Abstract
Objective
To explore the prognosis-predictive influence of human epidermal growth factor receptor 2 (HER2)-low status in breast cancer patients after neoadjuvant therapy (NAT).
Methods
Consecutive patients with invasive breast cancer who underwent NAT and surgery from January 2009 to December 2020 at multiple centers were included. A modified CPS+EG scoring system that integrates HER2-low status, CPS+EGHlow was developed. Multiple scoring systems were compared via receiver operating characteristic curves with the area under curve (AUC), the Akaike information criterion, the C-index, and calibration curves.
Results
A total of 2,141 patients were included: 1,074, 640, and 427 patients in the training, internal validation, and external validation groups, respectively. HER2-low patients had a significantly better breast cancer-specific survival (BCSS, P=0.008) and recurrence-free interval (RFI, P=0.030) compared to HER2-zero patients (P=0.038) but inferior outcomes than HER2-amplified ones (BCSS, P=0.002; RFI, P<0.001). The CPS+EGHlow (AUC: 0.846, 0.817, 0.901) could stratify patients according to BCSS in training, internal validation, and external validation group, respectively, overperforming pathological stage (PS) (AUC: 0.746, 0.779, 0.754), CPS+EG (AUC: 0.771, 0.752, 0.748), and Neo-Bioscore (AUC: 0.783, 0.777, 0.786, all P<0.05).
Conclusions
HER2-low status showed a significant prognostic value in breast cancer patients after NAT. The CPS+EGHlow model significantly outperformed PS, CPS+EG, and Neo-Bioscore in clinical outcome prediction, which may guide further therapy targeting HER2-low.
Keywords: Breast cancer, HER2-low, neoadjuvant therapy, prognosis, CPS+EGHlow
Introduction
Breast cancer (BC) is highly heterogeneous and can be classified into several different subtypes according to hormone receptor (HR) and human epidermal growth factor receptor 2 (HER2) status (1,2). In recent decades, plenty of anti-HER2 drugs have been developed for HER2-amplified/overexpressed BC (3) and have brought tremendous improvement in the prognosis of HER2-positive BC (4-6). On the contrary, HER2-negative BCs showed no response to most traditional anti-HER2 drugs (7,8). Recent findings have changed this predicament. As low to moderate levels of HER2 immunohistochemical (IHC) expression can be found in approximately half of BCs, these so-called HER2-low BCs (9) showed good response to novel anti-HER2 antibody-drug conjugates (ADCs) (10-12). Two phase 3 trials, DESTINY-Breast04 and DESTINY-Breast06, have currently evaluated the efficacy of trastuzumab deruxtecan (T-DXd) in advanced HER2-low BCs (13). Besides, HER2-low BC is also considered to have a favorable prognosis compared to HER2-zero BC. In a pooled analysis of four prospective clinical trials including 2,310 early BC patients, Denkert et al. demonstrated that HER2-low patients had significantly longer survival than HER2-zero ones (14). Our previous study also suggested that HER2-low BC patients have a longer recurrence-free interval (RFI) compared to HER2-zero ones (15). HER2-low BC is becoming increasingly important in the new era of precision medication.
As another important part of precision medication of BC, neoadjuvant therapy (NAT) is currently administered to patients with locally advanced BC or early-stage BC having an indication of systemic therapy to increase breast conserving rates and to test the response of the tumor to systemic therapy in vivo (16,17). What’s more, the response to NAT also carries important prognostic information and can guide further adjuvant therapy for HER2-amplified or triple-negative BC patients with residual disease. Apart from the most well-acknowledged pathological complete response (pCR) rate, which indicates favorable long-term outcomes, there are several staging systems to evaluate the prognostic risk categories according to tumor response to NAT. For instance, the clinicopathological staging system incorporating estrogen receptor (ER)-negative disease, the histological grade 3 tumor (CPS+EG scoring system) (18) and its updated version with HER2 status, Neo-Bioscore (19), are two scoring systems that incorporate aspects of tumor biology into the staging system (20). However, with the growing attention to HER2-low in not only its biological characteristics but also its distinct clinical value, it’s necessary to develop an accurate staging system taking HER2-low status into account.
In the current study, we aimed to develop and validate a modified scoring system including HER2-low status, namely CPS+EGHlow, to better stratify the risk of recurrence in BC patients receiving NAT. The prognostic performance of CPS+EGHlow was compared with other current scoring systems to select an optimal model to guide further treatment.
Materials and methods
Study population and design
Consecutive BC patients who received NAT and underwent subsequent surgery between January 2009 and December 2020 were retrospectively included from the Shanghai Jiao Tong University Breast Cancer Database (SJTU-BCDB), a multi-center database including more than 80,000 early BC cases from 20 medical centers in China (Supplementary Table S1). The inclusion criteria were as follows: 1) female; 2) receiving at least 4 cycles of NAT; 3) histologically confirmed invasive BC by core needle biopsy before NAT; and 4) complete pre-treatment clinical and post-NAT surgical histopathological information. Exclusion criteria were as follows: 1) occult BC patients with no lesions in the breast; 2) patients with unknown NAT regimens; 3) patients receiving neoadjuvant endocrine therapy alone; 4) patients receiving no full course of NAT; or 5) patients with missing clinicopathological information or follow-up data (Figure 1, Supplementary Table S2). Patients treated initially in 5 main centers, namely Ruijin Hospital, Jiaxing Maternity and Child Health Hospital, Shanxi Bethune Hospital, Zhejiang Provincial People’s Hospital, and Guangdong Women and Children Hospital, were included as internal group and were further randomly divided into the training group and the internal validation group at a ratio of 7:3. Patients treated in the remaining 15 medical centers were included as the external validation group (Supplementary Table S1).
Figure 1.
Flowchart of internal group (A) and external group (B). SJTU-BCDB, Shanghai Jiao Tong University Breast Cancer Database; HER2, human epidermal growth factor receptor 2; IHC, immunohistochemistry.
Written informed consent was obtained from the patient for publication of this case report and accompanying images, which is available for reasonable request. This study was approved by the independent Ethical Committees of Ruijin Hospital, Shanghai Jiao Tong University School of Medicine ((No. 2020-309). All procedures were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Clinical and pathological assessment
Baseline clinical tumor and nodal status were determined through physical examination or diagnostic ultrasound. The pre-treatment clinical TNM staging was based on the 8th American Joint Committee on Cancer (AJCC) staging manual (21).
Histopathological and immunohistochemistry (IHC) evaluation was accomplished on pre-therapeutic core biopsy samples and post-NAT surgical specimens by at least two independent experienced pathologists in each study center. The criteria for ER and progesterone receptor (PR) IHC evaluation followed the latest American Society of Clinical Oncology/College of American Pathologists (ASCO/CAP) guidelines, in which the positivity of ER/PR was defined as 1% or more invasive tumor cells stained positive by IHC (22). Ki67 assessment followed the standard procedure recommended by the International Ki67 in Breast Cancer Working Group (23).
The algorithms for HER2 testing were performed following the latest ASCO/CAP guidelines (24). HER2 protein expression was first determined by IHC as 0, 1+, 2+, and 3+. An extra fluorescence in situ hybridization (FISH) test was conducted in IHC 2+ samples by using a HER2/CEP17 dual probe to detect HER2 gene amplification. FISH-positive was defined as dual-probe HER2/CEP17 ratio of ≥2.0 with an average HER2 copy number ≥4.0 signals/cell, or dual-probe HER2/CEP17 ratio of <2.0 with an average HER2 copy number ≥6.0 signals/cell. Based on the outcomes of IHC and FISH, HER2 status was classified as HER2-amplified (IHC 3+, or IHC 2+ and FISH-positive), HER2-low (IHC 1+, or IHC 2+ and FISH-negative), and HER2-zero (HER2 IHC 0) (25).
Prognostic models and outcomes
The CPS+EG score was determined for each patient according to the previously published staging system (Table 1) (18). The score was calculated twice, once using 1% or higher as the cutoff for ER positivity and again using 10% or higher as the cutoff. Considering HER2-low status as a prognostic risk factor, which refers to superior prognosis compared to HER2-zero status, while refers to inferior prognosis compared to HER2-amplified ones, we assigned HER2-low as 1 score and HER2-zero as 2 scores in our CPS-EGHlow staging system. The detailed information of CPS-EGHlow staging system is shown in Table 1.
