Percentage of ER-/PR+ patients
Among the 10494 included patients, there were 445 ER-/PR+ patients (4.24%), 7129 ER+ patients (67.93%) and ER-/PR-2920 patients (27.83%). The last follow-up time was in November 2021. We reviewed the annual percentage of the ER-/PR+ group during the study period. The percentage of ER-/PR+ patients was the highest in 2014, accounting for 9.78%, and the lowest in 2009, accounting for 1.55% (Fig. 2a).
ESR1 mRNA expression in ER-/PR+ patients
We compared ESR1 mRNA expression levels in ER-/PR+ patients with ER+ patients. We found that the ESR1 mRNA expression level in 14 ER-/PR+ patients was significantly lower than the ESR1 mRNA expression level in ER+ patients(P=0.004) (Fig. 2b), proving ER-/PR+ tumours were not caused by false-negative staining of ER. In addition, the percentage of PR expression cells in ER-/PR+ tumours was significantly lower than that in ER+/PR+ tumours (16.75±17.76% vs 53.73±31.9%, P=1.9E-8). 82.92% of ER-/PR+ tumours expressed PR≤20%, indicating and most of the tumour cells may be ER-/PR-, and the tumours in this group had tremendous intra-tissue heterogeneity(Fig. 2c).
Demographic, clinicopathological features and treatment mode
As shown in Table 1, ER-/PR+ patients were more frequent in individuals 55 years or older (31.2% vs 24.9%) and were likely to be postmenopausal patients (51.7% vs 41.2%) when compared to ER+ patients. They also had higher TNM stage (stage III 34.6% vs 27.5%), a higher proportion of HER2 positive status (43.8% vs 19.8%), higher Ki67 expression (Ki67≥30%) (77.9% vs 50.7%), likely to be CK5/6 positive (45.2% vs 5.8%). More importantly, they had more differentiation grade 3 (78.1% vs 44.7%) tumours compared to ER+ patients, indicating that ER-/PR+ tumours have more aggressive biological characteristics than ER+ patients. Fewer patients of ER-/PR+ received endocrine therapy (63.1% vs 91%) and medical ovarian suppression (2.7% vs 15.6%) compared to ER+ patients, but more patients received chemotherapy (96.6% vs 92.8%).
Compared with ER-/PR- patients, ER-/PR+ patients showed similar age characteristics, educational characteristics, menopausal, HER2, Ki67%, and CK5/6 status but were likely to have stage III disease (34.6% vs 29.1%). However, the proportion of grade 3 was smaller (78.1% vs 83.5%) than ER-/PR- patients. In terms of treatment mode, more ER-/PR+ patients received endocrine therapy than ER-/PR- patients (63.1% vs 7.6%) (Table 1).
Table 1
Characteristics of patients with ER-/PR+, ER+ or ER-/PR- BC.
|
|
ER-/PR+
(N=445)
|
ER+ (N=7129)
|
ER-/PR- (N=2920)
|
ER-/PR+
Vs ER+ P
|
ER-/PR+ Vs ER-/PR-P
|
age
|
<40
|
67 (15.1%)
|
1241 (17.4%)
|
454 (15.5%)
|
0.01
|
0.898
|
|
40-54
|
239 (53.7%)
|
4114 (57.7%)
|
1584 (54.2%)
|
|
|
|
>=54
|
139 (31.2%)
|
1774 (24.9%)
|
882 (30.2%)
|
|
|
Education(years)
|
0-6
|
88 (20.2%)
|
1173 (16.7%)
|
554 (19.2%)
|
0.067
|
0.75
