DOI: https://doi.org/10.21203/rs.3.rs-1699310/v1
Objective: To avoid surgical over-treatment of the axilla in patients with lymph node (LN) conversion following neoadjuvant chemotherapy (NAC), high-performing axilla staging procedures are needed. This study is designed to develop a convenient modality to predict the axillary response to NAC in breast cancer patients.
Methods: In this retrospective study, a total of 1046 patients with breast cancer who received NAC followed by axillary lymph node dissection (ALND) between 2015 and 2021 were identified from a maintained database. The training set included 607 breast cancer patients with biopsy proven positive LNs at initial diagnosis, and receiving NAC followed by ALND. Clinicopathologic and ultrasound (US) characteristics were analyzed, and a nomogram was generated to predict the probability of axillary LNs residual metastasis. The predictive performances of models were assessed using multivariate logistic regression and receiver operator characteristic curve (ROC) analyses. The nomogram integrating clinicopathological and US characteristics was validated with an external cohort of 242 patients.
Results: In this study, 49.75% and 32.23% patients achieved axillary pathological complete response (pCR) after NAC in the training and external validation sets, respectively. Multivariate analysis indicated that expression of estrogen receptor (ER), human epidermal growth factor receptor 2 (HER2), Ki-67 score, and clinical nodal stage were independently significant factors for predicting the nodal response to NAC. Location and radiological response of primary tumors, cortical thickness and shape of LNs on US were also significantly associated with nodal pCR. The area under the ROC curve (AUC), estimating the ability of clinicopathologic model to determine axillary status after NAC, was 0.72 and that of US model was 0.81 in the training cohort. AUCs of the nomogram based on clinicopathologic and US characteristics for the training and validation sets were 0.86 and 0.82, respectively.
Conclusions: Nomogram incorporating routine clinicopathologic and US characteristics can predict nodal pCR in node-positive breast cancer patients receiving NAC and may be a feasible modality to aid clinicians in treatment decisions.
Table 1 . Comparison of clinicopathologic and US characteristics between the training and validation cohorts
Characteristics |
SYSUCC n=607 |
FMUCH N=242 |
p-value |
Age |
47.16±10.48 |
49.86±9.52 |
<0.001 |
Menopausal status |
|
|
0.019 |
Pre/perimenopausal |
377 (62.1%) |
171 (70.7%) |
|
Postmenopausal |
230 (37.9%) |
71 (29.3%) |
|
Tumor stage |
|
|
0.074 |
1 |
49 (8.1%) |
21 (8.7%) |
|
2 |
314 (51.7%) |
141 (58.3%) |
|
3 |
142 (23.4%) |
37 (16.3%) |
|
4 |
102 (16.8%) |
43 (17.8%) |
|
Nodal stage |
|
|
<0.001 |
1 |
180 (29.7%) |
160 (66.1%) |
|
2 |
226 (37.2%) |
58 (24.0%) |
|
3 |
201 (33.1%) |
24 (9.9%) |
|
ER status |
|
|
0.708 |
Negative |
205 (33.8%) |
85 (35.1%) |
|
Positive |
402 (66.2%) |
157 (64.9%) |
|
PR status |
|
|
0.572 |
Negative |
248 (40.9%) |
104 (43.0%) |
|
Positive |
359 (59.1%) |
138 (57.0%) |
|
HER2 status |
|
|
0.001 |
Negative |
327 (53.9%) |
162 (66.9%) |
|
Positive |
280 (46.1%) |
80 (33.1%) |
|
Ki-67 |
38.73±21.67 |
42.22±23.53 |
0.041 |
Location of breast tumor |
|
|
0.142 |
No residual disease on the lateral part of breast |
130 (21.4%) |
41 (16.9%) |
|
Residual tumor on the lateral part of breast |
477 (78.6%) |
201 (83.1%) |
|
Number of primary tumor |
|
|
0.054 |
Solitary disease |
497 (81.9%) |
184 (76.0%) |
|
Multifocal disease |
110 (18.1%) |
58 (24.0%) |
|
Tumor diameter before NAC(mm) |
42.43±19.84 |
36.94±18.25 |
<0.001 |
Tumor diameter after NAC(mm) |
23.93±16.26 |
22.37±14.71 |
0.339 |
Radiological response of breast tumor (%)* |
41.32±35.95 |
33.48±50.19 |
0.006 |
Long diameter(mm) |
14.04±6.97 |
12.29±4.69 |
0.002 |
Short diameter(mm) |
6.91±4.11 |
6.30±2.63 |
0.078 |
Ratio of long /short diameter |
2.15±0.75 |
2.03±0.61 |
0.120 |
Cortical thickness |
|
|
0.373 |
≤3mm |
303 (49.9%) |
129 (53.3%) |
|
>3mm |
304 (50.1%) |
113 (46.7%) |
|
Shape |
|
|
0.006 |
Oval |
513 (84.5%) |
188 (77.7%) |
|
Round |
32 (5.3%) |
10 (4.1%) |
|
Irregular |
62 (10.2%) |
44 (18.2%) |
|
Margin |
|
|
<0.001 |
Clear |
591 (97.4%) |
217 (89.7%) |
|
Obscure |
16 (2.6%) |
36 (10.3%) |
|
Fatty hilum |
|
|
0.002 |
Absence |
169 (27.8%) |
43 (17.8%) |
|
Presence |
438 (72.2%) |
199 (82.2%) |
|
Microcalcification |
|
|
0.546 |
Absence |
578 (95.2%) |
228 (94.2%) |
|
Prsence |
29 (4.8%) |
14 (5.8%) |
|
Values are expressed as the mean ± standard deviation or number.
