Clinical and pathological characteristics of the BC cohort and quantification of BATMs and BTSMs
The distribution of clinical and pathological parameters in our total cohort and in the BATM/BTSM-high and BATM/BTSM-low subgroup is displayed in Table 1. Not all pathological parameters could be obtained from every patient, which is why the numbers of patients in some subgroups is lower than the number of all cases analyzed. In the overall cohort, 75.8% of all patients were older than 50 years. Most tumors were ER positive (80.9%), PR positive (57.0%), did not show an HER2 amplification (88.3%) and had low proliferation rates (ki-67 ≤ 14%: 56%). 55.7% of all tumors were defined as luminal A-like, 20.1% were luminal B-like and 12.8% were TNBC. 8.1% of all patients had HER2 amplified luminal-like BC and only 2.7% were HER2 amplified non-luminal like. Tumor grading was only available in 54.3% of all cases, so this parameter must be regarded with limited reliability (Table 1).
Table 1
Distribution of BATMs and BTSMs compared to clinical and pathological parameters in our BC cohort
Parameters
|
Total
|
BATMs-low
|
BATMs-high
|
BTSMs-low
|
BTSMs-high
|
Number of cases (%)
|
Number of cases (%)
|
Number of cases (%)
|
Number of cases (%)
|
Number of cases(%)
|
Age
|
≥ 50 years
|
226 (75.8)
|
123(75.0)
|
41(25.0)
|
31(17.2)
|
149(82.3)
|
|
< 50 years
|
72 (24.2)
|
47(83.9)
|
9(16.1)
|
11(16.7)
|
51(77.3)
|
|
Chi square p value
|
|
0.169
|
0.926
|
Molecular subtype
|
Luminal A-like
|
166 (55.7)
|
99(78.0)
|
28(22.0)
|
37(26.8)
|
101(73.2)
|
|
Luminal B-like
|
60 (20.1)
|
39 (79.6)
|
10(20.4)
|
2(3.8)
|
50(96.2)
|
|
Triple negative
|
38 (12.8)
|
16(66.7)
|
8(33.3)
|
0(0.0)
|
30(100.0)
|
|
HER2 amplified luminal-like
|
24 (8.1)
|
11(78.6)
|
3(21.4)
|
1(6.3)
|
15(93.7)
|
|
HER2 amplified non luminal-like
|
8 (2.7)
|
5 (83.3)
|
1(16.7)
|
2(33.3)
|
4(66.7)
|
|
Chi square p value
|
|
0.779
|
0.000011***
|
Grading
|
G1
|
15 (5)
|
7(63.6)
|
4(36.4)
|
4(30.8)
|
9(69.2)
|
|
G2
|
103 (34.6)
|
56(77.8)
|
16(22.2)
|
12(15.0)
|
68(85.0)
|
|
G3
|
44 (14.8)
|
25(78.1)
|
7(21.9)
|
7(20.6)
|
27(79.4)
|
|
Chi square p value
|
|
0.594
|
0.337
|
Tumor focis
|
Unifocal
|
161 (54.0)
|
87(74.4)
|
30(25.6)
|
21(16.0)
|
110(84.0)
|
|
Multifocal/multicentric
|
137 (46.0)
|
83(80.6)
|
20(19.4)
|
21 (18.9)
|
90 (81.1)
|
|
Chi square p value
|
|
0.272
|
0.554
|
Tumor size
|
pT1
|
194 (65.1)
|
114(80.3)
|
28(19.7)
|
31(20.0)
|
124(80.0)
|
|
pT2
|
87 (29.2)
|
49(75.4)
|
16(24.6)
|
8(11.1)
|
64(88.9)
|
|
pT3
|
4 (1.3)
|
1(33.3)
|
2(66.7)
|
0(0.0)
|
3(100.0)
|
|
pT4
|
13 (4.4)
|
6(60.0)
|
4(40.0)
|
3(25.0)
|
9(75.0)
|
|
Chi square p value
