities has been evaluated among patients with a complete clinical response, but these methods are not sensitive enough to select patients for omission of surgery, as the negative predictive value for pCR is low13. Kuerer and colleagues14 have recently shown that it is possible to identify patients with a pCR in whom significant residual disease is unlikely by using improved imaging techniques, such as extensive image-guided vacuum-assisted core biopsy and fine-needle aspiration biopsy of the tumour bed. Similar prospective studies are being initiated internationally15–20. If the concept of omission of surgery in the exceptional NACT responders moves forward into clinical practice, these patients will still receive whole-breast radiotherapy.
As the number of patients with TNBC who are eligible for non-surgical management increases, it becomes increasingly important to identify and understand the factors that are predictive of residual disease after NACT. Over the past decade, a number of studies have been performed to identify predictors of pCR. Several have focused specifically on TNBC (Table 1). Most of these did not specify the breast and lymph node status in the definition of pCR, nor the role of residual in situ disease2–8,11. In a study of 117 patients with TNBC, Park and co-workers10 found that 57·3 per cent achieved ypT0 status, whereas 42·7 per cent achieved ypTis (residual ductal carcinoma in situ (DCIS) only). The triple-negative subtype and complete radiological response on MRI were predictive of ypT0 status, whereas the mammographic and MRI presentations of the primary lesion, and the presence of microcalcifications and residual disease on mammography or ultrasonography after NACT were not.
Table 1.
Studies evaluating predictors of pathological complete response in patients with triple-negative breast cancer
| Reference | Population | pCR rate (%) | pCR definition | pCR predictors | No correlation with pCR |
|---|---|---|---|---|---|
| Nogi et al.9 | 28 stage II–III | 18 | Not defined | EGFR-negative status (molecular marker) | |
| Huober et al.3 | 509 | 38.9 | ypT0/Tis ypN0 | Age < 40 years* | Tumour grade, cT category, histological tumour type |
| Masuda et al.8 | 33 any T, N0–2, M0 | 36 | ypT0/Tis ypN0 | High Ki-67, non-basal-like phenotype, CK5/6 | Age, T category, N status, grade, HR, p53, androgen receptor |
| Keam et al.5 | 105 | 13·3 | ypT0/Tis ypN0 | High Ki-67 | |
| Li et al.7 | 41 locally advanced, T2–4a, any N stage | 34 | ypT0/Tis ypN0 | Early T category, clinical response after 2 cycles, negative basal-like*, negative EGFR, high Ki-67*, positive nm23-H1* | Age, menopausal status, ECOG performance, N status, grade, HER2 status, CK5/6, cyclin D1 |
| Kraus et al.6 | 56 | 34 | ypT0/Tis ypN0 | None | Basal phenotype markers (CK5, CK14, CK17, EGFR), cell adhesion marker E-cadherin, proliferation marker Ki-67, DCIS |
| Gerber et al.1 | 678 cT1c–4d | 39·3 with bevacizumab 27·9 without bevacizumab | ypT0 ypN0 | Bevacizumab, lower tumour stage, grade 3 tumours | Age, cN category, histological type |
| Tan et al.11 | 183 HR-negative (122 HR–/HER2+, 61 TN) | 19·1 | ypT0/Tis ypN0 | High Ki-67*, tumour grade, HER2 status, LN status | Age, menopausal status, T category, NACT regimen |
| Humbert et al.2 | 50 clinical stage II–III | 42 | ypT0/Tis ypN0 | High Ki-67, negative EGFR*, high ΔSUVmax* | Age, menopausal status, T size, stage, LN status, tumour inflammation, histological grade, CA15.3 value, CEA value, baseline tumour metabolism |
| Jung et al.4 | 143 | 46·3 | ypT0/Tis ypN0 | High number of TILs (P = 0·007), absence of clear cytoplasm (P = 0·008), low necrosis (P = 0·018), high histological grade (P = 0·039) | Age, cT category, mitotic count, retraction artefact, small cell-like feature, DCIS component, fat invasion, lymphocytes in normal glands |
| Park et al.10 | 117 | 57·3 ypT0 42·7 ypTis | ypT0 or ypTis ypNany | For ypT0: TN subtype and complete response on MRI | Presentation of main lesion on MMG, MMG-associated microcalcifications, US shape and posterior features, US calcification, MRI presentation of main lesion, residual disease after NACT on MMG or US |
Independent predictors in multivariable analysis. pCR, pathological complete response; EGFR, epidermal growth factor receptor; CK, cytokeratin; HR, hormone receptor; ECOG, Eastern Cooperative Oncology Group; HER2, human epidermal growth factor receptor 2; DCIS, ductal carcinoma in situ; TN, triple negative; LN, lymph node; NACT, neoadjuvant chemotherapy; ΔSUVmax, change in standardized maximal uptake value; CA, cancer antigen; CEA, carcinoembryonic antigen; TIL, tumour-infiltrating lymphocyte; MMG, mammogram; US, ultrasound.
The optimal candidate for non-surgical management is a patient with no residual invasive carcinoma or DCIS left in the breast that could serve as a nidus for recurrence. The goal of this study was to identify the baseline clinicopathological characteristics associated with residual disease in patients with TNBC after NACT, and to evaluate the effectiveness of NACT in eradicating the invasive and DCIS components of TNBC.
