Controversy and Consensus: Short perspective Translating the science of cancer dormancy to the clinic.

Abstract

The paradigm of metastasis has been significantly remodeled by the incorporation of cancer dormancy as a mechanism to explain long-term remission intervals followed by relapse. There is overall consensus on the potential impact of better understanding dormancy. Key cancer cell autonomous and microenvironmental mechanisms might explain this biology and, in turn, the timing of metastasis. However, the approach and feasibility to apply this biology to clinical trials has been controversial. The discussion here provides insight into how these controversies are being resolved by the development of active clinical trials, thus bringing to reality opportunities to target cancer dormancy.

Consensus: cancer dormancy is an important biological process to understand and target.

It is clear that the concept of cancer dormancy(1), and probably more so among those studying metastasis(2), is now common in discussions, articles and research programs addressing the mechanisms of relapse. For some time the National Cancer Institute and other funding and advisory agencies such as the Department of Defense and the National Breast Cancer Coalition have had the biology of cancer dormancy on their radar (3) and provocative questions, RFAs and topics in study sections include this theme. Further, as evidenced in the recent Cancer Grand Challenge initiative by the Cancer Research UK and the NCI, cancer dormancy is seen by the leadership in these organizations as a main problem in cancer that must be solved.

In the last ten years publications on cancer dormancy have grown almost exponentially covering three main areas, angiogenic dormancy, cellular dormancy and immune-mediated dormancy (1). Recently, the most prevalent topics focus on cancer dormancy at the cellular level, that is cancer cells entering a prolonged growth (G0/G1) arrest despite carrying pro-mitogenic and pro-survival mutations, and how this process is controlled by stromal, vascular and other cellular and extracellular matrix niches(4). Thus, the data points to an inextricable connection between cancer cells and the microenvironment in controlling cancer cell dormancy. Further, a more detailed analysis is developing around the relationship between innate and adaptive immunity and cancer cell dormancy (4).

Effort has been placed on using in vivo and/or in vitro 3D culture systems to discover ways to model cancer cell dormancy, induce dormancy of cancer cells or reveal vulnerabilities that would render them sensitive to new dormancy-specific therapies such as unfolded protein response inhibitors or when blocking adhesion signaling, to conventional antiproliferative therapies(5,6). These include targeting processes such as autophagy, metabolism, stress signaling, nuclear receptors, adhesion signaling and using drugs that reprogram the epigenome to induce and maintain dormancy (1). The available body of work and current active projects on dormancy clearly indicate that investigators see this stage of the metastatic cascade as a new opportunity to target the metastatic process. This is because the discovery that disseminated cancer cells (DCCs) can originate very early in cancer evolution(7) and that early or late evolved DCCs can enter decades-long periods of dormancy, reshaped our notion of metastasis targeting. For the first time we gained insight into an unexpected biology, that is that cancer pauses its progression, revealing a window of opportunity for early intervention. Importantly, across several experimental and human studies it has become clear that even stage IV patients carry still dormant cancer cells that co-exist with growing lesions in the same or separate target organs(1). The implication is that even if a detectable lesion or lesions respond to therapy, undetectable DCCs evade therapy and persist only to later activate and complicate the patient’s future prognosis. Thus, targeting dormant cancer may not only be applicable in the adjuvant setting but also in stage IV complementing anti-proliferative, radiation and/or surgical therapies. Metastasis has always been seen as the last battle for life and this notion has been focused mainly on the clinical reality that once lesions are overt they need to be removed by surgeons or targeted as best as possible by radiation, chemo, immune or targeted therapies(7). While success stories always exist and there is continued progress, the hard truth is that metastatic disease once manifested as overt in a systemic fashion is very difficult to stop.

Controversy: can we ever design a cancer dormancy-targeting clinical trial to prevent relapse?

The question is how and when can we harness the biology of cancer dormancy to prevent and treat metastatic disease. Several questions emerge in these discussions: what biomarkers, systemic or cancer cell-specific can be used to find dormant cancer cells or niches that are in pro-dormancy vs pro-reactivation states? How do you design a dormancy-targeting clinical trial? How do we make sure we are “hitting” the target? How long does it take to read out a dormancy trial? The latter is the most troubling question for those in the clinical and pharmaceutical space because of the common paralyzing misconception that cancer dormancy trials must be only adjuvant trials that take many years or a decade to complete and thus will be incompatible with industry timelines to support testing a new treatment.