|
|
0.094
|
0.276
|
Axillary lymph node status
|
pN0
|
164 (55.0)
|
91(77.8)
|
26(22.2)
|
23(17.6)
|
108(82.4)
|
pN1
|
124 (41.6)
|
73(76.8)
|
22(23.2)
|
18(17.6)
|
84(82.4)
|
|
pN2
|
4 (1.3)
|
1(33.3)
|
2(66.7)
|
0(0.0)
|
4(100.0)
|
|
Chi square p value
|
|
0.224
|
0.796
|
ER status
|
Negative
|
57 (19.1)
|
28(75.7)
|
9(24.3)
|
4(8.9)
|
41(91.1)
|
|
Positive
|
241 (80.9)
|
142(77.6)
|
41(22.4)
|
38(19.3)
|
159(80.7)
|
|
Chi square p value
|
|
0.799
|
0.096
|
PR status
|
Negative
|
128 (43.0)
|
67(75.3)
|
22(24.7)
|
20(25.0)
|
80(75.0)
|
|
Positive
|
170 (57.0)
|
103(78.6)
|
28(21.4)
|
22(15.5)
|
120(84.5)
|
|
Chi square p value
|
|
0.561
|
0.362
|
HER2 status
|
Negative
|
263 (88.3)
|
154(77.4)
|
45(22.6)
|
39(17.8)
|
180(82.2)
|
|
Positive
|
33 (11.1)
|
16(76.2)
|
5(23.8)
|
3(13.0)
|
20(87.0)
|
|
Chi square p value
|
|
1.000
|
0.774
|
Expression of Ki-67
|
≤ 14%
|
167 (56.0)
|
100(78.1)
|
28(21.9)
|
37(26.6)
|
102(73.4)
|
|
> 14%
|
60 (20.1)
|
39(79.6)
|
10(20.4)
|
2(3.8)
|
50(96.2)
|
|
Chi square p value
|
|
0.832
|
0.001**
|
Expression of EP3
|
Low (IRS ≤ 1 )
|
87 (29.2)
|
43 (68.3)
|
20 (31.7)
|
16 (22.2)
|
56 (77.8)
|
|
High (IRS > 1 )
|
201 (67.4)
|
124 (80.5)
|
30 (19.5)
|
26 (15.5)
|
142 (84.5)
|
|
Chi square p value
|
|
0.051
|
0.208
|
BATMs, Breast adipose tissue macrophages; BTSMs, Breast tumor-stroma macrophages; ER, Estrogen receptor; PR, Progesterone receptor; HER2, Human epidermal growth factor receptor 2; EP3, Prostaglandin E receptor 3 .
|
A successful staining of BATMs was achieved in 220/298 patient samples (due to technical issues). 22.7% of these cases showed high populations of BATMs while the remaining 77.3% showed a low amount of BATMs (Table 1). BTSMs could be stained in 242/298 samples. High amounts of BTSMs were detected in 82.7% of these cases and a low quantity of BTSMs in 17.3% (Table 1). The quantities of BATMs and BTSMs correlated strongly with each other (r = 0.5, p = 2.98E-15) (Fig. 1e).
Association of BATMs and BTSMs with clinical and pathological parameters in BC
The distribution of clinical and pathological parameters in the BATMs/BTSMs high and low subgroups is displayed in Table 1; significant associations of BATMs/BTSMs with clinical and pathological parameters in Fig. 1.
The subgroups did not correlate to patient age, however, the total quantity of BATMs was significantly higher in patients older than 50 years compared to younger patients (p = 0.0325, Fig. 1a, 1b, 1f). No further correlations of BATMs and clinical or pathological parameters could be found, neither when analyzing the subgroups, nor when regarding the total quantity of BATMs.
