We read with great interest the work from Sharma and colleagues concluding that anti–CTLA-4 (αCTLA-4) therapy does not deplete intratumoral FOXP3+ Tregs in bladder, prostate, and melanoma human cancers (1). We believe this conclusion is supported by the data presented by the authors but warrants further discussion.
First of all, in humans like in mice, a minority of intratumoral FOXP3+CD4+T-cells (Tregs) express membrane-CTLA-4 (2). Therefore, an anti–CTLA-4 therapy with an antibody isotype that is prone to perform ADCC/ADCP (antibody-derived cell-cytotoxicity/phagocytosis) such as ipilimumab (IgG1) and tremelimumab (IgG2), should only deplete cells with membrane-CTLA-4 expression and not all the FOXP3+ cells. However, the ability to identify such population of membrane-CTLA-4+ Tregs can be challenging upon IHC staining because CTLA4 is usually also highly expressed within the cytoplasm of CD4 T cells.
We and others (including some of the authors from ref. 1) have demonstrated that, in mice, the in vivo antitumor activity of αCTLA-4 therapy does not rely on CTLA-4 blockade but relies on the selective, FcyR-dependent, depletion of CTLA-4+ (tumor specific) Tregs (2–4). Most importantly, these mouse studies and the human study from Romano and colleagues suggest that the intratumoral depletion of CTLA-4+ Tregs is associated to tumor response (5). Therefore, it would have been interesting to know within the dataset reported by Sharma and colleagues, which tumor lesions were in response versus progression. Because less than 20% of patients have objective responses with αCTLA-4 monotherapy, only a handful of patients were potentially responders in the small cohorts reported in ref. 1.
Furthermore, the kinetics of Tregs depletion upon αCTLA-4 seems extremely rapid [within hours in human in vitro assays (5), within days in vivo in mice (2–4)]. In Sharma and colleagues (1), the posttreatment biopsies were collected 5 to 18 weeks after the last injection of ipilimumab; therefore, the timing of CTLA4+ Tregs depletion may not have been adequately captured.
In mice, αCTLA-4 works in “inflamed” tumors harboring infiltration with FCγR+ innate immune cells (3). In humans, CD16high monocytes might be pivotal (5). A high activator (CD64, CD16, CD32a) to inhibitory (CD32b) FCγR ratio seems critical for innate effectors to efficiently perform ADCC/ADCP upon αCTLA-4 therapy, and some polymorphisms (CD16-V158F, CD32a-H131) have been associated to response to ipilimumab/tremelimumab (3). It would be interesting to conduct future studies on this question to describe the proportion and phenotype of all ADCC performing cells and their FCγR repertoire in the tumor microenvironment.
See the Response, p. 3469
Disclosure of Potential Conflicts of Interest
A. Marabelle reports receiving other commercial research support from Bristol-Myers Squibb, reports receiving speakers bureau honoraria from Bristol-Myers Squibb and AstraZeneca, and is a consultant/advisory board member for AstraZeneca/Medimmune. No potential conflicts of interest were disclosed by the other authors.