Purpose:

Tumor-infiltrating immune cells play a key role in tumor progression. The purpose of this study was to analyze whether the immune infiltrate predicts benefit from postoperative radiotherapy in a large randomized breast cancer radiotherapy trial.

Experimental Design:

In the SweBCG91RT trial, patients with stage I and II breast cancer were randomized to breast-conserving surgery (BCS) and postoperative radiotherapy or to BCS only and followed for a median time of 15.2 years. The primary tumor immune infiltrate was quantified through two independent methods: IHC and gene expression profiling. For IHC analyses, the absolute stromal area occupied by CD8+ T cells and FOXP3+ T cells, respectively, was used to define the immune infiltrate. For gene expression analyses, immune cells found to be prognostic in independent datasets were pooled into two groups consisting of antitumoral and protumoral immune cells, respectively.

Results:

An antitumoral immune response in the primary tumor was associated with a reduced risk of breast cancer recurrence and predicted less benefit from adjuvant radiotherapy. The interaction between radiotherapy and immune phenotype was significant for any recurrence in both the IHC and gene expression analyses (P = 0.039 and P = 0.035) and was also significant for ipsilateral breast tumor recurrence in the gene expression analyses (P = 0.025).

Conclusions:

Patients with an antitumoral immune infiltrate in the primary tumor have a reduced risk of any recurrence and may derive less benefit from adjuvant radiotherapy. These results may impact decisions regarding postoperative radiotherapy in early breast cancer.

Translational Relevance

Adjuvant radiotherapy after breast-conserving surgery (BCS) is an important treatment modality to reduce ipsilateral breast tumor recurrences (IBTR). Still, around 10% of patients experience an IBTR within the first decade of diagnosis. There is a need for predictive biomarkers to aid in the individualization of radiotherapy treatment. In this study, we show that a favorable local immune infiltrate is predictive of less benefit from adjuvant radiotherapy. Thus, the antitumoral immune response may be a factor to integrate in the individualization of adjuvant breast radiotherapy. Neoadjuvant radiotherapy may be more suitable for patients with an effective antitumoral immune response.

The immune compartment of the tumor microenvironment consists of both anti- and protumoral cells and both groups affect prognosis and tumor progression (1). Stromal tumor-infiltrating lymphocytes (henceforth referred to as TIL) correlate positively with an antitumoral immune infiltrate and a favorable prognosis in breast cancer (2). TILs represent a heterogeneous combination of cells primarily composed of T cells (3). The primary effector cell of the cell-mediated antitumoral immune response is the CD8+ cytotoxic T lymphocyte and a greater infiltration is associated with an increased survival in triple-negative and HER2-positive breast cancer (4, 5). T regulatory cells (Treg), which can be identified through IHC staining for FOXP3, are immunosuppressive and function as antagonists of the antitumoral immune response (Supplementary Fig. S1). A greater Treg infiltration is associated with a poor prognosis in breast cancer (6). TILs and a high CD8:FOXP3 ratio are treatment predictive for chemotherapy and anti-HER2 treatment in breast cancer (2, 7).

Radiotherapy has been shown to interact with the immune system in a variety of cancers (8), but the effects in the adjuvant setting remain largely unexplored despite great clinical importance (9). Adjuvant chemoradiation in breast cancer and radiotherapy in combination with checkpoint inhibitors in metastatic breast cancer show promising results and could suggest a favorable interaction between radiotherapy and the immune system (8, 10).

Several genomic predictors for recurrence and radiotherapy benefit have been developed, with ARTIC being the only predictor validated in a randomized cohort (11). These classifiers generally predict an unfavorable prognosis (11) and, therefore, fail to identify potential patients with an improved prognosis who may lack benefit from radiotherapy. We have previously shown that a high TIL infiltrate in the primary tumor confers an improved prognosis and could predict a decreased benefit from adjuvant radiotherapy (12).

The purpose of this study was to investigate how the primary tumor immune infiltrate affects the benefit from adjuvant radiotherapy in a retrospective analysis of a phase III trial cohort with the prespecified hypothesis that a favorable immune infiltrate is predictive of less benefit from postoperative radiotherapy.

Patients and study design

Patients from the SweBCG91RT trial were included and have been described elsewhere (13). In summary, 1,178 patients with lymph node-negative (N0), stage I or IIA breast cancer were randomly assigned between 1991 and 1997 to breast-conserving surgery with or without whole-breast radiotherapy and followed for a median time of 15.2 years. Radiotherapy was given with tangential opposed fields of 4 or 6 MV photons, with a prescribed dose of 48–54 Gy in 24–27 fractions, to the remaining breast parenchyma. No patients received a radiotherapy boost. Tumor blocks from the initial surgery were retrieved and used for IHC staining and gene expression analysis (ref. 12; Fig. 1; Supplementary Fig. S2). The trial and follow-up study were conducted in accordance with the Declaration of Helsinki. Informed oral consent was obtained from all patients, which was determined appropriate and approved by the regional ethical review board for the original study and for this study (approval nos., 2010/127 and 2015/548).

Figure 1.

CONSORT diagram for patients included in IHC analyses.

Figure 1.

CONSORT diagram for patients included in IHC analyses.

Close modal

IHC analyses

TILs were assessed on whole-tissue sections as semicontinuous values (<1%, 1%–9%, 10%–49%, 50%–74%, and ≥75%), as described previously (12). Staining for antitumoral CD8+ T cells and protumoral FOXP3+ Tregs was performed on tissue microarrays (TMA) and evaluated as the proportion of TILs occupied by the respective cell type and recorded using the same semicontinuous values as those used to assess TILs. The median value was multiplied with the median stromal area occupied by TILs to obtain a score of the calculated absolute stromal area occupied by CD8 and FOXP3, respectively. Two board-certified pathologists evaluated the IHC slides simultaneously, but independently, using a double-headed microscope. If the assessments differed, the evaluation was repeated until a consensus was reached. This method was used to reduce the risk of interobserver and intraobserver bias. Two TMAs (from two different cores) per patient were analyzed. The highest value for CD8 and FOXP3, respectively, was used to infer the infiltration of CD8+ T cells and Tregs. The kappa value for the concordance between TMAs was 0.38 for FOXP3 staining (71% agreement) and 0.28 for CD8 staining (56% agreement). Predetermined cutoffs of 5% and 2.5% were used to create categories with high and low values for CD8 and FOXP3, respectively. The choice of cutoffs was based on the distributions (Supplementary Fig. S3). A cutoff twice as high for CD8+ T cells as for FOXP3+ Tregs has been used previously (14). The categories were combined to create groups reflecting the mutual balance between the two cell types. Groups with the following order of proposed increasing immunocompetency were created: CD8low/FOXP3low, CD8highFOXP3high, and CD8highFOXP3low (no tumors had the combination CD8lowFOXP3high).

All patients with tumor tissue of sufficient quality to be evaluated via IHC were included. Clinicopathologic variables did not differ significantly between excluded and included patients, except for tumor size, which was significantly smaller among excluded patients (median 11 vs. 12 mm among excluded and included patients, respectively; Supplementary Table S1). In total, 927 patients were included in the multivariable analyses.

