Tumor-infiltrating lymphocytes predict improved overall survival after post-mastectomy radiotherapy: a study of the randomized DBCG82bc cohort

Figures & data

Figure 1. Hematoxylin–eosin (HE) staining’s showing examples of tumors with (A) low level of stromal tumor-infiltrating lymphocytes (TILs) and (B) high level of TILs using a cut off of 30%.

Table 1. Patient characteristics.

	All	Low TILs (<30%)		High TILs (≥30%)
	(n = 1011)	No PMRT (n = 458)	PMRT (n = 447)	No PMRT (n = 47)	PMRT (n = 59)
	n	n (%)	n (%)	n (%)	n (%)	p-value*
Age (years)
Median (range)	53 (25–69)	52 (25–69)	53 (30–69)	53 (31–68)	53 (27–67)
Age (number)
<50 years years	409	203 (44%)	161 (36%)	21 (45%)	24 (41%)	.68
≥50 years years	602	255 (56%)	286 (64%)	26 (55%)	35 (59%)
Menopausal status/protocol
Premenopausal/CMF	553	258 (56%)	236 (53%)	26 (55%)	33 (56%)	.92
Postmenopausal/Tam	458	200 (44%)	211 (47%)	21 (45%)	26 (44%)
Tumor size
≤20mm	362	166 (36%)	160 (36%)	17 (36%)	19 (32%)	.64
21–50mm	506	216 (47%)	238 (53%)	22 (47%)	30 (51%)
>50mm	143	76 (17%)	49 (11%)	8 (17%)	10 (17%)
Positive nodes
0	59	28 (6%)	24 (5%)	3 (6%)	4 (7%)	.73
1–3	462	199 (43%)	218 (49%)	21 (45%)	24 (41%)
≥4	490	231 (50%)	205 (46%)	23 (49%)	31 (53%)
Malignancy grade
Grade 1	231	111 (24%)	113 (25%)	3 (6%)	4 (7%)	<.0001
Grade 2	538	260 (57%)	240 (54%)	18 (38%)	20 (34%)
Grade 3	234	83 (18%)	91 (20%)	26 (55%)	34 (58%)
Not graded	8	4 (1%)	3 (1%)	0 (0%)	1 (2%)
Histology type
Ductal carcinoma	898	386 (84%)	410 (92%)	46 (98%)	56 (95%)	<.0001
Lobular carcinoma	92	61 (13%)	31 (7%)	0 (0%)	0 (0%)
Other carcinomas	21	11 (2%)	6 (1%)	1 (2%)	3 (5%)
ER status
Negative < 1%	274	101 (22%)	106 (24%)	30 (64%)	37 (63%)	<.0001
Positive ≥ 1%	737	357 (78%)	341 (76%)	17 (36%)	22 (37%)
HER2 status
Negative	787	371 (81%)	356 (80%)	26 (55%)	34 (58%)	<.0001
Positive	222	86 (19%)	90 (20%)	21 (45%)	25 (42%)
Unknown	2	1 (0%)	1 (0%)	0 (0%)	0 (0%)
Subtypes
ER + or PR+^a/HER2–/Ki67 low^b or unknown	557	280 (61%)	258 (58%)	10 (21%)	9 (15%)	<.0001
ER + or PR+/HER2–/Ki67 high^b	106	44 (10%)	53 (12%)	4 (9%)	5 (8%)
ER + or PR+/HER2+/Any Ki67	119	50 (11%)	51 (11%)	6 (13%)	12 (20%)
ER– and PR–/HER2+/Any Ki67	103	36 (8%)	39 (9%)	15 (32%)	13 (22%)
ER– and PR–/HER2–/Any Ki67	124	47 (10%)	45 (10%)	12 (26%)	20 (34%)
Unknown	2	1 (0%)	1 (0%)	0 (0%)	0 (0%)
DBCG RT-profile^c
High LRR risk	190	80 (72%)	88 (78%)	11 (100%)	11 (92%)	.034
Low LRR risk	57	31 (28%)	25 (22%)	0 (0%)	1 (8%)

*p-value is from Fisher’s exact test for distribution among low TILs vs. high TILs (irrespective of treatment)).

^aCut-off value for estrogen receptor (ER), progesterone receptor (PR): negative <1%, positive ≥1%.

^bCut-off value for Ki67: low <20%, high ≥20%.

^cData from 247 patients included in the formalin-fixed, paraffin-embedded cohorts in Tramm et al. [28].

Abbreviations: CMF: cyclophosphamide, methotrexate and fluorouracil; Tam: tamoxifen.

Table 2. Uni- and multivariate Cox regression.

