Immunotherapy has revolutionized cancer treatment, however, a significant fraction of patients fail to respond to therapy and may suffer serious side effects. Predicting and monitoring therapeutic efficacy remains an important challenge. Techniques to achieve these goals without multiple biopsies or other invasive methods will be critical. The efficacy of the various forms of immunotherapy is a direct result of changes evoked in the tumor microenvironment (TME). Noninvasive monitoring of events occurring in the TME can be used to tackle this issue.

We and others have recently developed PET imaging approaches to detect immune cells in living animals. A key question that remains is whether we can monitor therapeutic efficacy and predict a response. To noninvasively monitor the distribution of immune cells and mapping the immunological landscape in the TME more generally, we use the smallest antibody-derived protein binding domain format that retains antigen-binding capability. Camelids produce a subset of immunoglobulins that are composed of heavy chains only, with a single heavy chain variable domain (nanobody or VHH) that binds antigen in the absence of a light chain. These VHHs can be expressed recombinantly. To identify VHHs, we immunize alpacas with (a) protein(s) of interest. The resulting repertoire of immune VHHs is then cloned from peripheral blood lymphocytes and used to generate a phage or yeast display library. VHHs specific for the antigen of interest are recovered by panning against the immobilized antigen. The relevant VHHs are then expressed recombinantly, and equipped with C-terminal sortase and (His)₆ tags to facilitate purification and site-specific labeling.

Having established robust methods of performing immuno-PET using nanobodies, our experiments have broken new ground and laid the foundation for more sophisticated ways of exploring the TME. We can now predict, in the B16 melanoma and MC38 colorectal models, the response to immunotherapy based on differences in the distribution of CD8⁺ T cells. This raises the question of how these differences in CD8⁺ T cell distribution arise and are maintained. Using single-cell RNAseq performed on the immune infiltrating cells in responder and nonresponder animals, we have discovered that an effective response to checkpoint blockade is accompanied by a general reprogramming of the myeloid cells in the TME. The responders exhibit a dominant population of macrophages with an M1-like signature, whereas the immune cells in the nonresponders display an immunosuppressive M2-like status.

We are now delving deeper into the immune events that occur in response to immunotherapy to answer some fundamental questions: what is behind the therapy’s heterogeneous response? What are the molecular mechanisms that lead to shaping the immune landscape of tumors? How does the immune infiltrate status change in response to therapy? What causes immunotherapy to succeed or to fail? How do CD8+ and CD4+ T cells, myeloid cells, and macrophages behave in response to therapy? Is it possible to predict the outcome of therapy or stratify responders from nonresponders early on after initiation of therapy? What is the role of tumor-specific (antigen-specific) T cells? Can we monitor the dynamics of tumor-specific T cells (and CAR T cells)? How about different cytokines and chemokines that attract or repel T cells? Does the tumor-infiltrating T-cell repertoire change during the response to treatment? Results from these studies may shed light on our understanding of the response to immunotherapy and will help in designing more effective treatments, including personalized treatments. In the next few years, by combining our expertise in molecular biology, chemical biology and immunology, a unique combination, we hope to address some of these challenges.