|
|
6-12
|
242 (55.5%)
|
3866 (55%)
|
1654 (57.4%)
|
|
|
|
>=12
|
106 (24.3%)
|
1994 (28.4%)
|
672 (23.3%)
|
|
|
BMI, kg/m2
|
<24
|
303 (68.6%)
|
4568 (64.6%)
|
1856 (64.3%)
|
0.098
|
0.094
|
|
>=24
|
139 (31.4%)
|
2508 (35.4%)
|
1029 (35.7%)
|
|
|
Residence
|
Rural
|
129 (29%)
|
1851 (26%)
|
769 (26.3%)
|
0.181
|
0.262
|
|
Urban
|
316 (71%)
|
5269 (74%)
|
2151 (73.7%)
|
|
|
Menopausal status
|
Pre-
|
215 (48.3%)
|
4192 (58.8%)
|
1462 (50.1%)
|
<.001
|
0.523
|
|
Post-
|
230 (51.7%)
|
2937 (41.2%)
|
1458 (49.9%)
|
|
|
TNM stage
|
0
|
2 (0.4%)
|
92 (1.3%)
|
50 (1.7%)
|
0.005
|
0.024
|
|
I
|
85 (19.1%)
|
1588 (22.3%)
|
639 (21.9%)
|
|
|
|
II
|
204 (45.8%)
|
3487 (48.9%)
|
1381 (47.3%)
|
|
|
|
III
|
154 (34.6%)
|
1962 (27.5%)
|
850 (29.1%)
|
|
|
HER2 status
|
Negative
|
208 (46.7%)
|
4745 (66.6%)
|
1403 (48%)
|
<.001
|
0.863
|
|
Uncertain
|
42 (9.4%)
|
971 (13.6%)
|
262 (9%)
|
|
|
|
Positive
|
195 (43.8%)
|
1413 (19.8%)
|
1255 (43%)
|
|
|
Ki67
|
<30%
|
96 (22.1%)
|
3417 (49.3%)
|
604 (21.5%)
|
<.001
|
0.821
|
|
>=30%
|
339 (77.9%)
|
3521 (50.7%)
|
2211 (78.5%)
|
|
|
CK5/6
|
Negative
|
207 (54.8%)
|
5809 (94.2%)
|
1350 (55.5%)
|
<.001
|
0.822
|
|
Positive
|
171 (45.2%)
|
360 (5.8%)
|
1081 (44.5%)
|
|
|
Grade
|
1-2
|
73 (21.9%)
|
3282 (55.3%)
|
374 (16.5%)
|
<.001
|
0.019
|
|
3
|
261 (78.1%)
|
2656 (44.7%)
|
1895 (83.5%)
|
|
|
Chemotherapy
|
No
|
15 (3.4%)
|
513 (7.2%)
|
97 (3.3%)
|
0.003
|
1
|
|
Yes
|
430 (96.6%)
|
6616 (92.8%)
|
2823 (96.7%)
|
|
|
Radiotherapy
|
No
|
308 (69.2%)
|
4994 (70.1%)
|
2121 (72.6%)
|
0.748
|
0.149
|
|
Yes
|
137 (30.8%)
|
2135 (29.9%)
|
799 (27.4%)
|
|
|
ET
|
No
|
164 (36.9%)
|
642 (9%)
|
2699 (92.4%)
|
<.001
|
<.001
|
|
Yes
|
281 (63.1%)
|
6487 (91%)
|
221 (7.6%)
|
|
|
Surgical castration
|
No
|
445 (100%)
|
7121 (99.9%)
|
2920 (100%)
|
1
|
NA
|
|
Yes
|
0 (0%)
|
8 (0.1%)
|
0 (0%)
|
|
|
Ovarian suppression
|
No
|
433 (97.3%)
|
6015 (84.4%)
|
2884 (98.8%)
|
<.001
|
0.027
|
|
Yes
|
12 (2.7%)
|
1114 (15.6%)
|
36 (1.2%)
|
|
|
Survival analysis of ER-/PR+, ER+ and ER-/PR- patients
The median follow-up time for the prospective cohort was 65.3 months. 455 patients had BCSS endpoints, 618 patients had LRFS endpoints, and 1026 patients had DDFS endpoints. Compared with ER-/PR+ patients, ER+ patients presented more favourable BCSS (hazard ratio(HR) 0.42, 95%CI 0.28-0.62, P=2.20E-05), LRFS (HR0.4, 95%CI 0.29-0.55, P=1.90E-08), and DDFS (HR 0.52, 95%CI 0.4-0.68, P=1.10E-06). However, there were no significant differences in BCSS (HR0.7, 95%CI 0.46-1.1, P=0.09) and DDFS (HR0.97, 95%CI 0.79-1.2, P=0.77) in ER-/PR- group compared with ER-/PR+, but LRFS endpoints risk significantly decreased (HR 0.65, 95%CI 0.46 -0.9, P=0.01) of ER-/PR- group. The multivariate Cox regression analysis adjusted age, education, BMI, residence, menopausal status, TNM stage, HER2 status, radiotherapy, chemotherapy and ET (Fig. 3a-c, table 2).
