1208.Full-Text.Pdf

Published OnlineFirst June 19, 2019; DOI: 10.1158/2159-8290.CD-18-1454

RESEARCH ARTICLE

Interferon Signaling Is Diminished with Age and Is Associated with Immune Checkpoint Blockade Effi cacy in Triple-Negative Breast Cancer

Jaclyn Sceneay 1 ,2 , Gregory J. Goreczny 1 ,2 , Kristin Wilson 1 , Sara Morrow 1 , Molly J. DeCristo 1 ,2 , Jessalyn M. Ubellacker 1 ,2 , Yuanbo Qin 1 ,2 , Tyler Laszewski 1 , Daniel G. Stover 3 , Victor Barrera 4 , John N. Hutchinson 4 , Rachel A. Freedman 5 ,6 , Elizabeth A. Mittendorf 6 ,7 , and Sandra S. McAllister 1 ,2 ,8 ,9

ABSTRACT Immune checkpoint blockade (ICB) therapy, which targets T cell–inhibitory receptors, has revolutionized cancer treatment. Among the breast cancer subtypes, evaluation of ICB has been of greatest interest in triple-negative breast cancer (TNBC) due to its immunogenicity, as evidenced by the presence of tumor-infi ltrating lymphocytes and elevated PD-L1 expression relative to other subtypes. TNBC incidence is equally distributed across the age spectrum, affecting 10% to 15% of women in all age groups. Here we report that increased immune dysfunction with age limits ICB effi cacy in aged TNBC-bearing mice. The tumor microenvironment in both aged mice and patients with TNBC shows decreased IFN signaling and antigen presentation, suggesting failed innate immune activation with age. Triggering innate immune priming with a STING agonist restored response to ICB in aged mice. Our data implicate age-related immune dysfunction as a mechanism of ICB resistance in mice and suggest potential prognostic utility of assessing IFN-related genes in patients with TNBC receiving ICB therapy.

SIGNIFICANCE: These data demonstrate for the fi rst time that age determines the T cell–infl amed phenotype in TNBC and affects response to ICB in mice. Evaluating IFN-related genes from tumor genomic data may aid identifi cation of patients for whom combination therapy including an IFN pathway activator with ICB may be required.

1 Division of Hematology, Department of Medicine, Brigham and Women’s Note: Supplementary data for this article are available at Cancer Discovery Hospital, Boston, Massachusetts. 2 Department of Medicine, Harvard Online (http://cancerdiscovery.aacrjournals.org/). 3 Medical School, Boston, Massachusetts. Division of Medical Oncology, J. Sceneay and G.J. Goreczny contributed equally to this article. Ohio State University Comprehensive Cancer Center, Columbus, Ohio. 4 Department of Biostatistics, Harvard T.H. Chan School of Public Health, Corresponding Author: Sandra S. McAllister, Harvard Medical School, Boston, Massachusetts. 5 Department of Medical Oncology, Dana-Farber Brigham & Women’s Hospital, 4 Blackfan Circle, HIM 742, Boston, MA Cancer Institute, Boston, Massachusetts. 6 Breast Oncology Program, 02115. Phone: 617-525-4929; Fax: 617-355-9093; E-mail: smcallister1@ Dana-Farber/Brigham and Women’s Cancer Center, Boston, Massachusetts. bwh.harvard.edu 7 Division of Breast Surgery, Department of Surgery, Brigham and Cancer Discov 2019;9:1208–27 8 Women’s Hospital, Boston, Massachusetts. Broad Institute of Harvard and doi: 10.1158/2159-8290.CD-18-1454 MIT, Cambridge, Massachusetts. 9 Harvard Stem Cell Institute, Cambridge, Massachusetts. ©2019 American Association for Cancer Research.

1208 | CANCER DISCOVERY SEPTEMBER 2019 www.aacrjournals.org

INTRODUCTION FDA approval of atezolizumab as first-line therapy for metastatic TNBC. ICB represents an improvement over standard Recent clinical successes using immune checkpoint block- therapeutic strategies for TNBC; however, the reasons that ade (ICB) have led to FDA approval of antibody-based ther- ICB therapy has had selected efficacy for patients with TNBC apeutics that target T cell–inhibitory receptors, including (14) are unknown. CTLA4, PD-1, and PD-L1. These therapies have revolution- Age is associated with increasing immune dysfunction that ized cancer treatment, particularly for melanoma and non– includes significant changes to both innate and adaptive immu- small cell lung cancer (NSCLC; refs. 1–5). Focus has recently nity (15–17). Most notably, hematopoiesis is skewed toward mye- shifted to applying ICB to other tumor types including breast loid production whereas lymphopoiesis retracts with age. Some cancer. Breast cancers that do not express hormone receptors of the age-related changes specific to peripheral T-cell popula- (estrogen receptor, progesterone receptor) and lack amplifi- tions include fewer naïve T cells, increased numbers of terminally cation of HER2 are classified as triple-negative breast cancer differentiated memory T cells, reduced expression of the costim- (TNBC), which lacks targeted therapy and is the most aggres- ulatory molecule CD28, and increased exhaustion, evidenced by sive subtype of the disease (6). Evaluation of ICB has been of PD-1 expression (18). Antigen-presenting cells including den- greatest interest in TNBC due to higher numbers of tumor- dritic cells (DC) and macrophages also have decreased activity, infiltrating lymphocytes (TIL) and elevated PD-L1 expression, whereas immunosuppressive populations, such as regulatory which correlate with response to ICB, relative to other breast T cells (Treg) and myeloid-derived suppressor cells (MDSC), have cancer subtypes (7–10). ICB monotherapy has proven effec- increased activity with age in patients with cancer (19). Although tive for approximately 10% of patients with TNBC (11, 12). most aspects of innate and adaptive immunity are different The IMpassion130 trial recently reported that patients with between young and elderly individuals, the clinical implications metastatic TNBC whose tumors express PD-L1 significantly of these age-related changes are unknown. benefited from anti–PD-L1 (atezolizumab) when administered Age-related immune dysfunction could therefore present with the chemotherapy nab-paclitaxel (13), leading to recent particular challenges to efficacy of immunotherapy in breast

SEPTEMBER 2019 CANCER DISCOVERY | 1209

RESEARCH ARTICLE Sceneay et al. cancer, as more than 50% of patients are older than 60 years tumors grew similarly in young and aged BALB/c mice and of age at diagnosis (20). However, most clinical studies either McNeuA tumors grew more slowly in aged than in young FVB exclude or under-enroll older patients (21) and do not include mice (Supplementary Fig. S1C and S1D), suggesting that age- tissue-based analyses that can help predict for efficacy. Given related host physiology may affect tumor growth. the limited number of older patients enrolled on clinical We next assessed tumor-related changes to the peripheral trials evaluating ICB in TNBC, it is not known whether age immune system with age. Relative to cancer-free animals, cir- affects response to therapy. We therefore initiated a study in culating lymphocytes were reduced in young tumor-bearing murine models of TNBC to understand whether age-related mice, and this reduction was even more enhanced in aged immune dysfunction influences ICB efficacy in TNBC. tumor-bearing mice (Fig. 1G and H; Supplementary Fig. S1E). Young and aged tumor-bearing BALB/c and FVB mice exhib- RESULTS ited neutrophilia, which was more pronounced in the aged cohort than in the young cohort (Fig. 1I and J; Supplementary Age-Dependent Effects on Peripheral Immunity, Fig. S1F). In young BALB/c mice, we observed a reduction in TNBC Progression, and Tumor Immune circulating monocytes with tumorigenesis; however, in aged Microenvironment BALB/c mice, which already had lower monocyte counts, these To identify murine models correlating with age-associated levels were unchanged with tumorigenesis (Fig. 1K; Supple- immune changes seen in humans, we first assessed age- mentary Fig. S1G). In the FVB mice, neither age nor tumori- associated differences in peripheral immune cells from young genesis resulted in altered monocyte levels (Fig. 1L). (8–10 weeks) and aged (>12 months) BALB/c and FVB mice. We next examined the TME by IHC. There were no sig- Lymphocytes comprised the majority of circulating white nificant age-dependent differences in the percentage of Ki67+ blood cells (WBC) in both strains of mice, irrespective of age proliferating cells or myofibroblasts α( -SMA) in 4T1 or Met1 (Fig. 1A and B). However, total lymphocyte counts, includ- tumors (Supplementary Fig. S1H and S1I). Macrophage ing both CD4+ and CD8+ T cells, were significantly reduced (F4/80+) infiltration was significantly reduced in 4T1 tumors in both strains of aged mice compared with young mice from aged BALB/c mice relative to young, but unchanged in (Fig. 1A–D). The circulating myeloid cell compartment was Met1 tumors from FVB mice (Supplementary Fig. S1H and significantly expanded in both strains with age (Fig. 1A and S1I). We observed an increase in the overall percentage of B). Within the myeloid compartment, neutrophil counts intratumoral neutrophils (MPO+) in 4T1 tumors compared were elevated in both strains whereas monocyte counts were with Met1 tumors (Supplementary Fig. S1H and S1I), but altered in a strain-specific manner (Fig 1A and B). no difference when comparing young and aged mice in the Given that aged mice reflected the myeloid bias and decreased respective models. All mice also exhibited tumor-dependent lymphocyte production observed in humans with age (22–24), neutrophilia in the blood (Supplementary Fig. S1H and S1I). we proceeded to test whether age also affects tumor progres- Intratumoral infiltration of CD3+ and CD8+ T cells was also sion and development of the tumor microenvironment (TME). unchanged with age in both strains of mice (Fig. 1M–P). To do so, we used 4T1 and Met1 transplantable orthotopic Numbers of Tregs were significantly reduced in 4T1 tumors models of TNBC in BALB/c and FVB mice, respectively. from aged mice and increased in the aged Met1 tumor- In the BALB/c mice, 4T1 mammary carcinoma onset and bearing mice (Fig. 1M–P); nevertheless, these changes were incidence (100%) were similar between young and aged cohorts not sufficient to significantly alter the intratumoral CD8:Treg (Fig. 1E). Tumor growth kinetics were modestly increased in ratio in either strain (Fig. 1M–P). the aged cohort during the early stages of tumor development, but ultimately were similar in both young and aged cohorts Differential Responses to ICB Therapy with Age by late-stage disease (Fig. 1E; Supplementary Fig. S1A). In The age- and tumor-dependent changes to peripheral the FVB mice, Met1 tumor onset and incidence (100%) were immunity and the TME that we observed raised the ques- the same in both cohorts; however, tumor growth rates were tion of whether age influences response to immunotherapy, significantly reduced in the aged mice (Fig. 1F; Supplementary particularly checkpoint blockade. We first verified that the Fig. S1B). Testing various other mammary carcinoma cells 4T1 and Met1 cells express PD-L1 and significantly increased revealed similar strain-specific differences; 4T07 and 67NR cell-surface PD-L1 and MHC I expression in response to IFNγ

Figure 1. Age-dependent effects on hematopoiesis, TNBC tumor growth, and TILs. Hemavet analysis of peripheral blood from young and aged cancer-free (CF) BALB/c (A) and FVB (B) mice. Values represent percentage of total white blood cells (n = 5 mice/cohort). Flow cytometric analysis of peripheral blood CD4+ and CD8+ T lymphocyte populations as a percentage of viable cells in young and aged cancer-free BALB/c (C) or FVB (D) mice; (n = 5 mice/cohort). Growth kinetics of 4T1 tumors in young and aged BALB/c mice (n = 5 mice/cohort; E) and Met1 tumors in young and aged FVB mice (n = 9 mice/cohort; F). Representa- tive of two independent experiments. Hemavet analysis of lymphocytes in the BALB/c (G) and FVB (H) as a percentage of total WBCs in cancer-free and tumor- bearing mice (n = 4–5 BALB/c mice/cohort; n = 10 FVB mice/cohort). Hemavet analysis of neutrophils in the BALB/c (I) and FVB (J) as a percentage of total WBCs in cancer-free and tumor-bearing mice (n = 4–5 BALB/c mice/cohort; n = 10 FVB mice/cohort). Hemavet analysis of monocytes in the BALB/c (K) and FVB (L) as a percentage of total WBCs in cancer-free and tumor-bearing mice (n = 4–5 BALB/c mice/cohort; n = 10 FVB mice/cohort). M, Representa- tive immunofluorescence images of 4T1 tumors from young and aged BALB/c mice stained for CD3 (green), FOXP3 (red), and DAPI (blue).N, Average number of CD3+, CD3+/FOXP3, and CD8+ T cells and ratio of CD8+ T cells:Tregs per 4T1 tumor from young and aged BALB/c mice; n = 11 tumors/cohort from two independent experiments—each data point represents average count from 4–5 fields/tumor.O, Representative immunofluorescence images of Met1 tumors from young and aged FVB mice stained for CD3 (green), FOXP3 (red), and DAPI (blue). P, Average number of CD3+, CD3+/FOXP3, and CD8+ T cells and ratio of CD8+ T cells:Tregs per Met1 tumor from young and aged FVB mice; n = 5–6 tumors/cohort–each data point represents average count from 4–5 fields/tumor. Two-sided unpaired t test (N and P) with Welch correction (A–L). Error bars, SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

1210 | CANCER DISCOVERY SEPTEMBER 2019 www.aacrjournals.org

Age and Response to Immune Checkpoint Blockade in TNBC RESEARCH ARTICLE

A BALB/c C E 4T1 (BALB/c) 800 Young Aged 5 5 ) Young * 3 Aged 4 4 600 3 3 *

2 2 lume (mm 400 * (% of viable) (% of viable) + + 1 24.43% Neutrophils ** 33.6% Neutrophils 1 200 CD8 CD4 69.32% Lymphocytes * 61.13% Lymphocytes 0 0 mor vo 6.06% Monocytes * 3.85% Monocytes Tu 0.16% Eosinophils 1.2% Eosinophils Young Aged Young Aged 0 0.03% Basophils 0.23% Basophils 0 51015202530 Time (days)

B FVB D F Young Aged 2,000 Met1 (FVB)

