Age Correlates with Response to Anti-PD1, Reflecting Age-Related Differences in Intratumoral Effector and Regulatory T-Cell Populations

Abstract

Purpose: We have shown that the aged microenvironment increases melanoma metastasis, and decreases response to targeted therapy, and here we queried response to anti-PD1.Experimental Design: We analyzed the relationship between age, response to anti-PD1, and prior therapy in 538 patients. We used mouse models of melanoma, to analyze the intratumoral immune microenvironment in young versus aged mice and confirmed our findings in human melanoma biopsies.Results: Patients over the age of 60 responded more efficiently to anti-PD-1, and likelihood of response to anti-PD-1 increased with age, even when we controlled for prior MAPKi therapy. Placing genetically identical tumors in aged mice (52 weeks) significantly increased their response to anti-PD1 as compared with the same tumors in young mice (8 weeks). These data suggest that this increased response in aged patients occurs even in the absence of a more complex mutational landscape. Next, we found that young mice had a significantly higher population of regulatory T cells (Tregs), skewing the CD8⁺:Treg ratio. FOXP3 staining of human melanoma biopsies revealed similar increases in Tregs in young patients. Depletion of Tregs using anti-CD25 increased the response to anti-PD1 in young mice.Conclusions: While there are obvious limitations to our study, including our inability to conduct a meta-analysis due to a lack of available data, and our inability to control for mutational burden, there is a remarkable consistency in these data from over 500 patients across 8 different institutes worldwide. These results stress the importance of considering age as a factor for immunotherapy response. Clin Cancer Res; 24(21); 5347-56. ©2018 AACR See related commentary by Pawelec, p. 5193.

Figure 1.

Age-related efficacy of pembrolizumab in melanoma patients. A, Pie charts showing the percentages of patients with the indicated responses between younger and older melanoma patients combined using 62 years old as the most statistically significant cutoff. Significance was determined using a Fisher exact test. B, Graph indicating the probability of a patient progressing on pembrolizumab treatment given their age and determined using a logistic regression analysis. C, Percentage of patients from A who progressed on treatment, separated by gender, and using 62 as the most statistically significant cutoff. Best responsesof patients without prior MAPKi therapy (D) and with prior MAPKi therapy using most statistically significant cutoffs (E). Statistical cutoffs determined using Fisher exact test. F, Graph indicating the probability of patients progressing on pembrolizumab treatment, from D and E, given their age and determined usinga logistic regression analysis.

Figure 2.

Differential response to anti-PD1 in young and aged mice. Mice bearing BSC9AJ2 tumors treated with 3 doses of 300 μg rat IgG2AK (n = 5) or 4 doses of 300 μg anti-PD1 (n = 5) every 5 days, starting on day 0. Response rates in young (8- to 10-week-old; A) and aged (10-month-old; B) female mice, and in young (8- to 10-week-old; C) and aged (10-month-old; D) male mice. All experiments used a linear mixed-effect model to determine significance. Error bars, SEM.

Figure 3.

Effect of age on intratumoral immune populations in mice. FACS analysis of lymphocyte populations within tumors from young (8- to 10-week-old) andaged (10-month-old) C57/Bl6 male mice bearing Yumm1.7 murine melanomas (A) and females bearing BSC9AJ2 murine melanomas (B). Experiments were performed in duplicate, and melanomas were injected intradermally and harvested when tumor volumes were between 500 and 1,000 mm³. Significance determined by two-way ANOVA. Error bars, = 95% CI. CD8b⁺ levels as a percentage of live cells in young and aged mice in both models (C). Significance determined by two-tailed Student t test assuming unequal variance. Error bars, 95% CI. Ratios of CD45⁺CD3⁺CD8a⁺ cells and CD45⁺CD3⁺CD4⁺FoxP3⁺ cells within individual tumors from young and aged male mice bearing Yumm1.7 tumors (D) and female mice bearing BSC9AJ2 tumors (E). Significance determined by two-tailed t test assuming unequal variance. Error bars, 95% CI. CD45⁺CD8b⁺ cells from Yumm1.7 (F) and BSC9AJ2 (G) tumors expressing TNFα and IFNγ following 5-hour incubation with PMA and ionomycin prior to FACS analysis. Significance determined by two-tailed Student t test assuming unequal variance. Error bars, 95% CI. Ratios of CD45⁺CD3⁺CD8a⁺ cells to CD45⁺CD3⁺CD4⁺FoxP3⁺ cells within the spleens of mice bearing Yumm1.7 (H) or BSC9AJ2 tumors (I). Error bars, 95% CI.

Figure 4.

Lymphocyte populations in patient melanomas. A, Representative images of IHC staining of FOXP3⁺ cells in patient melanoma tumors at 40× magnification. B, Quantification of FOXP3⁺ cells performed by board-certified pathologists identifying the densest area of immune infiltrate within tumors and counting the positively stained cells at 40× magnification. Significance from two-sample Wilcoxon rank-sum (Mann–Whitney) tests. Error bars, 95% CI. C, Graph indicating the percentage of patients in each age group, with CD8⁺ cells representing less than 20% of the total immune cell infiltrate. Significance from Fisher exact test. Error bars, 95% CI. D, Quantification of CD8⁺ cells following outlier removal. Cells counted and significance determined as in C. Outliers determined and removed as identified by ROUT (Q = 0.1%) outlier test. Error bars, 95% CI. E, Percentages of patients from D whose CD8:FOXP3 ratio is low (<median − 1), med (median ± 1), or high (>median + 1) using median age of 60 as the cutoff.

Figure 5.

Depletion of Tregs with anti-CD25 increases response to anti-PD1. Young female mice were treated with anti-CD25 or an IgG1A isotype 5 days prior to subdermal of BSC9AJ2 cells with treatments continuing every 5 days until sacrifice. Anti-PD1 was administered every 3–4 days, starting on day 9 posttumoral injection. A, Anti-CD25–depleted Tregs in the peripheral blood of mice. B, Statistical analysis of tumor growth curves comparing anti-CD25, anti-PD1, and the combination of both antibodies in young mice. C, Tumor growth curves for above. All experiments used a linear mixed-effect model to determine significance. Error bars, SEM.