Table 1. Point assignments for CPS+EG, Neo-Bioscore, and CPS+EGHlow staging systems.
| Cancer stage | CPS+EG | CPS+EGHlow | Neo-Bioscore |
| ER, estrogen receptor; HER2, human epidermal growth factor receptor 2. | |||
| Clinical stage | |||
| I | 0 | 0 | 0 |
| IIA | 0 | 0 | 0 |
| IIB | 1 | 1 | 1 |
| IIIA | 1 | 1 | 1 |
| IIIB | 2 | 2 | 2 |
| IIIC | 2 | 2 | 2 |
| Pathologic stage | |||
| 0 | 0 | 0 | 0 |
| I | 0 | 0 | 0 |
| IIA | 1 | 1 | 1 |
| IIB | 1 | 1 | 1 |
| IIIA | 1 | 1 | 1 |
| IIIB | 1 | 1 | 1 |
| IIIC | 2 | 2 | 2 |
| Tumor marker | |||
| ER-negative | 1 | 1 | 1 |
| Grade 3 | 1 | 1 | 1 |
| HER2-zero | − | 2 | 1 |
| HER2-low | − | 1 | 1 |
| HER2-amplified | − | 0 | 0 |
Clinical stage (CS) and pathological stage (PS) were determined according to the 8th American Joint Committee on Cancer (AJCC) guidelines. The CPS+EG score was calculated according to pre-treatment CS, post-NAT, PS, ER status, and grade following previous publications (26). Neo-Bioscore was evaluated as described by Mittendorf et al. (19).
Breast cancer-specific survival (BCSS) was adopted as the primary endpoint of survival analysis in line with the initial development of the CPS+EG score (18). According to the Standardized Definitions for Efficacy End Points (STEEP) criteria (27), BCSS was calculated from the date of diagnosis to the date of death related to BC or censored at the date of last follow-up or death unrelated to BC. RFI and overall survival (OS) were also calculated as alternative survival endpoints, with the former defined as the time from the date of diagnosis to the first proven tumor relapse, death from BC, and the latter as the time from the date of diagnosis to death from any cause, respectively. The last follow-up was completed by June of 2022.
Statistical analysis
Statistical analysis and image construction were performed using IBM SPSS Statistics (Version 25; IBM Corp., NewYork, USA), R packages (Version 3.4.2; https://cran.r-project.org/), and GraphPad Prism (Version 8.0; GraphPad Software, CA, USA). A two-sided P value of <0.05 was considered statistically significant.
The Chi-squared test was used to compare categorical variables across groups. The 5-year BCSS, RFI, and OS were calculated using the Kaplan-Meier method for patient subgroups defined using multiple scoring systems (CS, PS, CPS+EG, Neo-Bioscore, and CPS-EGHlow). Within each scoring system, BCSS, RFI, and OS among subgroups were compared using the log-rank test. The predictive value of disease outcome was compared by receiver operating characteristic (ROC) curves with area under curve (AUC) between multiple scoring systems using the pROC R package (28). A further time-dependent ROC test was conducted to compare the prognostic value of different scoring systems in terms of 5-year BCSS and RFI using the timeROC R package (29).
Fits of the Cox proportional hazards model for these prognostic models were compared using the Akaike information criterion (AIC). AIC determines the best prognostic model from a set of models by selecting the one with the highest likelihood under the constraint of the smallest number of predictors. The lowest AIC value corresponds to the best model. Discrimination was evaluated by using the concordance index (C-index). C-index is the probability that given two randomly selected patients, and the patient with the most pejorative outcome will in fact have the most pejorative predicted outcome. A C-index value between 0.9 and 1.0 corresponds to an excellent discriminative power, while a C-index value of 0.5 corresponds to a worthless test. Its discriminative power is considered poor 0.6−0.7, fair 0.7−0.8, and good 0.8−0.9. C-index can be used for censored data. A bootstrap method with 500 resamples was used to determine confidence intervals.
Results
Patient cohorts and clinicopathologic characteristics
A total of 2,771 and 708 early BC patients receiving NAC and undergoing surgery in Ruijin Hospital affiliated by Shanghai Jiao Tong University School of Medicine between January 2009 to December 2020 were recruited as internal group and external group, respectively, of which 1,714 and 434 patients met the inclusion criteria were enrolled in the current study. Finally, 1,074, 640, and 427 were divided into training group, internal validation group, and external validation group (Figure 1).
Baseline clinical features and pathological characteristics in the training group, internal validation group, and external validation group are shown in Table 2. No significant difference in the baseline clinical and post-treatment pathological characteristics was observed among three groups of patients. In the overall population, the median age was 50 (range, 21−86) years old. Invasive ductal carcinoma (IDC) was diagnosed in 2,032 (94.9%) patients, and grade III tumors were found in 547 (25.5%) patients. HER2-low tumor was found in 995 (46.5%) patients. A total of 121 (5.7%) patients were classified as clinical stage I disease and 497 (23.2%) were PS I.
Table 2. Baseline clinicopathological characteristics.
| Characteristics | n (%) | P | ||
| Training group (N=1,074) |
Internal validation group (N=640) |
External validation group (N=427) |
||
| IDC, invasive ductal carcinoma; ER, estrogen receptor; PR, progesterone receptor; NA, not available; HER2, human epidermal receptor 2; TNBC, triple negative breast cancer; BCS, breast conversing surgery; SLNB, sentinel lymph node biopsy; ALND, axillary lymph node dissection; NAT, neo-adjuvant treatment. *, HER2 status is according to core needle biopsy specimen before NAC. | ||||
| Age (year) | 0.196 | |||
| <50 | 495 (46.1) | 318 (49.7) | 215 (50.4) | |
| ≥50 | 579 (53.9) | 322 (50.3) | 212 (49.6) | |
| Menstruation | 0.360 | |||
| Pre-menopausal | 560 (52.1) | 352 (55.0) | 237 (55.5) | |
| Post-menopausal | 514 (47.9) | 288 (45.0) | 190 (44.5) | |
| Histology | 0.319 | |||
| IDC | 1,027 (95.6) | 603 (94.2) | 402 (94.1) | |
| Non-IDC | 47 (4.4) | 37 (5.8) | 25 (5.9) | |
| Grade | 0.450 | |||
| I | 171 (15.9) | 92 (14.4) | 64 (15.0) | |
| II | 624 (58.1) | 374 (58.4) | 266 (62.3) | |
| III | 279 (26.0) | 174 (27.2) | 97 (22.7) | |
| ER | 0.364 | |||
| Negative | 348 (32.4) | 207 (32.3) | 123 (28.8) | |
| Positive | 726 (67.6) | 433 (67.7) | 304 (71.2) | |
| PR | 0.252 | |||
| Negative | 432 (40.2) | 248 (38.8) | 152 (35.6) | |
| Positive | 642 (59.8) | 392 (61.3) | 275 (64.4) | |
| Ki67 | 0.543 | |||
| <14% | 138 (12.8) | 71 (11.1) | 52 (12.2) | |
| ≥14% | 925 (86.1) | 565 (88.2) | 368 (86.2) | |
| NA | 11 (1.1) | 4 (0.7) | 7 (1.6) | |
| HER2* | 0.754 | |||
| HER2-zero | 223 (20.8) | 123 (19.2) | 87 (20.4) | |
| HER2-low | 496 (46.2) | 293 (45.8) | 206 (48.2) | |
| HER2-amplified | 355 (33.1) | 224 (35.0) | 134 (31.4) | |
| Primary phenotype | 0.550 | |||
| Luminal-A | 75 (7.0) | 30 (4.7) | 26 (6.1) | |
| Luminal-B/HER2- negative |
472 (43.9) | 282 (44.1) | 205 (48.0) | |