Methods
Institutional review board approval was obtained to query the prospectively managed Breast Cancer Management System Database at the University of Texas MD Anderson Cancer Center (Houston, Texas, USA). A consecutive cohort of patients with TNBC was identified with T1–2 N0–1 disease who had been treated with standard anthracycline- and taxane-based or carboplatin-based NACT, followed by surgical resection (breast-conserving surgery or mastectomy) with sentinel node or axillary node dissection. The patients were treated between January 2010 and December 2015. At diagnosis, all patients underwent diagnostic mammography combined with breast and lymph node ultrasound imaging. Patients with an abnormal axillary lymph node underwent percutaneous biopsy to verify metastasis. Breast MRI was used at the discretion of the treating physician and with input from a radiologist. For MRI, a 1.5- or 3-T GE Signa™ scanner was used (General Electric, Milwaukee, Wisconsin, USA) with the patient positioned prone, with the addition of contrast medium (Magnevist®; Bayer HealthCare Pharmaceuticals, Wayne, New Jersey, USA) and using delayed postcontrast images with fat suppression in the axial plane.
Data analysis and endpoints
The primary endpoint was to identify baseline clinicopathological variables associated with finding residual disease in the breast (not ypT0), lymph nodes (not ypN0) or both (not ypT0N0). As a secondary endpoint, the effect of NACT on the invasive and DCIS components of TNBC was evaluated. Demographic and clinical characteristics were analysed, including the presence of initially malignant-appearing microcalcifications on the mammogram, and the presence of DCIS on the initial diagnostic core needle biopsy (CNB). TNBC status was defined as oestrogen receptor and progesterone receptor positivity below 10 per cent by routine immunohistochemical analysis, and lack of human epidermal growth factor 2 (HER2) amplification by fluorescence in situ hybridization or a HER2 score of 0 by immunohistochemical analysis, according to the standard protocol used for TNBC at MD Anderson Cancer Center.
The tumour volumes before and after NACT measured by each imaging method (mammography, ultrasonography and MRI) were recorded. The radiological response in the breast was calculated as the ratio of post-NACT tumour volume to the initial volume. The response was classified according to the Response Evaluation Criteria in Solid Tumours (RECIST)21: complete response (complete resolution of the mass), partial response (at least 30 per cent decrease in tumour volume), progressive disease (20 per cent increase or more in tumour volume) or stable disease (no change in tumour volume). For a radiological complete response, complete resolution of the tumour on all subsequent imaging was required. Complete resolution of the mass on ultrasound imaging, but with the presence of residual calcifications on mammography, was classified as a partial response. For patients initially diagnosed with N1 biopsy-proven lymph node metastasis, the node response was recorded and classified as no response, decrease in size or number or metastatic nodes, or complete radiological node resolution. For the analysis, patients without a pCR in the breast and/or lymph nodes were compared with those who had a pCR.
Statistical analysis
The data were analysed for factors associated with residual disease after NACT. Independent-sample t tests were performed for comparison of means, and univariable analysis with χ2 and Fisher's exact tests was used to compare differences in percentages between groups. Two-sided P < 0·050 was considered significant. All variables with P < 0·100 in the univariable analysis were included in a binary logistic regression model for multivariable analysis. Statistical analysis was undertaken using SPSS® software version 24.0 (IBM, Armonk, New York, USA).
Results
The baseline characteristics of the 328 patients in the study population are shown in Table 2. The mean age was 51 (median 52, range 26–78) years. The predominant tumour type was invasive ductal carcinoma (94·5 per cent), and the majority of patients had clinical T2 disease (86·0 per cent). NACT comprised a standard anthracycline- and taxane-based regimen in 311 patients, and a carboplatin-based regimen in 17. Some 121 patients (36·9 per cent) had a pCR (no residual invasive disease or in situ disease, ypT0), and 30 (9·1 per cent) a pCR with residual DCIS only (ypTis) after NACT (Table 2).
Table 2.
Baseline characteristics, imaging response and pathological outcome after neoadjuvant chemotherapy of 328 patients with triple-negative breast cancer
| No. of patients (n = 328) | |
|---|---|
| Age (years) | |
| ≤ 40 | 72 (22·0) |
| > 40 | 256 (78·0) |
| Ethnicity | |
| Black | 61 (18·6) |
| White | 188 (57·3) |
| Hispanic | 52 (15·9) |
| Other | 27 (8·2) |
| Imaging by mammography and ultrasonography | |
| Mass/architectural distortion | 245 (74·7) |
| Malignant microcalcifications ± mass | 83 (25·3) |
| Tumour histological type | |
| Invasive ductal | 310 (94·5) |
| Other | 18 (5·5) |
| Nuclear grade | |
| 2 | 39 (11·9) |
| 3 | 289 (88·1) |
| DCIS (with invasive disease) seen on initial CNB | |
| No | 250 (76·2) |
| Yes | 78 (23·8) |
| Clinical T category | |
| T1 | 46 (14·0) |
| T2 | 282 (86·0) |
| Clinical/initial node status (cytology/histology) | |
| N0 | 208 (63·4) |
| N1, biopsy-proven | 120 (36·6) |
| Ki-67 (%)* | |
| ≤ 35 | 29 (13·7) |
| > 35 | 182 (86·3) |
| Necrosis | |
| No | 248 (75·6) |
| Yes | 80 (24·4) |
| Lymphovascular invasion | |
| No | 310 (94·5) |
| Yes | 18 (5·5) |
| Radiological response in breast after NACT† | |
| Complete response | 39 (12·5) |
| Partial response | 251 (80·7) |
| Stable disease/progressive disease | 21 (6·8) |
| Radiological lymph node response‡ | |
| No response | 7 (6·4) |
| Decrease in size or number | 75 (68·2) |
| Documented complete resolution | 28 (25·5) |
| Radiological combined breast and lymph node response† | |
| Complete response | 32 (10·3) |
| No complete response | 279 (89·7) |
| Pathological response in breast after NACT | |
| Residual invasive disease + DCIS | 86 (26·2) |
| Residual invasive only | 91 (27·7) |
| Residual DCIS only (ypTis) | 30 (9·1) |
| No residual invasive or in situ disease (ypT0) | 121 (36·9) |
| Pathological lymph node status after NACT | |
| Residual node disease | 80 (24·4) |
| No residual node disease (ypN0) | 248 (75·6) |
| Pathological status in breast and lymph nodes after NACT | |
| Residual disease present | 214 (65·2) |
| No residual disease (ypT0 ypN0) | 114 (34·8) |
Values in parentheses are percentages.