Gene expression analyses

GeneChip Human Exon 1.0 ST Arrays (Thermo Fisher Scientific) were used to obtain gene expression data (Gene Expression Omnibus with accession no., GSE119295), as described previously (11). To test our broader biological hypothesis of an inverse relationship between the effectiveness of the antitumoral immune response in the primary tumor and the benefit from radiotherapy, and to create robust definitions of the immune infiltrate, a more general quantification of the immune compartment (including additional prognostic cells beyond CD8+ T cells and FOXP3+ Tregs, see below) was performed using xCell (15). Analyses of the prognostic impact of different immune cells were performed in the Metabric, Mainz, and van de Vijver datasets (refs. 16–18; described in the Supplementary Data). Prognostic immune cells in the public datasets, where the prognostic effect was supported by the literature, were then scaled and summed to create groups of antitumoral and protumoral cells, respectively. In the antitumoral group, TILs (except for Tregs and Th cells) and plasma cells were included. In the protumoral group, tumor-associated macrophages, Tregs, Th1 cells, and natural killer T cells were included. Cutoffs for antitumoral and protumoral immune cells, respectively, were determined on the basis of the ability of the cutoffs to define prognostically distinct groups in the public datasets (Supplementary Fig. S4). Groups were created by combining the antitumoral group (high or low) with the protumoral group (high or low) with the following order of proposed increasing immunocompetency: antitumorallow/protumorallow, antitumorallow/protumoralhigh, antitumoralhigh/protumoralhigh, and antitumoralhigh/protumorallow. In total, 744 samples had sufficient mRNA quality and were included in multivariable analysis. Tumor size differed significantly between included and excluded patients (Supplementary Table S1).

Immune cell quantification with xCell

xCell was used to quantify the immune infiltrate from gene expression data. It performs well compared with other deconvolutional methods (15, 19) and it is, to our knowledge, the only tool which utilizes a correction method (termed spillover matrix) to correct for similarities in gene expression between similar cell types. Because similar cell types can have different/opposing effects (e.g., M1 and M2 macrophages; ref. 20), we deemed this to be beneficial for predicting prognosis accurately. All 5,079 genes used by xCell were found in the SweBCG91RT dataset.

Statistical analysis

Time to any recurrence as the first event within 10 years from date of diagnosis was used as primary endpoint. Secondary endpoints were time to ipsilateral breast tumor recurrence (IBTR) as first event within 10 years and distant metastasis within 10 years. The aims were to analyze the interaction between radiotherapy and immune phenotype on risk of any recurrence, IBTR, and distant metastasis, as well as the prognostic effects of immune phenotypes. A P < 0.05 was considered significant. P values reported for other analyses, which were not part of the main hypothesis, were not adjusted for multiplicity and should be interpreted with caution. HRs were calculated with cause-specific Cox proportional hazards regression to reflect the biologic effect of radiotherapy in the presence of competing risks. Other recurrences and death were considered competing risks for IBTR and any death was considered a competing risk for any recurrence and distant metastasis. Figures of cumulative incidence were created according to the method of Fine and Gray (21) and based on the multivariable Cox models of subhazards for the different endpoints. P values for the sub-HR between compared groups were denoted as PCIF in the plots. Clinical variables were tested in univariable analysis and then, if significant, in multivariable analysis. Tumor size did not remain significant in multivariable analysis for any recurrence or IBTR and was not included in the final models for these endpoints. Subtype was kept in multivariable analysis for all analyses, despite not being significant in univariable analysis, because of the biologic relevance. In addition, systemic treatment was included in all multivariable analyses. Multivariable Cox regression models with or without an interaction term between immune phenotype and radiotherapy were compared using the likelihood ratio test to analyze the predictive effect of immune phenotype on radiotherapy benefit. Differences in distributions between groups were compared with two-sided Fisher exact test.

The proportional hazards assumption was checked using the Schoenfeld residuals. For any recurrence, it was violated for radiotherapy, histologic grade, and subtype in both the IHC and gene expression analyses. In addition, immune phenotype did not meet the proportional hazards assumption criteria for the IHC analysis. The HRs for these variables should, therefore, be interpreted as the mean over the follow-up period of 0–10 years. STATA 16.0 was used for analysis (StataCorp. 2017, Stata: Release 15, Statistical Software, StataCorp LLC).

IHC

Descriptive results

Patient characteristics are described in Table 1. FOXP3 was strongly associated with unfavorable clinicopathologic variables, such as ER negativity, higher histologic grade, and younger age. CD8 was not as strongly correlated with these variables. The overall number of events of IBTRs, distant metastases, and all recurrences were 114, 95 and 173, respectively (Supplementary Table S5).

Table 1.

Distribution of clinicopathologic variables between immune groups.