| Luminal-B/HER2- positive |
217 (20.2) | 140 (21.9) | 87 (20.4) | |
| HER2-amplified | 138 (12.8) | 84 (13.1) | 47 (11.0) | |
| TNBC | 172 (16.0) | 104 (16.3) | 62 (14.5) | |
| Breast surgery | 0.212 | |||
| Mastectomy | 845 (88.0) | 545 (85.2) | 375 (87.8) | |
| BCS | 129 (12.0) | 95 (14.8) | 52 (12.2) | |
| Axillary surgery | 0.036 | |||
| SLNB | 90 (8.4) | 74 (11.6) | 32 (7.5) | |
| ALND±SLNB | 984 (91.6) | 566 (88.4) | 395 (92.5) | |
| Clinical stage | 0.315 | |||
| I | 59 (5.5) | 38 (5.9) | 24 (5.6) | |
| IIA | 269 (25.0) | 171 (26.7) | 121 (28.3) | |
| IIB | 414 (38.5) | 249 (38.9) | 158 (37.0) | |
| IIIA | 158 (14.7) | 88 (13.8) | 67 (15.7) | |
| IIIB | 54 (5.0) | 42 (6.6) | 14 (3.3) | |
| IIIC | 120 (11.2) | 52 (8.1) | 43 (10.1) | |
| Pathologic stage | 0.207 | |||
| 0 | 242 (22.5) | 157 (24.5) | 98 (23.0) | |
| I | 156 (14.5) | 101 (15.8) | 80 (18.7) | |
| IIA | 247 (23.0) | 167 (26.1) | 90 (21.1) | |
| IIB | 94 (8.8) | 41 (6.4) | 24 (5.6) | |
| IIIA | 183 (17.0) | 97 (15.2) | 72 (16.9) | |
| IIIB | 27 (2.5) | 12 (1.9) | 9 (2.1) | |
| IIIC | 125 (11.6) | 65 (10.2) | 54 (12.6) | |
| Neo-adjuvant anti-HER2 therapy | 0.445 | |||
| Yes | 355 (33.1) | 224 (35.0) | 134 (31.4) | |
| No | 720 (66.9) | 417 (65.0) | 293 (68.6) | |
| NAT cycle | 0.225 | |||
| 4−6 | 645 (60.1) | 403 (63.0) | 275 (64.4) | |
| >6 | 429 (39.9) | 237 (37.0) | 152 (35.6) | |
In terms of NAT regimens, all HER2-amplified patients were treated with anti-HER2 therapy. Moreover, 1,322 (61.7%) patients received 4−6 cycles of NAT, and 818 (38.2%) were treated with more than six cycles. After NAT, mastectomy and axillary lymph node dissection were carried out in 1,765 (82.4%) and 1,945 (90.8%) patients, respectively.
Adjuvant treatment information is listed as Supplementary Table S3, no significant difference was observed between patients treated before and after 2015 in appliance of adjuvant chemotherapy, radiation therapy, anti-HER2 therapy in HER2-amplified patients, and endocrine therapy in HR-positive patients.
Outcomes by HER2 status
At a median follow-up time of 34.43 (range, 7.53−155.60) months, 99 patients died, of which 91 were related to BC, and breast cancer-specific events were observed in 199 patients. Kaplan-Meier curves of the training group patients according to different HER2 statuses are shown in Figure 2. HER2-low patients had a significantly better BCSS compared to HER2-zero patients (P=0.038, Figure 2A), while inferior to HER2-amplified ones (P=0.034). The estimated 5-year BCSS of HER2-zero, HER2-low, and HER2-amplified patients were 83.6%, 89.3%, and 95.5%, respectively. Similarly, HER2-low patients had a significantly higher estimated 5-year RFI compared to HER2-zero patients (RFI, 83.4% vs. 71.1%, P=0.020, Figure 2B). The RFI of the HER2-low patients was significantly inferior to HER2-amplified ones (RFI, 83.4% vs. 90.0%, P=0.043, Figure 2B). Multivariate analysis showed that HER2 is independently associated with BCSS (P=0.011, Supplementary Table S4) and RFI (P=0.001, Supplementary Table S5) after adjusted with ER, PR status, and grade. In detail, HER2-low showed superior BCSS [hazard ratio (HR)=0.55, 95% confidence interval (95% CI), 0.31−0.89, P=0.025] and RFI (HR=0.65, 95% CI, 0.40−0.93, P=0.046) compared to HER2-zero ones, while showed inferior BCSS (HR=1.96, 95% CI, 2.26−8.09, P=0.007) and RFI (HR=2.03, 95% CI, 1.43−5.16, P<0.001) compared to HER2- amplified ones.
Figure 2.
Survival and HER2 status in whole population. (A) BCSS; (B) RFI. HER2, human epidermal growth factor receptor 2; HR, hazard ratio; 95% CI, 95% confidence interval; amp, amplified; BCSS, breast cancer-specific survival; RFI, recurrence-free interval.
Then we evaluated the relationship between clinical outcomes and HER2 status in the whole population. HER2 status is significantly associated with BCSS (P<0.001) and RFI (P<0.001). Similarly, HER2-low cases showed significantly better BCSS (5-year 89.6% vs. 82.2%, P=0.008, Supplementary Figure S1A) and RFI (5-year 80.7% vs. 70.6%, P=0.030, Supplementary Figure S1B) compared to HER2-zero patients, whereas they showed significantly worse BCSS (5-year 89.6% vs. 97.1%, P=0.002, Supplementary Figure S1A) and RFI (5-year 80.7% vs. 90.1%, P<0.001, Supplementary Figure S1B) compared to HER2-amplified ones.
Outcomes comparisons by CS, PS, CPS+EG, Neo-Bioscore, and CPS+EGHlow scoring systems
Demographics of patients stratified by CPS+EG, Neo-Bioscore, and CPS+EGHlow scoring systems are shown in Table 3. CPS+EGHlow was distributed similarly among three groups, while more patients in external validation groups had a CPS+EG score ≤3 compared to training and internal validation group (88.3% vs. 84.7% & 84.4%). Patients with score 7 and 8 of CPS+EGHlow were combined due to the relatively small number of patients. The estimated 5-year BCSS and RFI outcomes by CS, PS, CPS+EG, Neo-Bioscore, and CPS+EGHlow scoring system are summarized in Table 4 and Supplementary Table S6. When the CPS+EGHlow was applied, 1,708 (79.8%) patients shifted from the CPS+EG score, which reflected the percentage of HER2-zero and HER2-low tumors. The CPS+EGHlow stratified patient well in line with their BCSS (adjust HR=2.02, 95% CI, 1.74−2.42, P<0.001, Figure 3A) and RFI (adjust HR=1.89, 95% CI, 1.69−2.12, P<0.001, Figure 3B) after adjusted with significant prognostic clinicopathological features, including age, ER, PR status, and grade, in multivariate Cox regression model. The 5-year BCSS decreased progressively with increasing CPS-EGHlow score, ranging from 100% (score=0) to 22.3% (score=7−8).
Table 3. Data of multiple staging systems.
| Staging systems | n (%) | ||
| Training group (N=1,074) | Internal validation group (N=640) | External validation group (N=427) | |
| CPS+EG score | |||
| 0 | 82 (7.6) | 72 (11.3) | 45 (10.5) |
| 1 | 268 (25.0) | 156 (24.4) | 104 (24.4) |
| 2 | 311 (29.0) | 175 (27.3) | 141 (33.0) |
| 3 | 248 (23.1) | 137 (21.4) | 87 (20.4) |
| 4 | 117 (10.9) | 64 (10.0) | 38 (8.9) |
| 5 | 37 (3.4) | 30 (4.7) | 10 (2.3) |
| 6 | 11 (1.0) | 6 (0.9) | 2 (0.5) |
| CPS+EGHlow score | |||
| 0 | 39 (3.6) | 27 (4.2) | 16 (3.7) |
| 1 | 135 (12.6) | 101 (15.8) | 64 (15.0) |
| 2 | 204 (19.0) | 132 (20.6) | 100 (23.4) |
| 3 | 286 (26.6) | 142 (22.2) | 100 (23.4) |
| 4 | 227 (21.1) | 140 (21.9) | 83 (19.4) |
| 5 | 126 (11.7) | 60 (9.4) | 46 (10.8) |
| 6 | 44 (4.1) | 26 (4.1) | 14 (3.3) |
| 7 | 10 (0.9) | 12 (1.9) | 3 (0.7) |
| 8 | 3 (0.3) | 0 (0) | 1 (0.2) |
| Neo-Bioscore | |||
| 0 | 40 (3.7) | 28 (4.4) | 16 (3.7) |
| 1 | 148 (13.8) | 110 (17.2) | 67 (15.7) |
| 2 | 248 (23.1) | 142 (22.2) | 113 (26.5) |
| 3 | 294 (27.4) | 166 (25.9) | 116 (27.2) |
| 4 | 216 (20.1) | 118 (18.4) | 75 (17.6) |
| 5 | 91 (8.5) | 49 (7.7) | 29 (6.8) |
| 6 | 32 (3.0) | 23 (3.6) | 9 (2.1) |
| 7 | 5 (0.5) | 4 (0.6) | 2 (0.5) |
Table 4. Five-year BCSS outcomes by CS, PS, CPS+EG, Neo-Bioscore, and CPS+EGHlow staging system in whole population*.
