Trends
- Initially considered as an antiangiogenic therapy, MC is in fact a multi-targeted anti micro environment therapy.
- The fact that MC can directly target cancer cells and cancer stem cells has been neglected.
- MC can impact cancer cell metabolism.
- Several types of cell death, including immunogenic cell death, can be induced by MC.
- MC can impact the ecology of cancer clones without high selective pressure.
- Metronomic chemotherapy (MC) was initially described as an antiangiogenic therapy more than 15 years ago. Over the past few years, additional data have highlighted the impact of MC on the microenvironment beyond angiogenesis, with, most importantly, a potential impact on the immune system. Here, we review and reappraise the fact that MC might be able to directly kill cancer cells. Although long neglected, this question is of critical importance both fundamentally and clinically, especially when considering future associations with immunotherapies.
Metronomic Chemotherapy Is a Multi-Targeted Therapy
MC (see Glossary) has gained research interest over the past 15 years as a new strategy that could overcome resistance to chemotherapy [1, 2, 3, 4, 5, 6]. The introduction of MC also opened avenues for the development of oral, inexpensive, and well-tolerated treatments that might prevent tumor progression for an extended period of time [5. Although MC was initially reported to act through antiangiogenic mechanisms [2, 3], additional anticancer properties have since been unveiled. These include the stimulation of the antitumor immune response [7 and prevention of stromal activation [8; thus, MC is now regarded as a form of multi-targeted therapy [4, 5]. MC can be rationally combined with targeted and/or immunological therapies, further maintaining interest from both the scientific and medical communities.
One of the major paradigm shifts introduced by MC is that, instead of directly targeting cancer cells, this treatment could target the tumor microenvironment. However, both recent and old neglected evidence suggest that MC also directly targets and kills cancer cells. This might be important in the ever-changing landscape of medical oncology, in which targeted and immune therapies are gaining center stage [9 and could lead to a more rationale use of MC [3, 4, 5] (Figure 1, Key Figure).
Influence of Chemotherapy Dose and Schedule
When Skipper et al. [10 first introduced and demonstrated the log-kill effect for several cytotoxic agents 50 years ago, they concluded based on their experimental work in vitro that a large dose and/or short time schedule was superior to a chronic low-dose schedule with a similar or larger total dose. Of note, this was claimed to be true if the aim of treatment was to achieve complete eradication of the disease. The oncology community rapidly adopted the concept, which gave rise to the maximum-tolerated dose (MTD) paradigm and led to high-dose chemotherapy protocols that were made possible thanks to supportive care, such as peripheral blood stem cell support, broad-spectrum antibiotics, blood cell transfusions, and antiemetics [11. Nevertheless, the authors of the seminal paper never claimed that low doses of chemotherapy had no effect in vitro but only that the MTD approach was the best strategy to eradicate all cancer cells in vitro. Furthermore, they stated that the dose–effect concept was only valid for exponentially growing, non-mutagenic cells when anticancer agents acting independently of the cell cycle were applied. This implies that: (i) the model would not hold for all anticancer agents and would be of limited usefulness in vivo; (ii) the optimal protocol would depend on the mechanism(s) of action of the chemotherapeutic drugs used; and (iii) the duration of cancer cell exposure to treatment is critical [12. In this context, Kamen et al. proposed high-time therapy, a strategy that aimed to maximize the drug exposure duration at a given fixed drug concentration [13. In fact, the cell kill formula e−ket of the log-kill effect has both a concentration (c) and a time (t) component in the exponent. By modulating these two parameters, this formula can be applied for both MTD (short exposure and high concentrations) and MC (long exposure and low concentrations). Therefore, according to this formula, it is not surprising that longer drug exposure would contribute to increased cell killing even when using lower drug concentrations. As such, low-dose chemotherapy can have direct cytotoxic effects on cancer cells, as predicted by mathematical modeling and suggested by in vitro experiments [10, 12].