) Young 50 15 3 * ** * Aged 40 1,500 30 10

lume (mm 1,000 20 (% of viable) (% of viable) 5 + 11.85% Neutrophils * 16.04% Neutrophils 10 + 86.07% Lymphocytes ** 80.6% Lymphocytes 500 mor vo CD4 1.88% Monocytes 2.86% Monocytes 0 CD8 0 0.17% Eosinophils 0.35% Eosinophils Young Aged Young Aged Tu 0.03% Other 0.15% Other 0 0510 15 20 25 30 Time (days)

H I J * L G * **** K 80 * 80 ** 80 8 * 6 ** 100 ** * **** * **** 60 60 60 6 75 ** **** 4 40 ** 40 4 40 50 * 2 20 20 20 2 % Monocytes % Monocytes

25 % Neutrophils % Neutrophils (of total WBCs) (of total WBCs) (of total (of total WBCs) (of total WBCs) (of total (of total WBCs) (of total WBCs) (of total % Lymphocytes % Lymphocytes 0 0 0 0 0 0

Aged CF Aged CF Aged CF Aged 4T1 Aged CF Aged CF Aged CF Young CF Aged 4T1 Young CF Aged Met1 Young CF Young CF Aged Met1 Young CF Aged 4T1 Young CF Aged Met1 Young 4T1 Young Met1 Young 4T1 Young Met1 Young 4T1 Young Met1 Day 0Day 15 Day 0Day 15 Day 0Day 15 Day 0Day 15 Day 0Day 15 Day 0Day 15

M N O P CD3/FOXP3/DAPI 50 40 CD3/FOXP3/DAPI 10 8 40 8 /field /field 30 /field /field 6 + + + + 30 6 20 4 . CD3 . CD8 . CD3 20 4 . CD8 2 Young 10 Young . no . no . no 10 2 . no Av Av Av 00 0 Av 0 Young Aged Young Aged Young Aged YoungAged 4T1

25 20 Met1 6 15

/field /field *

+ *** + 20 15 :Treg :Treg CD3 CD3 + 4 + 10 + 15 + 10 Aged Aged 10 2 5 5 . FOXP3 100 µm . FOXP3 5 100 µm Ratio CD8 Ratio CD8 . no . no 0 0 0 0 Young Aged Young Aged Young Aged YoungAged Av Av

SEPTEMBER 2019 CANCER DISCOVERY | 1211

RESEARCH ARTICLE Sceneay et al.

(Supplementary Fig. S2A–S2D), indicating their potential to BALB/c mice, and 4T1 tumors demonstrated more T-cell be recognized by immune cells (CD8+ T cells). activity in general in the TME compared with Met1 tumors in We generated orthotopic 4T1 and Met1 TNBC in cohorts FVB mice, which grew more slowly and showed very minimal of young (8–10 weeks) and aged (>12 months) mice. Once T-cell and myeloid cell infiltration in the TME (Fig. 1; Sup- tumors reached approximately 5 mm3 in volume, mice were plementary Fig. S1). Accordingly, we collected tumors from randomized into treatment cohorts and administered four young and aged BALB/c mice 19 days after tumor cell injec- doses of anti–PD-L1, anti-CTLA4, or isotype antibody control tion, a time point at which significant differences in response (Supplementary Fig. S2E–S2G). to anti–PD-L1 and anti-CTLA4 were manifest. In four independent experiments in the BALB/c mice, only We first analyzed age- and treatment-associated changes in the young cohort exhibited significantly decreased 4T1 tumor T-cell phenotypes in peripheral blood cells by flow cytometry growth after either anti–PD-L1 or anti-CTLA4 treatment com- (Supplementary Fig. S3A). We focused on analysis of naïve pared with isotype controls; neither treatment was effective T cells (no prior antigen exposure), memory phenotypes in the aged mice (Fig. 2A and B; Supplementary Fig. S2H). In (prior exposure to antigen), and cell-surface checkpoint pro- separate experiments to measure survival, anti–PD-L1 and anti- tein expression (indicating active immune responses as well CTLA4 each improved overall survival of young mice relative to as loss of both proliferative potential and effector function). the young isotype controls; however, neither treatment affected Age alone was associated with approximately 3-fold increased survival of the aged mice (Fig. 2C; Supplementary Fig. S2I). numbers of circulating effector memory (EM) CD8+ cells We had earlier observed that Met1 tumor growth was (CD45+/CD3+/CD8+/CD44+/CD62L−) and approximately attenuated in aged mice relative to young mice (Fig. 1F); 2-fold increases in CD8+/PD-1+ T cells relative to young therefore, we did not directly compare young and aged FVB isotype-treated cohorts (Fig. 3A; Supplementary Fig. S3B). cohorts to one another and instead made comparisons only Anti-CTLA4 treatment resulted in approximately 2-fold within each age cohort. In four independent experiments, increased numbers of circulating naïve CD8+ T cells (CD45+/ anti–PD-L1 and anti-CTLA4 treatment significantly slowed CD3+/CD8+/CD44−/CD62L+) and approximately 2-fold the rate of Met1 tumor growth in young mice, although no increases in PD-1+/CD8+ cells in the young cohort, but not in complete regressions were observed (Fig. 2D and E; Supple- the aged cohort (Fig. 3A; Supplementary Fig. S3B). mentary Fig. S2J). In the aged mice from these experiments, We similarly analyzed spleen cells. Isotype-treated aged anti–PD-L1 treatment resulted in a slight reduction in tumor mice had approximately 4-fold higher numbers of splenic growth, although this was not statistically significant, and CD8+/PD-1+/TIM3+ T cells, approximately 7-fold higher the aged mice did not respond to anti-CTLA4 (Fig. 2F and G; numbers of CD8+ EM cells, and approximately 4-fold higher Supplementary Fig. S2J). In separate experiments to measure numbers of PD-1+ EM cells than young mice (Fig. 3B; Supple- survival, anti–PD-L1 and anti-CTLA4 treatment significantly mentary Fig. S3C). In response to ICB, the numbers of PD-1+- improved overall survival in young mice (Fig. 2H; Supple- naïve T cells (CD45+/CD3+/CD8+/CD44−/CD62L+/PD-1+) mentary Fig. S2K). Aged FVB mice gained significant overall were significantly reduced only in spleens of young mice (Fig. survival benefit from anti–PD-L1 but not anti-CTLA4 therapy 3B; Supplementary Fig. S3C). In response to anti–PD-L1, the (Fig. 2I; Supplementary Fig. S2K). numbers of CD3+/PD-1+/TIM3+ cells were enhanced approxi- To determine whether adaptive immunity is necessary for mately 2.5-fold in the aged mice but not in the young mice response to ICB, we first orthotopically injected 4T1 or (Fig. 3B; Supplementary Fig. S3C). Met1 TNBC cells into young and aged nude mice, which Collectively, our analysis of blood and spleen lymphocytes lack mature T cells, and administered ICB according to our indicated that age is associated with enhanced CD8+ T-cell EM protocol (Supplementary Fig. S2E). Aged nude mice bearing and exhaustion phenotypes, as previously reported in tumor- 4T1 tumors did not respond to anti–PD-L1 treatment (Fig. free aging models of immune dysfunction (25). Importantly, 2J; Supplementary Fig. S2L). The young nude mice showed peripheral T-cell phenotypes with ICB treatment were reflec- a statistically significant reduction in tumor volume with tive of active immune response in young but not aged mice. anti–PD-L1 treatment only at the experimental endpoint; Given these results, we tested the proliferation capacity of however, this reduction was modest relative to immunocom- CD8+ T cells from spleen and lymph nodes of young and aged petent mice (Fig. 2J; Supplementary Fig. S2L). Likewise, both mice. CD8+ T-cell proliferation in response to CD3/CD28 young and aged cohorts of Met1 tumor-bearing nude mice stimulation ex vivo was not significantly different between failed to respond to anti–PD-L1 or anti-CTLA4 treatment cohorts (Fig. 3C). However, the number of checkpoint pro- (Supplementary Fig. S2M–S2O). By neutralizing CD8+ T cells teins (PD-1, CTLA4, LAG3, and/or TIM3) expressed per cell in young BALB/c mice (Supplementary Fig. S2P and S2Q), was significantly increased on CD8+ T cells from aged mice, we validated that CD8+ T cells were necessary for response to but not young mice, after stimulation (Fig. 3D and E). anti–PD-L1 (Fig. 2K; Supplementary Fig. S2R). Among the CD8+ populations from aged mice, PD-1+/TIM3+, PD-1+/CTLA4+, and PD-1+/CTLA4+/TIM3+ cells underwent Age-Dependent Increase in T-cell Exhaustion the greatest expansion in response to stimulation (Fig. 3F). Phenotypes and Restriction of TILs The PD-1+/CTLA4+ CD8 cells from young mice expanded Given that the adaptive immune response is necessary most significantly following stimulation (Fig. 3F). Collec- for effective ICB therapy, we characterized peripheral and tively, these results demonstrated that CD8+ T cells from tumor-infiltrating lymphocytes in 4T1 tumor-bearing mice. aged mice are capable of proliferation but also concurrently We focused efforts on the 4T1 model due to the fact that increase expression of checkpoint proteins more so than tumor growth was similar between young and aged untreated T cells from young mice in response to stimulation.

1212 | CANCER DISCOVERY SEPTEMBER 2019 www.aacrjournals.org

Age and Response to Immune Checkpoint Blockade in TNBC RESEARCH ARTICLE

A B C

) 500 0.8 ** 3 4T1 *** * 100 4T1 Yo ung isotype ** Young isotype 400 Aged isotype * Aged isotype * * 0.6 * Yo ung anti–PD-L1 * * Young anti–PD-L1 * * * * 300 Aged anti–PD-L1 * Aged anti–PD-L1 * Yo ung anti-CTLA4 0.4 Young anti-CTLA4 * 50 * 200 Aged anti-CTLA4 * Aged anti-CTLA4 * 0.2

100 survial Percent Tumor weight (g) Tumor

Tumor volume (mm volume Tumor 0 0 0 0510 15 20 Young Aged Young Aged Yo ung Aged 100 20 30 40 50 60 70 80 Isotype Anti–PD-L1 Anti-CTLA4 Time (days) Time (days) H D E **** ) 800 Met1 1.5 Met1 3 **** Yo ung isotype * 100 * * Young isotype Yo ung anti–PD-L1 * * * * 600 * Young anti–PD-L1 * Yo ung anti-CTLA4 1.0 Young anti-CTLA4

400 50 0.5 200 Percent survival Percent Tumor weight (g) Tumor

Tumor volume (mm volume Tumor 0 0 0 0510 15 20 100 20 30 40

Time (days) Isotype Time (days) Anti–PD-L1Anti-CTLA4 Young F G I

) 800 Met1 1.5 3 Met1 Aged isotype 100 Aged isotype Aged anti–PD-L1 Aged anti–PD-L1 * 600 Aged anti-CTLA4 1.0 Aged anti-CTLA4 400 50 0.5 200 Percent survival Percent Tumor weight (g) Tumor

Tumor volume (mm volume Tumor 0 0 0 0510 15 20 100 20 30 40

Isotype Time (days) Anti–PD-L1Anti-CTLA4 Time (days) Aged J K ) 500 4T1-Nude ) 500 3 3 4T1-BALB/c Yo ung isotype Isotype 400 * 400 * Yo ung anti–PD-L1 * Anti–PD-L1 * * 300 Aged isotype 300 Anti–PD-L1 + anti-CD8 * Aged anti–PD-L1 200 200 100 100

Tumor volume (mm volume Tumor 0 (mm volume Tumor 0 0510 15 20 0510 15 20 25

Time (days) Time (days)

Figure 2. Differential responses to ICB therapy with age. Average growth (A) and endpoint mass (B) of 4T1 tumors from young and aged BALB/c mice treated with either 200 μg isotype control antibody, 200 μg anti–PD-L1, or 200 μg anti-CTLA4 on days 7, 10, 13, and 16 (arrows) postinjection (n = 4–7 mice/cohort); representative of four independent experiments. C, Kaplan–Meier survival analysis of young and aged BALB/c mice bearing 4T1 tumors that were treated with either 200 μg isotype control, 200 μg anti–PD-L1, or 200 μg anti-CTLA4 on days 7, 10, 13, and 16 (arrows) postinjection (n = 6–8 mice/cohort from 1 experiment). Average growth (D) and endpoint mass (E) of Met1 tumors from young FVB mice treated with either 200 μg isotype control, 200 μg anti–PD-L1, or 200 μg anti-CTLA4 on days 7, 10, 13, and 16 (arrows) postinjection (n = 7–8 mice/cohort); representative of four independent experiments. Average growth (F) and endpoint mass (G) of Met1 tumors from aged FVB mice treated with either 200 μg isotype control, 200 μg anti–PD-L1, or 200 μg anti-CTLA4 on days 7, 10, 13, and 16 (arrows) postinjection (n = 7 mice/cohort); representative of four independent experiments. H, Kaplan– Meier survival analysis of young FVB mice bearing Met1 tumors that were treated with either 200 μg isotype control, 200 μg anti–PD-L1, or 200 μg anti- CTLA4 on days 7, 10, 13, and 16 (arrows) postinjection (n = 7–9 mice/cohort). I, Kaplan–Meier survival analysis of aged FVB mice bearing Met1 tumors that were treated with either 200 μg isotype control, 200 μg anti–PD-L1, or 200 μg anti-CTLA4 on days 7, 10, 13, and 16 (arrows) postinjection (n = 6–8 mice/cohort). J, Average growth of 4T1 tumors in young and aged nude mice treated with either 200 μg isotype or 200 μg anti–PD-L1 on days 6, 9, 12, and 15 postinjection (n = 5–10 mice/cohort representing 1 experiment). K, Growth kinetics of 4T1 tumors in young BALB/c mice treated with 100 μg isotype control, 100 μg anti–PD-L1, or 100 μg anti–PD-L1 + 100 μg anti-CD8 according to the dosing regimen in Supplementary Fig. S3E (n = 5–6 mice/cohort representing 1 experiment). Two-way ANOVA followed by the Sidak multiple comparison test (A and J), one-way ANOVA followed by multiple comparison test (B, D, E, G, H, K, and L), and log-rank (Mantel–Cox) test (C, F, I). Error bars, SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

SEPTEMBER 2019 CANCER DISCOVERY | 1213

RESEARCH ARTICLE Sceneay et al.