VariablesCD8low/FOXP3lowCD8high/FOXP3highCD8high/FOXP3lowCorrelation P
TILs <10% 676 (99.6%) 0 (0%) 0 (0%) 676 (70.9%)  
 10%–49% 3 (0.4%) 65 (48.1%) 137 (98.6%) 205 (21.5%)  
 >50% 0 (0%) 70 (51.9%) 2 (1.4%) 72 (7.6%)  
 Total 679 (100%) 135 (100%) 139 (100%) 953 (100%) <0.001 
Immune groups for gene expression analyses Antitumorallow/protumorallow 151 (28.4%) 11 (9.8%) 18 (16.5%) 180 (23.9%)  
 Antitumorallow/protumoralhigh 102 (19.2%) 12 (10.7%) 15 (13.8%) 129 (17.1%)  
 Antitumoralhigh/protumoralhigh 117 (22.0%) 70 (62.5%) 39 (35.8%) 226 (30%)  
 Antitumoralhigh/protumorallow 162 (30.5%) 19 (17%) 37 (33.9%) 218 (29%) <0.001 
 Total 532 (100.1%) 112 (100%) 109 (100%) 753 (100%)  
Subtype Luminal A 434 (65.3%) 24 (18.5%) 86 (62.3%) 544 (58.3%)  
 Luminal B 181 (27.2%) 40 (30.8%) 28 (20.3%) 249 (26.7%)  
 HER2 positive 31 (4.7%) 22 (16.9%) 10 (7.2%) 63 (6.8%)  
 Triple negative 19 (2.9%) 44 (33.8%) 14 (10.1%) 77 (8.3%)  
 Total 665 (100%) 130 (100%) 138 (100%) 933 (100%) <0.001 
Grade 126 (18.7%) 2 (1.5%) 16 (11.6%) 144 (15.3%)  
 II 446 (66.3%) 44 (33.3%) 76 (55.1%) 566 (60%)  
 III 101 (15%) 86 (65.2%) 46 (33.3%) 233 (24.7%)  
 Total 673 (100%) 132 (100%) 138 (100%) 943 (100%) <0.001 
ER status ER positive 644 (96.6%) 76 (57.6%) 121 (87.1%) 841 (89.7%)  
 ER negative 23 (3.4%) 56 (42.4%) 18 (12.9%) 97 (10.3%)  
 Total 667 (100%) 132 (100%) 139 (100%) 938 (100%) <0.001 
Tumor size (mm) 1–10 287 (51.5%) 32 (54.1%) 45 (52.9%) 364 (52%)  
 10–15 246 (22.1%) 47 (13.1%) 53 (17.1%) 346 (20.2%)  
 16–20 96 (19%) 33 (19.3%) 26 (20.2%) 155 (19.2%)  
 > 20 47 (7.4%) 20 (13.5%) 15 (9.9%) 82 (8.6%) <0.001 
 Total 676 (100%) 132 (100%) 139 (100%) 938 (100%)  
Age (median)  60 56 58  0.004 
Systemic treatment Yes 48 (7.1%) 20 (14.8%) 11 (7.9%) 79 (8.3%) 0.01 
 No 631 (92.9%) 115 (85.2%) 128 (92.1%) 874 (91.7%)  
VariablesCD8low/FOXP3lowCD8high/FOXP3highCD8high/FOXP3lowCorrelation P
TILs <10% 676 (99.6%) 0 (0%) 0 (0%) 676 (70.9%)  
 10%–49% 3 (0.4%) 65 (48.1%) 137 (98.6%) 205 (21.5%)  
 >50% 0 (0%) 70 (51.9%) 2 (1.4%) 72 (7.6%)  
 Total 679 (100%) 135 (100%) 139 (100%) 953 (100%) <0.001 
Immune groups for gene expression analyses Antitumorallow/protumorallow 151 (28.4%) 11 (9.8%) 18 (16.5%) 180 (23.9%)  
 Antitumorallow/protumoralhigh 102 (19.2%) 12 (10.7%) 15 (13.8%) 129 (17.1%)  
 Antitumoralhigh/protumoralhigh 117 (22.0%) 70 (62.5%) 39 (35.8%) 226 (30%)  
 Antitumoralhigh/protumorallow 162 (30.5%) 19 (17%) 37 (33.9%) 218 (29%) <0.001 
 Total 532 (100.1%) 112 (100%) 109 (100%) 753 (100%)  
Subtype Luminal A 434 (65.3%) 24 (18.5%) 86 (62.3%) 544 (58.3%)  
 Luminal B 181 (27.2%) 40 (30.8%) 28 (20.3%) 249 (26.7%)  
 HER2 positive 31 (4.7%) 22 (16.9%) 10 (7.2%) 63 (6.8%)  
 Triple negative 19 (2.9%) 44 (33.8%) 14 (10.1%) 77 (8.3%)  
 Total 665 (100%) 130 (100%) 138 (100%) 933 (100%) <0.001 
Grade 126 (18.7%) 2 (1.5%) 16 (11.6%) 144 (15.3%)  
 II 446 (66.3%) 44 (33.3%) 76 (55.1%) 566 (60%)  
 III 101 (15%) 86 (65.2%) 46 (33.3%) 233 (24.7%)  
 Total 673 (100%) 132 (100%) 138 (100%) 943 (100%) <0.001 
ER status ER positive 644 (96.6%) 76 (57.6%) 121 (87.1%) 841 (89.7%)  
 ER negative 23 (3.4%) 56 (42.4%) 18 (12.9%) 97 (10.3%)  
 Total 667 (100%) 132 (100%) 139 (100%) 938 (100%) <0.001 
Tumor size (mm) 1–10 287 (51.5%) 32 (54.1%) 45 (52.9%) 364 (52%)  
 10–15 246 (22.1%) 47 (13.1%) 53 (17.1%) 346 (20.2%)  
 16–20 96 (19%) 33 (19.3%) 26 (20.2%) 155 (19.2%)  
 > 20 47 (7.4%) 20 (13.5%) 15 (9.9%) 82 (8.6%) <0.001 
 Total 676 (100%) 132 (100%) 139 (100%) 938 (100%)  
Age (median)  60 56 58  0.004 
Systemic treatment Yes 48 (7.1%) 20 (14.8%) 11 (7.9%) 79 (8.3%) 0.01 
 No 631 (92.9%) 115 (85.2%) 128 (92.1%) 874 (91.7%)  

Note: P values were calculated with the Spearman rank correlation. In total, 669 (70.9%) patients included in the IHC analyses had TILs <10%, 204 (21.6%) had TILs 10%–49%, and 72 (7.6%) had TILs ≥50%. The majority (99.5%) of immune-depleted tumors (CD8low/FOXP3low) had TILs <10%, while all immunogenic tumors (CD8high/FOXP3low and CD8high/FOXP3high) had TILs ≥10%. No tumors had the combination of CD8low/FOXP3high. CD8high/FOXP3high was most strongly associated with negative prognostic variables. CD8high/FOXP3high was also the group with the highest TIL values with 51.9% of patients having TILs ≥50% in contrast to CD8high/FOXP3low, in which only 1.4% had TILs ≥50%. The IHC immune groups were largely redistributed when the four immune groups for the gene expression analyses were created through xCell because of the lower cutoffs that were used.

Univariable analysis

High levels of FOXP3 (CD8high/FOXP3high) conferred an unfavorable prognosis regarding any recurrence [HR, 1.7, 95% confidence interval (CI), 1.2–2.4; P = 0.002] and distant metastasis (HR, 1.8; 95% CI, 1.1–3.0; P = 0.020) in univariable analysis. Tumors with high levels of CD8, but low levels of FOXP3 (CD8high/FOXP3low), showed a trend toward a reduced risk of IBTR (HR, 0.55; 95% CI, 0.30–1.0; P = 0.052; Table 2; Fig. 2).

Table 2.

Univariable Cox regression analysis of prognostic effect on risk of IBTR and any recurrence.