| CS | BCSS (95% CI) |
PS | BCSS (95% CI) |
CPS+EG | BCSS (95% CI) |
Neo-Bioscore | BCSS (95% CI) |
CPS+EGHlow | BCSS (95% CI) |
| BCSS, breast cancer-specific survival; CS, clinical stage; PS, pathologic stage; 95% CI, 95% confidence interval. *, 1,891 patients with mature follow-up information were involved in survival analysis. | |||||||||
| I (N=121) | 100 | 0 (N=497) | 97.9 (95.4−100) | 0 (N=199) | 94.1 (83.6−100) | 0 (N=84) | 100 | 0 (N=82) | 100 |
| IIA (N=561) | 96.1 (92.8−99.4) | I (N=337) | 97.9 (95.8−100) | 1 (N=528) | 98.0 (95.7−100) | 1 (N=325) | 96.9 (91.0−100) | 1 (N=300) | 97.7 (90.5−100) |
| IIB (N=821) | 93.3 (90.4−96.3) | IIA (N=504) | 92.3 (88.2−96.7) | 2 (N=627) | 97.0 (94.9−99.1) | 2 (N=503) | 97.7 (95.4−100) | 2 (N=436) | 95.6 (92.4−99.0) |
| IIIA (N=313) | 84.9 (79.3−90.8) | IIB (N=159) | 89.9 (82.7−97.8) | 3 (N=472) | 84.1 (79.4−89.0) | 3 (N=576) | 95.3 (92.8−97.9) | 3 (N=528) | 94.3 (91.0−97.8) |
| IIIB (N=110) | 80.0 (69.1−92.7) | IIIA (N=352) | 87.3 (81.9−93.1) | 4 (N=219) | 79.6 (72.4−87.6) | 4 (N=409) | 84.2 (79.4−89.3) | 4 (N=450) | 88.0 (83.3−93.0) |
| IIIC (N=215) | 83.3 (76.3−91.0) | IIIB (N=48) | 96.3 (89.4−100) | 5 (N=77) | 83.8 (73.4−95.8) | 5 (N=169) | 73.3 (64.5−83.4) | 5 (N=232) | 76.3 (68.6−84.7) |
| IIIC (N=244) | 73.7 (66.0−82.2) | 6 (N=19) | 31.2 (15.6−86.0) | 6 (N=75) | 42.9 (10.3−87.7) | 6−8 (N=113) | 30.4 (9.3−81.1) | ||
Figure 3.
Survival analysis according to CPS+EGHlow by Kaplan-Meier curves in whole population. (A) BCSS; (B) RFI. BCSS, breast cancer-specific survival; RFI, recurrence-free interval; HR, hazard ratio; 95% CI, 95% confidence interval. *, Adjusted with ER, PR, histology grade by multivariate Cox regression model.
Model comparison of CS, PS, CPS+EG, Neo-Bioscore, and CPS+EGHlow scoring systems in outcome prediction
Finally, we wanted to test whether the CPS+EGHlow was more prognostic than other established scoring systems. In terms of outcome prediction, ROCs showed that the CPS+EGHlow (AUC: 0.846, 0.817, 0.901, Figure 4) could stratify patients according to BCSS in all three groups, which demonstrated a significantly better prognostic value compared to CS (all P<0.001), PS (all P<0.001), CPS+EG (all P<0.001), and Neo-Bioscore (all P<0.001) no matter in the training group, the internal validation group, and the external validation group. Similar findings were also observed for RFI (all P<0.001, Supplementary Figure S2).
Figure 4.
BCSS assessed using full AUCs for pre-treatment CS, post-treatment PS, CPS+EG, NBS, and CPS+EGHlow in training (A), internal validation (B), and external validation (C) group. BCSS, breast cancer-specific survival; AUC, area under curve; CS, clinical stage; PS, pathological stage; NBS, Neo-Bioscore.
Next, we compared the value of different scoring systems in predicting 5-year BCSS (Table 5) and RFI (Supplementary Table S7) by time-dependent ROC. The AUCs of CPS+EGHlow on BCSS were 0.815 (0.742−0.888), 0.762 (0.640−0.884), and 0.712 (0.551−0.872) in the training, internal validation, and external validation group. The CPS+EGHlow showed significantly higher AUC compared to CS, PS, CPS+EG, and Neo-Bioscore in all three groups of patients (all P<0.05, except for the comparison with Neo-Bioscore in the internal validation group, P=0.254). However, the other two prognostic staging systems, either CPS+EG (0.715 vs. 0.725) or Neo-Bioscore (0.716 vs. 0.725), showed no constant superiority in prognosis prediction compared to PS in the training group.
Table 5. Five-year BCSS assessed by AUC in multiple staging systems in training, internal validation, and external validation group.
| BCSS | Training group | Internal validation group | External validation group | |||||||||||||
| AUC (95% CI) | Pa | Pb | Pc | AUC (95% CI) | Pa | Pb | Pc | AUC (95% CI) | Pa | Pb | Pc | |||||
| BCSS, breast cancer-specific survival; AUC, area under the curve; CS, clinical stage; PS, pathologic stage; 95% CI, 95% confidence interval; a, Contrast estimate with PS as reference; b, Contrast estimate with CPS+EG as reference; c, Contrast estimate with Neo-Bioscore as reference. | ||||||||||||||||
| CS | 0.701 (0.615−0.788) |
0.650 (0.532−0.769) |
0.583 (0.429−0.737) |
|||||||||||||
| PS | 0.725 (0.647−0.803) |
0.700 (0.578−0.817) |
0.662 (0.517−0.818) |
|||||||||||||
| CPS+EG | 0.715 (0.641−0.789) |
0.999 | 0.637 (0.516−0.758) |
0.003 | 0.636 (0.478−0.795) |
0.788 | ||||||||||
| Neo-Bioscore | 0.716 (0.639−0.794) |
0.999 | 0.999 | 0.730 (0.606−0.801) |
0.630 | 0.031 | 0.678 (0.523−0.833) |
0.896 | 0.010 | |||||||
| CPS+EGHlow | 0.815 (0.742−0.888) |
0.038 | 0.007 | 0.004 | 0.762 (0.640−0.884) |
0.022 | <0.001 | 0.038 | 0.712 (0.551−0.872) |
0.040 | 0.009 | 0.049 | ||||
We also assessed the prognostic performance of CS, PS, CPS+EG, Neo-Bioscore, and CPS+EGHlow by calculating AIC, and discrimination analysis was conducted using the C-index test. In the whole population, CPS+EGHlow was associated with the lowest AIC among the multiple staging systems in the prediction of BCSS (Supplementary Figure S3A) and RFI (Supplementary Figure S3E), corresponding to better prognostic performance. C-index values were good for CPS+EGHlow (0.82, 95% CI, 0.77−0.87, Supplementary Figure S4A) and fair for CPS+EG (0.72, 95% CI, 0.68−0.77) and Neo-Bioscore (0.75, 95% CI, 0.71−0.79) in prediction of BCSS. With regards to the prediction of RFI, C-index values were good for CPS+EGHlow (0.82, 95% CI, 0.79−0.85, Supplementary Figure S4E), fair for CPS+EG (0.78, 95% CI, 0.74−0.82) and Neo-Bioscore (0.79, 95% CI, 0.75−0.83). Constantly, similar trends of AIC and C-index were observed in the training group, the internal validation group, and the external validation group.