A Fold change in blood T cells BCFold change in spleen T cells

CD8+ 8 CD8+ 10 100 CD8+ CM CD8+ CM + + 7 9 CD8 EM PD-1 + +

CD8 CM PD-1 w CD8+ EM 8 75 CD8+ EM PD-1+ 6 CD8+ naïve CD8+ revertant CD8+ naïve PD-1+ + 7 CD8+ revertant PD-1 5 + 50

CD8 EM erated CD8) + CD8 naïve + 6 + CD8+ EM PD-1 CD8+ naïve PD-1 4 % CFSE lo 25 + + CD8+ revertant CD8 /PD-1 5 (prolif + − + CD8 /PD-1 3 CD8+ revertant EM PD-1 + + 4 CD8 /TIM3 CD8+ PD-1+ 0 + + + Young Aged CD8 /TIM3 PD-1 2 + + + CD8+/TIM3+ PD-1− CD8 /PD-1 TIM3 3 + + − CD8+/TIM3− PD-1+ CD8 /PD-1 TIM3 − − 1 2 CD8+/TIM3 PD-1 CD8+/PD-1+ TIM3+ CD8+/PD-1+ TIM3− 1 CD8+/TIM3+ Isotype Isotype Anti–PD-L1Anti-CTLA4Anti–PD-L1Anti-CTLA4 Young Aged Isotype Isotype Anti–PD-L1Anti-CTLA4Anti–PD-L1Anti-CTLA4 Young Aged D CD8+ T cells from young BALB/c mice

75.81% 0 Checkpoints 0 Checkpoints 68.68% 18.71% 1 Checkpoint 1 Checkpoint 26.55% 5.46% 2 Checkpoints 2 Checkpoints 4.27% 0.02% 3 Checkpoints 3 Checkpoints 0.46% 4 Checkpoints 0.04%

Unstimulated Stimulated

E CD8+ T cells from aged BALB/c mice

71.56% 0 Checkpoints 0 Checkpoints 61.70% 22.03% 1 Checkpoint 1 Checkpoint 27.64% 6.39% 2 Checkpoints 2 Checkpoints 8.40% 0.02% 3 Checkpoints 3 Checkpoints 2.03% 4 Checkpoints 0.22%

Unstimulated Stimulated

F Young-unstimulated Young-unstimulated Young-unstimulated Aged-unstimulated 25 Aged-unstimulated 5 Aged-unstimulated 100 Young-stimulated Young-stimulated * Young-stimulated Aged-stimulated Aged-stimulated ** *** Aged-stimulated 80 20 ** 4

T cells T cells * T cells + 60 + 15 + 3 ***

40 10 ** 2 20 5 1 *** * ** % of the CD8 % of the CD8 % of the CD8 0 0 0 − − + − + +

− CTLA4 − CTLA4 − CTLA4 + CTLA4 − CTLA4 + CTLA4

− TIM3 − TIM3 − TIM3 − TIM3 − TIM3 − TIM3

− LAG3 + LAG3 − LAG3 + LAG3 + LAG3 + LAG3

PD-1 PD-1 PD-1 PD-1 PD-1 PD-1

0 Checkpoints 1 Checkpoints 2 Checkpoints 3 Checkpoints

Figure 3. Age- and ICB-dependent differences in TILs and exhaustion phenotypes. Heat maps representing fold changes in numbers of cells expressing indicated checkpoint proteins in peripheral blood lymphocytes (A) and splenocytes (B) associated with age and ICB therapy; numbers were analyzed by flow cytometry and are represented relative to the young isotype-treated mice n( = 4–5 mice/cohort). C, Percentage of CD8+ T cells isolated from spleen and lymph nodes of indicated mice that proliferated in vitro 72 hours after stimulation with anti-CD3/CD28 beads (2 T cells:1 bead); n = 3 mice per condition with three technical replicates per mouse. Number of immune checkpoint proteins (PD-1, CTLA4, TIM3, LAG3) expressed per stimulated and unstimulated CD8+ T cell from young (D) and aged (E) mice (n = 3 mice per condition with three technical replicates). F, Repertoire of CD8+ T-cell exhaustion phenotypes (by indicated checkpoint protein profile) resulting from unstimulated and stimulated CD8+ T-cell cultures (n = 3 mice per condition with three technical replicates for stimulated CD8+ T cells). (continued on following page)

1214 | CANCER DISCOVERY SEPTEMBER 2019 www.aacrjournals.org

Age and Response to Immune Checkpoint Blockade in TNBC RESEARCH ARTICLE

G CD3/FOXP3/DAPI Young Aged

H I

120 ** + 40 4T1 * 100 * Young isotype *** Young isotype 30 80 Young anti–PD-L1 ** Young anti–PD-L1 cells/field /FOXP3 + Young anti-CTLA4 + Young anti-CTLA4 60 Aged isotype 20 Aged isotype 40 Aged anti–PD-L1 * Aged anti–PD-L1 Aged anti-CTLA4 cells/field 10 Aged anti-CTLA4

Anti-CTLA4 Anti–PD-L1 Isotype 20 erage CD3 erage CD3 Av 0 Av 0

J CD8 Young Aged Isotype

KL 25 o 4

4T1 * Young isotype Young isotype 20 * Anti–PD-L1 Young anti–PD-L1 3 Young anti–PD-L1

reg rati *

cells/field Young anti-CTLA4 Young anti-CTLA4 :T

+ 15 Aged isotype + 2 ** Aged isotype 10 Aged anti–PD-L1 Aged anti–PD-L1 Aged anti-CTLA4 1 Aged anti-CTLA4 5 erage CD8 Anti-CTLA4 erage CD8 Av 0 Av 0

MN4T1 Young Aged Fold change in expression of immune checkpoints 100 Cold + + 5.0 Minor S/T/P CD8 /PD-1 4.5 Restricted to P + + 75 CD8 CM PD-1 4.0 Medium S/T/P CD8+ EM PD-1+ 3.5 Hot CD8+ naïve PD-1+ 3.0 50 2.5 CD8+ revertant PD-1+

equency (%) 2.0 + +

Fr CD8 /TIM3 25 1.5 CD8+/PD-1+ TIM3+ 1.0 + − + 0.5 CD8 /PD-1 TIM3 0 0.0

Isotype Isotype Isotype Isotype Anti–PD-L1Anti-CTLA4Anti–PD-L1Anti-CTLA4 Anti–PD-L1Anti-CTLA4 Anti–PD-L1Anti-CTLA4 Young Aged

Figure 3. (Continued) G, Representative images of 4T1 tumor sections stained for CD3 (green), FOXP3 (red), and nuclei (blue). Scale bar, 100 μm. Average number of CD3+ lymphocytes (H) and CD3+/FOXP3+ Tregs (I; n = 5–11 4T1 tumors) and 4–5 fields/tumor for each cohort representing two independent experiments. J, Representative images of 4T1 tumor sections stained for CD8 (red) and nuclei stained with hematoxylin (blue). Scale bar, 100 μm. K, Average number of CD8+ lymphocytes (n = 5–10 tumors) and 4–5 fields/tumor for each cohort from two independent experiments.L, Ratio of intratumoral CD8+ lymphocytes to Tregs; n = 5–9 tumors analyzed from two independent experiments. M, CD8+ T-cell localization in 4T1 tumors as blindly assessed by two independent investigators. S, stromally restricted; T, intratumoral; P, peripherally restricted; localization stratified into 5 groups: “cold” (no CD8+ cells), “restricted” (P or S only), “minor infiltration” (S or P> T), “equivalent infiltration” (T≈S), and “hot” (T > S or P); n = 5–10 tumors analyzed per cohort from two independent experiments. N, Heat map representing fold changes in absolute numbers of cells expressing indicated checkpoint proteins in TILs associated with age and treatment; values were determined using flow cytometry and represented relative to the young isotype-treated mice; (n = 4–5 mice/cohort). Two-sided unpaired t-test with Welch correction (A, B, H, I, K, L, and N). Two-sided unpaired Student t test (C–F). Error bars, SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001. CM, central memory (CD45+/ CD3+/CD8+/CD44+/CD62L+); EM, effector memory (CD45+/CD3+/CD8+/CD44+/ CD62L−); naïve (CD45+/CD3+/CD8+/CD44−/CD62L+); revertant (CD45+/CD3+/CD8+/CD44−/CD62L−).

SEPTEMBER 2019 CANCER DISCOVERY | 1215

RESEARCH ARTICLE Sceneay et al.

We next characterized TIL populations [CD3+, FOXP3+ “minor infiltration” (S or P> T), “medium infiltration” (T≈S), (Tregs), and CD8+ T cells] in 4T1 tumor sections by IHC. Total and “hot” (T > S or P). Tumors from both young and aged iso- numbers of CD3+/FOXP3+ Tregs were lower in aged mice and type-treated mice displayed mixed phenotypes by this scoring unchanged with treatment (Fig. 3G–I). Moreover, CD8+ T cells method and were not significantly different from one another in these tumor sections were not elevated with ICB treatment (Fig. 3M). After ICB treatment, tumors from young mice were in the aged mice (Fig. 3J and K). These results translated to a approximately 3-fold enriched for the “medium” phenotype, significantly increased CD8+:Treg ratio in the aged mice (Fig. whereas “cold” and “minor” phenotypes disappeared (Fig. 3L), but when considered collectively, our results suggested 3M). In sharp contrast, tumors from aged 4T1 tumor-bearing this ratio was more a function of lower Treg numbers rather mice treated with anti–PD-L1 or anti-CTLA4 failed to expand than a robust CD8+ T-cell response. In contrast, both CD8+ T the “medium” phenotype, and in fact emergence of “cold” cells and Tregs increased with ICB therapy, particularly with phenotypes lacking TILs became obvious (Fig. 3M). anti-CTLA4 treatment, in young mice (Fig. 3G–L). Further characterization of TILs revealed that PD-1+/TIM3+ We therefore scored intratumoral (T), stromally restricted CD8+ T cells increased approximately 3 to 5-fold with ICB (S), and peripherally restricted (P) localization of CD8+ T cells treatment in young mice (Fig. 3N; Supplementary Fig. S3D). in tumor sections from all cohorts and stratified sections into Although ICB treatment did not enhance the PD-1+/TIM3+ 5 groups: “cold” (no CD8+ cells), “restricted” (P or S only), CD8+ T-cell population in aged mice, these cells were already

A Young PD-L1 vs. young isotype B Young CTLA4 vs. young isotype Enrichment plot: Enrichment plot: Enrichment plot: Enrichment plot: HALLMARK_INTERFERON_GAMMA_RESPONSE HALLMARK_INFLAMMATORY_RESPONSE HALLMARK_INTERFERON_GAMMA_RESPONSE HALLMARK_INFLAMMATORY_RESPONSE 0.40 0.45 0.6 NOM P = 0.000 0.6 NOM P = 0.000 0.35 NOM P = 0.000 0.40 NOM P = 0.000 0.30 FDR q = 0.014 0.35 FDR q = 0.000 0.5 FDR q = 0.000 0.5 FDR q = 0.000 0.25 0.30 0.4 0.4 0.20 0.25 0.3 0.3 0.15 0.20 0.15 0.2 0.2 0.10 ES = 0.46 ES = 0.63 ES = 0.63 ES = 0.39 0.10 0.1 0.1 0.05 0.05

Enrichment score (ES) NES = 1.42 Enrichment score (ES) NES = 1.69 NES = 1.89 NES = 1.88 0.00 0.00 Enrichment score (ES) 0.0 Enrichment score (ES) 0.0

1.5 1.5 “5” (positively correlated) “5” (positively correlated) “3” (positively correlated) “3” (positively correlated) 1.0 1.0 1 1 0.5 0.5 0.0 0.0 0 0 –0.5 Zero cross at 14111 –0.5 Zero cross at 14111 Zero cross at 12111 Zero cross at 12473 –1.0 –1.0 –1 –1 –1.5 “4” (negatively correlated) –1.5 “4” (negatively correlated) “4” negatively correlated) “4” (negatively correlated) –2 –2 02,500 5,000 7,500 10,000 12,500 15,000 17,500 02,500 5,000 7,500 10,000 12,500 15,000 17,500 0 2,500 5,000 7,500 10,000 12,500 15,000 17,500 02,500 5,000 7,500 10,000 12,500 15,000 17,500 ed list metric (Signal2Noise ) ed list metric (Signal2Noise ) ed list metric (Signal2Noise ) ed list metric (Signal2Noise ) Rank in ordered dataset Rank in ordered dataset Rank in ordered dataset Rank in ordered dataset Ranking metric scores Ranking metric scores Ranking metric scores Ranking metric scores Rank Enrichment profile Hits Rank Enrichment profile Hits Rank Rank Enrichment profile Hits Enrichment profile Hits C Aged PD-L1 vs. aged isotype D Aged CTLA4 vs. aged isotype Enrichment plot: Enrichment plot: Enrichment plot: Enrichment plot: HALLMARK_INTERFERON_GAMMA_RESPONSE HALLMARK_INFLAMMATORY_RESPONSE HALLMARK_INTERFERON_GAMMA_RESPONSE HALLMARK_INFLAMMATORY_RESPONSE NOM P = 0.96 0.35 NOM P = 0.348 0.30 ES = 0.321 NOM P = 0.356 0.30 ES = 0.316 NOM P = 0.421 0.25 0.30 0.25 0.25 FDR q = 0.421 0.20 FDR q = 0.96 0.25 FDR q = 0.348 NES = 1.04 FDR q = 0.356 NES = 1.02 0.20 0.20 0.15 0.20 0.10 0.15 0.15 0.15 0.05 0.10 ES = 0.36 0.10 0.10 chment score (ES)

ES = 0.28 0.05 chment score (ES) 0.05 chment score (ES) 0.05 chment score (ES) 0.00 NES = 1.04

