Discovery of the programmed death molecule 1, PD-1, on T cells was a major breakthrough that led to the development of immune checkpoint receptors for cancer treatment, and consequently the 2018 Nobel Prize in Physiology or Medicine for Tasuku Honjo. Anti-PD-1 immunotherapy is now widely used for the treatment of several cancers to boost patients’ CD8+ T cell immunity. Yet, effects of anti-PD-1 therapy on other key immune cell subsets remain understudied. Of particular interest is the effect of anti-PD-1 treatment on circulating T follicular helper (cTFH) cells, classically expressing PD-1 and playing an important part in germinal center (GC) reactions within lymphoid tissues following infection and vaccination. In this issue of Nature Immunology, Herati et al.1 defined quantitative and qualitative effects of anti-PD-1 therapy on cTFH cells and humoral responses to influenza vaccination. Robust cTFH cell responses were also observed in a subset of patients on anti-PD-1 therapy with immune-related adverse events (irAE) associated with immunotherapy.

In 2008, Wrammert and colleagues2 first reported the rise and fall of plasmablasts, also called antibody-secreting cells (ASCs), in response to influenza vaccination in blood samples obtained prior to vaccination, at 1 week post-vaccination (peak of plasmablast response) and 1 month later (peak of antibody response). Since then, studies have shown that plasmablast and cTFH cell frequencies concurrently peaked at 1 week, and correlated with antibody and memory B cell responses following influenza vaccination3,4,5. The role of cTFH cells is also evident in influenza virus infection6 and more recently, SARS-CoV-2 infection and/or mRNA vaccination7,8,9,10. Herati and colleagues1 have now used the influenza vaccination model with similar blood sampling timepoints in two independent anti-PD-1 immunotherapy cohorts, including patients with renal cell or urothelial carcinoma (cohort 1 includes cancer patients with and without anti-PD-1 treatment) and melanoma patients (cohort 2 also includes cancer patients with anti-PD-1 treatment and healthy adults for comparative analyses).

Activated memory CXCR5+ cTFH cells are typically characterized by their co-expression of ICOS and PD-1. Since PD-1 expression is lacking in anti-PD-1-treated patients, activated cTFH cell populations were characterized by their co-expression of ICOS and CD38; ICOS+PD-1+ and ICOS+CD38+ cTFH subsets have been used interchangeably with similar outcomes in the previous infection and/or vaccination studies mentioned above. Herati and colleagues showed that patients with anti-PD-1 immunotherapy had higher seroconversion and numerically higher fold-change increases in ICOS+CD38+ cTFH responses at 1 week post-vaccination compared to patients not treated with anti-PD-1 or to healthy control adults. Robust cTFH responses in anti-PD-1-treated patients coincided with numerical increases in plasmablasts as well as plasma concentrations of CXCL13, a biomarker of GC activity in lymphoid tissue (Fig. 1). However, functional analyses of the antibody response revealed qualitative impairments in anti-PD-1-treated patients, with lower affinity and less galactosylated/sialylated hemagglutinin-specific immunoglobulin G1 (IgG1) antibodies found at baseline compared to healthy controls. Fc glycosylation events are of importance as they can regulate antibody function.

Fig. 1: Anti-PD-1 immunotherapy impacts immune responses to influenza vaccines in cancer patients.
figure 1

a,b, Frequencies of circulating T follicular helper (cTFH) cells, plasmablasts (also known as antibody-secreting cells (ASCs)) and plasma CXCL13 concentrations are increased in patients receiving anti-PD-1 immunotherapy 1 week after influenza vaccination (b) compared to baseline (a). Anti-PD-1 was also associated with altered antibody responses. In patients receiving anti-PD-1 therapy, proliferation and cell cycle genes are upregulated in cTFH cells and plasmablasts, but cytokine signaling pathway genes are downregulated. These transcriptional profiles are enriched in anti-PD-1-treated patients with immune-related adverse events (irAE). aIgG4 MFI, aglycosylated immunoglobulin G4 median fluorescence intensity levels.