Discussion
In the current study, we developed a modified prognostic scoring system, CPS+EGHlow, based on CPS+EG scoring system and integrating HER2-low status into account. HER2-low tumor was considered as a favorable outcome compared to HER2-zero but inferior prognosis to HER2-amplified tumor. We demonstrated that CPS+EGHlow had a significant improvement of AUC for predicting 5-year BCSS compared to PS scoring system, whereas neither CPS+EG nor Neo-Bioscore staging system had a significant better prediction of 5-year BCSS. And we further validated the prognostic value of CPS+EGHlow in an internal group and an external group of patients recruited from other 15 top medical centers all around China. To the best of our knowledge, few existed model added HER2-low status as an important biomarker in a model in predicting prognosis of early BC patients after NAC. Considering the well response to novel anti-HER2 ADCs of HER2-low BC, our CPS+EGHlow would provide more prognosis information in the new era of anti-HER2 ADCs therapy.
HER2-low BC is a newly raised clinical entity because of its significant influence on prognosis and treatment response. Our previous study demonstrated the significant prognostic value of HER2-low status in early BC with (15) and without (30) NAC. In the current study, we found in a larger, multicenter cohort and an independent validation cohort that HER2-low BC patients experienced superior outcomes to HER2-zero ones, regardless of HR status. Our finding is in line with the existed evidence. Horisawa et al. analyzed data from 4,918 Japanese BC patients receiving upfront surgery that HER2-low is associated with better disease-free survival and OS in both ER-positive and ER-negative subgroups (31). Moreover, Denkert et al. also reported that HER2-low BC had a significantly better prognosis than HER2-zero BC, especially in HR-negative patients (14). What’s more, HER2-low BC was considered a distinct clinical entity not only because of its prognostic value but also due to its predictive significance in response to novel therapeutics. The current promising results from DESTINY-Breast04 and DAISY trials have shown that HER2-low BC responded well to T-DXd, indicating HER2-low status as a more important factor for favorable disease outcome in the new ADCs era (13,32).
Hence, we incorporate tripartite HER2 status into the CPS+EG staging system as a modified CPS+EGHlow scoring system to provide more accurate prognostic risk classification. We demonstrated that CPS+EG and Neo-Bioscore staging systems exhibited overall expected outcomes of BCSS and RFI in agreement with other reports (33,34). Furthermore, we demonstrated that CPS+EGHlow had a significant improvement of 5-year BCSS AUC compared to the PS scoring system (0.815 vs. 0.725), whereas neither CPS+EG (0.715 vs. 0.725) nor Neo-Bioscore staging system (0.716 vs. 0.725) had a significantly better prediction of 5-year BCSS in the training group. The same consequence was observed in the internal validation group and the external validation group as well. Comparing CPS+EG, Neo-Bioscore, and CPS+EGHlow altogether, CPS+EGHlow showed consistent significantly superiority to CPS+EG and Neo-Bioscore in 5-year BCSS and RFI in the training, internal validation, and external validation group. The notable improvement of CPS+EGHlow may contribute to the significant prognostic value of HER2-low status. To note, all of our patients were treated without novel ADCs, and the advantage of CPS+EGHlow might be greater once T-DXd could be used routinely in daily practice. Although HER2-low BC’s significant influence on prognosis and treatment response possibly supports it to be a distinct clinical entity, there is still an ongoing debate on whether it is a distinct biological entity with distinct characteristics. In our previous studies, we found that HER2-low patients had significantly higher ER and/or PR positivity rates than HER2-zero patients (15,30), which was consistent with studies conducted by Denkert et al. and Tarantino et al (14,35). And, these results were also supported on the RNA expression level that a higher rate of luminal-like tumors among HER2-low tumors and a higher rate of basal-like tumors among HER2-zero tumors by PAM50-based intrinsic subtyping (36).
To our knowledge, our study incorporated HER2-low status in prognostic risk staging in early-stage BC patients undergoing NAC and we firstly developed the CPS+EGHlow system. Strength of our work comes from the high-quality multicenter data with large amounts of patients. The main limitation of the current study is that the retrospective nature of our study may lead to an unavoidable diagnosis and selection mistakes, and central HER2 expression revision of all cases was not planned. Besides, lacking of the golden standard in diagnosing HER2-low status may limit the usefulness of CPS+EGHlow system in further clinical trials. Another limitation is represented by the heterogeneity of NAC.
Conclusions
We developed a modified CPS+EGHlow prognostic scoring system by integrating CPS+EG scoring system and HER2-low status, which had a higher accuracy performance compared with PS, CPS+EG, and Neo-Bioscore scoring system. With the adventure of targeting HER2-low ADC agents, we may use this CPS-EGHlow scoring system to guide further adjuvant ADC therapy for those HER2-low BC patients.
Figure S1.
Survival and HER2 status in whole population. (A) BCSS; (B) RFI. BCSS, breast cancer-specific survival; RFI, recurrence-free interval; HER2, human epidermal growth factor receptor 2; HR, hazard ratio; 95% CI, 95% confidence interval; amp, amplified.
Figure S2.
RFI assessed using full AUCs for pre-treatment CS, post-treatment PS, CPS+EG, NBS, and CPS+EGHlow in training (A), internal validation (B), and external validation (C) group. RFI, recurrence-free interval; AUC, area under curve; CS, clinical stage; PS, pathological stage; NBS, Neo-Bioscore.
Figure S4.
C-index for CS, PS, CPS+EG, NBS and CPS+EGHlow in whole population, training group, internal validation group, and external validation group. (A−D) BCSS in whole population, training group, internal validation group, and external validation group, respectively; (E−H) RFI in whole population, training group, internal validation group, and external validation group, respectively. CS, clinical stage; PS, pathological stage; NBS, Neo-Bioscore; BCSS, breast cancer-specific survival; RFI, recurrence-free interval.
Table S1. Sample size of all centers.
| Centers | No. of patients | % |
| Internal group | ||
| Ruijin Hospital | 523 | 24.4 |
| Jiaxing Maternity and Child Health Care Hospital | 114 | 5.3 |
| Shanxi Bethune Hospital | 155 | 7.2 |
| Zhejiang Provincial People’s Hospital | 468 | 21.9 |
| Guangdong Women and Children Hospital | 454 | 21.2 |
| External group | ||
| The First People’s Hospital of Foshan | 59 | 2.8 |
| Henan Cancer Hospital | 53 | 2.5 |
| The First Affiliated Hospital of Fujian Medical University | 51 | 2.4 |
| Quanzhou First Hospital | 44 | 2.1 |
| The First Affiliated Hospital of Nanchang University | 40 | 1.9 |
| The First People’s Hospital of Zunyi | 34 | 1.6 |
| International Peace Maternity & Child Health Hospital of China | 29 | 1.4 |
| Ningbo First Hospital | 22 | 1.0 |
| Shaoxing Shangyu People’s Hospital | 17 | 0.8 |
| Lishui People’s Hospital | 17 | 0.8 |
| Suzhou Ninth People’s Hospital | 16 | 0.7 |
| Fujian Fuding Hospital | 14 | 0.7 |
| Taizhou Central Hospital | 12 | 0.6 |
| Wenzhou People’s Hospital | 10 | 0.5 |
| Rui’an People’s Hospital | 9 | 0.4 |
| Total | 2,141 | 100 |
Table S2. Clinicopathological information of excluded patients.