The above questions have answers (or it is known where to look for them) and there are opportunities to start probing these concepts without running into cripplingly large budgets and long timelines of exclusively adjuvant trials. Nevertheless, more work is needed to check all the boxes to advance the approach of targeting cancer dormancy. While labs push the boundaries of research forward finding new targets in dormant cancer cells, critical effort will be needed to provide human sample correlate analysis (beyond the customary Kaplan-Meier curve) for these targets on DCCs or host compartments in residual disease patients. The limited analysis of samples during residual disease seems to be more common in solid cancer analysis than in hematological malignancies where probing the bone marrow to profile residual multiple myeloma or leukemia cells is a more common practice. Thus, as we discover targets and markers of dormancy or signals from the niches that induce or reverse dormancy these should be tested in human specimens. Some of these examples have been reviewed recently(4). For example, a recent study showed (8) that detection of the nuclear receptor NR2F1, a dormancy marker and potential target identified in experimental head and neck, prostate and breast cancer systems, can stratify DCCs into NR2F1^high and NR2F1^low profiles. This profiling of residual disease in breast cancer patients revealed that patients that had NR2F1^high DCCs had longer bone metastasis-free periods than patients that carried NR2F1^low DCCs. This could be simply attributed to the growth arrest function of NR2F1, because its expression was inversely correlated with Ki67 expression. However, Ki67 detection in parallel samples could not inform on the rate of metastatic relapse. Thus, dormancy markers appear to provide information on not only proliferative capacity but also on additional transcriptional programs that maintain DCC phenotypes for long periods. Another approach to validate in patients’ data from experimental dormancy signatures from a pancreatic cancer mouse models was recently reported (9). The authors performed singe cell RNAseq from primary tumors and from liver samples resected during primary lesion surgery that were apparently normal and that contained DCCs. This approach, although in a small number of patients, showed that dormancy signatures in the experimental models were traceable in human solitary DCCs in patients. These examples show how the biology of dormancy from research efforts can be probed and validated in human samples. Another approach is testing whether the host is able to produce the signals that promote dormancy and if present or absent, how do these inform on patient outcomes. In a recent study, it was demonstrated that NG2+/Nestin+ mesenschymal stem cells (MSCs) that control hematopoietic stem cell (HSC) dormancy in mice control dormancy of breast cancer DCCs by producing produce BMP7 and TGFβ2. Analysis of patient samples from a clinical trial where bone marrow supernatants were collected through the course of treatment(10)showed that patients that had bone metastatic relapse, more commonly had no detectable levels of TGFβ2 and BMP7 in the bone marrow supernatant. Analysis of the metastasis-free proportions showed that breast cancer patients positive for BMP7 in the bone marrow supernatant lived longer without metastasis than those that had undetectable BMP7. These data further support that not only DCCs may be a source of information but surrogate measures of functional pro-dormancy niches (producing TGFβ2 and BMP7 for example) may also inform on the state of dormant cancer cells in patients(10). Understanding how the niches that harbor DCCs change under specific circumstances (e.g., aging, immunosuppression, inflammation) is key to designing strategies to correct the niche function.