Ki-67 measurement was not performed in 117 patients (35·7 per cent).
Radiological breast response was not evaluated in 17 patients (5·2 per cent).
Among 110 patients with initial N1 biopsy-proven disease after neoadjuvant chemotherapy (NACT); radiological node response was not evaluated in ten patients (3·0 per cent). DCIS, ductal carcinoma in situ; CNB, core needle biopsy.
Residual disease in the breast after neoadjuvant chemotherapy
Patients with residual breast disease were more likely to present with malignant-appearing microcalcifications (P = 0·034), histological grade 2 disease (P = 0·033), T2 tumour (P = 0·022) and low Ki-67 (35 per cent or less) (P = 0·039) (Table 3). There was a trend towards statistical significance for tumour type other than invasive ductal cancer (P = 0·080), associated DCIS on CNB (P = 0·081) and N1 status (P = 0·097). On multivariable analysis, low Ki-67 (odds ratio (OR) 3·14, 95 per cent c.i. 1·19 to 8·30; P = 0·021) and large tumour size (OR 2·58, 1·20 to 5·58; P = 0·016) were independently associated with residual disease in the breast (Table 4).
Table 3.
Univariable analyses of clinicopathological variables associated with residual disease in the breast, lymph nodes, and both after neoadjuvant chemotherapy in 328 patients with T1/T2 triple-negative breast cancer
| Residual breast or in situ disease | Residual node disease | Residual breast or node disease | |||||||
|---|---|---|---|---|---|---|---|---|---|
| No (ypT0) | Yes | No (ypN0) | Yes | No (ypT0 ypN0) | Yes | ||||
| (n = 121) | (n = 207) | P † | (n = 248) | (n = 80) | P † | (n =114) | (n = 214) | P † | |
| Age (years) | 0·131 | 0·879 | 0·165 | ||||||
| ≤ 40 | 21 (29) | 51 (71) | 55 (76) | 17 (24) | 20 (28) | 52 (72) | |||
| > 40 | 100 (39·1) | 156 (60·9) | 193 (75·4) | 63 (24·6) | 94 (36·7) | 162 (63·3) | |||
| Mean(s.d.) | 52(11·3) | 51(11·8) | 0·513‡ | 51(11·7) | 51(11·3) | 0·975‡ | 52(11·5) | 51(11·7) | 0·557‡ |
| Ethnicity | 0·309 | 0·009 | 0·235 | ||||||
| Black | 19 (31) | 42 (69) | 38 (62) | 23 (38) | 17 (28) | 44 (72) | |||
| Other | 102 (38·2) | 165 (61·8) | 210 (78·7) | 57 (21·3) | 97 (36·3) | 170 (63·7) | |||
| Imaging by mammography and ultrasonography | 0·034 | 0·658 | 0·023 | ||||||
| Mass/architectural distortion only | 99 (40·4) | 146 (59·6) | 187 (76·3) | 58 (23·7) | 94 (38·4) | 151 (61·6) | |||
| Malignant microcalcifications ± mass | 22 (27) | 61 (73) | 61 (73) | 22 (27) | 20 (24) | 63 (76) | |||
| Tumour histological type | 0·080 | 0·578 | 0·128 | ||||||
| Invasive ductal | 118 (38·1) | 192 (61·9) | 233 (75·2) | 77 (24·8) | 111 (35·8) | 199 (64·2) | |||
| Other | 3 (17) | 15 (83) | 15 (83) | 3 (17) | 3 (17) | 15 (83) | |||
| Nuclear grade | 0·033 | 0·045 | 0·020 | ||||||
| 2 | 8 (21) | 31 (79) | 24 (62) | 15 (38) | 7 (18) | 32 (82) | |||
| 3 | 113 (39·1) | 176 (60·9) | 224 (77·5) | 65 (22·5) | 107 (37·0) | 182 (63·0) | |||
| DCIS (with invasive disease) on initial CNB | 0·081 | 0·290 | 0·030 | ||||||
| No | 99 (39·6) | 151 (60·4) | 193 (77·2) | 57 (22·8) | 95 (38·0) | 155 (62·0) | |||
| Yes | 22 (28) | 56 (72) | 55 (71) | 23 (29) | 19 (24) | 59 (76) | |||
| Clinical T category | 0·022 | 0·464 | 0·021 | ||||||
| T1 | 24 (52) | 22 (48) | 37 (80) | 9 (20) | 23 (50) | 23 (50) | |||
| T2 | 97 (34·4) | 185 (65·6) | 211 (74·8) | 71 (25·2) | 91 (32·3) | 191 (67·7) | |||
| Clinical/initial node status (cytology/histology) | 0·097 | 0·001 | 0·011 | ||||||
| N0 | 84 (40·4) | 124 (59·6) | 193 (92·8) | 15 (7·2) | 83 (39·9) | 125 (60·1) | |||
| N1, biopsy-proven | 37 (30·8) | 83 (69·2) | 55 (45·8) | 65 (54·2) | 31 (25·8) | 89 (74·2) | |||
| Ki-67 (%)* | 0·039 | 0·009 | 0·094 | ||||||
| ≤ 35 | 6 (21) | 23 (79) | 16 (55) | 13 (45) | 6 (21) | 23 (79) | |||
| > 35 | 76 (41·8) | 106 (58·2) | 144 (79·1) | 38 (20·9) | 70 (38·5) | 112 (61·5) | |||
| Necrosis | 0·235 | 0·231 | 0·138 | ||||||
| No | 96 (38·7) | 152 (61·3) | 192 (77·4) | 56 (22·6) | 92 (37·1) | 156 (62·9) | |||
| Yes | 25 (31) | 55 (69) | 56 (70) | 24 (30) | 22 (27) | 58 (73) | |||
| Lymphovascular invasion | 0·218 | 0·003 | 0·315 | ||||||
| No | 117 (37·7) | 193 (62·3) | 240 (77·4) | 70 (22·6) | 110 (35·5) | 200 (64·5) | |||
| Yes | 4 (22) | 14 (78) | 8 (44) | 10 (56) | 4 (22) | 14 (78) | |||
Values in parentheses are percentages.