The BTSMs-subgroup correlated strongly to the molecular subtype (p = 0.000011, Table 1). All (100%) TNBC cases belonged to the “BTSMs-high” subgroup. The “BTSMs-high” subgroup was also strongly represented in luminal B-like tumors – in both HER2 negative and HER2 positive cases. Luminal A like tumors showed high BTSMs populations less frequently, which was significantly different to the amount in TNBC in a pairwise comparison (p = 0.0005, Fig. 1 g). The BTSMs-high subgroup was least represented in HER2 positive non-luminal cases, however, as this subgroup contained only 6 cases, this result must be regarded with limited reliability. Not only when grouping the quantity of BTSMs into a high and low population, but also when comparing the absolute quantity of BTSMs between the different molecular BC subtypes, a significant association could be found (p = 0.0003, Fig. 1 g). As the molecular subtypes are defined by surrogate immunohistochemical parameters, these results were consistent when not the subtype but the single parameter was analyzed: high populations of BTSMs occurred more frequently in the ki-67 > 14% group (96.2%) than in cases with an expression of ki-67 ≤ 14% (73.4%, p = 0.001) (Table 1). The absolute quantity of BTSMs in ER negative patients was significantly higher than in ER positive patients (p = 0.002, Fig. 1 h). No correlations between the amount of BTSMs and PR status, HER2 status or other clinical and pathological parameters was found (Table 1).
Both BATMs and BTSMs were negatively associated with OS
Patients in the BATMs-high (HR = 2.483, p = 0.000401, Fig. 2a) as well as in the BTSMs-high (HR = 2.445, p = 0.021, Fig. 2b) subgroup showed a significantly impaired OS compared to the respective “low” subgroup. Median OS was 7.48 years in the BATM-high population (n = 50) while the median was not reached (NR) in our follow up period in the BATM-low (n = 170) population. 75% OS was 6.49 years in the BTSM-high (n = 200) versus 11.64 years in BTSM-low (n = 42) subgroup. Median OS was not reached in the BTSM-high as well as in the BTSM-low subgroup.
In a subgroup analysis, a high amount of BATMs was a negatively associated with OS independent of tumor foci (in unifocal tumors: supplementary Fig. 1a, n = 117, p = 0.006 and in multifocal and multicentric tumors: supplementary Fig. 1b, n = 103, p = 0.025) and independent of PR status (in PR positive tumors: supplementary Fig. 1d, n = 131, p = 0.041 and in PR negative tumors, supplementary Fig. 1e, n = 89, p = 0.005). High amounts of BATMs were also negatively associated with OS in the subgroups of patients aged older than 50 years (supplementary Fig. 1c, n = 164, p = 0.001), in Luminal A-like tumors (supplementary Fig. 1f, n = 127, p = 0.001), in TNBC (supplementary Fig. 1 g, n = 24, p = 0.049), in tumors smaller than 2 cm in size (supplementary Fig. 1 h, n = 142, p = 0.008), in BC patients with negative lymph node status (supplementary Fig. 1i, n = 117, p = 0.00021), in ER positive tumors (supplementary Fig. 1j, n = 183, p = 0.003), in HER2 negative tumors (supplementary Fig. 1k, n = 199, p = 0.001) and in tumors with a low proliferation rate (Ki-67 ≤ 14% ) (supplementary Fig. 1l, n = 128, p = 0.001). No significant correlations between the amount of BATMs and OS were seen in in the respective other subgroups (data not shown).
In the subgroup analysis, a high amount of BTSMs was a negatively associated with OS in the subgroup of patients aged older than 50 years (supplementary Fig. 2a, n = 180, p = 0.029), in Luminal A-like tumors (supplementary Fig. 2b, n = 138, p = 0.046), in multifocal and multicentric tumors (supplementary Fig. 2c, n = 111, p = 0.026), in tumors smaller than 2 cm in size (supplementary Fig. 2d, n = 155, p = 0.031), in BC patients with negative lymph node status (supplementary Fig. 2e, n = 131, p = 0.029), in ER positive tumors (supplementary Fig. 2f, n = 197, p = 0.034), in HER2 negative tumors (supplementary Fig. 2 g, n = 219, p = 0.026) and in tumors with a low proliferation rate (Ki67 ≤ 14% ) (supplementary Fig. 2 h, n = 139, p = 0.039). No significant correlations between the amount of BTSMs and OS were seen in the respective other subgroups (data not shown).