The influence of exposure duration on cell killing has been demonstrated, for instance, by the work of Shimoyama using sarcoma cells in culture. In these experiments, increasing exposure duration influenced the slope, threshold, and maximal cytotoxicity of all the agents studied [14, 15]. Although duration of exposure has the greatest impact when using cell cycle phase-sensitive anticancer agents, similar effects, albeit less intense, have been reported with a range of anticancer agents [12. For instance, Raymond et al. reported that both short-term (1 h) and prolonged exposure (14 days) to paclitaxel exerted significant concentration-dependent effects on the growth inhibition of ovarian, breast, and non-small cell lung cancer cells [16. Thus, prolonged exposure to 0.29-μM paclitaxel induced a cytotoxic effect that was three times stronger than a 1-h exposure to 2.9-μM paclitaxel, indicating that exposure duration is an important factor in the anticancer activity of paclitaxel in human tumors and long-term exposure may increase it. Similarly, a protracted (metronomic) schedule for the administration of topoisomerase 2 inhibitors has been demonstrated to be clinically relevant and sometimes superior to the MTD schedule [17. Thus, 1-h exposure to 2.2 and 22-μM topotecan resulted in 10% and 25% response rates in a clonogenic assay, respectively, while continuous exposure to 0.22 and 2.2-μM topotecan led to 34% and 76% response rates, respectively. These results suggested that topotecan was more active with long-term incubation and were confirmed by Winter et al., who explored the potential of metronomic topotecan and melphalan in four retinoblastoma cell lines [18. Using a metronomic schedule that enabled the authors to decrease the IC50 (i.e., concentration inhibiting 50% of cell viability) by a median of 13-fold compared with MTD treatment in vitro and in mice, continuous metronomic topotecan resulted in significantly smaller tumor volumes compared with conventional treatment (P <0.05).
Interestingly, the mechanism of action of a given anticancer agent can significantly differ when it is given metronomically compared with when it is administered following an MTD schedule. For instance, Harstrick et al. reported that the main mechanisms of 5-FU cytotoxicity depended on the mode of administration. Incorporation of fluorouridine triphosphate into RNA appeared to be the most important mechanism of action for 5-FU bolus schedules, while inhibition of thymidylate synthase became more important as the infusion time was prolonged [19. Other chemotherapy agents can display different mechanisms of action when administered following a metronomic or MTD schedule. Thus, gemcitabine can cause telomere shortening in HeLa cells by stabilizing telomeric repeat-binding factor 2 (TRF2) [20. Furthermore, Cadamuro et al. recently demonstrated that low-dose paclitaxel decreased nuclear expression of calcium-binding protein S100A4 in a model of cholangiocarcinoma [21. Interestingly, this regimen did not affect cell proliferation, apoptosis, or cytoskeletal integrity, but reduced cancer cell invasiveness in vitro and metastatic spread in vivo.
While the mechanisms involved in intrinsic and/or acquired resistance of cancer cells to MC remain limited [5, 22, 23], differences in mechanisms of action may, at least in part, explain the lack of systematic cross-resistance between MTD and metronomic schedules observed in animal models [3, 4] patients [5, 24, 25, 26], or in vitro [22, 23, 27, 28]. Chow et al. reported that in prostate cancer cell lines, PC-3 variants showed stable resistance to metronomic cyclophosphamide in vivo yet retained in vitro sensitivity to 4-hydroperoxy-cyclophosphamide (a precursor of cyclophosphamide) and other chemotherapeutic agents, such as doxorubicin or docetaxel [27. Sobrero et al. provided evidence of the lack of cross-resistance in human colon cancer cells exposed to short-term or continuous treatment with 5FU in vitro [28. Consistently, in women with breast cancer, rechallenge with metronomic capecitabine can lead to a response after standard capecitabine dosing has failed [25. This lack of systematic cross-resistance between MC and MTD chemotherapy may be important in the clinic, for instance, when considering the selection of patients who will receive metronomic maintenance. As an illustration, in the CAIRO3 study, patients who progressed on conventional capecitabine plus oxaliplatin plus bevacizumab (CAPOX-B) may have been retrospectively wrongly excluded from the subsequent randomization in which patients either were treated with metronomic maintenance capecitabine plus bevacizumab or had their treatment stopped [29.