En ri 0.00 NES = 0.82 En ri 0.00 En ri 0.00

En ri –0.05

1.5 “2” (positively correlated) 1.5 “2” (positively correlated) 1.5 “0” (positively correlated) 1.5 “0” (positively correlated) 1.0 1.0 1.0 1.0 0.5 0.5 0.5 0.5 0.0 0.0 0.0 Zero cross at 13104 0.0 Zero cross at 13104 Zero cross at 12673 Zero cross at 12673 –0.5 –0.5 –0.5 –0.5 –1.0 –1.0 –1.0 –1.0 “1” (negatively correlated) “1” (negatively correlated) –1.5 “1” (negatively correlated) –1.5 “1” (negatively correlated) –1.5 –1.5 ed list metric (Signal2Noise ) ed list metric (Signal2Noise ) 02,500 5,000 7,500 10,000 12,500 15,000 17,500 ed list metric (Signal2Noise ) 02,500 5,000 7,500 10,000 12,500 15,000 17,500 ed list metric (Signal2Noise ) 02,500 5,000 7,500 10,000 12,500 15,000 17,500 0 2,500 5,000 7,500 10,000 12,500 15,000 17,500 Rank in ordered dataset Rank in ordered dataset Rank in ordered dataset Rank in ordered dataset Rank Rank Enrichment profile Hits Ranking metric scores Rank Enrichment profile Hits Ranking metric scores Rank Enrichment profile Hits Ranking metric scores Enrichment profile Hits Ranking metric scores E Young PD-L1 responders vs. nonresponders F Young CTLA4 responders vs. nonresponders Enrichment plot: Enrichment plot: Enrichment plot: Enrichment plot: HALLMARK_INTERFERON_GAMMA_RESPONSE HALLMARK_INTERFERON_ALPHA_RESPONSE HALLMARK_INTERFERON_GAMMA_RESPONSE HALLMARK_INTERFERON_ALPHA_RESPONSE 0.5 0.5 0.45 0.45 NOM P = 0.000 NOM P = 0.000 0.40 NOM P = 0.000 0.40 NOM P = 0.005 0.4 FDR q = 0.001 0.4 FDR q = 0.004 0.35 FDR q = 0.002 0.35 FDR q = 0.010 0.30 0.30 0.3 0.3 0.25 0.25 0.20 0.20 0.2 0.2 0.15 0.15 ES = 0.50 0.10 ES = 0.46 0.10 0.1 ES = 0.50 0.1 ES = 0.45 NES = 1.54 0.05 NES = 1.74 0.05 Enrichment score (ES) NES = 1.64 Enrichment score (ES) 0.00 Enrichment score (ES) 0.00 NES = 1.59 Enrichment score (ES) 0.0 0.0

“1” (positively correlated) “1” (positively correlated) 1.5 “1” (positively correlated) 1.5 “1” (positively correlated) 1 1 1.0 1.0 0.5 0.5 0 0 Zero cross at 10040 Zero cross at 15040 0.0 Zero cross at 11111 0.0 Zero cross at 11010 –1 –1 –0.5 –0.5 “0” (negatively correlated) “0” (negatively correlated) –1.0 “0” (negatively correlated) –1.0 “0” (negatively correlated) –2 –2 ed list metric (Signal2Noise ) ed list metric (Signal2Noise ) ed list metric (Signal2Noise )

ed list metric (Signal2Noise ) 0 2,500 5,000 7,500 10,000 12,500 15,000 17,500 02,500 5,000 7,500 10,000 12,500 15,000 17,500 02,500 5,000 7,500 10,000 12,500 15,000 17,500 0 2,500 5,000 7,500 10,000 12,500 15,000 17,500 Rank in ordered dataset Rank in ordered dataset Rank in ordered dataset Rank in ordered dataset Rank Rank Rank

Rank Enrichment profile Hits Ranking metric scores Enrichment profile Hits Ranking metric scores Enrichment profile Hits Ranking metric scores Enrichment profile Hits Ranking metric scores

Figure 4. TNBC tumors of aged mice are deficient for IFN production and response and lack an inflamed phenotype.A–F, GSEA of RNA-seq data showing significantly enriched hallmark signatures when comparing anti–PD-L1 to isotype-treated young mice n( = 10 and 7 mice, respectively; A), anti-CTLA4 to isotype-treated young mice (n = 12 and 7 mice, respectively; B), anti–PD-L1 to isotype-treated aged mice (n = 8 and 7 mice, respectively; C), anti- CTLA4 to isotype-treated aged mice (n = 8 and 7 mice, respectively; D), anti–PD-L1 responders to nonresponders in the young cohort (n = 6 and 4 mice, respectively; E), anti-CTLA4 responders to nonresponders in the young cohort (n = 7 and 5 mice, respectively; F). (continued on following page)

1216 | CANCER DISCOVERY SEPTEMBER 2019 www.aacrjournals.org

Age and Response to Immune Checkpoint Blockade in TNBC RESEARCH ARTICLE approximately 7-fold elevated in tumors from aged isotype- (GSE130472). Multiple genes were significantly differentially treated mice relative to young isotype-treated mice (Fig. 3N; expressed between young and aged cohorts in response to Supplementary Fig. S3D). PD-1+ TILs in 4T1 tumors from treatment, and hierarchical clustering of these genes revealed aged mice also demonstrated a memory-like phenotype par- strong associations based on age and treatment, indicating ticularly after ICB therapy (Fig. 3N; Supplementary Fig. S3D). consistency of effect (Supplementary Fig. S4A and S4B; Sup- Overall these data demonstrated that ICB therapy does very lit- plementary File S1). Gene Set Enrichment Analysis (GSEA) tle to change the immune contexture of the TME in aged mice. revealed that some of the enriched gene sets shared in common between young and aged mice included UV response and Age-Dependent Deficiency in IFN epithelial–mesenchymal transition (EMT) with anti–PD-L1 Production and Response (Supplementary Fig. S4C and S4D) and TNFα signaling and To gain further insights into the tumor immune microen- EMT with anti-CTLA4 (Supplementary Fig. S4E and S4F). vironment at the molecular level that could explain lack of The most significantly enriched signatures in young mice response to ICB therapy in aged mice, we performed whole- treated with anti–PD-L1 or anti-CTLA4 compared with the tissue RNA sequencing (RNA-seq) on tumors from young and isotype control cohorts included gene sets involved in IFNγ res aged mice treated with isotype, anti–PD-L1, or anti-CTLA4 ponse and inflammatory response (Fig. 4A and B). In contrast,

G Age Age 3 Treatment Aged B cell 2 Young NK cell 1 Treatment 0 PD-L1 Th1 cell CTLA4 CD8+ T cell –1 Isotype T helper cell –2 T cell –3 Cytotoxic cell Tgd cell DC Mast cell pDC aDC NK CD56dim cell Macrophage Neutophil Th2 cell Eosinophil Tfh Treg TCM TEM iDC NK CD56bright cell Th17 cell Aged_CTLA4_1 Aged_CTLA4_2 Aged_CTLA4_3 Aged_CTLA4_4 Aged_CTLA4_5 Aged_CTLA4_6 Aged_CTLA4_7 Aged_CTLA4_8 Aged_Isotype_1 Aged_Isotype_2 Aged_Isotype_3 Aged_Isotype_4 Aged_Isotype_5 Aged_Isotype_6 Aged_Isotype_7 Aged_PD-L1_1 Aged_PD-L1_2 Aged_PD-L1_3 Aged_PD-L1_4 Aged_PD-L1_5 Aged_PD-L1_6 Aged_PD-L1_7 Aged_PD-L1_8 Young_CTLA4_1 Young_CTLA4_10 Young_CTLA4_11 Young_CTLA4_12 Young_CTLA4_2 Young_CTLA4_3 Young_CTLA4_4 Young_CTLA4_5 Young_CTLA4_6 Young_CTLA4_7 Young_CTLA4_8 Young_CTLA4_9 Young_Isotype_1 Young_Isotype_2 Young_Isotype_3 Young_Isotype_4 Young_Isotype_5 Young_Isotype_6 Young_Isotype_7 Young_PD-L1_1 Young_PD-L1_10 Young_PD-L1_2 Young_PD-L1_3 Young_PD-L1_4 Young_PD-L1_5 Young_PD-L1_6 Young_PD-L1_7 Young_PD-L1_8 Young_PD-L1_9

H CD8+/IFNγ+ I CD8+/GRZB+ 2,500 ** 15,000 * 2,000 10,000 1,500

1,000 5,000 Absolute number Absolute number 500

0 0

Young isotype Aged isotype Young isotype Aged isotype ung anti-CTLA4Aged anti–PD-L1 Aged anti–PD-L1 YoungYo anti–PD-L1 Aged anti-CTLA4 YoungYoung anti–PD-L1 anti-CTLA4Aged anti-CTLA4

Figure 4. (Continued) G, Heat map showing the proportions for different immune cell types (as the average expression of genes associated to the specific cell type) in tumors from the young and old isotype, anti–PD-L1, and anti-CTLA4–treated mice (7–12 tumors/cohort). Columns were clustered with supervision whereas rows were allowed to cluster without supervision. Expression is centered and scaled by row to highlight differences within each cell type. Average number of IFNγ+ (H) and GRZB+ (I) expressing CD8+ TILs; values were determined by flow cytometric intracellular staining n( = 4–5 mice/ cohort). Two-sided unpaired t test with Welch correction (H and I). Error bars, SEM. *, P < 0.05; **, P < 0.01.

SEPTEMBER 2019 CANCER DISCOVERY | 1217

RESEARCH ARTICLE Sceneay et al. these same signatures were not enriched in the tumors from with melanoma (defined as Gao_Signature; ref. 26). Again, aged mice with either ICB treatment (Fig. 4C and D). More we found that tumors from young isotype-treated mice over, when we stratified the young ICB treatment cohorts were enriched for this IFN response pathway compared with into responders and nonresponders, IFNγ response, IFNα tumors from aged mice (Fig. 5B). response, inflammatory response, and allograft rejection Processing and presentation of tumor antigens is also criti- were significantly enriched in the responders (Fig. 4E and F; cal for initiating antitumor immune responses. Therefore, we Supplementary Fig. S4G and S4H). also examined genes associated with antigen processing and We calculated 24 immune cell–related gene expression sig- presentation and found significant differences in expression natures via the nCounter Mouse PanCancer Immune Profil- of a number of these genes between young and aged isotype- ing Panel from NanoString and compared relative activity treated mice (Fig. 5C). of each signature across all samples (see Methods). Tumors Together, these results established that the critical medi- from young mice treated with either anti–PD-L1 or anti- ators of response to ICB therapy could be prospectively CTLA4 were largely enriched for both innate and adaptive identified in tumor gene expression data even prior to treat- immune cell signatures compared with the young isotype- ment. We therefore interrogated the METABRIC database treated mice (Fig. 4G). However, these immune signatures for patients with TNBC and stratified them into two groups were not enriched in the aged mice treated with anti–PD-L1 based on age at initial diagnosis: ≤40 and ≥65 years of age. or anti-CTLA4 compared with the respective isotype-treated Two distinct clusters of the most differentially expressed cohort (Fig. 4G). genes were observed in patients ≤40, whereas these genes did Intracellular flow cytometric analysis of CD8+ TILs was not cluster in the patients ≥65 (Supplementary Fig. S5). consistent with these findings. Specifically, tumors from In patients ≤40, IFNα response, IFNγ response, inflamma- aged mice had approximately 9-fold reduction in CD8+/ tory response, allograft rejection, IL6–JAK–STAT3 signaling, IFNγ+ cells than those from young mice, and numbers of and IL2–STAT5 signaling gene sets were the most highly these cells were not increased with ICB treatment as they enriched relative to those ≥65 (Fig. 5D), similar to our pre- were in young mice (Fig. 4H). Moreover, CD8+ T cells in clinical observations. To further understand which immune tumors from aged mice stored significantly more of the cell subtypes may be associated with these changes, we cytotoxic granule component granzyme B (GrzB) in response applied CIBERSORT to characterize 22 inferred immune to either ICB treatment, which was not evident in the young subsets for each patient (27). In patients ≥65 relative to ICB-treated mice (Fig. 4I). those ≤40, M1 macrophage and plasma cell signatures were Given that IFNγ production and release of GrzB are criti- significantly reduced whereas activated natural killer (NK) cal for antitumor immune responses, our data indicated that cells and M0 macrophage signatures were significantly ICB was less effective at triggering antitumor responses in enriched (Fig. 5E). aged mice. Moreover, our results established an important Taken together, these results established that the immune correlation between expression of inflammatory response and contexture of the TME in TNBC is significantly different IFN signaling pathways in the TME and the ability of TNBC between young and older patients, suggesting that age- tumors to respond to ICB therapy. related immune changes may lead to reduced efficacy of ICB in older individuals. Baseline IFN Production and Response Genes Stratify with Age in Mice and Patients with TNBC Stimulating IFN Signaling Improves Outcome We next asked whether age-dependent differences in and Response to ICB in Aged Mice critical ICB response pathways would be evident even prior Our results raised the question of whether stimulation of to treatment. Principal component analysis (PCA) of the IFN signaling, which was enriched in young mice but lack- RNA-seq data from murine tumors confirmed that tumors ing in aged mice, is sufficient to trigger response to ICB in from young and aged isotype-treated mice formed distinct aged mice. In particular, type I interferons (IFNα and IFNβ) clusters that stratified with age (Fig. 5A). Strikingly, GSEA play an important role in T-cell priming, and our results sug- revealed that the tumors from young isotype-treated mice gested that failed innate immune priming results in the lack were enriched for IFNα, IFNγ, and inflammatory response of response to ICB in aged mice. Activation of the stimulator genes relative to those from aged isotype-treated mice of interferon genes (STING) pathway promotes the transcrip- (Fig. 5B). We also applied a previously reported IFNγ pathway tion of type I IFNs and has previously been reported to pro- signature that defined response to anti-CTLA4 in patients mote antitumor immunity in other preclinical models (28).