US, ultrasound; SYSUCC, Sun Yat-Sen University Cancer Center; FMUCH, fujian medical university cancer hospital; ER, estrogen receptor; PR, progesterone receptor; HER2, human epidermal growth factor receptor 2; NAC, neoadjuvant chemotherapy.
*Relative change in the largest diameter of primary tumor on US according to RECIST 1.1.
Table 2 . pCR and non-pCR rates of breast or axillary node after NAC
|
SYSUCC |
FMUCH |
||||
|
B-pCR (n=173) |
(n=434) |
p |
B-pCR (n=54) |
(n=188) |
p |
N-pCR |
149 |
153 |
<0.001 |
30 |
48 |
<0.001 |
N-non pCR |
24 |
281 |
|
24 |
140 |
|
NAC, neoadjuvant chemotherapy; pCR, pathological complete response; SYSUCC, Sun Yat-Sen University Cancer Center; FMUCH, fujian medical university cancer hospital.
Table 3 . Univariate logistic regression analysis of characteristics predicting axillary response to NAC in node-positive breast cancer patients (n=607)
Characteristics |
OR |
95%CI |
p-value |
Age |
1.02 |
1.01-1.03 |
0.005 |
Menopausal status |
|
|
|
Pre/perimenopausal |
- |
- |
|
Postmenopausal |
1.08 |
0.81-1.43 |
0.591 |
Tumor stage |
|
|
|
1 |
- |
- |
|
2 |
0.72 |
0.40-1.32 |
0.29 |
3 |
1.28 |
0.67-2.45 |
0.47 |
4 |
1.62 |
0.81-3.21 |
0.17 |
Nodal stage |
|
|
|
1 |
- |
- |
|
2 |
4.02 |
2.62-6.15 |
<0.001 |
3 |
5.11 |
3.29-7.94 |
<0.001 |
ER status |
|
|
|
Negative |
- |
- |
|
Positive |
2.35 |
1.76-3.15 |
<0.001 |
PR status |
|
|
|
Negative |
- |
- |
|
Positive |
2.05 |
1.56-2.71 |
<0.001 |
HER2 status |
|
|
|
Negative |
- |
- |
|
Positive |
0.36 |
0.27-0.47 |
<0.001 |
Ki-67 |
0.99 |
0.98-1.00 |
0.003 |
Location of breast tumor |
|
|
|
No residual disease on the lateral part of breast |
- |
- |
|
Residual tumor on the lateral part of breast |
2.13 |
1.51-2.99 |
<0.001 |
Number of primary tumor |
|
|
|
Solitary tumor |
- |
- |
|
Multifocal disease |
1.45 |
1.03-2.05 |
0.035 |
Tumor diameter before NAC |
1 |
0.99-1.01 |
0.659 |
Tumor diameter after NAC |
1.03 |
1.02-1.04 |
<0.001 |
Radiological response of breast tumor* |
0.26 |
0.16-0.41 |
<0.001 |
Long diameter |
1.02 |
1.00-1.05 |
0.037 |
Short diameter |
1.22 |
1.15-1.3 |
<0.001 |
Ratio of long /short diameter |
0.49 |
0.40-0.60 |
<0.001 |
Cortical thickness |
|
|
|
≤3mm |
- |
- |
|
>3mm |
8.00 |
5.86-10.91 |
<0.001 |
Shape |
|
|
|
Oval |
- |
- |
|
Round |
2.33 |
1.19-4.55 |
0.014 |
Irregular |
11.25 |
5.59-22.63 |
<0.001 |
Margin |
|
|
|
Clear |
- |
- |
|
Obscure |
8.05 |
2.84-22.79 |
<0.001 |
Fatty hilum |
|
|
|
Presence |
- |
- |
|
Absence |
4.58 |
3.15-6.64 |
<0.001 |
Microcalcification |
|
|
|
Absence |
- |
- |
|
Prsence |
6.61 |
2.58-16.97 |
<0.001 |
OR, odds ratio; CI, confidence interval; ER, estrogen receptor; PR, progesterone receptor; HER2, human epidermal growth factor receptor 2; NAC, neoadjuvant chemotherapy.