Ki-67 measurement was not performed in 117 patients (35·7 per cent).
χ2 test, except ‡independent-samples t test.
Table 4.
Multivariable logistic regression analysis of baseline clinicopathological variables associated with residual disease after neoadjuvant chemotherapy for triple-negative breast cancer
| Odds ratio | P | |
|---|---|---|
| Residual disease in the breast (no ypT0) | ||
| Ki-67 (%) | 0·021 | |
| ≤ 35 | 3·14 (1·19, 8·30) | |
| > 35 | 1·00 (reference) | |
| Clinical T category | 0·016 | |
| T1 | 1·00 (reference) | |
| T2 | 2·58 (1·20, 5·58) | |
| Residual carcinoma in lymph nodes (no ypN0) | ||
| Clinical/initial node status | 0·001 | |
| N0 | 1·00 (reference) | |
| N1, biopsy-proven | 11·89 (5·32, 26·58) | |
| Histological grade | 0·067 | |
| 2 | 2·84 (0·93, 8·67) | |
| 3 | 1·00 (reference) | |
| Lymphovascular invasion | 0·016 | |
| No | 1·00 (reference) | |
| Yes | 5·63 (1·37, 23·14) | |
| Ethnicity | 0·016 | |
| Black | 3·03 (1·23, 7·48) | |
| Other | 1·00 (reference) | |
| Residual disease in the breast and lymph nodes (no ypT0 N0) | ||
| DCIS on CNB | 0·022 | |
| No | 1·00 (reference) | |
| Yes | 2·46 (1·14, 5·31) | |
| Ki-67 (%) | 0·083 | |
| ≤ 35 | 2·41 (0·89, 6·51) | |
| > 35 | 1·00 (reference) | |
| Clinical T category | 0·029 | |
| T1 | 1·00 (reference) | |
| T2 | 2·40 (1·09, 5·27) | |
| Clinical/initial node status | 0·030 | |
| N0 | 1·00 (reference) | |
| N1, biopsy-proven | 2·03 (1·07, 3·84) |
Values in parentheses are 95 per cent confidence intervals. DCIS, ductal carcinoma in situ; CNB, core needle biopsy.
Residual disease in lymph nodes after neoadjuvant chemotherapy
Residual nodal disease was associated with ethnicity (black) (P = 0·009), low histological grade (P = 0·045), N1 status (P = 0·001), low Ki-67 (P = 0·009) and lymphovascular invasion (P = 0·003) (Table 3). Black ethnicity (OR 3·03, 95 per cent c.i. 1·23 to 7·48; P = 0·016), N1 status (OR 11·89, 5·32 to 26·58; P = 0·001), histological grade 2 disease (OR 2·84, 0·93 to 8·67; P = 0·067) and lymphovascular invasion (OR 5·63, 1·37 to 23·14; P = 0·016) were independently associated with residual nodal disease on multivariable analysis (Table 4).
Overall combined residual breast and nodal disease after neoadjuvant chemotherapy
Residual breast and/or lymph node disease was associated with malignant microcalcifications (P = 0·023), histological grade 2 disease (P = 0·020), associated DCIS on the initial CNB (P = 0·030), T2 tumour (P = 0·021) and N1 status (P = 0·011). There was a trend towards statistical significance for low Ki-67 (P = 0·094) (Table 3). Except for the presence of malignant-appearing microcalcifications on baseline imaging and histological grade, all other variables remained statistically significant in multivariable analysis: presence of DCIS on CNB (OR 2·46, 95 per cent c.i. 1·14 to 5·31; P = 0·022), clinical T2 tumour (OR 2·40, 1·09 to 5·27; P = 0·029) and N1 status (OR 2·03, 1·07 to 3·84; P = 0·030). A trend towards significance remained for low Ki-67 (OR 2·41, 0·89 to 6·51; P = 0·083) (Table 4).