Figure 5. Age-dependent deficiencies in IFN production and response, inflammatory responses, and antigen presentation in mice and patients with TNBC. A, PCA of RNA-seq data from 4T1 tumors of young and aged isotype-treated mice (n = 7 tumors/cohort; 1 tumor per mouse). B, GSEA of significantly enriched hallmark signatures in young relative to aged isotype-treated mice n( = 7 tumors/cohort; 1 tumor per mouse). C, The log2 fold change and adjusted P value of differentially expressed genes related to antigen processing and presentation in tumors from young and aged isotype-treated mice. D, GSEA of the significantly enriched Hallmark pathways in tumor tissues from patients with TNBC≤ 40 years (n = 49) relative to patients ≥65 years (n = 68) from the METABRIC database. E, Quantification of the CIBERSORT immune cell signatures applied to the METABRIC database comparing tumors from patients with TNBC ≤40 years (n = 49) relative to those ≥65 years (n = 68). Two-tailed unpaired t test with Welch correction (E). Error bars, SEM. *, P < 0.05.

1218 | CANCER DISCOVERY SEPTEMBER 2019 www.aacrjournals.org

Age and Response to Immune Checkpoint Blockade in TNBC RESEARCH ARTICLE

A B Young isotype vs. aged isotype

Aged Enrichment plot: Enrichment plot: Young HALLMARK_INTERFERON_ALPHA_RESPONSE HALLMARK_INTERFERON_GAMMA_RESPONSE 0.45 0.45 0.40 NOM P < 0.000 0.40 NOM P < 0.000 0.35 0.35 0.30 FDR q < 0.004 0.30 FDR q < 0.000 0.25 0.25 10 0.20 0.20 0.15 0.15 0.10 0.10

Enrichment score (ES) ES = 0.44 ES = 0.45 0.05 Enrichment score (ES) 0.05 0.00 NES = 1.81 0.00 NES = 2.09 riance 3 3 ‘4’ (positively correlated) ‘4’ (positively correlated) 0 2 2 1 1 Zero cross at 7879 Zero cross at 7879 0 0 −1 −1 ‘1’ (negatively correlated) ‘1’ (negatively correlated) −2 −2 02,500 5,000 7,500 10,000 12,500 15,000 17,500 02,500 5,0007,500 10,000 12,500 15,000 17,500 Rank in ordered dataset Rank in ordered dataset Ranked list metric (Signal2Noise) Ranked list metric (Signal2Noise) Ranked PC2: 23% va PC2: −10 Enrichment profile Hits Ranking metric scores Enrichment profile Hits Ranking metric scores Enrichment plot: HALLMARK_INFLAMMATORY_RESPONSE Enrichment: GAO_SIGNATURE 0.25 0.35 NOM P < 0.05 0.30 NOM P < 0.006 0.20 0.25 0.15 FDR q < 0.155 FDR q < 0.006 0.20 0.10 0.15 −20 0.05 0.10 0.00 ES = 0.27 0.05 ES = 0.355 −0.05 0.00 Enrichment score (ES) Enrichment score (ES) 0.10 NES = 1.25 0.05 NES = 1.48 −30 −20 −10 010 − − PC1: 46% variance

3 3 ‘4’ (positively correlated) ‘4’ (positively correlated) 2 2 1 1 c (Signal2Noise) Zero cross at 7879 c (Signal2Noise) Zero cross at 7879 0 0 −1 −1 −2 ‘1’ (negatively correlated) −2 ‘1’ (negatively correlated) C 02,500 5,000 7,500 10,00012,500 15,000 17,500 0 2,500 5,0007,500 10,000 12,500 15,000 17,500 Rank in ordered dataset Rank in ordered dataset Ranked list met ri Ranked list met ri Ranked Young isotype vs. aged isotype Enrichment profile Hits Ranking metric scores Enrichment profile Hits Ranking metric scores

Gene Log2 Padj fold change P-value D Patients with triple-negative breast cancer ≤40 vs. ≥65 H2-DMa 0.599 0.01017 Enrichment plot: Enrichment plot: Enrichment plot: H2-K1 0.817 0.00068 HALLMARK_INTERFERON_ALPHA_RESPONSE HALLMARK_INTERFERON_GAMMA_RESPONSE HALLMARK_INFLAMMATORY_RESPONSE 0.6 0.7 NOM P < 0.000 NOM P < 0.000 0.35 NOM P < 0.003 H2-Q4 0.651 0.016797 0.6 0.5 0.30 FDR q < 0.000 FDR q < 0.000 0.25 FDR q < 0.015 H2-Q6 1.363 5.54E-06 0.5 0.4 0.20 0.4 0.15 H2-Q10 1.190 6.11E-05 0.3 0.10 0.3 0 05 0.2 . H2-T3 1.941 0.003503 0.2 0.00 0.1 −0.15 Enrichment score (ES) Enrichment score (ES) 0.1 ES = 0.70 ES = 0.58 Enrichment score (ES) ES = 0.36 −0.15 H2-T23 0.531 0.018999 0.0 NES = 2.61 0.0 NES = 2.40 −0.15 NES = 1.52 March1 −1.035 0.004837 March8 −0.372 0.046598 Mr1 0.556 0.007117 0.3 ‘Young’ (positively correlated) 0.3 ‘Young’ (positively correlated) 0.3 ‘Young’ (positively correlated) − 0.2 0.2 0.2 0.1 0.1 0.1 Nod1 0.757 0.019319 0.0 Zero cross at 9268 0.0 Zero cross at 9268 0.0 Zero cross at 9268 −0.1 −0.1 −0.1 Psmb9 0.811 0.027553 −0.2 −0.2 −0.2 −0.3 ‘Old’ (negatively correlated) −0.3 ‘Old’ (negatively correlated) −0.3 ‘Old’ (negatively correlated) Rab3c −2.044 0.010653 02,500 5,000 7,500 10,000 12,500 15,000 17,500 20,000 02,500 5,000 7,500 10,000 12,500 15,000 17,500 20,000 02,500 5,000 7,500 10,000 12,500 15,000 17,500 20,000 Rank in ordered dataset Rank in ordered dataset Ranked list metric (Signal2Noise) Ranked Ranked list metric (Signal2Noise) Ranked list metric (Signal2Noise) Ranked Rank in ordered dataset Rab6a −0.750 0.000348 Enrichment profile Hits Ranking metric scoresEnrichment profileHits Ranking metric scores Enrichment profile Hits Ranking metric scores Rab8b −0.729 0.000692 Enrichment plot: Enrichment plot: Enrichment plot: Tap1 0.879 0.025633 HALLMARK_ALLOGRAFT_REJECTION HALLMARK_IL6_JAK_STAT3_SIGNALING HALLMARK_IL2_STAT5_SIGNALING 0.40 0.50 0.35 Tap2 0.814 0.02283 0.45 NOM P < 0.000 0.4 NOM P < 0.002 NOM P < 0.000 0.30 0.40 FDR q < 0.000 FDR q < 0.005 0.25 FDR q < 0.010 0.35 0.3 Tapbp 0.702 0.027864 0.20 0.30 0.2 0.15 0.25 Tapbpl 0.508 0.009019 0.10 0.20 Antigen processing and presentation 0.1 0.05

chment score (ES) 0.15

chment score (ES) 0.00 Treml4 −1.002 0.021431 0.10

En ri 0.0 ES = 0.507 Enrichment score (ES) ES = 0.44 −0.05 ES = 0.38

0.05 En ri −0.10 0.00 0.1 NES = 2.08 − NES = 1.66 −0.15 NES = 1.57

0.3 ‘Young’ (positively correlated) 0.3 ‘Young’ (positively correlated) 0.3 ‘Young’ (positively correlated) 0.2 0.2 0.2 0.1 0.1 0.1 0.0 Zero cross at 9268 0.0 Zero cross at 9268 0.0 Zero cross at 9268 −0.1 −0.1 −0.1 −0.2 −0.2 −0.2 −0.3 ‘Old’ (negatively correlated) −0.3 ‘Old’ (negatively correlated) −0.3 ‘Old’ (negatively correlated) 02,500 5,000 7,500 10,000 12,500 15,000 17,500 20,000 02,500 5,000 7,500 10,000 12,500 15,000 17,500 20,000 02,500 5,000 7,500 10,000 12,500 15,000 17,500 20,000 Rank in ordered dataset Rank in ordered dataset Rank in ordered dataset Ranked list metric (Signal2Noise) Ranked Ranked list metric (Signal2Noise) Ranked Ranked list metric (Signal2Noise) Ranked Enrichment profile Hits Ranking metric scores Enrichment profile Hits Ranking metric scores Enrichment profile Hits Ranking metric scores E 25 ≤40 ≥65 *

20 *

15 *

* 5 Percentage of cell populations Percentage 0

+ T cells + T cells + T cells YδT-cells Monocytes Eosinophils CD8 Neutrophils + naïveNaïve T cells B cells Plasma cells Memory B cells Resting NK cells CD4 Resting mast cells MO macrophages Activated NK cells M2 macrophagesM1 macrophages Resting denritic cells Activated mast cells Follicular helper T cells Activated dendritic cells ated Regulatorymemory CD4 T cells (Tregs) Resting memory CD4 Activ

SEPTEMBER 2019 CANCER DISCOVERY | 1219

RESEARCH ARTICLE Sceneay et al.

AB C ** 800 4T1-aged mice 50 4T1-aged mice *** 100 ) Isotype ** Isotype 3 40 600 Anti–PD-L1 Anti–PD-L1 DMXAA * DMXAA DMXAA + anti–PD-L1 DMXAA + anti–PD-L1 30 4T1-aged mice 400 50 20 Isotype

Anti–PD-L1 **

200 * 10 DMXAA Percent survival Percent

Tumor volume (mm volume Tumor DMXAA + anti–PD-L1 0 0 0 0 246810 12 14 16 18 20 22 24 26 change (end of treatment) Fold 01020304050 Time (days) Time (days)

**** D E **** F 600 4T1-aged mice 100 **** 4T1-aged mice 100 Isotype **** Isotype ) 3

Anti-CTLA4 ** 80 Anti-CTLA4 DMXAA * DMXAA 400 DMXAA + anti-CTLA4 DMXAA + anti-CTLA4 60 50 4T1-aged mice 40 200 Isotype **

Anti-CTLA4 0.054 Percent survival Percent

20 0.06 DMXAA Tumor volume (mm volume Tumor DMXAA + anti-CTLA4 0 0 0 Fold change (end of treatment) Fold 024681012141618202224 01020304050 Time (days) Time (days)

GHI 500 4T1-young mice 40 * 15 ***

) Isotype 3 400 *** Anti-CTLA4 * 30

DMXAA * ** pression pression * 10 * 300 DMXAA + anti-CTLA4 20 200 5 10 100 Tumor volume (mm volume Tumor Relative Ifnb1 ex Relative Relative Ifna1 ex Relative 0 0 0 04 81216 Time (days) Vehicle DMXAA Vehicle DMXAA Vehicle DMXAA Vehicle DMXAA Young Aged Young Aged

Figure 6. Stimulating IFN signaling improves response to ICB and overall survival in aged mice. A, Growth kinetics of 4T1 tumors in aged BALB/c mice treated with isotype (100 μg), anti-CTLA4 (100 μg), DMXAA (250 μg), or the combination DMXAA (250 μg) + anti-CTLA4 (100 μg; n = 5–8 mice/ cohort); DMXAA monotherapy was performed in three independent experiments; combination treatment was performed twice. B, Fold change in tumor volume during the course of indicated treatment (n = 4–8 mice/cohort). C, Kaplan–Meier survival analysis of aged 4T1 tumor-bearing mice treated with isotype, anti-CTLA4, DMXAA, or combination DMXAA + anti-CTLA4 (n = 5–8 mice/cohort). D, Growth kinetics of 4T1 tumors in aged mice treated with isotype (100 μg), anti–PD-L1 (100 μg), DMXAA (250 μg), or the combination DMXAA (250 μg) + anti–PD-L1 (100 μg; n = 5–11 mice/ cohort). E, Fold change in tumor volume during the course of indicated treatment (n = 4–11 mice/cohort). F, Kaplan–Meier survival analysis of aged 4T1 tumor-bearing mice treated with isotype, anti–PD-L1, DMXAA, or combination DMXAA + anti–PD-L1 (n = 5–11 mice/cohort). Growth kinetics (G) of 4T1 tumors from young mice treated with isotype (100 μg), anti-CTLA4 (100 μg), DMXAA (250 μg), or DMXAA (250 μg) + anti-CTLA4 (100 μg). The combination therapy group only received 3 doses of anti-CTLA4 in this experiment (n = 5–7 mice per group, representing one experiment). H and I, qPCR analysis of the expression of Ifna1 (H) and Ifnb1 (I) relative to GAPDH from vehicle (DMSO) or DMXAA (250 μg) treated 4T1 tumor-bearing young and aged mice (n = 6 mice per group, representing one experiment). One-way ANOVA corrected for multiple corrections (A, B, D, E, and G), Two- sided unpaired t test with Welch correction (H and I) and log-rank (Mantel–Cox) test (C and F). Error bars, SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

To enhance IFN signaling, we used intratumoral injection of course of treatment (Fig. 6A, B, D, and E; Supplementary Fig. the mouse-specific STING agonist 5,6-dimethylxanthenone- S6E–S6H) but did not significantly improve overall survival 4-acetic acid (DMXAA). (Fig. 6C and F). Importantly, both DMXAA + anti–PD-L1 We first tested DMXAA monotherapy and combination and DMXAA + anti-CTLA4 combination therapies signifi- with either anti–PD-L1 or anti-CTLA4 in aged BALB/c mice cantly reduced 4T1 tumor growth, resulting in significantly with 4T1 TNBC, according to our protocol (Supplementary improved overall survival of aged mice (Fig. 6A–F; Supple- Fig. S6A–S6D). As observed repeatedly, neither anti-CTLA4 mentary Fig. S6E–S6H). nor anti–PD-L1 monotherapy was efficacious in aged mice Interestingly, in four independent experiments in young (Fig. 6A–F; Supplementary Fig. S6E–S6J). DMXAA mono- mice, DMXAA monotherapy did not alter tumor growth therapy significantly slowed tumor growth only during the relative to the isotype control cohort (Fig. 6G; Supplementary

1220 | CANCER DISCOVERY SEPTEMBER 2019 www.aacrjournals.org

Age and Response to Immune Checkpoint Blockade in TNBC RESEARCH ARTICLE

Isotype Yo ung ICB responders

Inflammatory response IFNγ signaling IFNα signaling IFNγ IFNα/β IFNγ IFNα/β Priming? Priming?