Table 2
The multivariate Cox regression analysis for survival outcomes among groups
|
|
BCSS
|
|
LRFS
|
|
DDFS
|
|
|
|
HR(95%CI)
|
P
|
HR(95%CI)
|
P
|
HR(95%CI)
|
P
|
All patients
|
|
|
|
|
|
|
|
|
ER-/PR+
|
Reference
|
|
Reference
|
|
Reference
|
|
|
ER+
|
0.42 (0.28-0.62)
|
2.20E-05
|
0.4 (0.29-0.55)
|
1.90E-08
|
0.52 (0.4-0.68)
|
1.10E-06
|
|
ER-/PR-
|
0.7 (0.46-1.1)
|
0.09
|
0.65 (0.46-0.9)
|
0.01
|
0.84 (0.64-1.1)
|
0.24
|
Patients received ET
|
|
|
|
|
|
|
|
|
ER-/PR+
|
Reference
|
|
Reference
|
|
Reference
|
|
|
ER+
|
0.41 (0.24-0.72)
|
0.002
|
0.43 (0.28-0.66)
|
0.0001
|
0.52 (0.38-0.71)
|
5.40E-05
|
|
ER-/PR-
|
1 (0.49-2.2)
|
0.94
|
0.97 (0.54-1.7)
|
0.91
|
0.98 (0.63-1.5)
|
0.92
|
ER-/PR+ patients
|
|
|
|
|
|
|
|
|
no ET
|
Reference
|
|
Reference
|
|
Reference
|
|
|
ET
|
0.43 (0.2-0.95)
|
0.038
|
0.52 (0.28-0.96)
|
0.037
|
1.1 (0.64-1.9)
|
0.74
|
Survival analysis of ER-/PR+, ER+, ER-/PR- patients treated with ET
We found that the ER-/PR+ group still presented a better BCSS (HR0.41, 95%CI 0.24-0.72, P=0.002), LRFS (HR 0.43, 95%CI 0.28-0.66, P=0.0001), and DDFS (HR=0.52, 95%CI 0.38-0.71, P=5.40E-05) after treatment with ET than that of the ER-/PR+ group after treatment with ET. However, there was no difference in BCSS (HR 1, 95%CI 0.49-2.2, P=0.94), LRFS (HR 0.97, 95%CI 0.54-1.7, P=0.91) and DDFS (HR 0.98, 95%CI 0.63-1.5, P=0.92) between ER-/PR+ group and ER-/PR- group after treatment of ET (Fig.3 d-f, table2). Age, education, BMI, residence, menopausal status, TNM stage, HER2 status, radiotherapy, and chemotherapy was adjusted for all the results of multivariate Cox regression analysis.
Survival analysis of ER-/PR+ patients treated with ET or not
After treatment of ET, ER-/PR+ patients showed a more favourable BCSS (0.038HR 0.43, 95%CI 0.2-0.95, P=0.038) and LRFS (HR 0.52, 95%CI 0.28-0.96, P=0.037) compared to those who did not receive ET. However, there was no difference in DDFS between the two groups (HR1.1, 95%CI 0.64-1.9, P=0.74) (Fig.3 g-i, table 2). Age, education, BMI, residence, menopausal status, TNM stage, HER2 status, radiotherapy, and chemotherapy was adjusted for all the results of multivariate Cox regression analysis.
Subgroup analysis
We then performed a subgroup analysis to mine which ER-/PR+ patients’ subgroup would benefit from ET treatment. Because the number of patients was limited, univariate Cox regression analysis was performed first, and menopausal status, HER2 status, PR level (cutoff 10%), and radiotherapy were taken as adjustment variables for the multivariate Cox regression analysis. For BCSS, patients with PR expression ≥10% could significantly benefit from ET (HR 0.19, 95%CI0.075−0.48, P=0.0005), but PR<10% patients could not benefit from ET treatment (HR 0.951, 95%CI 0.29−3.16, P=0.93) (P for interaction = 0.047). In addition, BMI<24, urban residence, non-radiotherapy patients, Ki67≥30%, CK5/6 negative, both HER2-negative and HER2-positive, age ≥40 years old, secondary education level could significantly benefit from ET, but the trend was consistent among subgroups (P for interaction > 0.05). Postmenopausal patients (HR=0.19, 95%CI0.06−0.49, P=0.0009) could benefit from endocrine therapy. Nevertheless, there was a different trend from premenopausal patients, but it did not reach statistical significance (P for interaction=0.062) (Fig. 4a).
For LRFS, subgroups aged 40-54 (HR 0.38 95%CI 0.15−0.97, P=0.043) and >55 (HR0.24, 95%CI 0.087−0.64, P=0.005) could benefit from endocrine therapy. However, the group of age<40 years did not benefit from endocrine therapy (HR 2.11, 95%CI 0.23−19.83, P=0.51) (P for interaction= 0.024). In addition, BMI≥24 and <24, urban residence, postmenopause, non-radiotherapy, grade1-2 differentiation, ki67<30%, CK5/6 negative, HER2 negative or positive, PR>10%, secondary education subgroups could benefit significantly from ET. However, the trend was consistent among all these subgroups (P for interaction > 0.05) (Fig. 4b).
For DDFS, we found that only postmenopausal patients benefited from ET (HR 0.5, 95%CI 0.26−0.96, P=0.036). However, premenopausal patients did not benefit from ET (HR1.96, 95%CI 0.73−5.27, P=0.18) (P for interaction=0.0292). Further, none other subgroups benefited from ET treatment (Fig. 3c).