*Relative change in the largest diameter of primary tumor on US according to RECIST 1.1.
Table 4 . Multivariate logistic regression analysis of characteristics predicting axillary response to NAC in node-positive breast cancer patients (n=607)
Characteristics |
OR |
95%CI |
p-value |
Age |
1 |
0.98-1.03 |
0.769 |
Nodal stage |
|
|
|
1 |
- |
- |
|
2 |
2.04 |
1.37-3.03 |
0.002 |
3 |
2.26 |
1.44-3.54 |
0.001 |
ER status |
|
|
|
Negative |
- |
- |
|
Positive |
2.67 |
1.81-3.93 |
<0.001 |
PR status |
|
|
|
Negative |
- |
- |
|
Positive |
1 |
0.57-1.75 |
0.995 |
HER2 status |
|
|
|
Negative |
- |
- |
|
Positive |
0.32 |
0.22-0.45 |
<0.001 |
Ki-67 |
0.99 |
0.98-1.00 |
0.041 |
Location of breast tumor |
|
|
|
No residual disease on the lateral part of breast |
- |
- |
|
Residual tumor on the lateral part of breast |
1.99 |
1.29-3.05 |
0.013 |
Number of breast tumor |
|
|
|
Solitary tumor |
- |
- |
|
Multifocal disease |
1.16 |
0.64-2.09 |
0.633 |
Tumor diameter after NAC |
1.00 |
0.97-1.02 |
0.716 |
Radiological response of breast tumor * |
0.36 |
0.22-0.61 |
0.017 |
Long diameter |
1.04 |
0.90-1.19 |
0.614 |
Short diameter |
1.05 |
0.80-1.38 |
0.717 |
Ratio of long /short diameter |
0.75 |
0.31-1.79 |
0.518 |
Cortical thickness |
|
|
|
≤3mm |
- |
- |
|
>3mm |
5.93 |
4.11-8.57 |
<0.001 |
Shape |
|
|
|
Oval |
- |
- |
|
Round |
3.09 |
1.15-8.30 |
0.003 |
Irregular |
4.88 |
2.15-11.03 |
<0.001 |
Margin |
|
|
|
Clear |
- |
- |
|
Obscure |
6.51 |
0.74-56.95 |
0.09 |
Fatty hilum |
|
|
|
Presence |
- |
- |
|
Absence |
1.43 |
0.72-2.80 |
0.305 |
Microcalcification |
|
|
|
Absence |
- |
- |
|
Prsence |
1.41 |
0.34-5.84 |
0.637 |
NAC, ; OR, odds ratio; CI, confidence interval; ER, estrogen receptor; PR, progesterone receptor; HER2, human epidermal growth factor receptor 2; NAC, neoadjuvant chemotherapy.
*Relative change in the largest diameter of primary tumor on US according to RECIST 1.1.
Table 5 . Predictive performance of clinicopathologic model, US model and combined model in the training cohort (n=607)
Models |
AUC |
Sensitivity |
Specificity |
Youden index |
Clinicopathologic model |
0.72 |
73.13 |
62.83 |
0.36 |
US model |
0.81 |
74.38 |
78.44 |
0.53 |
Combined model |
0.86 |
85.94 |
76.58 |
0.63 |
AUC, area under the receiver operating characteristic curve; US, ultrasound.
Table 6 . Predictive performance of clinicopathologic model, US model and combined model in the external validation cohort (n=242)
Models |
AUC |
Sensitivity |
Specificity |
Youden index |
Clinicopathologic model |
0.68 |
74.50 |
53.15 |
0.28 |
US model |
0.76 |
68.46 |
74.77 |
0.43 |
Combined model |
0.82 |
81.88 |
62.16 |
0.44 |
AUC, area under the receiver operating characteristic curve; US, ultrasound.