Cancer immunotherapy drugs blocking the PD-1 pathway – known as checkpoint inhibitors – are now FDA-approved for melanoma, lung cancer and several other types of cancer. These drugs are often described as “releasing the brakes” on dysfunctional T cells.

A new study from Emory Vaccine Center and Winship Cancer Institute researchers shows that even if the PD-1-imposed brakes are released, the tumor-specific T cells still need “fuel” to expand in numbers and restore effective immune responses. That fuel comes from co-stimulation through a molecule called CD28.

The results were published Thursday by the journal Science.

Despite the success of PD-1-targeting drugs, many patients’ tumors do not respond to them. The study’s findings indicate that CD28’s presence on T cells could be a clinical biomarker capable of predicting whether drugs targeting PD-1 will be effective. In addition, the requirement for CD28 suggests that co-stimulation may be missing for some patients, which could guide the design of combination therapies.

For the rest of our press release and quotes from authors Rafi Ahmed, Alice Kamphorst and Suresh Ramalingam, please go here. For some additional links and thoughts on PD-1 and CD28, read on:

1. It’s not just a two-way interaction between cancer cell and T cell. The requirement for CD28 co-stimulation points to the importance of nearby antigen-presenting cells providing CD28’s partner molecule B7, which is not present on most tumor cells.

2. A UCSF/Genentech companion paper, in Science and biorxiv, shows that CD28 signaling is a main target of PD-1 inhibition. Same topic, from a more biochemical angle.

3. The Emory/Winship Science paper makes a distinction between what T cells are doing in the tumor environment (accessible via invasive biopsy) vs in the blood (more readily available).

Some predictive biomarkers for PD-1-based immunotherapy are known, such as PD-L1 on cancer cells or tumor-infiltrating immune cells. That makes a lot of sense — if the tumor looks like it’s putting the T cells to sleep via PD-1/PD-L1 interactions, then blocking those interactions is more likely to have an effect. Still, as this Lancet Oncology review points out, PD-L1 enriches for clinical benefit, but is insufficient for patient selection for most cancers.

4. Immunologists already know a lot about CD28. Through B7, CD28 is the indirect target of abatacept and belatacept, drugs for rheumatoid arthritis and kidney transplant, respectively.

In humans, T cells tend to lose CD28 as individuals age, or with chronic stimulation. One way to increase the response rate to PD-1-based cancer immunotherapy drugs might be re-activate CD28 on T cells that have lost it. Certain cytokines such as IL-12 can reactivate CD28 in CD4 “helper” T cells, but does this work for CD8 T cells? Unclear.