Female BALB/c mice were supplied by Charles River UK at 6 to 8 weeks of age (young) or 60 to 72 weeks of age (aged). These mice were aged at Charles River UK specifically for the purpose of these studies and housing conditions were kept the same as for young mice. Mice were housed under specific pathogen-free conditions in Tecniplast Green Line Sealsafe Individually Ventilated Cages (IVC) changed weekly and holding a maximum of 6 animals with irradiated aspen chip bedding, Nestlets nesting material, a tunnel and wooden chew blocks. Mice were housed on a 12/12 light/dark cycle, at 20-24°C and 45-65% humidity with ad libitum UV-treated water and RM1 rodent diet (SAFE). Mice underwent a minimum of 5 days of acclimatization after arrival in the animal facility before study initiation.

The murine colon carcinoma cell line CT26 was obtained from ATCC and maintained in RPMI 1640 media supplemented with 10% FBS. It did not undergo any in vivo passaging and was maintained under limited passage from original stocks (typically under 5). It was reauthenticated using STR-based DNA profiling and multiplex polymerase chain reaction and tested for murine viruses and mycoplasma (IDEXX Bioresearch). For implantation, mice were shaved on the right flank and subcutaneously injected with 5 x 10⁵ CT26 cells in 100 µL PBS. Tumor volume was measured 3 times per week using electronic callipers and calculated using the formula (width² × length)/2. Mice were euthanized when they reached humane welfare limits pertaining to tumor volume (average diameter of 15mm) or tumor condition (ulceration of the skin above the tumor). Tumor growth rate was calculated as previously described (10) by fitting the tumor growth curve of each animal to the exponential model log10(tumor volume) = a + b * time + error, where a and b are parameters that correspond to the log initial volume and growth rate, respectively.

Mice were randomized to minimize differences in body weight and treated with either anti-mouse PD-L1 antibody mIgG1 D265A (AstraZeneca) intraperitoneally (IP) at 10 mg/kg twice weekly starting 4 days after tumor cell implantation for a total of 6 doses and/or anti-mouse CTLA-4 antibody (clone 9D9) mIgG1 (AstraZeneca) IP at 10 mg/kg twice weekly starting 7 days after tumor cell implantation for a total of 6 doses or anti-mouse OX40 antibody (clone OX86) mIgG2a (AstraZeneca) IP at 1 mg/kg 4 and 7 days after tumor cell implantation. Group sizes were determined using power analyses based on the variability of the tumor model in pilot studies.

Tumors were disaggregated using the gentleMACS™ Mouse Tumor Dissociation kit (Miltenyi Biotec). Tumor-draining lymph nodes (or matching inguinal lymph nodes in non-tumor bearing mice) were dissociated through a 70 µm nylon cell strainer. Cells were stained with a fixable viability dye (Thermo Fisher) and blocked with antibodies against murine CD16/CD32 (eBioscience) before staining with fluorescence-conjugated antibodies (Supplementary Table S1) in flow cytometry staining buffer with Brilliant Stain Buffer (BD Biosciences). Intracellular staining was performed using the FoxP3/Transcription Factor Staining Buffer Set (eBioscience) and cells were fixed in 3.7% formaldehyde. Counting beads (123Count eBeads; eBioscience) were added to the samples before acquisition on a BD LSRFortessa flow cytometer (BD Biosciences). Panel-specific FMO controls and single color controls were prepared for each experiment. Analysis was carried out using FlowJo (BD Biosciences). As an adjunct to conventional gating, for tumor samples the data was quality controlled using the flowAI plugin in FlowJo followed by T-distributed stochastic neighbor embedding (optSNE) dimensionality reduction and FlowSOM clustering analysis.

For time-to-endpoint analysis, groups were compared using a log-rank test in GraphPad Prism. Fold changes in tumor growth rates were compared using unpaired Mann-Whitney testing in GraphPad Prism. Cell populations and growth rates were compared using either two-way ANOVA with Sidak’s multiple comparisons testing or unpaired Mann-Whitney testing in GraphPad Prism.

