4-1BB and the critical importance of local control

Pathology stain of tertiary lymphoid structures (arrows) found on the margin of, and formed within, a colon carcinoma sample. https://doi.org/10.1016/j.labinv.2023.100063

Immunity is complex and can be dangerous when exploited clinically, as demonstrated by the lethal administration of TNF, or anti-CD40L antibody (Biogen) or CAR-T cells expressing the CD16 Fc-receptor (Unum), among many other examples. Many other targets of interest have produced therapeutics, including antibodies and fusion proteins, that have failed due to toxicities discovered in preclinical or clinical development. At Biogen we tabled anti-4-1BB antibodies early in development due to expected or demonstrated toxicities; this happened despite intensive efforts to get the signaling ‘right’, that is, to develop antibodies that bound to 4-1BB and activated the canonical signaling cascade, as triggered by the 4-1BB ligand (4-1BBL). Similar issues hindered the anti-LT-beta receptor program, the anti-Trail Receptor-2 program and other programs targeting proteins within the TNF-Receptor-related protein family.

Targeting 4-1BB remained of interest to the immuno-oncology field as cell culture experiments and tumor models in mice suggested that robust anti-tumor immune responses could be triggered by anti-4-1BB antibody therapies. However, in clinical trials, activating anti-4-1BB antibodies such as urelumab (BMS) and utomilumab (Pfizer) showed dose-limiting liver toxicities and minimal therapeutic activity. From such studies a hypothesis emerged that anti-4-1BB antibodies were binding to the 4-1BB target on liver cells and then transducing non-canonical and ultimately pathological signaling cascades due to antibody cross-linking or multimerization induced by the Fc-receptor binding component of the antibodies. This complex hypothesis has led to numerous attempts to localize antibodies away from the normal organs and tissues, including the liver, and instead trying to direct the anti-4-1BB activity to the site of the tumor or to the tumor cells themselves.

This is where things get interesting due to the complexity of the immune system and its regulation. Let’s back up a bit.

In humans, the 4-1BB protein is expressed on the surface of activated cytotoxic (CD8-positive) T cells. This is the T cell type most closely associated with anti-tumor immune responses. 4-1BB expression can also be detected on other activated immune cells including NK cells, B cells, monocytes, and dendritic cells. 4-1BB is an “inducible costimulatory molecule”, meaning the cells that express the protein are stimulated by other signals first, notably signals sent by the T-cell receptor (TCR) and the CD28 protein. 4-1BB enhances the response of activated T-cells to “foreign antigens”: protein fragments derived from infected, diseased or cancerous cells that are presented to T cells by antigen-presenting cells (APC). 4-1BBL is expressed on APCs including B cells, dendritic cells, and macrophages and the binding of 4-1BBL to 4-1BB enhances T-cell inflammatory cytokine production and supports T-cell survival. 4-1BBL expression has also been described on some tumor cells and on various normal cell types outside the immune system.

Animal models of tumor immunity, performed mainly in mice, have shown that the anti-tumor immunity induced by 4-1BB signaling is dependent on T-cell expression of 4-1BB on the CD8 cytotoxic T-cell subset and also the CD4 T-cell subset.

So, to summarize thus far:

• 4-1BB expression is “inducible” and depends on other stimulatory signals

• Expression of 4-1BB and 4-1BBL suggests that T cell/APC interaction is at the

• core of this pathway’s biology

• T-cell activation and survival are the consequence of 4-1BB signaling

• Systemic delivery (ie. throughout the body) of anti-4-1BB antibodies causes toxicities but not anti-tumor efficacy

… misreading the subtleties of 4-1BB pathway biology

I recently came across four 4-1BB-targeting abstracts at #ASCO24 that I found perplexing. The investigators are attempting to localize the 4-1BB signaling effect while misreading the subtleties of 4-1BB pathway biology. All three used an anti-PD-L1 antibody domain as the “anchor” in a bispecific antibody (BiAB: anti-4-1BB activating domain + anti-PD-L1 blocking domain) to localize activation of 4-1BB. The therapeutic hypothesis has several components:

• PD-L1 is absent from liver cells and this should prevent or reduce 4-1BB-activation mediated liver toxicity

• PD-L1 is expressed in the tumor microenvironment (TME) on cancer cells and on immune cells that have a T-cell suppressive activity, like M2-monocytes.

• Blocking PD-L1 binding to it’s receptor PD-1 will prevent T-cell immune-suppression and stimulate anti-tumor immunity

• Activating 4-1BB on the now stimulated T-cells will enhance anti-tumor immunity, synergistically with anti-PD-L1

On the surface, the hypothesis may appear reasonable but as discussed below, this is a hypothesis at least 4 years behind our current understanding of PD-1/PD-L1 pathway biology.