IBTRAny recurrenceDistant metastasis
VariablesHR (95% CI)PHR (95% CI)PHR (95% CI)P
IHC immune groups CD8low/FOXP3low 1.0 (ref.)  1.0 (ref.)a  1.0 (ref.)  
 CD8high/FOXP3high 1.2 (0.75–1.9) 0.45 1.7 (1.2–2.4) 0.002 1.8 (1.1–3.0) 0.020 
 CD8high/FOXP3low 0.55 (0.30–1.0) 0.052 0.79 (0.51–1.2) 0.28 1.0 (0.57–1.9) 0.98 
Gene expression immune groups Antitumorallow/protumorallow 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 Antitumorallow/protumoralhigh 1.2 (0.68–2.1) 0.52 1.2 (0.77–2.0) 0.51 1.5 (0.75–3.1) 0.24 
 Antitumoralhigh/protumoralhigh 1.2 (0.73–2.0) 0.48 1.0 (0.67–1.6) 0.11 1.4 (0.74–2.6) 0.30 
 Antitumoralhigh/protumorallow 0.90 (0.53–1.5) 0.69 1.3 (0.85–1.9) 0.56 1.1 (0.57–2.2) 0.74 
RT treatment No RT 1.0 (ref.)a  1.0 (ref.)a  1.0 (ref.)  
 RT 0.40 (0.28–0.58) <0.001 0.55 (0.42–0.73) <0.001 0.77 (0.51–1.16) 0.21 
Histologic grade 1.0 (ref.)a  1.0 (ref.)a  1.0 (ref.)  
 II 1.8 (0.98–3.2) 0.056 2.0 (1.2–3.2) 0.010 2.8 (1.1–7.8) 0.049 
 III 2.3 (1.2–4.4) 0.008 3.3 (2.0–5.6) <0.001 8.7 (3.1–24) <0.001 
Age <50 years 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 ≥50 years 0.56 (0.39–0.81) 0.002 0.58 (0.43–0.78) <0.001 0.55 (0.36–0.86) 0.008 
Subtype Luminal A 1.0 (ref.)  1.0 (ref.)a  1.0 (ref.)a  
 Luminal B 1.3 (0.85–1.9) 0.24 1.4 (1.0–2.0) 0.023 2.0 (1.6–4.0) <0.001 
 HER2-positive 1.5 (0.78–2.8) 0.23 1.8 (1.1–2.9) 0.023 2.8 (1.4–5.8) 0.004 
 Triple-negative 1.4 (0.77–2.6) 0.27 1.9 (1.2–2.9) 0.007 4.5 (2.5–8.0) <0.001 
Tumor size (mm) 1–10 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)a  
 11–15 0.97 (0.66–1.4) 0.88 1.3 (0.94–1.81) 0.11 2.6 (1.5–4.6) 0.001 
 16–20 1.2 (0.74–1.9) 0.46 1.6 (1.1–2.4) 0.019 3.2 (1.7–6.1) <0.001 
 >20 0.82 (0.41–1.7) 0.58 1.9 (1.2–3.0) 0.005 4.1 (2.1–8.3) <0.001 
Systemic treatment No 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 Yes 0.33 (0.12–0.89) 0.028 0.77 (0.45–1.3) 0.34 1.2 (0.59–2.3) 0.66 
IBTRAny recurrenceDistant metastasis
VariablesHR (95% CI)PHR (95% CI)PHR (95% CI)P
IHC immune groups CD8low/FOXP3low 1.0 (ref.)  1.0 (ref.)a  1.0 (ref.)  
 CD8high/FOXP3high 1.2 (0.75–1.9) 0.45 1.7 (1.2–2.4) 0.002 1.8 (1.1–3.0) 0.020 
 CD8high/FOXP3low 0.55 (0.30–1.0) 0.052 0.79 (0.51–1.2) 0.28 1.0 (0.57–1.9) 0.98 
Gene expression immune groups Antitumorallow/protumorallow 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 Antitumorallow/protumoralhigh 1.2 (0.68–2.1) 0.52 1.2 (0.77–2.0) 0.51 1.5 (0.75–3.1) 0.24 
 Antitumoralhigh/protumoralhigh 1.2 (0.73–2.0) 0.48 1.0 (0.67–1.6) 0.11 1.4 (0.74–2.6) 0.30 
 Antitumoralhigh/protumorallow 0.90 (0.53–1.5) 0.69 1.3 (0.85–1.9) 0.56 1.1 (0.57–2.2) 0.74 
RT treatment No RT 1.0 (ref.)a  1.0 (ref.)a  1.0 (ref.)  
 RT 0.40 (0.28–0.58) <0.001 0.55 (0.42–0.73) <0.001 0.77 (0.51–1.16) 0.21 
Histologic grade 1.0 (ref.)a  1.0 (ref.)a  1.0 (ref.)  
 II 1.8 (0.98–3.2) 0.056 2.0 (1.2–3.2) 0.010 2.8 (1.1–7.8) 0.049 
 III 2.3 (1.2–4.4) 0.008 3.3 (2.0–5.6) <0.001 8.7 (3.1–24) <0.001 
Age <50 years 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 ≥50 years 0.56 (0.39–0.81) 0.002 0.58 (0.43–0.78) <0.001 0.55 (0.36–0.86) 0.008 
Subtype Luminal A 1.0 (ref.)  1.0 (ref.)a  1.0 (ref.)a  
 Luminal B 1.3 (0.85–1.9) 0.24 1.4 (1.0–2.0) 0.023 2.0 (1.6–4.0) <0.001 
 HER2-positive 1.5 (0.78–2.8) 0.23 1.8 (1.1–2.9) 0.023 2.8 (1.4–5.8) 0.004 
 Triple-negative 1.4 (0.77–2.6) 0.27 1.9 (1.2–2.9) 0.007 4.5 (2.5–8.0) <0.001 
Tumor size (mm) 1–10 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)a  
 11–15 0.97 (0.66–1.4) 0.88 1.3 (0.94–1.81) 0.11 2.6 (1.5–4.6) 0.001 
 16–20 1.2 (0.74–1.9) 0.46 1.6 (1.1–2.4) 0.019 3.2 (1.7–6.1) <0.001 
 >20 0.82 (0.41–1.7) 0.58 1.9 (1.2–3.0) 0.005 4.1 (2.1–8.3) <0.001 
Systemic treatment No 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 Yes 0.33 (0.12–0.89) 0.028 0.77 (0.45–1.3) 0.34 1.2 (0.59–2.3) 0.66 

Abbreviation: RT, radiotherapy.

aThe proportional hazards assumption was violated.

Figure 2.

A–I, Cumulative incidence of IBTR, any recurrence, and distant metastasis among immune groups created by IHC assessment. From left to right, the groups are arranged in order of increasing proposed immunocompetency. The most immune-depleted group (antitumorallow/protumorallow) showed the greatest benefit from radiotherapy (RT).

Figure 2.

A–I, Cumulative incidence of IBTR, any recurrence, and distant metastasis among immune groups created by IHC assessment. From left to right, the groups are arranged in order of increasing proposed immunocompetency. The most immune-depleted group (antitumorallow/protumorallow) showed the greatest benefit from radiotherapy (RT).

Close modal
Figure 3.

A–L, Cumulative incidence of IBTR, any recurrence, and distant metastasis among immune groups created by gene expression. From left to right, the groups are arranged in order of increasing proposed immunocompetency. The most immune-depleted group (antitumorallow/protumorallow) showed the greatest benefit from radiotherapy (RT).

Figure 3.

A–L, Cumulative incidence of IBTR, any recurrence, and distant metastasis among immune groups created by gene expression. From left to right, the groups are arranged in order of increasing proposed immunocompetency. The most immune-depleted group (antitumorallow/protumorallow) showed the greatest benefit from radiotherapy (RT).

Close modal

Multivariable analysis

The following analyses for any recurrence and IBTR were adjusted for histologic grade, age, subtype, and systemic therapy. Regarding distant metastasis, the analysis was adjusted for histologic grade, age, subtype, tumor size, and systemic therapy. Among patients who did not receive radiotherapy, the CD8high/FOXP3low group showed a decreased risk of IBTR (HR, 0.41; 95% CI, 0.19–0.86; P = 0.018), any recurrence (HR, 0.41; 95% CI, 0.21–0.77; P = 0.006), and distant metastasis (HR, 0.44; 95% CI, 0.18–1.10; P = 0.078) in multivariable analysis compared with the CD8low/FOXP3low group (HR, 1.0; data not shown). No significant differences in risk of distant metastasis among unirradiated patients were observed between the groups in multivariable analysis.