Only BATMs but not BTSMs were negatively associated with DFS
In the overall patient cohort, the BATM-high subgroup showed a significantly impaired DFS compared to patients with low amounts of BATMs (median DFS 5.35 years in the BATM-high (n = 50) vs. 9.53 years in the BATM-low (n = 170) population, HR = 1.800, p = 0.032, Fig. 2c). There was no significant association of the amount of BTSMs with DFS, neither in the overall cohort (HR = 1.285 p = 0.431, Fig. 2d), nor in a subgroup analysis (data not shown).
A subgroup analysis revealed that BATMS were negatively associated with DFS in the subgroups of patients aged older than 50 years (supplementary Fig. 1m, n = 164, p = 0.015) and in patients with unifocal tumors (supplementary Fig. 1n, n = 117, p = 0.016). No significant correlations between the amount of BATMs and DFS were seen in the respective other subgroups (data not shown).
BATMS were independently associated with OS in BC
Univariate COX regression analysis showed that the BATMs subgroup (p = 0.001, HR = 2.483, 95%CI: 1.474–4.182), the BTSMs subgroup (p = 0.025, HR = 2.445, 95%CI: 1.117–5.354), the molecular subtype (p = 0.03, HR = 1.213, 95%CI :1.081–1.444), grading (p = 0.003, HR = 1.763, 95%CI: 1.056–2.945), tumor size (p = 5.57E-13, HR = 2.064, 95%CI: 1.695–2.513), axillary lymph node status (p = 0.002, HR = 1.859, 95%CI: 1.256–2.749) and ER status (p = 0.026, HR = 0.589, 95%CI: 0.369–0.940) were significantly associated with OS (Table 2).
Multivariate analysis was performed in three different models. All models included the univariate significant parameters age, molecular subtype, grading, tumor size and axillary lymph node status. Multivariate analysis model 1 was performed including BTSMs but without BATMs, to determine if BTSMs were independently associated with OS in the whole patient cohort. Multivariate model 2 was performed including BATMs but without BTSMs, to determine if BATMs were independently associated with OS in the whole patient cohort. Multivariate analysis model 3 was performed with both BATMs and BTSMs, to determine if BATMs, BTSMs or both were independently associated with OS when both subtypes of macrophages were considered. As shown in supplementary table 1, the multivariate COX regression analysis model 1 revealed that tumor size (p = 0.001, HR = 1.873, 95%CI: 1.304–2.689) was independently associated with OS, but BTSMs were not. The multivariate COX regression analysis model 2 revealed that BATMs (p = 0.002, HR = 4.259, 95%CI: 1.666–10.887) and tumor size (p = 0.001, HR = 1.847, 95%CI: 3.113-1093.217) were independently associated with OS. The multivariate COX regression analysis model 3 also showed that BATMs (p = 0.004, HR = 4.464, 95%CI: 1.624–12.269) and tumor size (p = 0.001, HR = 1.827, 95%CI: 1.269–2.631) were independently associated with OS. So even when BTSMs were taken into consideration, BATMs remained associated with OS of BC patients (Table 2).
Table 2
Univariate and multivariate Cox regression analyses of OS including various prognostic parameters in patients with BC
|
Univariate analysis
|
Multivariate analysis model 3 (with both BATM and BTSM)
|
|
p
|
HR
|
95% CI
|
p
|
HR
|
95% CI
|
Age (< 50y vs. ≥50y)
|
0.055
|
1.746
|
0.987–3.088
|
0.741
|
0.843
|
0.305–2.326
|
Molecular subtype (LuA-like vs.LuB-like vs. TNBC vs. HER2 Lu-like vs. HER2 nonLu-like)
|
0.03*
|
1.213
|
1.018–1.444
|
0.185
|
1.351
|
0.866–2.107
|
Grading (G1 vs. G2 vs. G3)
|
0.003*
|
1.763
|
1.056–2.945
|
0.405
|
1.403
|
0.633–3.110
|
Tumor foci (unifocal vs. multifocal and muticentric)
|
0.889
|
0.971
|
0.642–1.469
|
n.i.