Integrating imaging response after neoadjuvant chemotherapy and predicting pathological response
The final overall imaging response after NACT was evaluated and integrated as a means of providing information to clinicians and patients on the likelihood of finding residual disease after NACT. The combined imaging response after NACT showed an accuracy of 71·7 (95 per cent c.i. 66·3 to 76·6) per cent, a false-negative rate of 4·4 (2·0 to 8·2) per cent and a negative predictive value of 76·9 (60·7 to 88·9) per cent for predicting residual breast and lymph node disease (Table 5). A strong correlation was found between stable or progressive disease on imaging and the presence of residual breast or lymph node disease (in all 21 patients in whom imaging revealed stable or progressive disease, this was confirmed on final pathology; P < 0·001). The correlation between a complete or partial response on imaging and the final pathology was less clear. Imaging response information was therefore added as a clinical test to help determine the probability of residual disease in patients with a partial or complete radiological response. In multivariable analysis, not having a complete radiological response (OR 5·07, 95 per cent c.i. 1·88 to 13·67; P = 0·001), N1 status (OR 3·26, 1·59 to 6·66; P = 0·001) and DCIS on initial CNB (OR 2·46, 1·07 to 5·68; P = 0·035) were significantly associated with residual breast and lymph node disease (Table 6).
Table 5.
Performance of combined imaging in predicting residual breast and lymph node disease after neoadjuvant chemotherapy for triple-negative breast cancer
| Breast | Lymph nodes | Breast and lymph nodes | |
|---|---|---|---|
| Accuracy (%) | 70·4 (65·0, 75·4) | 33·1 (27·9, 38·6) | 71·7 (66·3, 76·6) |
| Sensitivity (%) | 96·0 (92·2, 98·2) | 94·5 (86·6, 98·5) | 95·6 (91·8, 98·0) |
| Specificity (%) | 26·7 (18·9, 35·7) | 14·5 (10·3, 19·6) | 27·3 (19·2, 36·6) |
| False-negative rate | 4·0 (1·8, 7·8) | 5·5 (1·5, 13·4) | 4·4 (2·0, 8·2) |
| Positive predictive value (%) | 69·1 (63·3, 74·5) | 25·1 (20·1, 30·6) | 70·9 (65·2, 76·2) |
| Negative predictive value (%) | 79·5 (63·5, 90·7) | 89·7 (75·8, 97·1) | 76·9 (60·7, 88·9) |
Values in parentheses are 95 per cent confidence intervals. Based on combined imaging (mammography, ultrasound imaging and MRI) in 311 of 328 patients (94·8 per cent) defined as complete radiological response or no complete radiological response. Only 24 patients (7·3 per cent) underwent a response evaluation by MRI.
Table 6.
Multivariable logistic regression analysis of the association between baseline initial clinicopathological variables, combined with final additional imaging response information, and residual disease after neoadjuvant chemotherapy for triple-negative breast cancer
| Odds ratio | P | |
|---|---|---|
| Radiological response in breast | 0·001 | |
| Complete response | 1·00 (reference) | |
| No complete response | 5·07 (1·88, 13·67) | |
| Histological type | 0·127 | |
| Invasive ductal carcinoma | 3·49 (0·70, 17·34) | |
| Other | 1·00 (reference) | |
| Clinical/initial node status | 0·001 | |
| N0 | 1·00 (reference) | |
| N1, biopsy-proven | 3·26 (1·59, 6·66) | |
| DCIS on CNB | 0·035 | |
| No | 1·00 (reference) | |
| Yes | 2·46 (1·07, 5·68) |
Values in parentheses are 95 per cent confidence intervals. DCIS, ductal carcinoma in situ; CNB, core needle biopsy.
Effect of neoadjuvant chemotherapy on the ductal carcinoma in situ component
The second goal of this study was to evaluate the effect of NACT on the DCIS component of TNBC, as it is important to eradicate both the invasive and in situ components in patients who will not undergo surgery. Patients were significantly more likely to have residual disease in the breast after NACT if DCIS was present on the initial core biopsy. Of 328 patients in the cohort, 78 (23·8 per cent) had a DCIS component on the initial CNB. The pathological outcome of these patients after NACT was evaluated.
NACT had a significant effect on the DCIS component, although DCIS was present on final pathology after NACT in 43 of the 78 patients (55 per cent). Twenty-two of 78 patients (28 per cent) achieved a pCR (no residual invasive cancer or DCIS in the breast). In comparison, of the 250 patients without a DCIS component, 99 (39·6 per cent) had a pCR (no residual invasive cancer or DCIS) after NACT. DCIS on the initial CNB was predictive of the presence of residual DCIS after NACT, independently of the presence of residual invasive cancer (P < 0·001).
Discussion
The identification of initial clinicopathological variables associated with residual disease can help in the selection of patients for clinical trials testing the safety of omitting surgery in patients with TNBC who experience an exceptional response to NACT. In this study, novel variables significantly associated with residual disease included the presence of malignant-appearing microcalcifications on initial mammography and the presence of DCIS on CNB.
There is no clear consensus on the definition of a pCR. Some trials have defined pCR as pathological eradication of the invasive component only, others as the eradication of both the invasive and residual DCIS components, and some have included the axillary nodes22–25. Although there is evidence that the presence of residual DCIS has no effect on survival, it affects local treatment and surgical planning after NACT.26 It is therefore important to identify the variables associated with pCR, defined as having no residual invasive or in situ disease (ypT0), particularly when evaluating the potential for non-operative management of TNBC.