With risk of IBTR and any recurrence as dependent variables, the immune-depleted group was the only group with a significant benefit from radiotherapy regarding IBTR (HR, 0.37; 95% CI, 0.24–0.57; P < 0.001) and any recurrence (HR, 0.49; 95% CI, 0.34–0.69; P < 0.001). The group with the best prognosis among unirradiated patients (i.e., CD8high/FOXP3low) showed the least benefit from radiotherapy with HR of 0.60 for IBTR (95% CI, 0.18–2.0; P = 0.41), HR of 1.50 for any recurrence (95% CI, 0.68–3.4; P = 0.30), and HR of 1.9 for distant metastasis (95% CI, 0.62–5.7; P = 0.26). The test for interaction between radiotherapy and immune phenotype was significant for any recurrence (P = 0.039), but not for IBTR (P = 0.68) or distant metastasis (P = 0.36).

Gene expression analyses

Descriptive results

Immune group classification by IHC showed an overall concordance with classification by gene expression (Table 1). Because of lower cutoffs for the gene expression analyses, the immune-depleted IHC group (CD8lowFOXP3low, n = 669) was largely redistributed and the corresponding immune-depleted group for the gene expression analyses (antitumorallow/protumorallow) consisted of 184 patients. Both the anti- and protumoral immune cell populations were positively correlated with unfavorable clinicopathologic characteristics (Supplementary Table S2).

Univariable analysis

The following HRs represent the change in risk per one unit increase in SD. In the public datasets, the antitumoral immune cell variable was associated with an improved prognosis in the Metabric (HR, 0.91; 95% CI, 0.85–0.99; P = 0.028), Mainz (HR, 0.60; 95% CI, 0.40–0.90; P = 0.014), and van de Vijver (HR, 0.76; 95% CI, 0.61–0.95; P = 0.015) datasets. The protumoral immune cell variable was associated with a poor prognosis in the Metabric (HR, 1.21; 95% CI, 1.13–1.3; P < 0.001), Mainz (HR, 1.36; 95% CI, 1.02–1.82; P = 0.034), and van de Vijver (HR, 1.21; 95% CI, 0.99–1.49; P = 0.058) datasets.

In our cohort, the antitumoral immune cell variable did not significantly affect the risk of any endpoint. The protumoral immune cell variable predicted an increased risk of any recurrence (HR, 1.05; 95% CI, 1.02–1.36; P = 0.029) and distant metastasis (HR, 1.36; 95% CI, 1.09–1.68; P = 0.0048) in univariable analysis.

Multivariable analysis

The following multivariable analyses for any recurrence and IBTR were adjusted for histologic grade, age, and subtype. Regarding distant metastasis, the analysis was adjusted for histologic grade, age, subtype, and tumor size. In a multivariable analysis of unirradiated patients, none of the immune groups had a significantly reduced risk of IBTR, any recurrence, or distant metastasis, but a trend of higher HRs in the immune-depleted group could be observed.

The immune-depleted group was the only group with a significant benefit from radiotherapy regarding any recurrence (HR, 0.22; 95% CI, 0.10–0.49; P < 0.001) in adjusted analysis. Patients in this group also showed the greatest benefit from radiotherapy regarding IBTR (HR, 0.11; 95% CI, 0.032–0.36; P < 0.001). Radiotherapy did not significantly reduce the risk of metastasis in any group (Table 3, Fig. 3). The interaction between radiotherapy and the immune groups was P = 0.025 for IBTR, P = 0.035 for any recurrence, and P = 0.43 for distant metastasis (Table 3, Fig. 3).

Table 3.

Multivariable Cox regression analysis of predictive effect of immune phenotype on risk of IBTR, any recurrence, and distant metastasis with radiotherapy treatment, including grade, age, subtype, size, systemic treatment, and the interaction between immune phenotype and radiotherapy treatment.

IBTRAny recurrenceDistant metastasis
VariablesHR (95% CI)PHR (95% CI)PHR (95% CI)P
IHC immune groups CD8low/FOXP3low, no RT 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 CD8low/FOXP3low, RT 0.37 (0.24–0.57) <0.001 0.49 (0.34–0.69) <0.001 0.83 (0.50–1.4) 0.49 
 CD8high/FOXP3high, no RT 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 CD8high/FOXP3high, RT 0.50 (0.19–1.3) 0.16 0.57 (0.30–1.09) 0.089 0.73 (0.30–1.8) 0.49 
 CD8high/FOXP3low, no RT 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 CD8high/FOXP3low, RT 0.60 (0.18–2.0) 0.41 1.5 (0.68–3.4) 0.30 1.9 (0.62–5.7) 0.26 
 Interaction immune phenotype x RT  0.68  0.039  0.36 
Gene expression immune groups Antitumorallow/protumorallow, no RT 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 Antitumorallow/protumorallow, RT 0.11 (0.032–0.36) <0.001 0.22 (0.10–0.49) <0.001 0.59 (0.20–1.6) 0.31 
 Antitumorallow/protumoralhigh, no RT 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 Antitumorallow/protumoralhigh, RT 0.57 (0.24–1.4) 0.20 0.78 (0.38–1.6) 0.48 0.93 (0.34–2.5) 0.89 
 Antitumoralhigh/protumorallow, no RT 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 Antitumoralhigh/protumorallow, RT 0.51 (0.21–1.1) 0.10 0.80 (0.44–1.5) 0.47 1.7 (0.68–4.2) 0.26 
 Antitumoralhigh/protumoralhigh, no RT 1.0 (ref.)  1.0 (ref.)  1.0 (ref)  
 Antitumoralhigh/protumoralhigh, RT 0.70 (0.36–1.4) 0.30 0.68 (0.39–1.2) 0.16 0.73 (0.31–1.71) 0.46 
 Interaction Immune phenotype × RT  0.025  0.035  0.43 
IBTRAny recurrenceDistant metastasis
VariablesHR (95% CI)PHR (95% CI)PHR (95% CI)P
IHC immune groups CD8low/FOXP3low, no RT 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 CD8low/FOXP3low, RT 0.37 (0.24–0.57) <0.001 0.49 (0.34–0.69) <0.001 0.83 (0.50–1.4) 0.49 
 CD8high/FOXP3high, no RT 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 CD8high/FOXP3high, RT 0.50 (0.19–1.3) 0.16 0.57 (0.30–1.09) 0.089 0.73 (0.30–1.8) 0.49 
 CD8high/FOXP3low, no RT 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 CD8high/FOXP3low, RT 0.60 (0.18–2.0) 0.41 1.5 (0.68–3.4) 0.30 1.9 (0.62–5.7) 0.26 
 Interaction immune phenotype x RT  0.68  0.039  0.36 
Gene expression immune groups Antitumorallow/protumorallow, no RT 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 Antitumorallow/protumorallow, RT 0.11 (0.032–0.36) <0.001 0.22 (0.10–0.49) <0.001 0.59 (0.20–1.6) 0.31 
 Antitumorallow/protumoralhigh, no RT 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 Antitumorallow/protumoralhigh, RT 0.57 (0.24–1.4) 0.20 0.78 (0.38–1.6) 0.48 0.93 (0.34–2.5) 0.89 
 Antitumoralhigh/protumorallow, no RT 1.0 (ref.)  1.0 (ref.)  1.0 (ref.)  
 Antitumoralhigh/protumorallow, RT 0.51 (0.21–1.1) 0.10 0.80 (0.44–1.5) 0.47 1.7 (0.68–4.2) 0.26 
 Antitumoralhigh/protumoralhigh, no RT 1.0 (ref.)  1.0 (ref.)  1.0 (ref)  
 Antitumoralhigh/protumoralhigh, RT 0.70 (0.36–1.4) 0.30 0.68 (0.39–1.2) 0.16 0.73 (0.31–1.71) 0.46 
 Interaction Immune phenotype × RT  0.025  0.035  0.43 