Herati and colleagues very clearly and elegantly delved further to transcriptionally profile cTFH cell and ASC populations in healthy adults and anti-PD-1-treated patients using bulk mRNA sequencing methods. Gene ontology and gene set enrichment analyses of the rich dataset revealed upregulation of proliferation and cell cycle genes in cTFH cells and ASCs at 1 week after influenza vaccination in anti-PD-1-treated patients compared to healthy controls, which aligned with the numerical fold-increases by flow cytometry. However, genes associated with leukocyte activation and cytokine signaling pathways were downregulated in anti-PD-1-treated patients but were enriched in the healthy adults, perhaps representing an “exhausted” phenotype similarly observed in CD8+ T cells of cancer patients on anti-PD-1 immunotherapy. Therefore, it seems that PD-1 blockade may actually dysregulate cTFH cell and humoral responses by unblocking their proliferative response, while dampening the overall immune response such as downregulating key cTFH cytokine circuits to adequately facilitate humoral responses.

Defining cellular or molecular biomarkers for predicting better or worse outcomes in clinical settings has been the holy grail for physicians, immunologists, and the like. One major drawback of PD-1 checkpoint blockade in cancer immunotherapy is the development of irAE, which leads to morbidity and discontinuation of therapy. Differentially expressed genes were indeed already identified in ICOS+CD38+ cTFH cells at baseline between anti-PD-1-treated patients who developed irAE and non-irAE patients1. Baseline cTFH cells from irAE patients were more activated and proliferative but were blunted in cytokine signaling pathways compared to non-irAE patients, with higher cell surface ICOS and aglycosylated immunoglobulin G4 (aIgG4) median fluorescence intensity levels, suggesting that a higher cTFH activation state was associated with the development of irAE. However, since only half of irAE patients had higher fold-change increases in ICOS+CD38+ cTFH cells at 1 week following influenza vaccination, questions remain about whether the other ~50% of patients were transcriptionally similar to non-irAE patients on anti-PD-1 immunotherapy, although the numbers are too small for such comparisons. Therefore, these observations need to be confirmed in larger cohorts and future studies. But for now, these highly activated cTFH cell populations in the context of influenza vaccination serve as a very promising tool for predicting development of irAE in cancer immunotherapy patients.

Circulating CXCR5+ TFH cells can be further characterized by expression of CXCR3 and/or CCR6 to define type 1 cTFH1 (CXCR3+), type 2 cTFH2 (CXCR3CCR6) and interleukin-17-expressing cTFH17 (CCR6+) subsets. Influenza vaccination, typically with protein-based formulations, and influenza virus infection can elicit more robust cTFH1 cell responses compared to cTFH2 and cTFH17 subsets3,5,6. Enrichment of SARS-CoV-2-specific cTFH1 responses over cTFH2 and cTFH17 have also been observed following SARS-CoV-2 infection and mRNA vaccination9, as well as following COVID-19 vaccination with a protein-based Spike-clamp vaccine candidate from an earlier phase I clinical trial11. Tucked away nicely in Extended Data Fig. 1, Herati and colleagues showed that at 1 week after influenza vaccination, frequencies of CXCR3+ICOS+CD38+ cTFH1 cells were equally increased in both healthy adults and anti-PD-1-treated patients, whereas cTFH2 and cTFH17 frequencies decreased. Numerically, if ICOS+CD38+ cTFH1 cells were compared, would there have been fold-change differences between healthy and anti-PD-1-treated cohorts experiencing irAE or not? If cell numbers and sequencing costs were not an issue in this day and age, it would be tantalizing to further tease out these cTFH cell subsets transcriptionally to see whether the re-wiring effects of anti-PD-1 immunotherapy was preferentially affecting type 1 responses, particularly as the authors report downregulation of more type 1 cytokine signaling circuits in the parent cTFH populations (that is, IFNγ, IL-2/STAT5, TNF/NF-κB, IL-6/JAK/STAT3).

Overall, Herati and colleagues have proven that measuring real-time human immune responses to influenza vaccination remains a highly valuable tool, not only to measure immune protection from severe influenza disease with annual influenza vaccine updates, but also to decipher key mechanisms underlying the re-wiring immunological effects of anti-PD-1 treatment in cancer immunotherapy patients. It remains to be seen whether dysregulation of high activation/proliferation and blunted cytokine pathways on cTFH cells and plasmablasts, and ultimately GC reactions, will persist long-term after anti-PD-1 treatment has ceased.