|
n.i.
|
n.i.
|
Tumor size (pT1 vs. pT2 vs. pT3 vs. pT4)
|
5.573E-13*
|
2.064
|
1.695–2.513
|
0.001*
|
1.827
|
1.269–2.631
|
Axillary lymph node status (pN0 vs. pN1 vs.pN2)
|
0.002*
|
1.859
|
1.256–2.749
|
0.285
|
1.650
|
0.659–4.131
|
ER status (ER- vs. ER+)
|
0.026*
|
0.589
|
0.369–0.940
|
0.604
|
0.718
|
0.205–2.515
|
PR status (PR- vs. PR+)
|
0.088
|
0.697
|
0.461–1.054
|
n.i.
|
n.i.
|
n.i.
|
HER2 status (HER2- vs. HER2+)
|
0.079
|
1.667
|
0.942–2.952
|
n.i.
|
n.i.
|
n.i.
|
Expression of ki-67 (ki-67 ≤ 14% vs. ki-67 > 14%)
|
0.891
|
1.040
|
0.588–1.840
|
n.i.
|
n.i.
|
n.i.
|
BATMs (low vs. high)
|
0.001*
|
2.483
|
1.474–4.182
|
0.004*
|
4.464
|
1.624–12.269
|
BTSMs (low vs. high)
|
0.025*
|
2.445
|
1.117–5.354
|
0.737
|
0.813
|
0.243–2.721
|
Multivariate analysis model 3 was performed with both BATMs and BTSMs, to determine if BATMs, BTSM or both were independently associated with OS when both subtypes of macrophages were considered. ER, Estrogen receptor; PR, Progesterone receptor; HER2, Human Epidermal growth factor receptor 2; LuA-like, Luminal A-like; LuB-like, Luminal B-like; TNBC, Triple negative breast cancer, HER2 Lu-like, HER2 amplified Luminal -like; HER2 nonLu like, HER2 amplified non luminal-like; BATMs, Breast adipose tissue macrophages; BTSMs, Breast tumor-stroma macrophages; HR hazard ratio, CI, confidence interval; n.i., not included in multivariate model as p > 0.05 in univariate analysis; *, significant (p-value < 0.05).
|
BATMs were independently associated with DFS in BC
Univariate COX regression analysis showed that the BATMs subgroup (p = 0.035, HR = 1.800, 95%CI: 1.042–3.108), tumor grade (p = 0.03, HR = 1.669, 95%CI: 1.050–2.654), tumor size (p = 0.002, HR = 1.493, 95%CI: 1.159–1.922) and lymph node status (p = 0.01, HR = 1.696, 95%CI: 1.137–2.528) were independently associated with DFS a in our BC cohort (Table 3). BTSMs was not associated with DFS in the univariate analysis.
The multivariate analysis included the parameters age, grading, tumor size, axillary lymph node status and the BATM subgroup. As shown in Table 3, the multivariate COX regression analysis revealed that the BATM subgroup (p = 0.005, HR = 3.240, 95%CI: 1.423–7.378), grading (p = 0.043, HR = 1.825, 95%CI: 1.018–3.271) and tumor size (p = 0.011, HR = 1.646, 95%CI: 1.120–2.418) were independently associated with DFS in patients with BC (Table 3).