Here, the presence of microcalcifications on initial imaging and DCIS on initial CNB, together with histological grade, clinical T category, clinical node status and low Ki-67, were found to be predictive of residual breast disease in cT1–2N0–1 TNBC. The presence of malignant-appearing microcalcifications on initial imaging lost its association with residual disease on multivariable analysis, probably because of the correlation with DCIS on CNB. T category was also found to be predictive of ypT0/Tis ypN0 in patients with TNBC by Li and colleagues7. The predictive value of clinical node status was not confirmed in other studies that focused on TNBC1–3,7–8; however, Ki-67 was found to be predictive in all2,5,7–8,11 but one study6 that evaluated it as a predictor.
At MD Anderson Center, breast MRI is used less frequently as an imaging tool than at other centres in the USA. Ultrasound examination of the breast and regional lymph nodes, along with mammography, is documented before commencing NACT. This includes examination of the axillary, supraclavicular, infraclavicular and internal mammary chain nodes. Image-guided fine-needle aspiration or core biopsy was performed on all suspicious lymph nodes in this cohort, as standard institutional practice27. Tadros and colleagues28 recently reported that a pCR in the breast was highly correlated with lymph node status after NACT in patients with TNBC and those with HER2-positive disease. Of patients who presented with N1 disease and had a documented pCR in the breast, 89·6 per cent were also found to have a pCR in the axilla; in patients with N1 and residual breast disease, the risk of residual lymph node disease was quite high at 57·5 per cent. This is valuable information for the appropriate selection of patients who may be able to avoid breast and axillary surgery.
On MRI, the in situ component can present as a non-mass enhancement. Of 117 patients with TNBC studied by Park et al.10, 67 had ypT0 and 50 ypTis disease after breast surgery. On mammography, microcalcifications were more common in patients with ypTis. After NACT, the main lesion in patients with ypT0 disease was more likely to present as a mass on MRI, whereas non-mass enhancement was more common in patients with ypTis disease. The sensitivity, specificity and accuracy for the detection of residual DCIS before surgery were 88·0, 38·8 and 59·8 per cent respectively for mammography; 82·0, 40·3 and 58·1 per cent for ultrasonography; and 68·0, 70·1 and 62·9 per cent for MRI. TNBC and radiological complete response on MRI after NACT were significant predictors of ypT0. In all but the HER2-positive subtype, breast MRI was predictive of ypT010. In a study evaluating breast imaging procedures for predicting a pCR, Schaefgen and colleagues29 found that MRI performance was superior in TNBC (negative predictive value 94 per cent, false-negative rate 5 per cent).
In the present study, all patients with stable or progressive disease on combined imaging after NACT had residual disease at final pathology. This has been shown to be associated with a higher risk of recurrence and death. Such patients should be offered novel agents, a principle applied successfully in the landmark CREATE-X trial30. In this trial, capecitabine given on the basis of residual disease was associated with a significant absolute increase of 8·5 per cent in overall survival of patients with TNBC, compared with those assigned not to receive capecitabine. Thus, patients with TNBC could benefit from additional systemic therapy if a biopsy after NACT demonstrates residual disease, but additional systemic therapy may be de-escalated and surgery might be avoided when no residual disease is identified; this should be explored in future trials. In patients with a complete or partial response on imaging after NACT, repeating image-guided CNBs should be integrated to identify patients with residual disease.
Recent results from Kuerer and co-workers14 revealed that the combination of vacuum-assisted core biopsy and fine-needle aspiration biopsy had an accuracy of 98 (95 per cent c.i. 87 to 100) per cent, a false-negative rate of 5 (0 to 24) per cent and a negative predictive value of 95 (75 to 100) per cent in predicting residual breast cancer. In the present study, the use of imaging after NACT had a clinically subpar accuracy of 71·7 (95 per cent c.i. 66·3 to 76·6) per cent, a false-negative rate of 4·4 (2·0 to 8·2) per cent and a negative predictive value of 76·9 (60·7 to 88·9) per cent for finding residual breast and lymph node disease.
Finding residual DCIS alone has been shown in many studies not to influence survival adversely22–24. However, for patients planned for inclusion in trials of no surgery and radiotherapy alone, having residual DCIS would not be advisable as it could serve as a potential nidus for recurrence. Wiechmann et al.31 evaluated the features of breast tumours based on molecular subtypes. An extensive intraductal component was more common in HER2-overexpressing tumours. Peintinger and colleagues32 evaluated factors that might affect the accuracy of the combination of mammography and ultrasonography in predicting residual tumour size. An extensive intraductal component was associated with an overestimation of pCR on univariable, but not multivariable, analysis. Kraus and co-workers6 and Jung et al.4 did not identify the presence of DCIS as a predictor of pCR among patients with TNBC, although this could be due to smaller patient populations (56 and 143 patients respectively). In the present investigation, however, an association was found between the presence of DCIS on CNB and residual DCIS (irrespective of the invasive component) in the final surgical specimen.
As understanding of different breast cancer subtypes and their chemosensitivity profiles evolves, knowledge of both the baseline clinicopathological variables and radiological imaging properties associated with a pCR can help in counselling patients who may be candidates for omission of surgery trials. Subtype-specific prediction models that incorporate the presence or absence of residual DCIS might provide further guidance towards non-surgical breast cancer management in selected patient subgroups. The use of standard breast imaging provides essential information, based on which patients with stable and progressive TNBC might benefit from alternative systemic regimens; however, imaging alone is inadequate for the identification of patients with a clinical and partial radiological response who might be candidates for non-surgical management. The inclusion of image-guided biopsy is necessary after NACT to select patients for trials of non-surgical treatments.