Indeed, all four studies seemingly failed. These failures highlight the need to understand not only where different immune proteins may be expressed on cells, but, critically, where their biological functions are manifested within the anti-tumor immune response.

Four #ASCO abstracts

Other therapeutics in this space include INBRX-105 (Inhibrx, Inc, discontinued), PRS-344 (Pieris, licensed to Servier BioInnovation LLC), MCLA-145 (Merus Inc, development suspended last week), BH3120 (Hamni Pharmaceutical), ND-021 (Numab), FS222 (invoX, which bought F-Star) and many others.

I predict that all of these programs will fail.

The NIMBY problem

I think the “Cancer Immunity Cycle” as originally presented by Chen and Mellman (10.1016/j.immuni.2013.07.012) led many investigators down the wrong path, focusing intently on immune processes occurring within the tumor. As I’ve discussed previously, while the site of PD-1 inhibition had long been assumed to be within the TME, more recent literature has shown that secondary and tertiary lymphoid organs (lymph nodes and tertiary lymphoid structures or TLS) are the sites of ongoing and productive anti-tumor immunity. Mellman et al in their 2023 revision of the original model make this point clearly (10.1016/j.immuni.2023.09.011). This leaves us with a question:

What happens when you block PD-L1 activity and simultaneously stimulate 4-1BB activity within a lymph node or TLS?

A few observations:

• In syngeneic mouse models, anti-tumor activity via 4-1BB stimulation is abrogated if T cells remain trapped in lymph nodes (10.1172/jci.insight.124507 and references within this excellent 2024 review: https://doi.org/10.1038/s12276-023-01136-4).

• Chronic 4-1BB stimulation within lymph nodes causes B cell loss, germinal center dissolution, granuloma formation and failed immune responses. (https://www.nature.com/articles/s41423-020-00533-3).

• PD-L1 expression in the TME is a surprisingly poor biomarker for the clinical efficacy of anti-PD-1 and anti-PD-L1 blocking antibodies, despite the use of PD-L1-directed assays in determining patient treatment protocols.

• PD-L1 expression in tumor-associated lymph nodes shows the expected expression on macrophage subsets and particularly bright expression on lymph node dendritic cells (DC, https://doi.org/10.1016/j.ccell.2020.09.001). This observation may be important given the critical role of DC in tumor antigen presentation.

• Of keen interest, TLS, sites of active, ongoing anti-tumor immunity, are rich in activated PD-1-positive T-cells. TLS in association with tumors are closely correlated with patient outcome in response to anti-PD-1 or anti-PD-L1 treatment, regardless of PD-L1 expression within the TLS (https://www.nature.com/articles/s43018-021-00232-6).

The complexity of PD-1, PD-L1, 4-1BB and 4-1BBL expression and the diversity of potential cellular interactions within and outside of the TME, lymph nodes and TLS suggests that the simple hypothesis behind the PD-L1 x 4-1BB BiAb programs – that binding to PD-L1 will localize 4-1BB activation within the TME – is flawed. It seems more likely that unexpected interactions are taking place, including within the lymphoid organs. Orchestration of immune responses therein is highly complex and fiercely regulated. Thus, 4-1BB expression is controlled by signals triggered by the TCR and CD28, and normally transient. Further, coordinated interactions among these pathways leads to productive differentiation of T-cells, while temporally disconnected signals given in trans, as provided by activating anti-4-1BB antibodies, may be less productive (eg. https://www.nature.com/articles/s41467-021-27613-w). This is a NIMBY problem – which backyard is the right one? Immunologically, what immune responses are you really activating in patients receiving the BiAb therapies?

One more question.

Why do these programs trigger cytopenias, liver toxicities and (some) kidney toxicities? These adverse events suggest widespread, uncoordinated and nonspecific immune activation. The organ toxicities are, I think, especially unexpected, given the effort to either design conditional activation (anti-PD-L1 binds then anti-4-1BB becomes competent to bind) or differential affinity (PD-L1-high binding affinity, 4-1BB-low binding affinity). Liver toxicity, in this setting, may be a consequence of dysregulated immune responses in liver-draining lymph nodes (50% of lymph fluid originates in the liver). Since organ immunity is local immunity, T-cells and other lymphocytes activated by such dysregulation may very well traffic back into the liver and cause inflammatory damage. Justa thought.

Finally, there is an even larger amount of effort directed to anti-4-1BB x anti-tumor antigen BiAb programs eg. 4-1BB/Her2, 4-1BB/GPC3, 4-1BB/PSMA, 4-1BB/Claudin18.2 and so on. Will these programs fair better? I wouldn’t bet on it.

Stay tuned.