Considering therapy-induced senescence as a mechanism of tumour dormancy contributing to disease recurrence

Abstract

The capability of tumour cells to escape from therapy-induced senescence, as well as cell-non-autonomous functions of senescence, support the premise that senescence could serve as one pathway to tumour dormancy (among others that include quiescence and diapause) that is permissive for disease recurrence. Consequently, the pharmacologic targeting of senescent tumour cells could mitigate the risk for cancer resurgence, thereby enhancing the therapeutic efficacy of cancer chemotherapy.

Main

Two articles that were recently published in the Controversy and Consensus section of Cancer Research, as well as one in Cancer Discovery, address, either directly or indirectly, a fundamental issue in cancer therapy, that of tumour dormancy and disease recurrence. Here, we integrate these articles to indicate the potential involvement of senescence and escape from senescence in these processes. Senescence is a state of proliferative arrest characterised by metabolic dysregulation, increased lysosomal biogenesis (marked by the upregulation of the senescence-associated beta-galactosidase, SA-β-gal), epigenetic and gene expression alterations, and a secretory phase termed the senescence-associated secretory phenotype (SASP). The capability of tumour cells to undergo senescence is conserved, despite their immortality, and is precipitated by exposure to a wide spectrum of antitumor drugs, radiation and cellular stresses [1]. Accordingly, senescence represents a fundamental component of the tumour cell response to anticancer therapy, both in vitro and in vivo, and likely also in cancer patients.

Potential of therapy-induced senescent cells to escape the stable growth arrest

The scientific community has been somewhat hesitant in accepting the possibility that senescence is not a permanent form of growth arrest; this is largely due to the longstanding paradigm that senescence is irreversible. It is for this reason terms such as “pseudo-senescence” and “senescence-like” have entered the literature in studies where cells with clear hallmarks of senescence were shown to recover proliferative capacity. Wang and Demaria et al. cite several reports where drugs such as doxorubicin and the CDK4/6 inhibitor, abemaciclib, induce SA-β-gal, the standard indicator of senescence, but retain proliferative potential [2]. These authors argue that SA-β-gal alone is “not a reliable strategy to evaluate therapy-induced senescence (TIS)”, and that multiple senescence markers must be combined. While we fully agree with this premise, most studies in the literature where TIS is identified do, in fact, rely on a multiplicity of hallmarks mentioned above [1]. What is critical to emphasise, however, is that the ability of tumour cells to retain self-renewal capacity does not indicate that the phenomenon being observed is something other than “true” senescence. Studies from our own laboratories have demonstrated that tumour cells can escape/re-emerge from senescence induced by therapies as diverse as conventional chemotherapy, radiotherapy, targeted therapy and androgen deprivation [1]. Other laboratories have substantiated these observations in multiple models [1]. Furthermore, it is well established that cells escape from other forms of senescence, such as oncogene-induced senescence (OIS) by derepression of hTERT gene expression and chromosomal rearrangements that allow for expression of proliferative drivers [3], and from replicative senescence (RS) through telomere maintenance and loss of signalling elements, such as p53 and p16^INK4a, that otherwise maintain the stable senescent growth arrest [4].

In this context, the elegant studies by Duy et al. using primary patient acute myeloid leukemia (AML) cells firmly establish that what the authors conservatively term a “senescence-like resilient phenotype” permits the AML cells to “survive and repopulate leukemia” [5]. This work demonstrated that moderate-dose chemotherapy induces a senescent state in 15 primary patient-derived AML samples based on the enrichment for senescence/SASP transcriptomic signatures and significant SA-β-gal staining [5]. Yet, a proportion of those chemotherapy-induced senescent AML cells are able to regain the ability to replicate within 9 days of drug exposure [5]. While acknowledging the potential for the existence of different subsets of senescence, such as a “deep” form from which few cells can escape and a shallower form that is more permissive for escape, it is obligatory to discard the stance that TIS in tumour cells is irreversible if the field is to move forward.