Note: For IHC analyses, CD8+ T cells and FOXP3+ Tregs were analyzed. For gene expression analyses, prognostic immune cells, where the prognosis could be supported by literature and analyses in public datasets, were grouped into an antitumoral and a protumoral group. IHC and gene expression immune groups are arranged in proposed order of increasing immunocompetency. The same covariables (histologic grade, age, systemic therapy, and subtype for IBTR and any recurrence and histologic grade, age, subtype, systemic therapy, and tumor size for distant metastasis) were used for both analyses. Full tables are presented in the Supplementary Data (Supplementary Tables S3 and S4). For distant metastasis, tumor size was included in multivariable analysis yielding 922 patients in the analysis of immune groups derived from IHC staining, instead of 927 patients (due to missing information of tumor size among 5 patients).

Abbreviation: RT, radiotherapy.

Our study shows that patients with early-stage breast cancer with a favorable immune infiltrate in the primary tumor may derive less benefit from adjuvant radiotherapy regarding breast cancer recurrence. Patients with favorable immune infiltrates, representing around 14.5% of patients in the analyzed SweBCG91RT cohort, may have an improved prognosis, but a small or nonexistent benefit from adjuvant radiotherapy. To our knowledge, this is the first study to analyze the predictive effect of the immune infiltrate in the primary tumor, by two independent methods, among patients with breast cancer treated with adjuvant radiotherapy in a large randomized trial. In the SweBCG91RT trial, few patients received systemic adjuvant treatment (22) and the influence of the immune system on the effect of adjuvant radiotherapy can largely be studied without confounding effects from systemic treatment.

Our findings should be contrasted against studies on neoadjuvant chemoradiotherapy in other cancers where a favorable interaction with an activated antitumoral immune infiltrate has been observed (23–25). Radiotherapy is capable of altering the tumor microenvironment favoring activation of the immune infiltrate (refs. 9, 26, 27; Supplementary Fig. S1). Radiotherapy against a high tumor burden theoretically has a better potential for release of neoantigens (26, 28) compared with that in the adjuvant setting, where the tumor burden is low. The difference in neoantigen load could thus, explain why our findings do not conform with above-mentioned studies. Neoadjuvant, rather than adjuvant, radiotherapy could be more suitable for patients with an activated antitumoral immune response.

Radiotherapy can also have negative effects on the immune system and treatment with radiotherapy alone has been associated with immunosuppressive effects (9). Because of the radiosensitivity of lymphocytes, radiotherapy has the potential to suppress the immune infiltrate (29, 30). Our findings suggest that in the adjuvant setting, the immune-stimulating effects of radiotherapy may be weaker than in the neoadjuvant setting and instead, the immunosuppressive effects could be more pronounced. The effect of radiotherapy on the immune system also varies with different fractionation protocols (31). The immunosuppressive effects may occur primarily with conventional fractionated radiotherapy (32), as used in our study, and a different radiotherapy fractionation scheme may have produced different results. This study supports our previous findings of less benefit from adjuvant radiotherapy among patients with an effective antitumoral immune response (12). Two independent methods with different cutoffs were used to quantify the immune infiltrate, and both produced significant interactions with radiotherapy.

Given that the distribution of immune groups did not vary much based on known clinicopathologic variables (Table 1; Supplementary Table S2), a direct analysis of the immune infiltrate (e.g., CD8:Treg balance), in addition to standard clinicopathologic variables, could be necessary to identify subgroups with differing benefit from postoperative radiotherapy. Our findings, therefore, conform with large EBCTCG meta-analyses, which have failed to identify subgroups, not accounting for immune phenotype, where the proportional benefits of adjuvant radiotherapy vary (33). To our knowledge, most genomic classifiers, which have been developed to predict radiotherapy benefit, measure primarily tumor intrinsic characteristics. In light of the growing evidence of the immune system being an important determiner of radiotherapy benefit (34, 35), we believe that immunologic biomarkers can complement and improve these genomic classifiers. Our study further supports the use of IHC measurement of TILs as a predictor of adjuvant radiotherapy benefit, which could assist with decision-making in cases with borderline indications (12).

In our study, a low proportion of patients were treated with systemic therapy. Therefore, the rate of recurrence was likely higher than it would have been in the modern setting. In addition, systemic therapy may interact with the local immune infiltrate (36, 37). By including systemic therapy as a covariate, we attempted to adjust for this. However, future studies are needed to adequately assess whether our findings can be generalized to patients treated with systemic therapy. Another limitation is the use of deconvolutional methods to estimate immune cell infiltration from bulk microarray data. Previous studies show that these methods may be unreliable for distinguishing similar cell types (19). However, we believe that the broad classification of prognostic immune cells as protumoral or antitumoral created reliable measurements of the immune infiltrate compared with a transcriptome-based analysis of individual immune cells. Additional limitations are the use of TMAs, which may miss heterogeneity, the subjectiveness in IHC scoring, and our IHC-based method for calculating CD8 and Treg infiltration. Our study included low-risk patients and it is not clear whether our findings can be generalized to high-risk breast cancer. Radiotherapy against a high tumor burden may stimulate the local immune infiltrate (9), and in high-risk tumors, the tumor mutational burden and residual tumor burden after surgery may be greater. In such cases, the chances of adjuvant radiotherapy having a stimulatory effect on a local immune infiltrate may be larger than what was observed in our study. Finally, multiple comparison adjustments for analyses which were not part of the main hypothesis were not performed.

In conclusion, patients with early breast cancer and a favorable immune infiltrate are protected from breast cancer recurrence, but may not derive the same benefit from postoperative radiotherapy. We demonstrate an interaction between immune infiltrate and effect of adjuvant radiotherapy. To our knowledge, our study is among the first to show that the interaction between radiotherapy and the immune system could differ in the adjuvant setting compared with irradiation of a preexisting solid tumor. Our findings may have clinical implications concerning postoperative radiotherapy in breast cancer and could provide information for optimization of radiotherapy fractionation schemes. It remains to be determined whether the addition of checkpoint inhibitors produces different results.

