Immune evasion by dormant disseminated cancer cells: A Fermi paradox?

Abstract

Rare disseminated tumor cells (DTCs) can persist after treatment in patients for years, and the immune system does not eliminate them. Goddard et al. propose that immune evasion by rare dormant DTCs is due to an improbability of contact imposed by large distances separating effector T cells and DTCs.

Lethal metastases in secondary organs arise from disseminated tumor cells (DTCs), which can disseminate before a primary tumor is detectable and remain clinically dormant for years before reawakening. DTC dormancy is characterized by co-existing quiescence and survival programs commonly instructed by target organ niches.¹ Targeting dormant DTCs that predate treatments or that emerge as “persister” DTCs is a major clinical challenge as they evade anti-proliferative therapies. New approaches include the induction maintenance of DTC dormancy and/or targeting adaptive stress (unfolded protein response, UPR) and metabolic (autophagy) survival pathways.²

Immunotherapies such as immune checkpoint inhibitors, cancer vaccines, and CAR-T cells should in principle eliminate dormant DTCs. Immune surveillance is highly sensitive and should detect and eliminate rare events of cancer cells. In fact, vaccination is based on this principle. However, in cancer, growing evidence supports that quiescent DTCs can escape immune surveillance, which may explain the partial success of immunotherapies.

It is known that organ transplant recipients who were cancer free and received immune suppressants for organ engraftment can develop metastatic disease in the transplanted organ.¹ This suggests that rare clinically dormant cancer cells are kept in check by the immune system. Further, in immunization mouse models, extremely rare dormant-growth arrested acute myeloid leukemia cells actively evade immune detection by upregulating immune checkpoint proteins (PD-L1 and CTLA-4 ligands) to avoid T cell activation.³ Later, it was shown that pancreatic cancer DTCs downregulate major histocompatibility complex (MHC) class I through UPR activation, rendering dormant DTCs undetectable by CD8⁺ T cells.⁴ Recently, it was also shown that quiescent DTCs avoid natural killer (NK) cell clearing, and only upon awakening, STING activation makes them recognizable by the immune system.⁵ Even more complex, NK cells may induce DTC dormancy via active interferon (IFN) signaling.⁶ Thus, rare dormant cancer cells can display active mechanisms of immune evasion: but is this the only reason for their persistence?

The current paper by Goddard et al.⁷ proposes an intriguing alternate or complementary scenario to those proposed above that parallels the Fermi paradox (Figure 1). This paradox questions why we have not encountered intelligent life (that probabilistically could exist) in the universe and argues that one reason is that unsurmountable distances prevent any contact.

Figure 1. Distance between dormant DTCs and CD8⁺ T cells as a potential mechanism of immune evasion.

The putative model in humans proposes that disseminated tumor cells (DTCs) that reactivate and form metastasis detectable in M1 patients (A) could be surveilled by relatively fast-moving (10–25 μm/min) CD8⁺ T cells that, due to the high frequency of cancer cells, are in short range (average of <500 μm distance) for interaction and MHC class I-specific engagement. In M0 patients that have only residual disease post-surgery or minimal residual disease (MRD) patients that responded to a given therapy, only residual DTCs persist (B), and they are so rare that even fast-moving patrolling CD8⁺ T cells somehow are always at a radial distance larger than 500 μm and never encounter them, preventing their clearance (C). The authors propose that by increasing CD8⁺ T cell frequency, the distance is reduced, and DTCs can be cleared. It is possible that the distance problem is further complicated by immune evasion mechanisms that may be present in the rare DTCs (red box, C) or in the niches (green box) where these DTCs reside, such that even if low-frequency encounters occur, T cell activation is still precluded even if T cell therapies aiming at increasing the frequency of effector T cells are employed (C).

Goddard et al.⁷ elegantly demonstrate that in a model of DTC dormancy, using a strong antigen (GFP) and immunization strategies, MHC class I downregulation by single DTCs, while detected, is not strong enough to preclude antigen-specific MHC class I-directed CD8⁺ T cell attack (Figure 1). However, residual DTCs persist in very low numbers in the lung and likely less in the bone marrow (BM). Thus, they propose that immune evasion by rare DTCs is mainly about large distances separating the low numbers of remnant DTCs and antigen-specific CD8⁺ T cells, which they termed “relative scarcity”; they simply never meet (Figure 1).