Elsewhere, these different mechanisms of action may result in differential effects on cell death. While anticancer drugs commonly kill cancer cells via apoptosis, low-dose MC can induce different types of cell death. For instance, Cortes et al. reported that low doses of actinomycin D inhibited proliferation and induced apoptosis in vitro, as well as tumor regression in vivo, in a p53-dependent manner in a model of subcutaneously implanted neuroblastoma. However, a pan-caspase inhibitor only partially inhibited cell death induction, suggesting that the treatment could activate an apoptosis-independent cell death pathway [30.
MC alone or in combination can also induce immunogenic cell death (ICD) [31. Several anticancer agents can trigger ICD when used alone or when given in a MTD manner. Nevertheless, various traditional anticancer agents routinely used in the clinic (e.g., anthracyclines, cyclophosphamide, bleomycin, bortezomib, and oxaliplatin) can also trigger ICD when used in a metronomic fashion [5, 31]. Interestingly, the continuous ICD effect of MC might also be leveraged by the concomitant stimulation of the immune system obtained with MC. This paves the way for combinations with specific immunotherapies that can actively engage the immune system against cancer cells [5, 31, 32, 33, 34]. For instance, Liikanen et al. [33 showed that a combination of an oncolytic virus with MC, temozolomide, and cyclophosphamide led to ICD in vitro in prostate cancer PC3-MM2 and breast cancer MDA-MB-436 cell lines. The same combination given to patients with refractory tumors led to disease control in 67% of patients. Similar findings were obtained by Tongu et al. [34 when using cyclophosphamide or doxorubicin in a model of CT-26 carcinoma cells.
Elsewhere, Taschner-Mandl et al. reported that low-dose topotecan could lead to senescence in neuroblastoma cells [35. Notably, topotecan-treated senescent tumor cells acted as a growth inhibitor in a dose-dependent manner on non-senescent tumor cells and MYCN expression was significantly reduced in vitro and in vivo given that, in a mouse xenotransplant-model for MYCN-amplified neuroblastoma, metronomic topotecan led to senescence selectively in tumor cells, complete or partial remission, and prolonged survival. Similarly, daily low-dose hydroxyurea also induced senescence in MYCN-amplified neuroblastoma, suggesting a new mode-of-action for MC [36. Conventional treatments, such as chemotherapy and radiotherapy, can induce premature senescence instead of apoptosis; therefore, treatment-induced senescence could compensate for resistance to apoptosis through alternative signaling pathways.
Cancer Metabolism and Metronomic Chemotherapy
The field of energy metabolism in cancer biology is growing quickly. As a result, the regulation of metabolism is currently emerging as a strategy of choice to control tumor progression [37. Recent studies have shown that MC could impact tumor metabolism. For example, Yapp et al. demonstrated that metronomic gemcitabine decreased glucose metabolism in patient-derived xenografts of pancreatic cancer [38. Tumor metabolism was measured with PET scans to evaluate glucose consumption and metabolic activity as a marker of cell viability and proliferation. Interestingly, FDG uptake was reduced in tumors treated with metronomic gemcitabine and the levels of Ki67 and TUNEL staining in frozen sections in the metronomic treated group were also lower than those in the control group. Overall, these results suggest that metronomic gemcitabine exerts a cytostatic effect on the tumor where cancer cells are still alive, but are proliferating less. Elsewhere, it has been reported [39 that the combination of metronomic paclitaxel and the AKT inhibitor perifosine led to increased overall survival in mice with non-small cell lung cancer compared with controls or single-agent treatment. The efficacy of the paclitaxel–perifosine combination was linked to inhibition of the two major bioenergetic pathways: oxidative phosphorylation and glycolytic metabolism.
There is also growing interest in the role of cancer stem cell (CSC) metabolism in carcinogenesis and treatment. While there is still no consensus on the metabolic characteristics of CSCs (i.e., their glycolytic phenotype and oxidative state), they have become a major focus in cancer research, and significant efforts are being made towards discovering druggable targets [40 because MC may also target CSCs.