A. Stenmark Tullberg reports other from Prelude Dx outside the submitted work, as well as a patent for biomarker pending. E. Holmberg reports grants from PFS Genomics during the conduct of the study and Prelude Dx outside the submitted work, as well as a patent for biomarker pending. S.L. Chang reports other from PFS Genomics outside the submitted work, as well as a patent for transcriptomic profiling for prognosis of breast cancer to identify subjects who may be spared adjuvant systemic therapy pending and for methods and genomic classifiers for prognosis of breast cancer and predicting benefit from adjuvant radiotherapy pending. F.Y. Feng reports nonfinancial support from PFS Genomics during the conduct of the study; personal fees from Janssen, Blue Earth Diagnostics, Astellas, Myovant, Roivant, Celgene, Genentech, and Bayer; and other from PFS Genomics and Serimmune Inc. outside the submitted work, as well as a patent for transcriptomic profiling for prognosis of breast cancer to identify subjects who may be spared adjuvant systemic therapy pending, methods and genomic classifiers for prognosis of breast cancer and predicting benefit from adjuvant radiotherapy pending, and compositions and methods for the analysis of radiosensitivity issued to PFS Genomics. C. Speers reports other from PFS Genomics outside the submitted work. L.J. Pierce reports other from Up to Date and PFS Genomics outside the submitted work. D. Lundstedt reports grants from Varian Research outside the submitted work. P. Karlsson reports grants from PFS Genomics during the conduct of the study; grants from Prelude Dx outside the submitted work; a patent for biomarker pending; and Roche, conference report. No disclosures were reported by the other authors.

A. Stenmark Tullberg: Conceptualization, formal analysis, investigation, methodology, writing-original draft, writing-review and editing. H.A.J. Puttonen: Investigation, methodology. M. Sjöström: Conceptualization, formal analysis, supervision, investigation, visualization, methodology, writing-review and editing. E. Holmberg: Conceptualization, resources, data curation, software, formal analysis, supervision, investigation, visualization, methodology, writing-review and editing. S.L. Chang: Conceptualization, formal analysis, supervision, methodology, writing-review and editing. F.Y. Feng: Conceptualization, resources, supervision, methodology, writing-review and editing. C. Speers: Resources, supervision, investigation, writing-review and editing. L.J. Pierce: Resources, supervision, methodology, writing-review and editing. D. Lundstedt: Conceptualization, supervision, writing-review and editing. F. Killander: Conceptualization, resources, supervision, methodology, writing-review and editing. E. Niméus: Conceptualization, resources, supervision, methodology, writing-review and editing. A. Kovács: Conceptualization, resources, supervision, investigation, methodology, writing-review and editing. P. Karlsson: Conceptualization, resources, formal analysis, supervision, funding acquisition, investigation, methodology, project administration, writing-review and editing.