In this study, we sought to investigate the impact of age on the CT26 mouse syngeneic flank tumor model, which is a commonly-used preclinical model for assessment of novel anti-tumor therapies. To determine whether the characteristics of this model would be affected by age, we implanted CT26 tumor cells simultaneously in both aged (60-72 weeks of age) and young (6-8 weeks of age) female BALB/c mice. Correlations between mouse and human age suggest that this would correspond to 50-70 year-old and 15 year-old patients, respectively (9), and we confirmed that several hallmarks of ageing were observed in these mice at 60-72 weeks of age, such as thymic involution and increased adiposity (Supplementary Figure 1) (11, 12).

Given the known impact of age on the immune system, we first sought to determine whether aged animals were able to respond to the presence of a tumor by evaluating the changes triggered in the inguinal lymph node (the lymph node closest to the site of flank tumor implantation) after the addition of a tumor. We observed similar changes triggered in response to the tumor in both aged and young mice, namely a decrease in the frequency of CD4⁺ T cells and an increase in the frequency of B cells (Figure 1A), with the latter being a phenomenon previously described after the implantation of tumor cells (13).

Despite this similar response to the presence of a tumor, we found that the immune composition of the tumor-draining lymph node was still markedly different between young and aged mice, with a reduced frequency of both CD4⁺ and CD8⁺ T cells and an increased frequency of B cells in the aged cohort (Figure 1B). Similar differences were also seen in the lymph nodes of non-tumor bearing mice (Figure 1C) and in the spleens of aged mice, irrespective of tumor presence (Supplementary Figures 2A, B), suggesting that these differences are mainly driven by baseline differences in the immune status between young and aged mice, rather than a difference in the response elicited by the tumor. We also further analyzed the phenotype of CD4⁺ T cells and found that, consistent with known changes in immune homeostasis with ageing (3), the proportion of naïve T cells was significantly decreased in the inguinal lymph nodes (Figure 1D) and spleens (Supplementary Figures 2C, D) of aged mice. This decrease was also observed in tumor-draining lymph nodes, as well as increases in the central memory subset in the aged mice (Figure 1E). Taken together these findings suggest that the immune system of aged mice is able to respond to the presence of a tumor but that this response does not restore their immune phenotype to that seen in young mice, suggesting a potential for differential responses to immunotherapy in aged animals.

Having observed a similar immune response elicited by the tumor despite baseline differences in the immune compartment of the aged mice compared to young mice, we next assessed whether age would impact the tumor immune cell infiltrate. We found that aged mice implanted with CT26 tumors reached welfare endpoints (a function of both tumor volume and tumor condition) faster than the young mice, with a median time-to-endpoint of 20 days for the aged mice versus 23 days for the young mice (Figure 2A, p=0.0004). However, analysis of the tumor growth rates of individual animals across 4 independent experiments revealed no consistent difference in the kinetics of tumor growth between aged and young mice (Figure 2B, Supplementary Figure 3). The CT26 tumor model is prone to develop ulceration of the skin above the tumor, which can lead to sacrifice of the animal for welfare reasons. We found that the proportion of animals being sacrificed due to skin ulceration at tumor volumes under 500mm³ was higher in aged mice compared to young mice (25% versus 15%, respectively; data not shown), suggesting that the difference detected in the time-to-endpoint analysis is driven by differences in tumor condition rather than tumor volume.

Having determined that CT26 tumors grew similarly in aged and young mice, we next assessed the impact of age on the immune cell composition of the tumors. Flow cytometric analysis of tumor-infiltrating immune cells revealed a significant decrease in the size of the CD45⁺ immune cell infiltrate (expressed as the absolute cell count per mm³ of tumor) in aged mice (Figure 2C, p=0.0034). Unsupervised clustering analysis then enabled the detection and quantification of several lymphoid (Figure 2D) and myeloid (Figure 2E) cell subsets within the TME (Supplementary Figure 4). Significant differences were observed with tumors in the aged mice containing fewer NK cells, PD1⁺ CD8⁺ T cells and monocytic cells (all p<0.05) as well as increased macrophages (p<0.0001) and a trend to increased Treg (p=0.0557) (Figure 2F). Altogether these changes in the tumor-infiltrating immune cell composition would suggest the potential for dampened anti-tumor immune responses in the aged mice. Indeed, we observe a reduction in key cytotoxic cell types (CD8⁺ T cells and NK cells) that have been linked to improved prognosis in several cancer types and an increase in key immunosuppressive cells types (Treg and macrophages) that have been linked overall to poorer prognosis (14). These differences in the composition of the tumor immune infiltrate also resulted in a significant decrease in the ratio of CD8⁺ T cells to Treg in the tumors of aged mice (Figure 2G, p=0.0007), a variable known to correlate with responsiveness to immuno-oncology therapies (15).