Young CD8+T cell APC CD8+ T-cell CD8+ T cell APC STING activation IFN α /IFN β DMXAA IFNγ signaling IFNα signaling Checkpoint expression Inflammatory response Antigen presentation Aged CD8+ T cell

Priming? Priming?

CD8+ T cell APC CD8+ T- cell APC

Isotype Aged ICB

Figure 7. Proposed model for immune checkpoint blockade in young and aged mice. The TME in young mice with TNBC is characterized by inflammatory and IFN-related pathways that indicate innate immune priming and poise the immune system to respond to ICB. In aged mice, the TME shows a characteristically “cold” immune phenotype with distinct differences in TILs and poor innate immune priming as indicated by reduced expression of IFN- related pathway genes. Subsequently application of ICB fails to generate antitumor immune responses in aged mice. Addition of STING agonists such as DMXAA help to create a TME amenable to ICB therapy with age and suggest analysis of IFN-related genes taken from tumor genomic data may provide prognostic utility in TNBC.

Fig. S6I and S6K). Anti-CTLA4 and the combination DMXAA suggesting efficient innate priming of CD8+ T cells (Fig. 7, + anti-CTLA4 significantly reduced tumor growth to the top left). In young mice, these pathways are further enriched same degree in the young mice, suggesting there is no addi- with ICB therapy and together with increased infiltration tive effect of DMXAA in young mice (Fig. 6G; Supplementary of CD8+ T cells this results in a robust antitumor response Fig. S6J and S6L). (Fig. 7, top right). Tumors from aged mice and patients with To ensure that DMXAA treatment was affecting type I IFN TNBC >65 years lack enrichment for genes associated with expression, we performed qPCR on whole tumors one day antigen presentation, inflammation, or IFN response path- following intratumoral injection of DMXAA (Supplementary ways, suggesting that T-cell priming by antigen-presenting Fig. S6M and S6N). Importantly, expression of both IFNα cells is not as efficient as in young mice (Fig. 7, bottom and IFNβ was significantly enhanced after DMXAA injec- left). Aging also results in an immunologically cold tumor tion in both young and aged mice, with no age-associated microenvironment, defined by peripherally restricted TILs differences in relative expression observed (Fig. 6H and I). that have increased checkpoint protein expression (Fig. 7, Together, these results established that response to ICB in middle). The age-dependent tumor microenvironment fails aged mice with TNBC is rescued by stimulating type I IFNs to generate an antitumor response to ICB (Fig. 7, bottom within the tumor microenvironment. right). Stimulating type I IFN signaling with a STING agonist (DMXAA) promotes response to ICB in the aged mice (Fig. 7, bottom right). DISCUSSION To date, only a handful of studies have investigated age- This work revealed a critical role for physiologic aging in related immune dysfunction as a potential mechanism for reducing ICB efficacy in murine models of TNBC. Tumors resistance to ICB. In advanced NSCLC, where anti–PD-1 anti- from young mice with TNBC and patients with TNBC bodies are now standard-of-care for patients who progress on <40 years display enrichment of genes associated with anti- or after platinum-based chemotherapy, patients > 75 years of gen presentation, inflammation, and IFN response pathways, age demonstrated less benefit after nivolumab therapy compared

SEPTEMBER 2019 CANCER DISCOVERY | 1221

RESEARCH ARTICLE Sceneay et al. with younger patients; however, overall survival (OS) and the authors note the ORR was numerically higher in patients toxicity were not different (29–32). One meta-analysis of data who were >65 or postmenopausal. This difference was not as from more than 5,000 patients treated with ICB found no dif- apparent when a cutoff of >50 was used. Age stratification ference in OS between young and older patients in multiple in the Impassion130 trial showed that both young (18–40 cancer types (33). Another meta-analysis with greater repre- years) and older (>65 years) women performed better with sentation of patients more than 65 years of age also showed combination therapy (18–40: 3.7 vs. 3.6 months; ≥65: 9.1 vs. comparable efficacy of anti–PD-1 or anti–PD-L1 therapy in 6.2 months; ref. 13). However, as with nearly all breast cancer young and older patients (34). In addition, comparative ben- clinical trials, young and older women were underrepresented efit and toxicity analyses have shown only modestly higher (12.6% and 24%, respectively) in this trial. TNBC also presents rates of treatment discontinuation and immune-related unique challenges to ICB therapy due to age-associated dif- adverse events in patients with melanoma older than 80 years ferences in disease progression as well as the composition of of age when treated with ICB (35). Conversely, another recent the tumor microenvironment (42), as we, and others, have study showed increased rates of response to ICB in older reported (43). patients with melanoma (36). Correlative studies from TNBC trials of ICB that include It is worth noting that most of the studies of age and longitudinal molecular and histopathologic assessment have response to ICB, be it patient data or mouse models, have not yet been reported. Therefore, the baseline immune phe- been done in NSCLC or melanoma. These cancer types gen- notype of tumors from the reported studies is unknown, erally have a higher mutational load and increased immu- making interpretation of their results difficult in the context nogenicity compared with breast cancer (37), which, among of our data. Interpretation of these types of analyses are fur- other things, makes them more susceptible to ICB, which ther complicated by the fact that we do not actually know is now part of standard-of-care therapy in these cancers. where the appropriate cutoff for “old” is; therefore, grouping In fact, it is now well accepted that tissue-specific differ- all patients above or below a certain age may mask any actual ences in the immune microenvironment critically affect age-specific effects. In fact, a recent study demonstrated con- cancer progression and therapeutic responses (38). Likewise, siderable differences in the rate of immune aging between mouse strain and life span, tumor models, complexity of healthy individuals (24), supporting the notion that immune the immune microenvironment, and specific ICB therapy age does not necessarily reflect chronologic age, and that in are important considerations when drawing conclusions fact immunologic age may be a better predictor of response from preclinical studies. For example, decreased numbers to immunotherapy. of tumor-infiltrating Tregs were associated with improved The age-stratified responses to ICB that we observed in response to anti–PD-1 therapy in 10-month-old C57BL/6 our TNBC mouse models appear to be due to defective mice in a recent melanoma study (36). However, decreased innate immune priming with age, as tumors from aged mice Tregs in the 4T1 tumors from aged (>12 months) compared (without therapy) showed overall decreased antigen process- with young BALB/c mice did not lead to improved responses ing and presentation as well as decreased IFNα and IFNγ to anti–PD-L1 or anti-CTLA4 in our study. In fact, decreased signaling, resulting in an overall “cold” TME that was not Tregs with age appeared to be a function of a “cold” immune responsive to ICB monotherapy. Only the addition of the environment in our models, which is supported by our STING agonist DMXAA, which drives type I IFN signaling, observation of reduced IFN signatures with age. Impor- showed efficacy in combination with ICB in aged mice. Our tantly, the young mice bearing either 4T1 or Met1 tumors parallel findings in tumors from older patients with TNBC responded to ICB, unlike young mice in the melanoma in the METABRIC cohort suggest that older patients with study (36). Conversely, the 4T1 tumor model results in neu- TNBC exhibit a “cold” TME characterized by poor T-cell trophilia in BALB/c mice, which can also limit the effects infiltrate and a lack of IFN signaling. To the best of our of ICB in general, although we observed that intratumoral knowledge, these data demonstrate for the first time that neutrophils did not change in young compared with aged older age is a defining factor in determining the T cell– mice, whereas response to ICB did. Therefore, although inflamed phenotype in TNBC and could be an important neutrophilia may contribute to the overall limited immu- determinant of ICB therapy response. nogenicity of the 4T1 model, it does not explain the age- Intratumoral expression of IFN pathway genes is closely specific response to ICB we observed, or that intratumoral associated with response to ICB in the clinic. Intratumoral stimulation of type I IFNs could recover responses to ICB in loss of IFNγ signaling is associated with primary resistance aged mice. Clearly, however, the caveats to preclinical mod- to anti-CTLA4 therapy (26), whereas a T cell–inflamed TME els highlight the need for more detailed clinical data from characterized by active IFNγ signaling and antigen presenta- ICB trials in different cancers. tion is a common feature associated with tumors that are The application of ICB to TNBC is still under active inves- responsive to anti–PD-1 (44). In line with these studies, our tigation, with atezolizumab in combination with abraxane data suggest that analysis of IFN-related genes taken from recently receiving FDA approval for first-line treatment of tumor genomic data may provide prognostic utility for metastatic TNBC (13). ICB monotherapy in breast cancer patients selected to receive ICB monotherapy or potentially shows response rates of only 5% to10% in unselected cohorts ICB therapy in combination with an IFN pathway activator, (12, 39, 40). In selected cohorts of patients with PD-L1+ meta- such as a STING agonist, to generate a T-cell response in static TNBC in the phase II KEYNOTE 086 trial, pembroli- “cold” tumors. zumab as a first-line therapy showed an objective response rate Clinical trials utilizing intratumoral injection of synthetic (ORR) of 21.4% (41). Although this result was not significant, CDN STING agonists alone or in combination with anti–PD-1

1222 | CANCER DISCOVERY SEPTEMBER 2019 www.aacrjournals.org

Age and Response to Immune Checkpoint Blockade in TNBC RESEARCH ARTICLE are now under way. One of these trials (NCT03010176) using Cell Lines the STING agonist MK-1454 also includes patients with 4T1 tumor cells were provided by F. Miller (Wayne State University TNBC, and early reports suggest the combination with pem- School of Medicine) and maintained as described previously (48). The brolizumab is more effective than MK-1454 alone, which Met1 TNBC cell line was a gift from J. Joyce (University of Lausanne) lacked efficacy. Likewise, our study indicated that the mouse with permission from A. Borowsky (UC Davis School of Medicine) STING agonist DMXAA as monotherapy did not affect and maintained as described previously (49). The McNeuA cell line tumor growth in young mice, but significantly improved was provided by Michael Campbell (University of California, San response to ICB in aged mice. Given these results, one might Francisco) and maintained as described previously (50). The 4T07 and 67NR cell lines were provided by Robert Weinberg (Whitehead Insti- speculate that efficacy of STING agonists stratifies with tute for Biomedical Research and MIT) and maintained as described age or that patients whose tumors lack enrichment for IFN previously (51). Met1 cells were generated from a spontaneously signatures would benefit from these therapies. Nevertheless, arising tumor in an FVB/N-Tg (MMTV-PyVmT) mouse (49) and 4T1 the implications of our findings for patients with TNBC cells are a variant line from a single spontaneously arising mammary whose tumors reflect the biology we observed have yet to be tumor from a BALB/cfC3H mouse (52). All cells were routinely tested determined. for Mycoplasma and confirmed negative when used in experiments. A critical aspect of effective T-cell responses is antigen For IFNγ stimulation to assess PD-L1 and MHC I expression, presentation, primarily via professional antigen-presenting tumor cells were incubated for 24 hours with 1 ng/mL recombinant cells, including macrophages and DCs. Macrophages have mouse IFNγ (BioLegend, 575304) and expression of cell surface been shown to exhibit decreased antigen presentation and receptors was analyzed by flow cytometry. MHC Class II expression with age (45). Notably, we observed Tumor Models decreased expression of antigen processing and presentation Tumor cells were prepared in sterile PBS and injected orthotopically genes in tumors from aged mice. In line with these observa- into the fourth left mammary fatpad (1 × 105 4T1, 1 × 105 4T07, and tions, we provide gene expression–based evidence of reduced 1 × 105 67NR cells in BALB/c or nude mice; 2 × 105 Met1 or 1 × 105 tumor-infiltrating M1 macrophages in older patients with McNeuA in FVB or nude mice) on day 0. Tumors were measured every TNBC in our analysis of the METABRIC cohort, and in second or third day using digital calipers, and tumor volume was cal- 2 TNBC tumors from aged mice compared with young. The culated as π × length × width /6. For single-agent immune checkpoint ability of DCs to prime T cells (among other functions) has inhibition, 100 or 200 μg anti-CTLA4 (clone 9D9, BioXCell, BE0164), anti–PD-L1 (clone 10F.9G2, BioXCell, BE0101), or isotype (clone LTF-2, also been reported to decrease with age (46). This aspect of BioXCell, BE0090) antibodies in 100 or 200 μL sterile PBS were used. tumor immunology remains to be further investigated in our All ICB mAbs were injected intraperitoneally at three-day intervals for models. a total of four doses, starting at day 7 (4T1 and Met1) post-tumor cell Although ICB therapy presents an emerging treatment injection when tumors reached a volume of no more than approximately option for older patients with TNBC and older patients with 5 mm3 in all mice. For survival studies, mice were considered at an ethical cancer in general in terms of efficacy and safety compared endpoint when the tumors reached 800 mm3 in volume. with chemotherapy, how the aging immune system responds For CD8 T-cell depletion studies, 100 μg anti-CD8β (clone 53-5.8, in cancer and, more specifically, after ICB treatment, remains BioXCell, BE0223) was prepared in 100 μL sterile PBS and was poorly understood. Immune-related adverse events and ICB- administered intraperitoneally on days 6, 7, then weekly until experi- related toxicities may also worsen with age, but the lack of mental endpoint. Single-agent ICB was administered as described above. CD8 T-cell depletion from the blood was confirmed by flow elderly patients enrolled in clinical trials hinders progress in cytometry on day 8 post tumor cell injection. this area. With an aging global population predicted to rise For combination DMXAA and immune checkpoint inhibition from 10% in 2000 to more than 22% by 2050 (47), this is an experiments, 250 μg DMXAA (Invivogen, tlrl-dmx) was injected intra- increasingly pressing concern. Building on the preclinical tumorally on day 10 and 100 μg of either anti–PD-L1 or anti-CTLA4 results and data presented here, it may be possible to strat- was prepared in 100-μL sterile PBS and administered intraperito- ify patients with TNBC into those requiring combination neally every three days for a total of 4 doses starting on day 10 post therapy with an immune-stimulatory agent such as a STING tumor cell injection. For single-agent ICB, 100 μg of ICB mAb was agonist using tumor genomic data for IFN pathway genes, prepared in 100 μL of sterile PBS and injected intraperitoneally start- and therefore make ICB therapy applicable to patients with ing at day 7 post tumor cell injection unless otherwise stated. TNBC of all ages. For RNA-seq experiments, tumor cells were injected bilaterally into the fourth left and right mammary fat pads then treated with 3 doses of ICB. Tumors were excised one day after the third dose of treatment (day 14) to ensure the capture of an antitumor response. The mice were METHODS considered nonresponders when tumor growth increased regardless of age, partial responders when tumor growth remained similar with treat- Animals ment, and responders when tumor growth reduced with treatment. Female BALB/c ByJ, FVB/NJ, C57BL/6, and NCR-NU (nude) mice were purchased from Jackson Labs. BALB/c mice were aged Hemavet in house or provided by the National Institute on Aging at 12 Blood was collected by either retro-orbital or a terminal intracar- months of age. FVB/NJ and NCR-NU mice were aged in house. diac bleed into EDTA-coated tubes. Blood samples were analyzed For all experiments, young mice were defined as 8 to 12 weeks on a Hemavet 950 (Drew Scientific) according to the manufacturer’s of age; aged mice as >12 months of age. All experiments were instructions. conducted in accordance with regulations of the Children’s Hos- pital Institutional Animal Care and Use Committee (protocol # Flow Cytometry 12–11-2308R) and the Brigham and Women’s Hospital (protocol Tumors were excised on day 19 of tumor growth after receiving # 2017N000056). 4 doses of ICB, cut into small pieces, and digested with 0.2 mg/mL