P. Karlsson was supported by the Swedish state under the agreement between the Swedish government and the county councils, ALF-agreement grant ALFGBG-716711, Swedish Cancer Society grant Can- 2019/0081, and King Gustav V Jubilee Clinic Foundation grant 2019:248.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
He
B
,
Gao
R
,
Lv
D
,
Wen
Y
,
Song
L
,
Wang
Xi
, et al
The prognostic landscape of interactive biological processes presents treatment responses in cancer
.
EBioMedicine
2019
;
41
:
120
33
.
2.
Loi
S
,
Michiels
S
,
Salgado
R
,
Sirtaine
N
,
Jose
V
,
Fumagalli
D
, et al
Tumor infiltrating lymphocytes are prognostic in triple negative breast cancer and predictive for trastuzumab benefit in early breast cancer: results from the FinHER trial
.
Ann Oncol
2014
;
25
:
1544
50
.
3.
Gu-Trantien
C
,
Loi
S
,
Garaud
S
,
Equeter
C
,
Libin
M
,
de Wind
A
, et al
CD4+ follicular helper T cell infiltration predicts breast cancer survival
.
J Clin Invest
2013
;
123
:
2873
92
.
4.
Ali
HR
,
Provenzano
E
,
Dawson
S-J
,
Blows
FM
,
Liu
B
,
Shah
M
, et al
Association between CD8+ T-cell infiltration and breast cancer survival in 12,439 patients
.
Ann Oncol
2014
;
25
:
1536
43
.
5.
Pruneri
G
,
Vingiani
A
,
Denkert
C
. 
Tumor infiltrating lymphocytes in early breast cancer
.
Breast
2018
;
37
:
207
14
.
6.
Mao
Y
,
Qu
Q
,
Chen
X
,
Huang
Ou
,
Wu
J
,
Shen
K
. 
The prognostic value of tumor-infiltrating lymphocytes in breast cancer: a systematic review and meta-analysis
.
PLoS One
2016
;
11
:
e0152500
.
7.
Loi
S
,
Sirtaine
N
,
Piette
F
,
Salgado
R
,
Viale
G
,
Van Eenoo
F
, et al
Prognostic and predictive value of tumor-infiltrating lymphocytes in a phase III randomized adjuvant breast cancer trial in node-positive breast cancer comparing the addition of docetaxel to doxorubicin with doxorubicin-based chemotherapy: BIG 02–98
.
J Clin Oncol
2013
;
31
:
860
7
.
8.
Hu
ZI
,
Ho
AY
,
McArthur
HL
. 
Combined radiation therapy and immune checkpoint blockade therapy for breast cancer
.
Int J Radiat Oncol Biol Phys
2017
;
99
:
153
64
.
9.
Walle
T
,
Martinez Monge
R
,
Cerwenka
A
,
Ajona
D
,
Melero
I
,
Lecanda
F
. 
Radiation effects on antitumor immune responses: current perspectives and challenges
.
Ther Adv Med Oncol
2018
;
10
:
1758834017742575
.
10.
Formenti
SC
,
Golden
EB
,
Goldberg
JD
,
Li
X
,
Taff
J
,
Fenton-Kerimian
MB
, et al
Results of a phase I-II study of adjuvant concurrent carboplatin and accelerated radiotherapy for triple negative breast cancer
.
Oncoimmunology
2017
;
6
:
e1274479
.
11.
Sjöström
M
,
Chang
SL
,
Fishbane
N
,
Davicioni
E
,
Zhao
SG
,
Hartman
L
, et al
Clinicogenomic radiotherapy classifier predicting the need for intensified locoregional treatment after breast-conserving surgery for early-stage breast cancer
.
J Clin Oncol
2019
;
37
:
3340
9
.
12.
Kovács
A
,
Stenmark Tullberg
A
,
Werner Rönnerman
E
,
Holmberg
E
,
Hartman
L
,
Sjöström
M
, et al
Effect of radiotherapy after breast-conserving surgery depending on the presence of tumor-infiltrating lymphocytes: a long-term follow-up of the SweBCG91RT randomized trial
.
J Clin Oncol
2019
;
37
:
1179
87
.
13.
Malmström
P
,
Holmberg
L
,
Anderson
H
,
Mattsson
J
,
Jönsson
P-E
,
Tennvall-Nittby
L
, et al
Breast conservation surgery, with and without radiotherapy, in women with lymph node-negative breast cancer: a randomised clinical trial in a population with access to public mammography screening
.
Eur J Cancer
2003
;
39
:
1690
7
.
14.
Takada
K
,
Kashiwagi
S
,
Goto
W
,
Asano
Y
,
Takahashi
K
,
Takashima
T
, et al
Use of the tumor-infiltrating CD8 to FOXP3 lymphocyte ratio in predicting treatment responses to combination therapy with pertuzumab, trastuzumab, and docetaxel for advanced HER2-positive breast cancer
.
J Transl Med
2018
;
16
:
86
.
15.
Aran
D
,
Hu
Z
,
Butte
AJ
. 
xCell: digitally portraying the tissue cellular heterogeneity landscape
.
Genome Biol
2017
;
18
:
220
.
16.
Curtis
C
,
Shah
SP
,
Chin
S-F
,
Turashvili
G
,
Rueda
OM
,
Dunning
MJ
, et al
The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups
.
Nature
2012
;
486
:
346
52
.
17.
Schmidt
M
,
Bohm
D
,
von Torne
C
,
Steiner
E
,
Puhl
A
,
Pilch
H
, et al
The humoral immune system has a key prognostic impact in node-negative breast cancer
.
Cancer Res
2008
;
68
:
5405
13
.
18.
van de Vijver
MJ
,
He
YD
,
van 't Veer
LJ
,
Dai
H
,
Hart
AAM
,
Voskuil
DW
, et al
A gene-expression signature as a predictor of survival in breast cancer
.
N Engl J Med
2002
;
347
:
1999
2009
.
19.
Sturm
G
,
Finotello
F
,
Petitprez
F
,
Zhang
JD
,
Baumbach
J
,
Fridman
WH
, et al
Comprehensive evaluation of transcriptome-based cell-type quantification methods for immuno-oncology
.
Bioinformatics
2019
;
35
:
i436
45
.
20.
Ley
K
. 
M1 means kill; M2 means heal
.
J Immunol
2017
;
199
:
2191
3
.
21.
Fine
JP
,
Gray
RJ
. 
A proportional hazards model for the subdistribution of a competing risk
.
J Am Stat Assoc
1999
;
94
:
496
509
.
22.
Sjöström
M
,
Lundstedt
D
,
Hartman
L
,
Holmberg
E
,
Killander
F
,
Kovács
A
, et al
Response to radiotherapy after breast-conserving surgery in different breast cancer subtypes in the Swedish Breast Cancer Group 91 radiotherapy randomized clinical trial
.
J Clin Oncol
2017
;
35
:
3222
3229
.
23.
Teng
F
,
Mu
D
,
Meng
X
,
Kong
L
,
Zhu
H
,
Liu
S
, et al
Tumor infiltrating lymphocytes (TILs) before and after neoadjuvant chemoradiotherapy and its clinical utility for rectal cancer
.
Am J Cancer Res
2015
;
5
:
2064
74
.
24.
Matsutani
S
,
Shibutani
M
,
Maeda
K
,
Nagahara
H
,
Fukuoka
T
,
Nakao
S
, et al
Significance of tumor-infiltrating lymphocytes before and after neoadjuvant therapy for rectal cancer
.
Cancer Sci
2018
;
109
:
966
79
.
25.
Balermpas
P
,
Michel
Y
,
Wagenblast
J
,
Seitz
O
,
Weiss
C
,
Rödel
F
, et al
Tumour-infiltrating lymphocytes predict response to definitive chemoradiotherapy in head and neck cancer
.
Br J Cancer
2014
;
110
:
501
9
.
26.
Demaria
S
,
Formenti
SC
. 
Role of T lymphocytes in tumor response to radiotherapy
.
Front Oncol
2012
;
2
:
95
.
27.
Gupta
A
,
Probst
HC
,
Vuong
V
,
Landshammer
A
,
Muth
S
,
Yagita
H
, et al
Radiotherapy promotes tumor-specific effector CD8+ T cells via dendritic cell activation
.
J Immunol
2012
;
189
:
558
66
.
28.
Wattenberg
MM
,
Fahim
A
,
Ahmed
MM
,
Hodge
JW
. 
Unlocking the combination: potentiation of radiation-induced antitumor responses with immunotherapy
.
Radiat Res
2014
;
182
:
126
38
.
29.
Trowell
OA
. 
The sensitivity of lymphocytes to ionising radiation
.
J Pathol Bacteriol
1952
;
64
:
687
704
.
30.
Wasserman
J
,
Blomgren
H
,
Rotstein
S
,
Petrini
B
,
Hammarström
S
. 
Immunosuppression in irradiated breast cancer patients: in vitro effect of cyclooxygenase inhibitors
.
Bull N Y Acad Med
1989
;
65
:
36
44
.
31.
Grapin
M
,
Richard
C
,
Limagne
E
,
Boidot
R
,
Morgand
V
,
Bertaut
A
, et al
Optimized fractionated radiotherapy with anti-PD-L1 and anti-TIGIT: a promising new combination
.
J Immunother Cancer
2019
;
7
:
160
.
32.
Lee
Y
,
Auh
SL
,
Wang
Y
,
Burnette
B
,
Wang
Y
,
Meng
Y
, et al
Therapeutic effects of ablative radiation on local tumor require CD8+ T cells: changing strategies for cancer treatment
.
Blood
2009
;
114
:
589
95
.
33.
Early Breast Cancer Trialists' Collaborative Group (EBCTCG)
,
Darby
S
,
McGale
P
,
Correa
C
,
Taylor
C
,
Arriagada
R
, et al
Effect of radiotherapy after breast-conserving surgery on 10-year recurrence and 15-year breast cancer death: meta-analysis of individual patient data for 10,801 women in 17 randomised trials
.
Lancet
2011
;
378
:
1707
16
.
34.
Liang
H
,
Deng
L
,
Chmura
S
,
Burnette
B
,
Liadis
N
,
Darga
T
, et al
Radiation-induced equilibrium is a balance between tumor cell proliferation and T cell-mediated killing
.
J Immunol
2013
;
190
:
5874
81
.
35.
Deloch
L
,
Derer
A
,
Hartmann
J
,
Frey
B
,
Fietkau
R
,
Gaipl
US
. 
Modern radiotherapy concepts and the impact of radiation on immune activation
.
Front Oncol
2016
;
6
:
141
.
36.
Denkert
C
,
Loibl
S
,
Noske
A
,
Roller
M
,
Müller
BM
,
Komor
M
, et al
Tumor-associated lymphocytes as an independent predictor of response to neoadjuvant chemotherapy in breast cancer
.
J Clin Oncol
2010
;
28
:
105
13
.
37.
Ono
M
,
Tsuda
H
,
Shimizu
C
,
Yamamoto
S
,
Shibata
T
,
Yamamoto
H
, et al
Tumor-infiltrating lymphocytes are correlated with response to neoadjuvant chemotherapy in triple-negative breast cancer
.
Breast Cancer Res Treat
2012
;
132
:
793
805
.

Supplementary data