Therapy-induced senescence as a mechanism of tumour dormancy

Considering the quite extensive evidence that now exists in the literature for repopulation by subpopulations of senescent tumour cells, it is reasonable to argue that senescence is likely to represent one form of tumour cell dormancy. However, consideration of senescence as a mechanism of “cellular dormancy” does not, of itself, eliminate the potential for a quiescent state, or other phenotypes of prolonged arrest contributing to disease relapse when a subset of these cells escape the G₀ growth arrest [6]. Accordingly, a heterogenous surviving, but growth-arrested tumour cell population following chemotherapy or radiation that includes quiescent, senescent, diapause and “persister” cells can contribute to disease recurrence and patient relapse in a cell-autonomous fashion [7].

This concept is supported by molecular and functional characteristics of senescence that fit within current paradigms relating to tumour dormancy (Fig. 1). For example, while a state of senescence may facilitate the clearance of tumour cells by innate immune system components, recent evidence has suggested that the interaction between senescent cells and the immune system is more complex. Work by Muñoz et al. rigorously demonstrated that some senescent tumour cells escape immunosurveillance utilising MMP-dependent shedding of NKG2D ligands in an autocrine fashion, and suppressing NKG2D receptor-mediated immunosurveillance in a paracrine fashion, indicative of a role for the SASP in evading immune recognition [8]. A seemingly apparent contradiction to the plasticity of dormant cells is that while dormant cells are characterised by activating pluripotency programs orchestrated by chromatin (epigenetic), senescent cells primarily undergo repressive heterochromatic alterations. However, Milanovic et al. have shown that a  subset of therapy-induced tumour cells exhibited a significant upregulation of an adult tissue stem-cell-like transcriptomic signature, thereby acquiring an enhanced tumour initiation potential in vivo [9]. While impaired angiogenesis is an established component of the dormant state, it is currently uncertain whether senescent cells can alter the regional blood supply. Nevertheless, recent work by Li et al. showed that senescent cells contribute to impaired angiogenesis and endothelial dysfunction through the SASP [10]. Moreover, Aguirre-Ghiso et al. highlight ongoing clinical trials based on the finding that cancer cells activate autophagy during tumour dormancy [11]. Given that senescence is invariably induced in virtually all preclinical studies whenever autophagy is observed, it would appear likely that senescent cells survive, in part, through the maintenance of autophagy-associated metabolic activity. Furthermore, recent studies also suggest that cell-to-cell cannibalism might sustain the senescence arrest.

Fig. 1. Hypothetical model of the contribution of senescent tumour cells to tumour dormancy.

This figure provides a model depicting how senescent tumour cells can contribute to dormancy. After a period of arrested growth, and a stable metabolic state mediated by autophagy, senescent tumour cells can escape via recovery of self-renewal capacity. Alterations in the gene expression profile of senescence escapers can generate stem-cell-like characteristics and more aggressive phenotypes. Through the senescence-associated secretory phenotype (SASP), senescent tumour cells can promote the proliferation of adjacent, possibly quiescent, tumour cells and facilitate their acquisition of enhanced invasive and migratory traits in a cell-non-autonomous fashion. Moreover, senescent tumour cells are able to evade recognition and elimination by immune cells through disabling certain ligand-receptor interactions. Overall, senescent tumour cells can represent one form of cellular dormancy that is permissive for disease recurrence and facilitate the development of an aggressive state in senescence-naïve tumour cells through the SASP. This figure was generated through biorender.com.

The possibility that elimination of senescent tumour cells might delay disease recurrence

Extensive studies in the literature have now established that tumour cells exhibiting senescence-associated hallmarks are susceptible to elimination by senolytics [12]. If our premise that tumour dormancy is associated with senescence ultimately proves to be correct, then senolytic strategies may also serve as adjuvant therapy to clear dormant senescent cells that might otherwise recover self-renewal capacity. Current limitations to this approach, however, are that some highly effective senolytics are relatively toxic, while other less toxic alternative have not proven to be effective against senescent tumour cells [13]. Nevertheless, senolytics appear to hold promise for delaying or interfering with disease recurrence from tumour dormancy.

Author contributions

TS and DAG contributed to the conceptualisation and writing of the manuscript.

Funding

None.

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

Not applicable.

Consent to publish

Not applicable.