Having identified age-related differences in both the tumor-draining lymph node and tumor-infiltrating immune cells, we sought to investigate whether these changes would impact the response of aged mice to immunotherapy. We treated both aged and young animals with an anti-OX40 antibody and found that whilst time-to-endpoint was prolonged in both cohorts, the complete response rate was more than halved from 73% to 24% in aged mice (Figure 3A). This impact of age on the effect of treatment was confirmed by assessment of the tumor growth rate (Figure 3B, p<0.0001), with more profound anti-OX40-induced reductions in tumor growth rate observed in the young cohort, despite no age-related differences in expression of OX40 (Supplementary Figure 5A). Published work has previously associated the anti-tumor activity of OX40 agonism with depletion of Tregs within the TME (16) so we assessed whether the reduced efficacy of the anti-OX40 antibody in aged mice correlated with impaired Treg depletion within the tumor. We found that this was not the case and similar levels of Treg depletion were seen in the aged mice compared to the young mice (Figure 3C). We did not observe any impact of treatment on CD8⁺ T cell proportions (Supplementary Figure 5B) but we observed significant differences in the response to OX40 agonism within the CD4⁺ T cell compartment in aged mice. Whereas in young mice, treatment with the anti-OX40 antibody led to a reduction in the frequency of effector CD4⁺ T cells and an increase in effector memory CD4⁺ T cells, these changes were not seen in the aged mice after treatment (Figure 3D). Given the reduced induction of effector memory CD4⁺ T cells in the aged mice, we next investigated whether a memory response could also be generated after treatment. Both aged and young mice that had achieved complete tumor clearance after treatment with anti-OX40 antibody were rechallenged with CT26 tumor cells on the opposite flank 53 days later. We observed that growth of the second tumors was prevented in both aged and young mice, suggesting that memory responses had been successfully generated in the aged mice (Figure 3E).

Given the impact of age on the anti-tumor efficacy of the immune agonist OX40 treatment, we next sought to determine how age might affect the response to immune checkpoint blockade. We investigated CTLA-4 blockade and saw a significant impact of age on treatment response. Whereas young mice displayed a prolonged time to endpoint (p=0.0023) and a 18% complete response rate after treatment with anti-CTLA-4 antibody, this was not the case in the aged mice and we saw a reduced complete response rate of only 3% (Figure 4A). This dampening in the effect of CTLA-4 blockade in aged mice was also seen in the tumor growth rate analysis, where the fold change in growth rate compared to the untreated group was significantly less pronounced in the aged mice (Figure 4B, p=0.0104).

In clinical practice, it has been observed that addition of PD-1 or PD-L1 blockade to CTLA-4 inhibition can increase the response rate and durability of responses in patients (17–19). Therefore we assessed whether addition of an anti-PD-L1 antibody could overcome the reduced responsiveness of aged mice to CTLA-4 blockade. When we first assessed PD-L1 blockade monotherapy, we saw only modest anti-tumor activity and did not see a clear impact of age on response to this treatment (Supplementary Figure 6). However, after treatment with the combination of anti-PD-L1 and anti-CTLA-4 antibodies we saw striking age-related differences in activity of this immune checkpoint blockade regimen. Indeed, in young mice this combination led to strong anti-tumor efficacy with a 55% complete response rate (Figure 4C). In contrast, the complete response rate was only 11% in aged mice after treatment with the combination of anti-PD-L1 and anti-CTLA-4 antibodies (Figure 4C) and there was a significant decrease in the effect of the treatment on tumor growth rates (Figure 4D p=0.0002).