Acknowledgements
The authors thank H. Lin and B. P. Hobbs from the Department of Biostatistics for statistical assistance, and A. Sutton from the Department of Scientific Publications for editorial assistance with this manuscript. This work was supported by the Dutch Cancer Society Clinical KWF Fellowship (R.F.D.v.l.P.), the P. H. and Fay Etta Robinson Distinguished Professorship in Research Endowment (H.M.K.), a Cancer Center Support Grant from the National Institutes of Health (CA16672), and funding from the MD Anderson Clinical Research Funding Award Program (H.M.K.).
Disclosure: The authors declare no conflict of interest.
References
- 1. Gerber B, Loibl S, Eidtmann H, Rezai M, Fasching PA, Tesch Het al.; German Breast Group Investigators . Neoadjuvant bevacizumab and anthracycline–taxane-based chemotherapy in 678 triple-negative primary breast cancers; results from the GeparQuinto study (GBG 44). Ann Oncol 2013; 24: 2978–2984. [DOI] [PubMed] [Google Scholar]
- 2. Humbert O, Riedinger JM, Charon-Barra C, Berriolo-Riedinger A, Desmoulins I, Lorgis Vet al. Identification of biomarkers including 18FDG-PET/CT for early prediction of response to neoadjuvant chemotherapy in triple-negative breast cancer. Clin Cancer Res 2015; 21: 5460–5468. [DOI] [PubMed] [Google Scholar]
- 3. Huober J, von Minckwitz G, Denkert C, Tesch H, Weiss E, Zahm DMet al. Effect of neoadjuvant anthracycline–taxane-based chemotherapy in different biological breast cancer phenotypes: overall results from the GeparTrio study. Breast Cancer Res Treat 2010; 124: 133–140. [DOI] [PubMed] [Google Scholar]
- 4. Jung YY, Hyun CL, Jin MS, Park IA, Chung YR, Shim Bet al. Histomorphological factors predicting the response to neoadjuvant chemotherapy in triple-negative breast cancer. J Breast Cancer 2016; 19: 261–267. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Keam B, Im SA, Lee KH, Han SW, Oh DY, Kim JHet al. Ki-67 can be used for further classification of triple negative breast cancer into two subtypes with different response and prognosis. Breast Cancer Res 2011; 13: R22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Kraus JA, Beriwal S, Dabbs DJ, Ahrendt GM, McGuire KP, Johnson RRet al. Predictors of pathologic complete response after standard neoadjuvant chemotherapy in triple-negative breast carcinoma. Appl Immunohistochem Mol Morphol 2012; 20: 334–339. [DOI] [PubMed] [Google Scholar]
- 7. Li XR, Liu M, Zhang YJ, Wang JD, Zheng YQ, Li Jet al. CK5/6, EGFR, Ki-67, cyclin D1, and nm23-H1 protein expressions as predictors of pathological complete response to neoadjuvant chemotherapy in triple-negative breast cancer patients. Med Oncol 2011; 28(Suppl 1): S129–S134. [DOI] [PubMed] [Google Scholar]
- 8. Masuda H, Masuda N, Kodama Y, Ogawa M, Karita M, Yamamura Jet al. Predictive factors for the effectiveness of neoadjuvant chemotherapy and prognosis in triple-negative breast cancer patients. Cancer Chemother Pharmacol 2011; 67: 911–917. [DOI] [PubMed] [Google Scholar]
- 9. Nogi H, Kobayashi T, Suzuki M, Tabei I, Kawase K, Toriumi Yet al. EGFR as paradoxical predictor of chemosensitivity and outcome among triple-negative breast cancer. Oncol Rep 2009; 21: 413–417. [PubMed] [Google Scholar]
- 10. Park S, Yoon JH, Sohn J, Park HS, Moon HJ, Kim MJet al. Magnetic resonance imaging after completion of neoadjuvant chemotherapy can accurately discriminate between no residual carcinoma and residual ductal carcinoma in situ in patients with triple-negative breast cancer. PLoS One 2016; 11: e0149347. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Tan QX, Qin QH, Yang WP, Mo QG, Wei CY. Prognostic value of Ki67 expression in HR-negative breast cancer before and after neoadjuvant chemotherapy. Int J Clin Exp Pathol 2014; 7: 6862–6870. [PMC free article] [PubMed] [Google Scholar]
- 12. Swisher SK, Vila J, Tucker SL, Bedrosian I, Shaitelman SF, Litton JKet al. Locoregional control according to breast cancer subtype and response to neoadjuvant chemotherapy in breast cancer patients undergoing breast-conserving therapy. Ann Surg Oncol 2016; 23: 749–756. [DOI] [PubMed] [Google Scholar]
- 13. van la Parra RF, Kuerer HM. Selective elimination of breast cancer surgery in exceptional responders: historical perspective and current trials. Breast Cancer Res 2016; 18: 28. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Kuerer HM, Rauch GM, Krishnamurthy S, Adrada BE, Caudle AS, DeSnyder SMet al. A clinical feasibility trial for identification of exceptional responders in whom breast cancer surgery can be eliminated following neoadjuvant systemic therapy. Ann Surg 2017; [Epub ahead of print]. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. De Los Santos JF, Cantor A, Amos KD, Forero A, Golshan M, Horton JKet al. Magnetic resonance imaging as a predictor of pathologic response in patients treated with neoadjuvant systemic treatment for operable breast cancer. Translational Breast Cancer Research Consortium trial 017. Cancer 2013; 119: 1776–1783. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Molyneux R, Rea D, Jafri M, Herring K, Trivedi S, MacKenzie Met al. NOSTRA PRELIM: a non randomised pilot study designed to assess the ability of image guided core biopsies to detect residual disease in patients with early breast cancer who have received neoadjuvant chemotherapy to inform the design of a planned trial. Cancer Res 2017; 77(Suppl): Abstract P5-16-14. [Google Scholar]