Their motivation emerges from the well-known data that single DTCs that can be probed in the clinic in the BM are a very rare population.² They propose that this scarcity (>500 μm radius; Figure 1) decreases the probability of T cells encountering DTCs close enough to kill them (Figure 1). In support of this approach, two strategies tested in lungs aimed at increasing the T cell/DTC ratio as adoptive transfer of transgenic GFP-specific CD8⁺ T cells or T cells from vaccinated animals increased DTCs clearance. Interestingly, the effect of these strategies was independent of MHC class I levels as CAR-T cell therapy (MHC class I independent) toward engineered antigens in dormant DTCs was also able to clear them.

The DTC relative-scarcity hypothesis proposes an interesting explanation for the failure of DTC immune surveillance. However, it also raises new questions about how this model fits existing evidence of active immune evasion and immune surveillance (e.g., CD8⁺ T cells have a 10–25 mm/min speed) as a general process (Figure 1).

For example, is the relative-scarcity hypothesis applicable to all organs, or do active mechanisms play a role in specific niches? The authors also ponder this question. It is known that the lung, where the principle is truly tested, has a higher burden of DTCs than the BM. As the authors show, the BM niche still carries EGFP⁺ DTCs that co-exist with EGFP-specific T cells. It would be important to know if, in this niche, DTCs activate other immune evasive mechanisms. Also, because D2.0R cells injected in animals show 10-fold more prevalence in the lungs than in the BM,⁸ could this be informing on a threshold for the relative scarcity mechanism in different organs? It will be important to unravel the complementarity of these mechanisms in different niches.

As the authors acknowledge, the tested T cell-based immunotherapies left behind 2%–10% of the original DTC population in the lungs. Why do these DTCs persist despite antigen-specific targeting? They propose that by increasing the dose of the CD8⁺ or CAR-T effector cells, they could overcome relative DTC scarcity. Inoculation of 230 JEDI T cells was sufficient to induce significant DTC clearance in the lungs; interestingly, a 10×, 100×, and 1,000× increase in injected T cells showed additional, but not proportional, DTC clearance. Only when the highest dose of T cells was used (2.3 × 10⁶), DTCs were further cleared but without a proportional gain in effect. It is thus tempting to speculate that, while increasing the probability of encounter does improve clearance, other mechanisms are still at play. As discussed by the authors, it would be worthwhile testing if higher doses of T cell inoculation would achieve a 100% DTC elimination in lungs (and bones) or if it will plateau as already observed in the experiments. If the remnant MHC class I⁺ DTCs are more immunosuppressive, perhaps via the expression of immune checkpoint proteins, remains unexplored (Figure 1).³

It is also important to consider that Goddard et al.⁷ utilize experimental metastasis models where cell lines already predisposed to enter dormancy are injected in bulk into circulation and carry highly penetrant neoantigens such as EGFP, HA, and NY-ESO-1, which are readily targeted by engineered or immunized T cells. It would be important to test whether the relative-scarcity mechanism is operational when using genetically engineered or spontaneous models of cancer that proved the existence of immune surveillance⁹ where the cancer coevolves since early stages with the immune system. Another variable to consider is the evolutionary age of the DTCs as cancer cells that disseminate early carry less evolved genomes and in principle less neoantigens¹⁰ that likely render early DTCs less immunogenic.

The authors also test CAR-T cell killing with HER2, a more relevant breast cancer antigen. Anti-HER2 CAR-T cell therapy is currently in clinical trials (NCT03500991). This strategy cleared a maximum of 74% of DTC in the lung, but similar questions regarding the scarcity in different niches also apply to this third line of testing that used immune-compromised animals.

In summary, this study provides a tantalizing possibility that increasing the dosage of effector CD8⁺ T cells would be sufficient to drastically reduce minimal residual disease. This may already improve outcomes. Perhaps combining their approach with immune checkpoint inhibitors, STING activators, or blockers of immunosuppressive cytokines may improve dormant DTC eradication. For the CAR-T approach, identifying a strong neo-antigen will also be a rate-limiting step. Additionally, considering other immune cells (e.g., macrophages, dendritic cells [DCs], CD4⁺ T cells, NK cells) may need to be explored as the literature cited here suggests that they could also be leveraged to eradicate DTCs (Figure 1). Perhaps in the clinical setting, understanding the minimal residual disease burden quantitatively may allow determining if the approach proposed by Goddard et al.⁷ would be effective for specific patients. Overall, this study tests an intriguing new hypothesis to find new ways to target dormant DTCs, which would have an enormous impact for cancer patients.