Table 3
Univariate and multivariate Cox regression analyses of DFS including various prognostic parameters in patients with BC
|
Univariate analysis
|
Multivariate analysis
|
|
p
|
HR
|
95% CI
|
p
|
HR
|
95% CI
|
Age(<50y vs. ≥50y)
|
0.19
|
0.732
|
0.459–1.167
|
0.134
|
0.537
|
0.238–1.210
|
Molecular subtype (LuA-like vs.LuB-like vs. TNBC vs. HER2 Lu-like vs. HER2 nonLu-like)
|
0.329
|
1.093
|
0.914–1.307
|
n.i.
|
n.i.
|
n.i.
|
Grading (G1 vs. G2 vs. G3)
|
0.03*
|
1.669
|
1.050–2.654
|
0.043*
|
1.825
|
1.018–3.271
|
Tumor foci (unifocal vs. multifocal and muticentric)
|
0.370
|
1.214
|
0.794–1.857
|
n.i.
|
n.i.
|
n.i.
|
Tumor size (pT1 vs. pT2 vs. pT3 vs. pT4)
|
0.002*
|
1.493
|
1.159–1.922
|
0.011*
|
1.646
|
1.120–2.418
|
Axillary lymph node status (pN0 vs. pN1 vs.pN2)
|
0.01*
|
1.696
|
1.137–2.528
|
0.856
|
1.078
|
0.482–2.411
|
ER status (ER- vs. ER+)
|
0.771
|
0.926
|
0.550–1.557
|
n.i.
|
n.i.
|
n.i.
|
PR status (PR- vs. PR+)
|
0.249
|
1.291
|
0.836–1.994
|
n.i.
|
n.i.
|
n.i.
|
HER2 status (HER2- vs. HER2+)
|
0.511
|
1.228
|
0.666–2.262
|
n.i.
|
n.i.
|
n.i.
|
Expression of ki-67 (ki-67 ≤ 14% vs. ki-67 > 14%)
|
0.093
|
1.569
|
0.928–2.653
|
n.i.
|
n.i.
|
n.i.
|
BATMs (Low vs. High)
|
0.035*
|
1.800
|
1.042–3.108
|
0.005*
|
3.240
|
1.423–7.378
|
BTSMs (Low vs. High)
|
0.432
|
1.285
|
0.687–2.403
|
n.i.
|
n.i.
|
n.i.
|
ER, Estrogen receptor; PR, Progesterone receptor; HER2, Human Epidermal growth factor receptor 2; LuA-like, Luminal A-like; LuB-like, Luminal B-like; TNBC, Triple negative breast cancer, HER2 Lu-like, HER2 amplified Luminal -like; HER2 nonLu like, HER2 amplified non luminal-like; BTAMs, Breast adipose tissue macrophages; BTSMs, Breast tumor-stroma macrophages; HR hazard ratio, CI confidence interval, n.i., not included in multivariate model, as p > 0.05 in univariate analysis, *, significant (p-value < 0.05).
|
BATMs correlated negatively to EP3 expression and a combination of both parameters identified a subgroup with extremely poor OS and DFS
The quantity of BATMs (continuous variable) correlated negatively to the EP3 expression quantified by the IRS (continuous variable, spearman r = − 0.1977, p = 0.0034, Fig. 3a). In the BATM-high subgroup, the EP3 expression was significantly lower than in the BATM-low subgroup (p = 0.00392) (Fig. 3b). Similarly, when categorizing EP3 in an EP3-high (IRS > 1) and an EP3-low (IRS ≤ 1) expressing subgroup, there were higher quantities of BATMs in the EP3-low than in the EP3-high subgroup. (p = 0.0322, Fig. 3c). Comparing both categorized variables (BATM-high and -low with EP3-high and -low), a correlation of only borderline significance could be found (p = 0.051, Table 1). No correlations between EP3 expression and the quantity of BTSMs could be shown (data not shown).