- 17. Heil J, Kummel S, Schaefgen B, Paepke S, Thomssen C, Rauch Get al. Diagnosis of pathological complete response to neoadjuvant chemotherapy in breast cancer by minimal invasive biopsy techniques. Br J Cancer 2015; 113: 1565–1570. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Rea D, Tomlins A, Francis A. Time to stop operating on breast cancer patients with pathological complete response? Eur J Surg Oncol 2013; 39: 924–930. [DOI] [PubMed] [Google Scholar]
- 19. Nederlands Trial Register. Vrancken Peeters MJ. Towards Omitting Breast Cancer Surgery in Patients Without Residual Tumour After Upfront Chemotherapy; 2017. http://www.trialregister.nl/trialreg/admin/rctview.asp?TC=6120 [accessed 4 November 2017]. [Google Scholar]
- 20. Heil J, Schaefgen B, Sinn P, Richter H, Harcos A, Gomez Cet al. Can a pathological complete response of breast cancer after neoadjuvant chemotherapy be diagnosed by minimal invasive biopsy? Eur J Cancer 2016; 69: 142–150. [DOI] [PubMed] [Google Scholar]
- 21. Eisenhauer EA, Therasse P, Bogaerts J, Schwartz LH, Sargent D, Ford Ret al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009; 45: 228–247. [DOI] [PubMed] [Google Scholar]
- 22. Mazouni C, Peintinger F, Wan-Kau S, Andre F, Gonzalez-Angulo AM, Symmans WFet al. Residual ductal carcinoma in situ in patients with complete eradication of invasive breast cancer after neoadjuvant chemotherapy does not adversely affect patient outcome. J Clin Oncol 2007; 25: 2650–2655. [DOI] [PubMed] [Google Scholar]
- 23. Cortazar P, Zhang L, Untch M, Mehta K, Costantino JP, Wolmark Net al. Pathological complete response and long-term clinical benefit in breast cancer: the CTNeoBC pooled analysis. Lancet 2014; 384: 164–172. [DOI] [PubMed] [Google Scholar]
- 24. von Minckwitz G, Untch M, Blohmer JU, Costa SD, Eidtmann H, Fasching PAet al. Definition and impact of pathologic complete response on prognosis after neoadjuvant chemotherapy in various intrinsic breast cancer subtypes. J Clin Oncol 2012; 30: 1796–1804. [DOI] [PubMed] [Google Scholar]
- 25. Jones RL, Lakhani SR, Ring AE, Ashley S, Walsh G, Smith IE. Pathological complete response and residual DCIS following neoadjuvant chemotherapy for breast carcinoma. Br J Cancer 2006; 94: 358–362. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Kuerer HM, Newman LA, Smith TL, Ames FC, Hunt KK, Dhingra Ket al. Clinical course of breast cancer patients with complete pathologic primary tumour and axillary lymph node response to doxorubicin-based neoadjuvant chemotherapy. J Clin Oncol 1999; 17: 460–469. [DOI] [PubMed] [Google Scholar]
- 27. Krishnamurthy S, Sneige N, Bedi DG, Edieken BS, Fornage BD, Kuerer HMet al. Role of ultrasound-guided fine-needle aspiration of indeterminate and suspicious axillary lymph nodes in the initial staging of breast carcinoma. Cancer 2002; 95: 982–988. [DOI] [PubMed] [Google Scholar]
- 28. Tadros AB, Yang WT, Krishnamurthy S, Rauch GM, Smith BD, Valero Vet al. Identification of patients with documented pathologic complete response in the breast after neoadjuvant chemotherapy for omission of axillary surgery. JAMA Surg 2017; 152: 665–670. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Schaefgen B, Mati M, Sinn HP, Golatta M, Stieber A, Rauch Get al. Can routine imaging after neoadjuvant chemotherapy in breast cancer predict pathologic complete response? Ann Surg Oncol 2016; 23: 789–795. [DOI] [PubMed] [Google Scholar]
- 30. Masuda N, Lee SJ, Ohtani S, Im YH, Lee ES, Yokota Iet al. Adjuvant capecitabine for breast cancer after preoperative chemotherapy. N Engl J Med 2017; 376: 2147–2159. [DOI] [PubMed] [Google Scholar]
- 31. Wiechmann L, Sampson M, Stempel M, Jacks LM, Patil SM, King Tet al. Presenting features of breast cancer differ by molecular subtype. Ann Surg Oncol 2009; 16: 2705–2710. [DOI] [PubMed] [Google Scholar]
- 32. Peintinger F, Kuerer HM, Anderson K, Boughey JC, Meric-Bernstam F, Singletary SEet al. Accuracy of the combination of mammography and sonography in predicting tumour response in breast cancer patients after neoadjuvant chemotherapy. Ann Surg Oncol 2006; 13: 1443–1449. [DOI] [PubMed] [Google Scholar]