	All
	Univariate		Multivariate*
	HR (95% CI)	p-value	HR (95% CI)	p-value
Loco-regional recurrence
PMRT
PMRT vs. no PMRT	0.26 (0.19,0.35)	<.00001	0.23 (0.17,0.31)	<.00001
Menopausal status/protocol^a
Post- vs. premenopausal	1.00 (0.77,1.30)	1.00	1.06 (0.81,1.39)	.67
T-size
21–50mm vs. ≤20mm	1.52 (1.13,2.06)	.0058	1.41 (1.04,1.92)	.027
≥51mm vs. ≤20mm	3.04 (2.09,4.43)	<.00001	2.10 (1.42,3.11)	<.001
Positive nodes
1–3 vs. none	2.08 (0.84,5.14)	.11	2.63 (1.05,6.56)	.038
≥4 vs. none	6.38 (2.61,15.56)	<.0001	7.36 (3.00,18.05)	<.0001
Malignancy grade†
Grade 2 vs. grade 1	1.40 (0.98,1.98)	.064	1.17 (0.82,1.68)	.38
Grade 3 vs. grade 1	2.43 (1.66,3.56)	<.00001	1.76 (1.14,2.72)	.011
Histology type
Lobular carcinoma vs. ductal	0.98 (0.63,1.54)	.93	0.79 (0.49,1.28)	.34
Other carcinoma vs. ductal	0.87 (0.36,2.12)	.76	0.83 (0.26,2.61)	.74
HER2
Positive vs. negative	1.62 (1.20,2.18)	.0014	1.17 (0.85,1.61)	.35
TILs
≥30% vs. <30%	1.38 (0.95,2.01)	.090	1.00 (0.67,1.50)	1.00
Distant metastasis
PMRT
PMRT vs. no PMRT	0.73 (0.62,0.86)	<.001	0.73 (0.62,0.86)	<.001
Menopausal status/protocol^a
Post- vs. premenopausal	1.07 (0.91,1.26)	.40	1.08 (0.92,1.28)	.34
T-size
21–50mm vs. ≤20mm	1.45 (1.21,1.75)	<.0001	1.23 (1.02,1.48)	.034
≥51mm vs. ≤20mm	2.47 (1.93,3.16)	<.00001	1.71 (1.33,2.21)	<.0001
Positive nodes
1–3 vs. none	1.06 (0.69,1.64)	.79	1.20 (0.77,1.87)	.42
≥4 vs. none	3.49 (2.29,5.32)	<.00001	3.50 (2.28,5.36)	<.00001
Malignancy grade^b
Grade 2 vs. grade 1	1.41 (1.14,1.76)	.0019	1.20 (0.96,1.50)	.11
Grade 3 vs. grade 1	2.15 (1.68,2.74)	<.00001	1.74 (1.33,2.29)	<.0001
Histology type
Lobular carcinoma vs. ductal	1.01 (0.76,1.34)	.94	1.03 (0.77,1.37)	.86
Other carcinoma vs. ductal	0.46 (0.22,0.97)	.041	0.72 (0.32,1.62)	.43
HER2
Positive vs. negative	1.58 (1.31,1.91)	<.00001	1.34 (1.10,1.64)	.0043
TILs
≥30% vs. <30%	0.88 (0.66,1.16)	.35	0.62 (0.46,0.83)	.0013
Overall mortality
PMRT
PMRT vs. no PMRT	0.79 (0.69,0.91)	.0013	0.78 (0.68,0.91)	<.001
Menopausal status/protocol^a
Post- vs. premenopausal	1.41 (1.23,1.63)	<.00001	1.43 (1.24,1.65)	<.00001
T-size
21–50mm vs. ≤20mm	1.44 (1.23,1.69)	<.00001	1.25 (1.06,1.47)	.0074
≥51mm vs. ≤20mm	2.25 (1.81,2.79)	<.00001	1.75 (1.40,2.19)	<.00001
Positive nodes
1–3 vs. none	1.11 (0.79,1.57)	.54	1.28 (0.90,1.82)	.18
≥4 vs. none	2.78 (1.98,3.91)	<.00001	2.85 (2.02,4.03)	<.00001
Malignancy grade^b
Grade 2 vs. grade 1	1.28 (1.07,1.54)	.0062	1.11 (0.93,1.34)	.25
Grade 3 vs. grade 1	1.67 (1.36,2.06)	<.00001	1.42 (1.12,1.80)	.0033
Histology type
Lobular carcinoma vs. ductal	1.12 (0.89,1.42)	.34	1.11 (0.87,1.42)	.39
Other carcinoma vs. ductal	0.69 (0.41,1.17)	.17	0.86 (0.44,1.67)	.65
HER2
Positive vs. negative	1.48 (1.25,1.75)	<.00001	1.31 (1.10,1.58)	.0032
TILs
≥30% vs. <30%	0.84 (0.66,1.06)	.15	0.65 (0.50, 0.84)	<.001

*The multivariate analysis is stratified for estrogen receptor status (ER), and ER is therefore not listed among the variables in the table.

^aAccording to study protocol, premenopausal women were treated with CMF and postmenopausal women with tamoxifen.

^bExcluding 8 patients without tumor grade.

Figure 2. Cumulative incidence proportion plots for (A,B) risk of loco-regional recurrence (LRR) and (C,D) risk of distant metastasis (DM) and Kaplan–Meier plots for (E,F) overall survival (OS) for all 1,011 patients divided according to level of tumor-infiltrating lymphocytes (TILs) and stratified according to randomization to post-mastectomy radiotherapy (PMRT) or not (red – PMRT, blue – no PMRT). Actuarial risks at 5, 10 and 20 years and risk difference in actuarial risks at 20 years with 95% confidence intervals are provided, as well as p-values for Log Rank tests for Kaplan–Meier estimates. Interaction tests between TILs and PMRT are stated for both adjusted Cox analysis and risk differences.

Figure 3. Kaplan–Meier plots for overall survival (OS) for (A,B) estrogen receptor (ER) negative and (C,D) ER-positive tumors divided according to level of tumor-infiltrating lymphocytes (TILs) and stratified according to randomization to post-mastectomy radiotherapy (PMRT) or not (red – PMRT, blue – no PMRT). Actuarial risks at 5, 10 and 20 years and risk difference in actuarial risks at 20 years with 95% confidence intervals are provided, as well as p-values for Log Rank tests and interaction tests between TILs and PMRT (for both adjusted Cox analysis and risk differences).

Tramm T, Mohammed H, Myhre S, et al. Development and validation of a gene profile predicting benefit of postmastectomy radiotherapy in patients with high-risk breast cancer: a study of gene expression in the DBCG82bc cohort. Clin Cancer Res. 2014;20(20):5272–5280.

 PubMed Web of Science ®Google Scholar