Thread by @connor_rosen

Two great new papers out on unexpected demographic results in immunotherapy responses - older and obese patients (and animal models) have better responses to immunotherapy - out now in Clin Cancer Res @AACR and @NatureMedicine #Cancer #Immunotherapy - long thread ahead...

First paper in Clinical Cancer Research on efficacy of checkpoint blockade (CPI) by age (clincancerres.aacrjournals.org/content/24/21/…) - pretty clear from examination of ~500 melanoma patients across 8 centers that PFS was higher in older patients (over 62 years old).

PFS difference was driven by stable disease / partial responses, not more complete responses in the older cohort. It was independent of gender and occurred in both MAPKi naive and treated patients, with slightly stronger differences in the MAPKi naive patients.

Modeling in mice showed the same result - an anti-PD1 resistant tumor (BSC9AJ2) in young mice had growth slowed in mice over 10 months of age. This was associated with a decrease in %CD4+FoxP3+ cells and an increase in CD8+ cells in the tumor.

Similar observations in the Yumm1.7 model, suggesting in both cases an increase in the humoral CD8:Foxp3 ratio (CTL:Treg) might be be the cause of the better response in aged mice. This was tumor specific, as splenic CTL:Treg ratio was no higher (or even lower) in older mice.

Depletion of Tregs with anti-CD25 was able to synergies with anti-PD1 in young mice to provide some protection. Would’ve been nice to see the direct comparison with aged mice, but good result in keeping with the literature.

Finally, back to the patients, staining of melanoma samples revealed an age-dependent decrease in Foxp3 and increase in CD8 T cells, and more patients have High CTL:Treg ratios.

While the mechanism behind tumor-specific Treg exclusion in older patients and increased CD8 infiltration remains unclear, it matches the way we think about melanoma that this increase would provide enhanced benefits with anti-PD1 therapy. Definitely bears further investigation!

The authors discuss caveats about the human observations, which bear strong consideration - somatic mutations that accumulate with age, CMV and other viral infections (raised in a preview in the same issue), metabolic status of aged patients (leading on to the next paper)…

But overall the mouse data, with genetically identical tumors, are suggestive of a host-intrinsic phenomenon, rather than something to do with development of tumors. I think it’s overall good evidence that age (somehow!) plays a role in CPI response.

On to the next paper, focusing on obesity, in @NatureMedicine (nature.com/articles/s4159…). The beginning establishes in mice, NHPs, and humans that obesity (diet-induced, spontaneous, or BMI>30, respectively) results in impaired immune function - less Ki67 on T cells and more PD1.

Keeping the age paper in mind, this obesity-related dysfunction was visible in 6 month old mice only in the liver, but was systemic (blood, liver, spleen) at 11 months - and the control mice at 11 months looked more dysfunctional than control at 6 months (see the supplement)

Looking in the tumor setting of obese mice, there was substantial dysfunction of T cells in the tumor as well. Unlike with the aging paper, the authors don’t see increased Treg infiltration, although they don’t do absolute quantitation, only %Foxp3+ of CD4s - so no CTL:Treg ratio

This dysfunction was accompanied by more rapid tumor growth, and characterized by down regulation of effector genes (IFNG, TNF, CD69) and Ki67 as well as up regulation of multiple checkpoints (PD1, Tim3, Lag3).

Interestingly the tumor did not have higher PD-L1, suggesting an immune-intrinsic defect. RNA-seq indicated a role for pathways involved in leptin signaling, and observations in humans and mice indicated that leptin correlated with PD-1 expression on T cells.

This seemed to be to leptin signaling through T cells, as leptin-deficient mice did not have elevated PD-1 on their T cells, and adoptive transfer of WT and db/db (leptin receptor deficient) T cells into WT mice led to improved proliferation and lower PD-1 on the db/db T cells.

Further, leptin acts directly on T cells to increase PD-1 expression through STAT3 activation, as shown through pharmacological inhibition of STAT3 in in vitro T cell stimulations with leptin.

Leptin treatment of ob/ob (leptin-deficient) mice caused them to lose weight, but moderately increased PD-1 expression and tumor growth. More importantly, transfer of db/db T cells vs WT T cells into obese mice was able to protect against tumor growth.

All this together combines to suggest that the environment in the obese mice suppresses T cells through leptin signaling in the T cells, and this can impair anti-tumor responses. Removing leptin or the ability of T cells to sense leptin rescues the T cell defect in obese mice.

But, when the authors looked at response to PD-1 inhibition, the obese mice responded better - in both PD-1 resistant and PD-1 susceptible tumor models. This was accompanied by increased T cell infiltration into the tumors and fewer metastases, and no higher toxicity.

Back to humans - looking for differences in responses in obese melanoma patients, the authors saw elevated T cell exhaustion markers (PD1, Tim3, Lag3, Tigit, Tbet, Eomes), with no difference in PD-L1 or Foxp3 expression, and fewer CD8’s in the tumor in colorectal patients.

Pivotally, both overall and progression-free survival were improved in obese patients following anti-PD-1 or anti-PD-L1, with pretty impressive hazard ratios, and no increase in immune-related adverse events.

I think both these papers providing exciting and unexpected insights into what’s going on in the broader patient population following treatment with CPIs. Both are unexpected, with some differences in the mechanisms and room for more exploration, especially in the aged cohort.

More importantly, both hit one idea on the head - chronic inflammation, from obesity or aging, results in general immune dysfunction that can be exacerbated in a tumor setting. However, this dysfunction is more easily reversed with CPI treatment.

(thoughts continuing in a separate thread, I hit twitter's limit of linked tweets!)

I wonder if the underlying reason is that, in the setting of chronic inflammation, tumors have never “needed” to evolve extra immune-suppression mechanisms - the T cells they saw were already PD-1 positive, so up regulation of PD-L1 was sufficient to be protective.

How many of these tumors are PD-L1 negative but T cell infiltrated - the kind of “immunologically cold” tumors that pure CPI doesn’t seem as effective in? They’ve never had to pursue alternative mechanisms to suppress the T cells, and I suspect that’s why they respond.

Looking in other settings of chronic inflammation may reveal that systemic immune dysfunction prior to tumor growth may actually be broadly predictive of efficacy - since the tumor only has needed to develop one immune evasion mechanism, it’s more easily subverted.

A classic PD-1 papers shows that exhausted T cells from chronic viral infections are readily re-invigorated by anti-PD-1 treatment (nature.com/articles/natur…)

Does this suggest T cells driven to exhaustion by chronic viral infection or other chronic inflammation can be more easily reactivated than patients with tumor-driven exhaustion, where other mechanisms are likely also at play?

I don’t know if there are studies of, for example, liver cancer responses to CPI in chronic hepatitis patients (although that may become more difficult due to possible antigen-specific responses) as another place to look for similar effects.

Overall, these were two good studies revealing unexpected responses to CPI, and I look forward to further dissection of the mechanisms involved and ways to harness this to further improve the efficacy of checkpoint blockade therapy!