SEPTEMBER 2019 CANCER DISCOVERY | 1223

RESEARCH ARTICLE Sceneay et al. collagenase type IV (Worthington Biochemical Corp. LS004186) nextgen.readthedocs.org/en/latest/) and custom R scripts. Reads and 0.02 mg/mL DNAse I (Roche, 04536282001) for 1 hour at were aligned to the Genome Reference Consortium Mouse Refer- 37°C with agitation. Single-cell suspensions were generated from ence build number 38 (GRCm38) of the mouse genome (aka mm10) tumors, blood, and spleens and incubated with anti-CD16/32 for augmented with transcript information from Ensembl using STAR 20 minutes on ice to block Fc receptors, and relevant antibodies for (56) with soft trimming enabled to remove adapter sequences, other 30 minutes on ice. For intracellular staining, samples were fixed and contaminant sequences such as poly-A tails and low Phred quality permeabilized using the Foxp3 Fixation/Permeabilization kit (eBio- score sequences. Counts of reads aligning to known genes were gen- science, 00-5523-00) according to the manufacturer’s instructions. erated by featureCounts (57) for use in quality-control measures. In Nonviable cells were excluded on the basis of staining with 7-amino- parallel, transcripts per million measurements per isoform were gen- actinomycin D (BioLegend, 420404), or Zombie Yellow for intracel- erated by quasi-alignment using Salmon (58) for use in clustering lular staining (BioLegend, 423104). CountBright absolute counting and differential expression analyses. STAR alignments were checked beads (Life Technologies, C36950) were used to calculate absolute for evenness of coverage, ribosomal RNA content, genomic context cell numbers according to the manufacturer’s instructions. Flow of alignments (for example, alignments to exons and introns), com- cytometry acquisition was carried out on a FACS Fortessa 4 or LSR II plexity, and other quality checks using a combination of FastQC, (BD Biosciences). Data analysis was performed using FlowJo software Qualimap (59), MultiQC (https://github.com/ewels/MultiQC), and (Tree Star). Antibody details are listed in Supplementary Table S1. custom tools. Samples were clustered in an unsupervised manner by both PCA and hierarchical means using rlog-transformed reads T-cell Stimulation to identify potential outliers and technical artefacts. All samples CD8+ T cells were isolated from the spleens and lymph nodes of passed these quality-control measures. Differential expression at the young and aged mice using autoMACS and the CD8 T Cell Negative gene level was called with DESeq2 (60), using the counts per gene Selection Kit (Miltenyi Biotec, 130-104-075) according to the manu- estimated from the Salmon quasi-alignments by tximport (61) as facturer’s instructions. T cells were labeled with 5(6)-carboxyfluores- quantitating at the isoform level has been shown to produce more cein diacetate N-succinimidyl ester (Sigma, 21888) for 10 minutes accurate results at the gene level. at room temperature and 1 × 105 cells incubated at a 2:1 ratio with Heat maps of expression levels were generated with pheatmap (62), CD3/CD28 magnetic beads (Dynabeads, Gibco, 11452D) for 72 centered (mean of row subtracted from all values) and scaled (all hours before analysis by flow cytometry. centered values divided by the SD of the row), and then clustered via the ward.D2 algorithm (63). IHC and Microscopy Cutoff–free GSEA was also performed against Gene Ontology Dissected tissues were fixed in 4% paraformaldehyde for 24 hours, terms, restricting the analysis to those terms with at least 10 stored in 70% ethanol for 24 hours, paraffin-embedded, and sec- genes, using clusterProfiler (64) and the ranked log2 fold changes tioned. Sectioning and hematoxylin and eosin, F4/80, Ki67, and of all genes from the DESeq2 analyses. Functional gene sets were MPO staining were performed by the DF/HCC Specialized Histopa- considered enriched if their FDR-adjusted P value was less than thology Core at Brigham and Women’s Hospital. IHC was performed 0.05. as described previously (53), using citrate buffer for antigen retrieval The nCounter Mouse PanCancer Immune Profiling Panel from and staining with primary antibodies at 4°C overnight. Images were NanoString was used for the relative expression of immune cell sig- captured using a Nikon Eclipse Ni microscope, and a minimum of natures. For each cell type, the expression of all the associated genes four images per tumor were analyzed by two independent investiga- was averaged and used as cell type enrichment measure. These values tors in a blinded fashion. CD3+ T cells, CD8+ T cells, and FOXP3+ were used for the heat map plots. Tregs were counted using ImageJ; αSMA, F4/80, and MPO stains GSEA was performed using the Hallmark gene set using version 3 using CellProfiler as described previously (54); and Ki67 staining of the GSEA software (Broad). The number of permutations was set using the ImmunoRatio program (55). Antibody details are listed in to 1,000 and the permutation type was gene-set. Results were consid- Supplementary Table S1. ered significant with aP value and FDR value under 0.05. The localization of CD8+ T cells was scored blindly by two independent investigators on a 0–5 scale where 0 is cold (no TILs), 1 is qPCR restricted (peripheral or stromal only), 2 is minor infiltration (S or RNA was isolated from whole tumors as described above, and the P > T), 3–4 is medium infiltration (T≈S), and 5 is hot (significant TILs). concentration was determined using a Nanodrop (Thermo Fisher Sci- entific). cDNA was synthesized using the iScript Reverse Transcription RNA Isolation and Sequencing Supermix for qRT-PCR according to the manufacturer’s specifica- RNA was extracted from tumors no larger than approximately 150 tions (Bio-Rad, 170-8841) and generated on a T100 Thermal Cycler mm3 and stored in 1 mL RNALater (Qiagen, 1018087) at 4°C over- (Bio-Rad). qPCR amplification was performed using iTaq Universal night before being transferred to −80°C. Tumors were homogenized SYBR Green supermix (Bio-Rad, 1725124) and run on an ABI Prism using a homogenizer 150 (Thermo Fisher Scientific) in RLT buffer 7900HT Sequence Detection System (Applied Biosystems). Relative (Qiagen, 79216), processed through QIAshredder columns (Qiagen, gene expression was quantified using the ΔΔCt method with the follow- 79656), and then total RNA was isolated using an RNeasy Mini Kit ing primers: IFNA1: (forward-5′-GCTAGGCTCTGTGCTTTCCT-3′; (Qiagen, 74136) according to the manufacturer’s protocol. Quality of reverse-5′-TCCTGCGGGAATCCAAAGTC-3′); IFNB1: (forward-5′- total RNA was determined using a 2100 Bioanalyzer (Agilent). RNA- ACCACAGCCCTCTCCATCAA-3′; reverse-5′-TTGAAGTCCGCCCTG seq was performed using paired-end sequencing with a read length of TAGGT); STING: (forward-5′-CCAAGAACCCACAGACGGAA-3′; 2 × 75bp on the Illumina HiSeq 2000 platform by the Dana-Farber reverse-5′-GCTCTTGGGACAGTACGGAG-3′); GAPDH (forward-5′- Cancer Institute Molecular Biology Facility. All data are deposited in GGTGAAGGTCGGTGTGAACG-3′; reverse-5′-CTCGCTCCTGGAA the Gene Expression Omnibus (GSE130472). GATGGTG-3′). All reactions were performed in duplicate and expression levels were normalized to the housekeeping gene GAPDH. RNA-seq Differential Expression and Functional Enrichment Analysis Human Data Analysis All samples were processed and analyzed using an RNA-seq Normalized gene expression data and paired clinical feature data pipeline implemented in the bcbio-nextgen project (https://bcbio- were obtained from the METABRIC European Genome-phenome

1224 | CANCER DISCOVERY SEPTEMBER 2019 www.aacrjournals.org

Age and Response to Immune Checkpoint Blockade in TNBC RESEARCH ARTICLE

Archive (IDs EGAD00010000210 and EGAD0001000021). TNBC Received December 12, 2018; revised April 16, 2019; accepted June was defined as tumors negative for estrogen receptor, progester- 14, 2019; published first June 19, 2019. one receptor, and HER2 amplification by SNP6 analyses. Patients were stratified into “40 and younger” (age at diagnosis≤ 40) versus “65 and over” (age at diagnosis ≥65). GSEA was performed using the Hallmark gene set using version 3 of the GSEA software REFERENCES (Broad). The number of permutations was set to 1,000 and the . 1 Topalian SL, Sznol M, McDermott DF, Kluger HM, Carvajal RD, permutation type was gene-set. Results were considered signifi- Sharfman WH, et al. Survival, durable tumor remission, and long- cant with a P value and FDR value under 0.05. Inferred immune term safety in patients with advanced melanoma receiving nivolumab. subset proportion was determined via CIBERSORT (27), using the J Clin Oncol 2014;32:1020–30. LM22 dataset. 2. Powles T, Eder JP, Fine GD, Braiteh FS, Loriot Y, Cruz C, et al. MPDL3280A (anti-PD-L1) treatment leads to clinical activity in metastatic bladder cancer. Nature 2014;515:558–62. Statistical Analysis 3. Garon EB, Rizvi NA, Hui R, Leighl N, Balmanoukian AS, Eder JP, et al. Pembrolizumab for the treatment of non-small-cell lung cancer. The data are represented as mean ± SEM. The data was analyzed N Engl J Med 2015;372:2018–28. by two-way ANOVA with the Sidak multiple comparison correc- 4. Tumeh PC, Harview CL, Yearley JH, Shintaku IP, Taylor EJ, Robert tion, one-way ANOVA with Tukey multiple comparison test for L, et al. PD-1 blockade induces responses by inhibiting adaptive significance, or an unpaired two-tailedt test with Welch correction immune resistance. Nature 2014;515:568–71. as indicated in the figure legends. Kaplan–Meier survival curves 5. Larkin J, Chiarion-Sileni V, Gonzalez R, Grob JJ, Cowey CL, Lao CD, were analyzed using log-rank (Mantel–Cox) test. All the data were et al. Combined nivolumab and ipilimumab or monotherapy in analyzed using GraphPad Prism 7.0. P < 0.05 was considered statis- untreated melanoma. N Engl J Med 2015;373:23–34. tically significant. 6. Schneider BP, Winer EP, Foulkes WD, Garber J, Perou CM, Richardson A, et al. Triple-negative breast cancer: risk factors to potential targets. Clin Cancer Res 2008;14:8010–8. Disclosure of Potential Conflicts of Interest 7. Mittendorf EA, Philips AV, Meric-Bernstam F, Qiao N, Wu Y, E.A. Mittendorf is a consultant/advisory board member for Merck, Harrington S, et al. PD-L1 expression in triple-negative breast cancer. Genomic Health, Sellas Lifesciences, AstraZeneca, and Roche/Genen- Cancer Immunol Res 2014;2:361–70. tech. No potential conflicts of interest were disclosed by the other 8. Loi S, Michiels S, Salgado R, Sirtaine N, Jose V, Fumagalli D, authors. 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: Authors’ Contributions 1544–50. Conception and design: J. Sceneay, G.J. Goreczny, S.S. McAllister 9. Adams S, Gray RJ, Demaria S, Goldstein L, Perez EA, Shulman LN, Development of methodology: J. Sceneay, G.J. Goreczny, T. Laszewski, et al. Prognostic value of tumor-infiltrating lymphocytes in triple- negative breast cancers from two phase III randomized adjuvant J.N. Hutchinson breast cancer trials: ECOG 2197 and ECOG 1199. J Clin Oncol 2014; Acquisition of data (provided animals, acquired and managed 32:2959–66. patients, provided facilities, etc.): G.J. Goreczny, S. Morrow, 10. Denkert C, von Minckwitz G, Darb-Esfahani S, Lederer B, Heppner M.J. DeCristo, Y. Qin, T. Laszewski BI, Weber KE, et al. Tumour-infiltrating lymphocytes and progno- Analysis and interpretation of data (e.g., statistical analysis, sis in different subtypes of breast cancer: a pooled analysis of 3771 biostatistics, computational analysis): J. Sceneay, G.J. Goreczny, patients treated with neoadjuvant therapy. Lancet Oncol 2018;19: K. Wilson, T. Laszewski, D.G. Stover, V. Barrera, J.N. Hutchinson, 40–50. R.A. Freedman, E.A. Mittendorf, S.S. McAllister 11. Kwa MJ, Adams S.Checkpoint inhibitors in triple-negative breast Writing, review, and/or revision of the manuscript: J. Sceneay, cancer (TNBC): Where to go from here. Cancer 2018;124:2086–103. G.J. Goreczny, D.G. Stover, V. Barrera, J.N. Hutchinson, R.A. Freedman, 12. Dirix LY, Takacs I, Jerusalem G, Nikolinakos P, Arkenau HT, Forero- E.A. Mittendorf, S.S. McAllister Torres A, et al. Avelumab, an anti-PD-L1 antibody, in patients Administrative, technical, or material support (i.e., reporting with locally advanced or metastatic breast cancer: a phase 1b or organizing data, constructing databases): J.M. Ubellacker, JAVELIN Solid Tumor study. Breast Cancer Res Treat 2018;167: J.N. Hutchinson 671–86. Study supervision: S.S. McAllister 13. Schmid P, Adams S, Rugo HS, Schneeweiss A, Barrios CH, Iwata H, Other (conducted experiments): K. Wilson et al. Atezolizumab and nab-paclitaxel in advanced triple-negative breast cancer. N Engl J Med 2018;379:2108–21. 14. Chawla A, Philips AV, Alatrash G, Mittendorf E. Immune check- Acknowledgments points: a therapeutic target in triple negative breast cancer. Oncoim- The authors would like to thank the Dana-Farber Cancer Insti- munology 2014;3:e28325. tute Molecular Biology Facility for RNA-seq sample acquisition, 15. Rossi DJ, Bryder D, Weissman IL. Hematopoietic stem cell aging: mechanism and consequence. Exp Gerontol 2007;42:385–90. and the Harvard School of Public Health for bioinformatic sup- 16. Beerman I, Bhattacharya D, Zandi S, Sigvardsson M, Weissman port. J. Sceneay was a Glenn Foundation Fellow of the Life Sciences IL, Bryder D, et al. Functionally distinct hematopoietic stem cells Research Foundation. This work was also supported by funds modulate hematopoietic lineage potential during aging by a mecha- from the NIH NCI R21CA182614 and the Department of Defense nism of clonal expansion. Proc Natl Acad Sci U S A 2010;107: BCMRP Ear of Hope Scholar Award W81XWH-14-1-0191 (to 5465–70. S.S. McAllister). 17. Sudo K, Ema H, Morita Y, Nakauchi H. Age-associated characteristics of murine hematopoietic stem cells. J Exp Med 2000;192:1273–80. The costs of publication of this article were defrayed in part by 18. Pawelec G. Age and immunity: what is “immunosenescence”? Exp the payment of page charges. This article must therefore be hereby Gerontol 2018;105:4–9. marked advertisement in accordance with 18 U.S.C. Section 1734 19. Pawelec G. Does patient age influence anti-cancer immunity? Semin solely to indicate this fact. Immunopathol 2018;41:125–31.