Impact on Cancer Stem Cells and Clonal Heterogeneity
Indeed, long-term chronic exposure of prostate and colon cancer cells to paclitaxel and etoposide in vitro led to the generation of ‘drug-tolerant’ cells via epigenetic mechanisms [41. Interestingly, such drug-tolerant cells exhibited impaired tumorigenicity in vivo with fewer CD44+ cells compared with the parental cell population, demonstrating that MC could target CD44+ tumor-initiating cells. Furthermore, Folkins et al. demonstrated that metronomic cyclophosphamide could significantly reduce both primary and secondary glioma spheroids isolated from drug-treated patients [42. Vives et al. also reported that metronomic cyclophosphamide reduced the number of CD133+ precursor cells and triple-positive CD133+/CD44+/CD24+ cancer stem cells in human pancreatic tumour xenografts [43. Recently, Chan et al. showed that the same overall dose administered as MC (versus a MTD regimen) prevented therapy-induced stromal ELR+ chemokine paracrine signaling to trigger the phenotypic conversion of carcinoma cells into CSCs and promote their invasiveness [8. In turn, this translated into an enhanced treatment response and an increase in mouse survival.
Mathematical modeling of MC can also provide valuable information [44. Kareva recently proposed that CSCs can prevent the antitumor immune response by generating a ‘protective shield’ of non-stem cancer cells around them [45. This shield protects CSCs by creating a physical barrier between them and the immune system and by promoting competition for resources. The author further tested different therapeutic strategies in silico and proved that MC could lead to a sustained ‘peeling off’ of the outer layers of the tumor, which protects the CSCs. Given that MC does not annihilate the immune system, it provides immune cells with access to the tumor core, which contains CSCs.
Lastly, MC might also interfere with the Darwinian ecosystem driving the clonal evolution of tumor cells in which subclones compete and/or cooperate with each other for space and resources [46. Phenotypic resistance to chemotherapy in cancer cells is associated with the lower proliferation of resistant cells in the presence of ‘fitter’ chemosensitive cells [47. Hahnfeldt et al. [48 reanalyzed the issue of chemotherapy dose fractioning taking into account tumor heterogeneity. Considering two subpopulations of cancer cells with two distinct sensitivities to a given anticancer agent, with a possible transition rate from the less sensitive population to the most sensitive allowing for a resensitization effect, the authors demonstrated that more regularly spaced dosing of the drug led to increased antitumor responses compared with irregular spacing. MC might also impede ‘subclonal switching signals’ to prevent dormant subclones (such as CSCs) from shifting to dominating subclones that can, in turn, promote tumor progression or post-treatment recurrence [49. It was recently shown in a lung cancer model that MC could help restrain the proliferation of the most resistant clones [50, paving the way for strategies alternating metronomic and higher doses of chemotherapy in line with the chemoswitch concept [51, 52]. Similarly, Liao et al. developed a tool called a metronormogram to optimize the dose frequency of treatment for heterogeneous tumors to maximize the treatment benefit by controlling the composition of the cancer cell population [53]
Concluding Remarks
MC is defined as a multi-targeted therapy aimed at the tumor microenvironment. Its potential direct anticancer effects have been somewhat neglected. We do not wish to question here the anti-micro environment effects of MC but instead wish to shine light back on its direct effects on cancer cells and how these effects can impact our understanding of the mechanisms of actions of MC and its use in the clinic (see Outstanding Questions). For instance, we believe that this largely unexplored aspect of the mechanisms of action of MC is critical in the context of immunotherapies. The capacity of MC to deplete regulatory T cells and myeloid-derived suppressor cells, induce the maturation of dendritic cells, and trigger the liberation of tumor antigens through the induction of immunogenic cell death, makes it an ideal partner for combining with immune checkpoint inhibitors, as reported recently [54, and with other immunotherapies [4, 5, 9]. Additionally, because clonal heterogeneity acts as a frontier to targeted therapies [55, the broad capacity of MC to modulate clonal interactions, as well as its cytotoxic potential, should be re-evaluated.