| Characteristics | n (%) | P | |
| Included patients (N=2,141) | Excluded patients (N=1,338) | ||
| IDC, invasive ductal carcinoma; NA, not available; ER, estrogen receptor; PR, progesterone receptor; HER2, human epidermal receptor 2. | |||
| Age (year) | 0.292 | ||
| <50 | 1,028 (48.01) | 667 (49.85) | |
| ≥50 | 1,113 (51.99) | 671 (50.15) | |
| Menstruation | 0.490 | ||
| Pre-menopausal | 1,149 (53.67) | 702 (52.47) | |
| Post-menopausal | 992 (46.33) | 636 (47.53) | |
| Histology | <0.001 | ||
| IDC | 2,032 (94.91) | 1,210 (90.43) | |
| Non-IDC | 109 (5.09) | 128 (9.57) | |
| Grade | <0.001 | ||
| I | 327 (15.27) | 186 (13.90) | |
| II | 1,264 (59.04) | 631 (47.16) | |
| III | 550 (25.69) | 101 (7.55) | |
| NA | 0 (0) | 420 (31.39) | |
| ER | 0.721 | ||
| Negative | 678 (31.67) | 300 (22.42) | |
| Positive | 1,463 (68.33) | 667 (49.85) | |
| NA | 0 (0) | 371 (27.73) | |
| PR | <0.001 | ||
| Negative | 832 (38.86) | 296 (22.12) | |
| Positive | 1,309 (61.14) | 671 (50.15) | |
| NA | 0 (0) | 371 (27.73) | |
| Ki67 | 0.062 | ||
| <14% | 261 (12.19) | 90 (6.73) | |
| ≥14% | 1,858 (86.78) | 815 (60.91) | |
| NA | 22 (1.03) | 433 (32.36) | |
| HER2 | 0.199 | ||
| HER2-zero | 433 (20.22) | 178 (13.30) | |
| HER2-low | 995 (46.47) | 381 (28.48) | |
| HER2-positive | 713 (33.30) | 321 (23.99) | |
| NA | 22 (1.03) | 458 (34.23) | |
Table S3. Treatment information in all patients according to years of first diagnosis of breast cancer.
| Treatments | n (%) | P | ||
| Patients during 2009−2012 (N=119) | Patients during 2013−2016 (N=475) | Patients during 2017−2020 (N=1,533) | ||
| HER2, human epidermal receptor 2; ER, estrogen receptor; PR, progesterone receptor. | ||||
| Adjuvant chemotherapy | 57 (47.9) | 226 (47.6) | 902 (48.7) | 0.821 |
| Adjuvant radiation therapy | 85 (71.4) | 357 (75.2) | 1,410 (76.3) | 0.143 |
| Anti-HER2 therapy in HER2-amplified patients |
10/11 (90.9) | 124/139 (89.2) | 518/563 (92.0) | 0.279 |
| Endocrine therapy in ER/PR-positive patients |
55/67 (82.1) | 240/297 (80.8) | 919/1,099 (83.6) | 0.734 |
Table S4. Univariate and multivariate prognostic analysis on BCSS in training group of patients.
| Characteristics | Univariate | Multivariate | |||||
| OR | 95% CI | P | OR | 95% CI | P | ||
| BCSS, breast cancer-specific survival; IDC, invasive ductal carcinoma; ER, estrogen receptor; PR, progesterone receptor; HER2, human epidermal receptor 2; TNBC, triple negative breast cancer; BCS, breast conversing surgery; SLNB, sentinel lymph node biopsy; ALND, axillary lymph node dissection; OR, odds ratio; 95% CI, 95% confidence interval. | |||||||
| Age (year) | 0.618 | − | |||||
| <50 | 1.00 | ||||||
| ≥50 | 0.86 | 0.49−1.53 | |||||
| Menstruation | 0.916 | − | |||||
| Pre-menopausal | 1.00 | ||||||
| Post-menopausal | 1.03 | 0.58−1.83 | |||||
| Histology | 0.900 | − | |||||
| IDC | 1.00 | ||||||
| Non-IDC | 1.10 | 0.27−4.52 | |||||
| Grade | 0.002 | 0.029 | |||||
| I | 0.21 | 0.11−0.53 | <0.001 | 0.27 | 0.19−0.61 | ||
| II | 0.34 | 0.19−0.62 | <0.001 | 0.43 | 0.24−0.80 | 0.008 | |
| III | 1.00 | 1.00 | |||||
| ER | <0.001 | 0.059 | |||||
| Negative | 1.00 | 1.00 | |||||
| Positive | 0.28 | 0.15−0.50 | 0.47 | 0.22−1.03 | |||
| PR | <0.001 | 0.164 | |||||
| Negative | 1.00 | 1.00 | |||||
| Positive | 0.24 | 0.12−0.48 | 0.52 | 0.21−1.30 | |||
| Ki67 | 0.961 | − | |||||
| <14% | 1.00 | ||||||
| ≥14% | 1.02 | 0.45−2.30 | |||||
| HER2* | 0.032 | 0.011 | |||||
| HER2-zero | 1.00 | 1.00 | |||||
| HER2-low | 0.67 | 0.28−0.98 | 0.038 | 0.55 | 0.31−0.89 | 0.025 | |
| HER2-amplified | 0.32 | 0.14−0.92 | 0.033 | 0.28 | 0.11−0.70 | 0.007 | |
| Primary phenotype | <0.001 | − | |||||
| Luminal-A | 0.02 | 0.01−0.23 | <0.001 | ||||
| Luminal-B/HER2-negative | 0.27 | 0.14−0.51 | <0.001 | ||||
| Luminal-B/HER2-positive | 0.04 | 0.01−0.31 | 0.002 | ||||
| HER2-amplified | 0.35 | 0.14−0.86 | 0.022 | ||||
| TNBC | 1.00 | ||||||
| Breast surgery | 0.204 | − | |||||
| Mastectomy | 1.00 | ||||||
| BCS | 2.14 | 0.67−6.88 | |||||
| Axillary surgery | 0.638 | − | |||||
| SLNB | 1.00 | ||||||
| ALND±SLNB | 1.41 | 0.34−5.86 | |||||
Table S5. Univariate and multivariate prognostic analysis on RFI in training group of patients.
| Characteristics | Univariate | Multivariate | |||||
| OR | 95% CI | P | OR | 95% CI | P | ||
| RFI, recurrence-free interval; IDC, invasive ductal carcinoma; ER, estrogen receptor; PR, progesterone receptor; HER2, human epidermal receptor 2; TNBC, triple negative breast cancer; BCS, breast conversing surgery; SLNB, sentinel lymph node biopsy; ALND, axillary lymph node dissection; OR, odds ratio; 95% CI, 95% confidence interval. | |||||||
| Age (year) | 0.649 | − | |||||
| <50 | 1.00 | ||||||
| ≥50 | 1.10 | 0.73−1.65 | |||||
| Menstruation | 0.420 | − | |||||
| Pre-menopausal | 1.00 | ||||||
| Post-menopausal | 1.18 | 0.79−1.77 | |||||
| Histology | 0.735 | − | |||||
| IDC | 1.00 | ||||||
| Non-IDC | 1.19 | 0.44−3.24 | |||||
| Grade | <0.001 | <0.001 | |||||
| I | 0.03 | 0.01−0.20 | <0.001 | 0.04 | 0.01−0.28 | 0.001 | |
| II | 0.29 | 0.19−0.45 | <0.001 | 0.37 | 0.24−0.58 | <0.001 | |
| III | 1.00 | 1.00 | |||||
| ER | <0.001 | 0.540 | |||||
| Negative | 0.45 | 0.30−0.67 | 1.00 | ||||
| Positive | 1.00 | 0.85 | 0.51−1.42 | ||||
| PR | <0.001 | 0.001 | |||||
| Negative | 1.00 | 1.00 | |||||
| Positive | 0.32 | 0.21−0.50 | 0.45 | 0.28−0.71 | |||
| Ki67 | 0.971 | − | |||||
| <14% | 1.00 | ||||||
| ≥14% | 1.00 | 0.99−1.01 | |||||
| HER2* | 0.003 | 0.001 | |||||
| HER2-zero | 1.00 | 1.00 | |||||
| HER2-low | 0.60 | 0.38−0.95 | 0.031 | 0.65 | 0.40−0.93 | 0.046 | |
| HER2-amplified | 0.37 | 0.21−0.66 | 0.001 | 0.32 | 0.18−0.57 | <0.001 | |
| Primary phenotype | <0.001 | − | |||||
| Luminal-A | 0.07 | 0.01−0.53 | 0.009 | ||||
| Luminal-B/HER2-negative | 0.44 | 0.28−0.69 | <0.001 | ||||
| Luminal-B/HER2-positive | 0.15 | 0.06−0.35 | <0.001 | ||||
| HER2-amplified | 0.47 | 0.25−0.89 | 0.021 | ||||
| TNBC | 1.00 | ||||||
| Breast surgery | 0.128 | − | |||||
| Mastectomy | 1.00 | ||||||
| BCS | 1.75 | 0.61−6.76 | |||||
| Axillary surgery | 0.164 | − | |||||
| SLNB | 1.00 | ||||||
| ALND±SLNB | 2.27 | 0.72−7.20 | |||||
Table S6. Five-year RFI outcomes by CS, PS, CPS+EG, Neo-Bioscore, and CPS+EGHlow staging system in whole population*.