SEPTEMBER 2019 CANCER DISCOVERY | 1225

RESEARCH ARTICLE Sceneay et al.

20. Hutchins LF, Unger JM, Crowley JJ, Coltman CA Jr., Albain KS. 39. Emens LA, Cruz C, Eder JP, Braiteh F, Chung C, Tolaney SM, et al. Underrepresentation of patients 65 years of age or older in cancer- Long-term clinical outcomes and biomarker analyses of atezolizumab treatment trials. N Engl J Med 1999;341:2061–7. therapy for patients with metastatic triple-negative breast cancer: a 21. Freedman RA, Foster JC, Seisler DK, Lafky JM, Muss HB, Cohen HJ, phase 1 study. JAMA Oncol 2018;5:74–82. et al. Accrual of older patients with breast cancer to alliance systemic 40. Adams S, Schmid P, Rugo HS, Winer EP, Loirat D, Awada A, et al. therapy trials over time: protocol A151527. J Clin Oncol 2017;35: Pembrolizumab monotherapy for previously treated metastatic 421–31. triple-negative breast cancer: cohort A of the phase II KEYNOTE-086 22. Beerman I, Maloney WJ, Weissmann IL, Rossi DJ. Stem cells and study. Ann Oncol 2019;30:397–404. the aging hematopoietic system. Curr Opin Immunol 2010;22: 41. Adams S, Loi S, Toppmeyer D, Cescon DW, De Laurentiis M, Nanda 500–6. R, et al. Pembrolizumab monotherapy for previously untreated, 23. Pang WW, Price EA, Sahoo D, Beerman I, Maloney WJ, Rossi DJ, et al. PD-L1-positive, metastatic triple-negative breast cancer: Cohort Human bone marrow hematopoietic stem cells are increased in fre- B of the phase 2 KEYNOTE-086 study. Ann Oncol 2018;30: quency and myeloid-biased with age. Proc Natl Acad Sci U S A 2011; 405–11. 108:20012–7. 42. Azim HA Jr, Michiels S, Bedard PL, Singhal SK, Criscitiello C, 24. Alpert A, Pickman Y, Leipold M, Rosenberg-Hasson Y, Ji X, Gaujoux Ignatiadis M, et al. Elucidating prognosis and biology of breast cancer R, et al. A clinically meaningful metric of immune age derived arising in young women using gene expression profiling. Clin Cancer from high-dimensional longitudinal monitoring. Nat Med 2019;25: Res 2012;18:1341–51. 487–95. 43. Marsh T, Wong I, Sceneay J, Barakat A, Qin Y, Sjodin A, et al. Hemat- 25. Lee KA, Shin KS, Kim GY, Song YC, Bae EA, Kim IK, et al. Char- opoietic age at onset of triple-negative breast cancer dictates disease acterization of age-associated exhausted CD8(+) T cells defined aggressiveness and progression. Cancer Res 2016;76:2932–43. by increased expression of Tim-3 and PD-1. Aging Cell 2016;15: 44. Ayers M, Lunceford J, Nebozhyn M, Murphy E, Loboda A, Kauf- 291–300. man DR, et al. IFN-gamma-related mRNA profile predicts clinical 26. Gao J, Shi LZ, Zhao H, Chen J, Xiong L, He Q, et al. Loss of IFN- response to PD-1 blockade. J Clin Invest 2017;127:2930–40. gamma pathway genes in tumor cells as a mechanism of resistance to 45. Linehan E, Fitzgerald DC. Ageing and the immune system: focus on anti-CTLA-4 therapy. Cell 2016;167:397–404. macrophages. Eur J Microbiol Immunol 2015;5:14–24. 27. Newman AM, Liu CL, Green MR, Gentles AJ, Feng W, Xu Y, et al. 46. Agrawal A, Agrawal S, Gupta S. Role of dendritic cells in inflamma- Robust enumeration of cell subsets from tissue expression profiles. tion and loss of tolerance in the elderly. Front Immunol 2017;8:896. Nat Methods 2015;12:453–7. 47. Dorshkind K, Montecino-Rodriguez E, Signer RA. The ageing 28. Corrales L, Glickman LH, McWhirter SM, Kanne DB, Sivick KE, immune system: is it ever too old to become young again? Nat Rev Katibah GE, et al. Direct Activation of STING in the tumor micro- Immunol 2009;9:57–62. environment leads to potent and systemic tumor regression and 48. Miller FR. Tumor subpopulation interactions in metastasis. Invasion immunity. Cell Rep 2015;11:1018–30. Metast 1983;3:234–42. 29. Brahmer J, Reckamp KL, Baas P, Crino L, Eberhardt WE, 49. Borowsky AD, Namba R, Young LJ, Hunter KW, Hodgson JG, Poddubskaya E, et al. Nivolumab versus docetaxel in advanced Tepper CG, et al. Syngeneic mouse mammary carcinoma cell lines: squamous-cell non-small-cell lung cancer. N Engl J Med 2015;373: two closely related cell lines with divergent metastatic behavior. Clin 123–35. Exp Metast 2005;22:47–59. 30. Borghaei H, Paz-Ares L, Horn L, Spigel DR, Steins M, Ready NE, et al. 50. Campbell MJ, Wollish WS, Lobo M, Esserman LJ. Epithelial and Nivolumab versus docetaxel in advanced nonsquamous non-small- fibroblast cell lines derived from a spontaneous mammary carcinoma cell lung cancer. N Engl J Med 2015;373:1627–39. in a MMTV/neu transgenic mouse. In Vitro Cell Dev Biol Anim 31. Herbst RS, Baas P, Kim DW, Felip E, Perez-Gracia JL, Han JY, et al. 2002;38:326–33. Pembrolizumab versus docetaxel for previously treated, PD-L1- 51. Yang J, Mani SA, Donaher JL, Ramaswamy S, Itzykson RA, Come C, positive, advanced non-small-cell lung cancer (KEYNOTE-010): a et al. Twist, a master regulator of morphogenesis, plays an essential randomised controlled trial. Lancet 2016;387:1540–50. role in tumor metastasis. Cell 2004;117:927–39. 32. Rittmeyer A, Barlesi F, Waterkamp D, Park K, Ciardiello F, von Pawel 52. Aslakson CJ, Miller FR. Selective events in the metastatic pro- J, et al. Atezolizumab versus docetaxel in patients with previously cess defined by analysis of the sequential dissemination of sub- treated non-small-cell lung cancer (OAK): a phase 3, open-label, mul- populations of a mouse mammary tumor. Cancer Res 1992;52: ticentre randomised controlled trial. Lancet 2017;389:255–65. 1399–405. 33. Nishijima TF, Muss HB, Shachar SS, Moschos SJ. Comparison of 53. McAllister SS, Gifford AM, Greiner AL, Kelleher SP, Saelzler MP, Ince efficacy of immune checkpoint inhibitors (ICIs) between younger and TA, et al. Systemic endocrine instigation of indolent tumor growth older patients: a systematic review and meta-analysis. Cancer Treat requires osteopontin. Cell 2008;133:994–1005. Rev 2016;45:30–7. 54. Elkabets M, Gifford AM, Scheel C, Nilsson B, Reinhardt F, Bray MA, 34. Elias R, Giobbie-Hurder A, McCleary NJ, Ott P, Hodi FS, Rahma O. et al. Human tumors instigate granulin-expressing hematopoietic Efficacy of PD-1 & PD-L1 inhibitors in older adults: a meta-analysis. cells that promote malignancy by activating stromal fibroblasts in J Immunother Cancer 2018;6:26. mice. J Clin Invest 2011;121:784–99. 35. Friedman CF, Wolchok JD. Checkpoint inhibition and melanoma: 55. Tuominen VJ, Ruotoistenmaki S, Viitanen A, Jumppanen M, Isola J. Considerations in treating the older adult. J Geriatr Oncol 2017; ImmunoRatio: a publicly available web application for quantitative 8:237–41. image analysis of estrogen receptor (ER), progesterone receptor (PR), 36. Kugel CH III, Douglass SM, Webster MR, Kaur A, Liu Q, Yin X, et al. and Ki-67. Breast Cancer Res 2010;12:R56. Age correlates with response to anti-PD1, reflecting age-related differ- 56. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, ences in intratumoral effector and regulatory T-cell populations. Clin et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics Cancer Res 2018;24:5347–56. 2013;29:15–21. 37. Rizvi NA, Hellmann MD, Snyder A, Kvistborg P, Makarov V, Havel 57. Liao Y, Smyth GK, Shi W. featureCounts: an efficient general purpose JJ, et al. Cancer immunology. Mutational landscape determines sen- program for assigning sequence reads to genomic features. Bioinfor- sitivity to PD-1 blockade in non-small cell lung cancer. Science 2015; matics 2014;30:923–30. 348:124–8. 58. Patro R, Duggal G, Love MI, Irizarry RA, Kingsford C. Salmon pro- 38. Salmon H, Remark R, Gnjatic S, Merad M. Host tissue determinants vides fast and bias-aware quantification of transcript expression. Nat of tumour immunity. Nat Rev Cancer 2019;19:215–27. Methods 2017;14:417–9.

1226 | CANCER DISCOVERY SEPTEMBER 2019 www.aacrjournals.org

Age and Response to Immune Checkpoint Blockade in TNBC RESEARCH ARTICLE

59. Garcia-Alcalde F, Okonechnikov K, Carbonell J, Cruz LM, Gotz S, 62. Kolde R, Vilo J. GOsummaries: an R package for visual functional Tarazona S, et al. Qualimap: evaluating next-generation sequencing annotation of experimental data. F1000Res 2015;4:574. alignment data. Bioinformatics 2012;28:2678–9. 63. Murtagh F, Legendre P. Ward’s hierarchical agglomerative clustering 60. Love MI, Huber W, Anders S. Moderated estimation of fold change and method: which algorithms implement Ward’s Criterion? J Classif dispersion for RNA-seq data with DESeq2. Genome Biol 2014;15:550. 2014;31:274–95. 61. Soneson C, Love MI, Robinson MD. Differential analyses for 64. Yu G, Wang LG, Han Y, He QY. clusterProfiler: an R package for RNA-seq: transcript-level estimates improve gene-level inferences. comparing biological themes among gene clusters. OMICS 2012;16: F1000Res 2015;4:1521. 284–7.

SEPTEMBER 2019 CANCER DISCOVERY | 1227

Interferon Signaling Is Diminished with Age and Is Associated with Immune Checkpoint Blockade Efficacy in Triple-Negative Breast Cancer

Jaclyn Sceneay, Gregory J. Goreczny, Kristin Wilson, et al.

Cancer Discov 2019;9:1208-1227. Published OnlineFirst June 19, 2019.

Updated version Access the most recent version of this article at: doi:10.1158/2159-8290.CD-18-1454

Supplementary Access the most recent supplemental material at: Material http://cancerdiscovery.aacrjournals.org/content/suppl/2019/06/19/2159-8290.CD-18-1454.DC1

Cited articles This article cites 64 articles, 12 of which you can access for free at: http://cancerdiscovery.aacrjournals.org/content/9/9/1208.full#ref-list-1

Citing articles This article has been cited by 5 HighWire-hosted articles. Access the articles at: http://cancerdiscovery.aacrjournals.org/content/9/9/1208.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerdiscovery.aacrjournals.org/content/9/9/1208. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.