The above and other examples suggest that there are clear opportunities to probe the biology of dormancy in patient samples and obtain useful information. But have there been any clinical trials designed around the biology derived from cancer dormancy studies? The answer is yes. For example, a Phase II clinical trial was motivated by findings showing that that during dormancy, cancer cells can activate autophagy(1). This trial takes advantage of available FDA approved autophagy inhibitors (e.g., hydroxychloroquine (HCQ)) in combination with mTOR inhibitors to treat breast cancer patients during remission phases and thus prevent loco-regional or distant recurrences (Clinical trial identifier: clinicaltrials.gov identifier: NCT03032406). Another study (Clinical trial identifier: clinicaltrials.gov identifier: NCT04841148) will test safety and early efficacy of HCQ or a PD-L1 blocking antibody, with or without a CDK4/6 inhibitor, in early-stage ER+ breast cancer patients. The patients to be treated are those who harbor DCCs in the bone marrow after definitive surgery and standard adjuvant therapy and thus are at risk of relapse. This trial appears to be designed following the notion that autophagy is turned on in dormant cancer cells fulfilling a survival function and that there are available drugs to inhibit this pathway. Further the use of CDK4/6 inhibitors may allow creating a vulnerable population of DCCs to HCQ or PD-L1 targeting. There is ample evidence that mTOR inhibitors induce autophagy across various cancer types and thus, autophagy inhibitors may show improved combinatorial DCC killing effect, but only in cells that maintain this pathway active (as tested in clinicaltrials.gov identifier: NCT03032406). CDK4/6 inhibitors, although less studied as autophagy inducers, can also activate this catabolic pathway in some cancer types and thus, there may be promise in arresting still proliferative DCCs and then sensitizing them to cell death with HCQ or other autophagy inhibitors. A similar trial design could be used in a short term clinical trial by testing dormancy markers as intermediate clinical endpoints for ER+ breast cancer to resolve the question of whether for example treatment with letrozole and/or CDK inhibitors induce dormancy of residual CTCs and DCCs and whether dormancy cues that can be detected in circulation inform on potential pro-dormancy niches. These trials could be done by sampling patients under current standard of care in a year or less. Importantly, a separate clinical trial is attempting to screen for the presence of bone marrow DCCs in patients within 5 years after having being treated for breast cancer (Clinical trial identifier: clinicaltrials.gov identifier: NCT02732171). One of the trial’s goals is to have patients that have DCCs in their bone marrow and therefore at a higher risk of developing metastasis to be offered to participate in other trials targeting DCCs such as NCT03032406. Studies on dormancy and mobilization of DCCs from the bone marrow(1) led to a new Phase Ib trial (Clinical trial identifier: clinicaltrials.gov identifier: NCT04197999) for GMI-1359, a small molecule, which concomitantly blocks E-selectin and CXCR4 function, both molecules implicated in DCC BM retention as well as metastatic progression. While safety is the main objective of this trial, other secondary objectives include pre- and post-dose circulating tumor cells (CTC) enumeration to determine tumor cell mobilization. Such trials, could be adapted to understand the frequency of dormant DCCs in stage IV patients and to validate markers of dormancy discovered in the lab. Along the same lines this trial could also be used to test if GMI-1359 affects dormancy pathways. Other trials such as clinicaltrials.gov identifier: NCT00072020 or clinicaltrials.gov identifier: NCT00172068 where zoledronic acid was found to reduce bone metastasis and to improve disease outcomes for women with established menopause or to eliminate DCCs from the bone marrow with a concomitant decrease in bone metastasis, respectively may be useful to test for dormancy markers in collected biospecimens that carry DCCs or to probe for pro-dormancy cues produced by BM niche cells as discussed above(10). This way questions on how dormancy is modulated by current therapies may also be answered. Using similar approaches for biomarker readouts as in (8–10) in a relatively short clinical trial samples could be collected before and after initiating the letrozole and/or CDK inhibitor treatment. These are a few examples on how clinical trials can be designed following the biology revealed from studying cancer dormancy. Additional information on these trials can be obtained following the provided identifiers.

It was shown that dormancy is defined by a specific epigenetic state and that malignant DCCs can be reprogrammed into this phenotype ((8) and references within). These data led to a clinical trial where disseminated PCa residual disease that is not yet detectable by imaging (biochemical or PSA-detected recurrence) is reprogrammed into dormancy (Clinical trial identifier: clinicaltrials.gov identifier: NCT03572387). This clinical trial repurposed 5-azacytidine and all-trans retinoic acid (both FDA approved drugs) to treat PCa patients after hormonal ablation. This study will also enumerate CTCs and detect NR2F1 (a gene induced by the 5-azacytidine and all-trans retinoic acid therapy ((8) and references within) in these CTCs to determine the reprogramming success. In addition plasma levels of cues that induce dormancy(1) such as LIF, GAS6, BMP4 and BMP7 are also measured to determine the systemic pro-dormancy or pro-reactivation state of the host upon treatment. This is proof that with careful thinking, the basic science on cancer dormancy can developed into innovative trials that do not represent unsurmountable financial hurdles. These trials could be adapted to also use ctDNA or sequencing of DCC genomes to understand how dormancy and the used therapies shape the evolution of disseminated residual disease. Overall, identifying targets and how to read them out, developing markers and methods to detect the DCCs and dormancy- or reactivation- supportive niches during MRD in trials will help better understand how to target these residual cancer cells based on their unique biology. The ongoing work in this space will likely impulse a new era of clinical trials tailored to target dormant cancer cells in all their manifestations and put an end to the controversies associated with this path that is no longer theoretical and it is being reduced to practice.