Consistent with the previous study by our team (in which patients with DCIS were not excluded) (24), OS and DFS were superior in invasive BC patients with high expression of EP3 compared to patients with low EP3 expression (OS: p = 0.000019, HR = 0.407, 95%-CI 0.266–0.623), Fig. 3d and DFS: p = 0.000086, HR = 0.426, 95%-CI 0.274–0.660, Fig. 3e). We defined four subgroups using combinations of the categorized variables EP3-high/-low and BATM-high/low. Doing so, we could identify a subgroup with an extremely poor prognosis: patients in the subgroup “EP3-low/BATM-high” (n = 20) showed the worst survival rates - median OS 2.31 years and median DFS 2.18 years – compared to the group “EP3-high/BATM-low” (n = 124) with the best prognosis (OS: median OS 11.42 years, p = 0.000002, Fig. 3f and table 4, part A; DFS: median DFS NR, p = 0.000005, Fig. 3 g and table 4, part A). In other words, patients in the “EP3-low /BATMs-high” subgroup had a 1.756 times higher mortality risk and 1.922 times higher recurrence risk than patients in the favorable “EP3-high/BATM-low” subgroup (table 4 part A). Moreover, the association of BATMs with OS was different considering EP3 expression: in the BATMs-high subgroup, EP3-low-expressing patients had a 1.647 times higher mortality risk (p = 0.023) and 2.107 times higher recurrence risk (p = 0.004) than EP3-high-expressing patients (table 4, part B). But EP3 was also associated with survival in the BATM-low subgroup, where EP3-low-expressing patients had a 1.511 times higher mortality risk (p = 0.01) and 1.405 times higher recurrence risk (p = 0.028) than EP3-high-expressing patients (table 4, part C). Interestingly, BATMs on the other hand did not modify the positive prognostic association of a high EP3-expression: in the EP3-high subgroup, OS and DFS were not significantly different between BATM-low and BATM-high patients (OS: p = 0.059; DFS: p = 0.326, table 4 part D). Only in the EP3-low subgroup, BATMs were significantly correlated to OS and the BATM-high subgroup had a 2.722 times higher mortality risk (p = 0.007) and a 4.049 times higher recurrence risk (p = 0.002) than the BATM-low subgroup (table 4, part E).
Table 4
BC survival analysis using a combination of the prognostic factors EP3 and BATMs
|
OS%
|
p value
|
HR
|
95% CI
|
DFS%
|
p value
|
HR
|
95% CI
|
A.
|
|
|
|
|
|
|
|
|
EP3 high + BATM low
|
83.1
|
|
|
|
77.4
|
|
|
|
EP3 high + BATM high
|
70.0
|
0.059
|
2.128
|
0.973–4.658
|
66.7
|
0.326
|
1.437
|
0.697–2.962
|
EP3 low + BATM low
|
58.1
|
0.01*
|
1.511
|
1.102–2.070
|
60.5
|
0.028*
|
1.405
|
1.037–1.902
|
EP3 low + BATM high
|
35.0
|
0.000002*
|
1.756
|
1.391–2.217
|
60.0
|
0.000005*
|
1.922
|
1.453–2.544
|
B.
|
|
|
|
|
|
|
|
|
BATM high + EP3 high
|
70.0
|
|
|
|
66.7
|
|
|
|
BATM high + EP3 low
|
35.0
|
0.023*
|
1.647
|
1.070–2.536
|
60.0
|
0.004*
|
2.107
|
1.264–3.513
|
C.
|
|
|
|
|
|
|
|
|
BATM low + EP3 high
|
83.1
|
|
|
|
77.4
|
|
|
|
BATM low + EP3 low
|
58.1
|
0.01*
|
1.511
|
1.102–2.070
|
60.5
|
0.028*
|
1.405
|
1.037–1.902
|
D.
|
|
|
|
|
|
|
|
|
EP3 high + BATM low
|
83.1
|
|
|
|
77.4
|
|
|
|
EP3 high + BATM high
|
70.0
|
0.059
|
2.128
|
0.973–4.658
|
66.7
|
0.326
|
1.437
|
0.697–2.962
|
E.
|
|
|
|
|
|
|
|
|
EP3 low + BATM low
|
58.1
|
|
|
|
60.5
|
|
|
|
EP3 low + BATM high
|
35.0
|
0.007*
|
2.722
|
1.321–5.609
|
60.0
|
0.002*
|
4.049
|
1.665–9.847
|
BATMs, Breast adipose tissue macrophage; EP3, prostaglandin E receptor 3 ; OS, Overall survival; DFS, Disease-free survival; p, p value; HR, hazard ratio; CI confidence interval; *,significant (p-value < 0.05).
|