| CS | RFI (95% CI) |
PS | RFI (95% CI) |
CPS+EG | RFI (95% CI) |
Neo-Bioscore | RFI (95% CI) |
CPS+EGHlow | RFI (95% CI) |
| RFI, recurrence-free interval; CS, clinical stage; PS, pathologic stage; 95% CI, 95% confidence interval. *, 1,891 patients with mature follow-up information were involved in survival analysis. | |||||||||
| I (N=121) | 97.8 (93.7−100) | 0 (N=497) | 95.5 (92.3−98.7) | 0 (N=199) | 93.5 (85.7−100) | 0 (N=84) | 85.7 (63.3−100) | 0 (N=82) | 85.7 (63.3−100) |
| IIA (N=561) | 88.8 (83.6−94.4) | I (N=337) | 87.7 (79.9−95.0) | 1 (N=528) | 91.3 (85.1−97.8) | 1 (N=325) | 94.0 (88.4−99.9) | 1 (N=300) | 94.8 (91.9−97.9) |
| IIB (N=821) | 84.1 (80.0−88.3) | IIA (N=504) | 86.5 (81.9−91.4) | 2 (N=627) | 92.8 (89.8−96.0) | 2 (N=503) | 92.4 (88.0−97.1) | 2 (N=436) | 93.0 (86.4−100) |
| IIIA (N=313) | 79.1 (73.0−85.6) | IIB (N=159) | 83.3 (74.8−92.7) | 3 (N=472) | 72.0 (66.2−78.3) | 3 (N=576) | 91.0 (87.6−94.6) | 3 (N=528) | 90.2 (86.4−94.1) |
| IIIB (N=110) | 68.3 (56.2−83.1) | IIIA (N=352) | 75.3 (68.2−83.1) | 4 (N=219) | 68.3 (60.3−77.2) | 4 (N=409) | 72.0 (66.1−78.4) | 4 (N=450) | 80.5 (75.1−86.3) |
| IIIC (N=215) | 68.5 (59.1−79.3) | IIIB (N=48) | 85.4 (70.1−100) | 5 (N=77) | 49.6 (35.9−68.5) | 5 (N=169) | 56.2 (46.7−67.7) | 5 (N=232) | 60.6 (62.3−70.2) |
| IIIC (N=244) | 55.5 (46.3−65.3) | 6 (N=19) | 31.2 (15.6−86.0) | 6 (N=75) | 35.6 (10.3−87.7) | 6 (N=113) | 29.3 (9.3−62.3) | ||
Table S7. Five-year RFI assessed by AUC in multiple staging systems in training, internal validation, and external validation group.
| RFI | Training group | Internal validation group | External validation group | |||||||||||||
| AUC (95% CI) | Pa | Pb | Pc | AUC (95% CI) | Pa | Pb | Pc | AUC (95% CI) | Pa | Pb | Pc | |||||
| RFI, recurrence-free interval; AUC, area under the curve; CS, clinical stage; PS, pathologic stage. a, Contrast estimate with PS as reference; b, Contrast estimate with CPS+EG as reference; c, Contrast estimate with Neo-Bioscore as reference. | ||||||||||||||||
| CS | 0.629 (0.547−0.711) | 0.646 (0.551−0.742) | 0.563 (0.449−0.676) | |||||||||||||
| PS | 0.709 (0.638−0.779) | 0.711 (0.616−0.807) | 0.671 (0.556−0.786) | |||||||||||||
| CPS+EG | 0.688 (0.607−0.768) | 0.952 | 0.663 (0.566−0.760) | 0.014 | 0.656 (0.547−0.766) | 0.787 | ||||||||||
| Neo-Bioscore | 0.701 (0.622−0.780) | 0.998 | 0.624 | 0.737 (0.644−0.831) | 0.128 | 0.012 | 0.670 (0.561−0.779) | 0.984 | 0.480 | |||||||
| CPS+EGHlow | 0.760 (0.689−0.832) | 0.011 | 0.003 | 0.004 | 0.780 (0.701−0.858) | 0.024 | 0.030 | 0.254 | 0.701 (0.594−0.807) | 0.049 | 0.036 | 0.040 | ||||
SUPPLEMENTARY DATA
Supplementary data to this article can be found online.
Acknowledgements
The study was supported by the National Natural Science Foundation of China (No. 82072937 and 82072897) and Interdisciplinary Program of Shanghai Jiao Tong University (No. YG2024QNB05).
Acknowledgments
Footnote
Conflicts of Interest: The authors have no conflicts of interest to declare.
Contributor Information
Kunwei Shen, Email: kwshen@medmail.com.cn.
Xiaosong Chen, Email: chenxiaosong0156@hotmail.com.
References
- 1.Perou CM, Sørlie T, Eisen MB, et al Molecular portraits of human breast tumours. Nature. 2000;406:747–52. doi: 10.1038/35021093. [DOI] [PubMed] [Google Scholar]
- 2.Goldhirsch A, Winer EP, Coates AS, et al Personalizing the treatment of women with early breast cancer: highlights of the St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer 2013. Ann Oncol. 2013;24:2206–23. doi: 10.1093/annonc/mdt303. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Loibl S, Gianni L HER2-positive breast cancer. Lancet. 2017;389:2415–29. doi: 10.1016/s0140-6736(16)32417-5. [DOI] [PubMed] [Google Scholar]
- 4.Pernas S, Tolaney SM Management of early-stage human epidermal growth factor receptor 2-positive breast cancer. JCO Oncol Pract. 2021;17:320–30. doi: 10.1200/op.21.00020. [DOI] [PubMed] [Google Scholar]
- 5.Tarantino P, Trapani D, Curigliano G Mastering the use of novel anti-HER2 treatment options. JCO Oncol Pract. 2021;17:605–6. doi: 10.1200/op.21.00216. [DOI] [PubMed] [Google Scholar]
- 6.Xu H, Xu B Breast cancer: Epidemiology, risk factors and screening. Chin J Cancer Res. 2023;35:565–83.. doi: 10.21147/j.issn.1000-9604.2023.06.02. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Fehrenbacher L, Cecchini RS, Geyer CE Jr, et al NSABP B-47/NRG oncology phase III randomized trial comparing adjuvant chemotherapy with or without trastuzumab in high-risk invasive breast cancer negative for HER2 by FISH and with IHC 1+ or 2. J Clin Oncol. 2020;38:444–53. doi: 10.1200/jco.19.01455. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Gianni L, Lladó A, Bianchi G, et al Open-label, phase II, multicenter, randomized study of the efficacy and safety of two dose levels of pertuzumab, a human epidermal growth factor receptor 2 dimerization inhibitor, in patients with human epidermal growth factor receptor 2-negative metastatic breast cancer. J Clin Oncol. 2010;28:1131–7. doi: 10.1200/jco.2009.24.1661. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Tarantino P, Hamilton E, Tolaney SM, et al HER2-Low breast cancer: Pathological and clinical landscape. J Clin Oncol. 2020;38:1951–62. doi: 10.1200/jco.19.02488. [DOI] [PubMed] [Google Scholar]
- 10.Banerji U, van Herpen CML, Saura C, et al Trastuzumab duocarmazine in locally advanced and metastatic solid tumours and HER2-expressing breast cancer: a phase 1 dose-escalation and dose-expansion study. Lancet Oncol. 2019;20:1124–35. doi: 10.1016/s1470-2045(19)30328-6. [DOI] [PubMed] [Google Scholar]
- 11.Tamura K, Tsurutani J, Takahashi S, et al Trastuzumab deruxtecan (DS-8201a) in patients with advanced HER2-positive breast cancer previously treated with trastuzumab emtansine: a dose-expansion, phase 1 study. Lancet Oncol. 2019;20:816–26. doi: 10.1016/s1470-2045(19)30097-x. [DOI] [PubMed] [Google Scholar]
- 12.Modi S, Park H, Murthy RK, et al Antitumor activity and safety of trastuzumab deruxtecan in patients with HER2-low-expressing advanced breast cancer: Results from a phase Ib study. J Clin Oncol. 2020;38:1887–96. doi: 10.1200/jco.19.02318. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Modi S, Jacot W, Yamashita T, et al Trastuzumab deruxtecan in previously treated HER2-low advanced breast cancer. N Engl J Med. 2022;387:9–20. doi: 10.1056/NEJMoa2203690. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Denkert C, Seither F, Schneeweiss A, et al Clinical and molecular characteristics of HER2-low-positive breast cancer: pooled analysis of individual patient data from four prospective, neoadjuvant clinical trials. Lancet Oncol. 2021;22:1151–61. doi: 10.1016/s1470-2045(21)00301-6. [DOI] [PubMed] [Google Scholar]
- 15.Zhu S, Lu Y, Fei X, et al Pathological complete response, category change, and prognostic significance of HER2-low breast cancer receiving neoadjuvant treatment: a multicenter analysis of 2489 cases. Br J Cancer. 2023;129:1274–83. doi: 10.1038/s41416-023-02403-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Brandão M, Reyal F, Hamy AS, et al Neoadjuvant treatment for intermediate/high-risk HER2-positive and triple-negative breast cancers: no longer an ‘option’ but an ethical obligation. ESMO Open. 2019;4:e000515. doi: 10.1136/esmoopen-2019-000515. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Laas E, Labrosse J, Hamy AS, et al Determination of breast cancer prognosis after neoadjuvant chemotherapy: comparison of Residual Cancer Burden (RCB) and Neo-Bioscore. Br J Cancer. 2021;124:1421–7. doi: 10.1038/s41416-020-01251-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Jeruss JS, Mittendorf EA, Tucker SL, et al Combined use of clinical and pathologic staging variables to define outcomes for breast cancer patients treated with neoadjuvant therapy. J Clin Oncol. 2008;26:246–52. doi: 10.1200/JCO.2007.11.5352. [DOI] [PubMed] [Google Scholar]
- 19.Mittendorf EA, Vila J, Tucker SL, et al The Neo-Bioscore update for staging breast cancer treated with neoadjuvant chemotherapy: Incorporation of prognostic biologic factors into staging after treatment. JAMA Oncol. 2016;2:929–36. doi: 10.1001/jamaoncol.2015.6478. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Xiang H, Xin L, Ye J, et al A multicenter study on efficacy of dual-target neoadjuvant therapy for HER2-positive breast cancer and a consistent analysis of efficacy evaluation of neoadjuvant therapy by Miller-Payne and RCB pathological evaluation systems (CSBrS-026) Chin J Cancer Res. 2023;35:702–12. doi: 10.21147/j.issn.1000-9604.2023.06.13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Amin MB, Greene FL, Edge SB, et al The Eighth Edition AJCC Cancer Staging Manual: Continuing to build a bridge from a population-based to a more “personalized” approach to cancer staging. CA Cancer J Clin. 2017;67:93–9. doi: 10.3322/caac.21388. [DOI] [PubMed] [Google Scholar]
- 22.Hammond ME, Hayes DF, Dowsett M, et al American Society of Clinical Oncology/College Of American Pathologists guideline recommendations for immunohistochemical testing of estrogen and progesterone receptors in breast cancer. J Clin Oncol. 2010;28:2784–95. doi: 10.1200/jco.2009.25.6529. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Nielsen TO, Leung SCY, Rimm DL, et al Assessment of Ki67 in breast cancer: Updated recommendations from the International Ki67 in Breast Cancer Working Group. J Natl Cancer Inst. 2021;113:808–19. doi: 10.1093/jnci/djaa201. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Wolff AC, Hammond MEH, Allison KH, et al Human epidermal growth factor receptor 2 testing in breast cancer: American Society of Clinical Oncology/College of American Pathologists Clinical Practice Guideline Focused Update. J Clin Oncol. 2018;36:2105–22. doi: 10.1200/jco.2018.77.8738. [DOI] [PubMed] [Google Scholar]
- 25.Tarantino P, Gandini S, Nicolò E, et al Evolution of low HER2 expression between early and advanced-stage breast cancer. Eur J Cancer. 2022;163:35–43. doi: 10.1016/j.ejca.2021.12.022. [DOI] [PubMed] [Google Scholar]
- 26.Mittendorf EA, Jeruss JS, Tucker SL, et al Validation of a novel staging system for disease-specific survival in patients with breast cancer treated with neoadjuvant chemotherapy. J Clin Oncol. 2011;29:1956–62. doi: 10.1200/jco.2010.31.8469. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Tolaney S, Garrett-Mayer E, White J, et al Updated standardized definitions for efficacy end points (STEEP) in adjuvant breast cancer clinical Trials: STEEP version 2. 0. J Clin Oncol. 2021;39:2720–31. doi: 10.1200/jco.20.03613. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Robin X, Turck N, Hainard A, et al pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics. 2011;12:77. doi: 10.1186/1471-2105-12-77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Blanche P, Dartigues JF, Jacqmin-Gadda H Estimating and comparing time-dependent areas under receiver operating characteristic curves for censored event times with competing risks. Stat Med. 2013;32:5381–97. doi: 10.1002/sim.5958. [DOI] [PubMed] [Google Scholar]
- 30.Lu Y, Zhu S, Tong Y, et al HER2-low status is not accurate in breast cancer core needle biopsy samples: An analysis of 5610 consecutive patients. Cancers (Basel) 2022;14:6200. doi: 10.3390/cancers14246200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Horisawa N, Adachi Y, Takatsuka D, et al The frequency of low HER2 expression in breast cancer and a comparison of prognosis between patients with HER2-low and HER2-negative breast cancer by HR status. Breast Cancer. 2022;29:234–41. doi: 10.1007/s12282-021-01303-3. [DOI] [PubMed] [Google Scholar]
- 32.Mosele F, Deluche E, Lusque A, et al Trastuzumab deruxtecan in metastatic breast cancer with variable HER2 expression: the phase 2 DAISY trial. Nat Med. 2023;29:2110–20. doi: 10.1038/s41591-023-02478-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Abdelsattar JM, Al-Hilli Z, Hoskin TL, et al Validation of the CPS + EG staging system for disease-specific survival in breast cancer patients treated with neoadjuvant chemotherapy. Ann Surg Oncol. 2016;23:3206–11. doi: 10.1245/s10434-016-5324-y. [DOI] [PubMed] [Google Scholar]
- 34.Xu L, Liu Y, Fan Z, et al Assessment of CPS + EG, Neo-Bioscore and modified Neo-Bioscore in breast cancer patients treated with preoperative systemic therapy: A multicenter cohort study. Front Oncol. 2021;11:606477. doi: 10.3389/fonc.2021.606477. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Tarantino P, Jin Q, Tayob N, et al Prognostic and biologic significance of ERBB2-low expression in early-stage breast cancer. JAMA Oncol. 2022;8:1177–83. doi: 10.1001/jamaoncol.2022.2286. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Schettini F, Chic N, Brasó-Maristany F, et al Clinical, pathological, and PAM50 gene expression features of HER2-low breast cancer. NPJ Breast Cancer. 2021;7:1. doi: 10.1038/s41523-020-00208-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary data to this article can be found online.