Polytherapy and Targeted Cancer Drug Resistance

Abstract

A current challenge in cancer treatment is drug resistance. Even the most effective therapies often fail to produce complete and durable tumor response and ultimately give rise to therapy resistance and tumor relapse. How resistance arises in cancer remains incompletely understood. While drug resistance in cancer is thought to be driven by irreversible genetic mutations, emerging evidence also implicates reversible proteomic and epigenetic mechanisms in the development of drug resistance. Tumor microenvironment-mediated mechanisms and tumor heterogeneity can significantly contribute to cancer treatment resistance. Here, we discuss the diverse and dynamic strategies that cancers use to evade drug response, the promise of upfront combination and intermittent therapies and therapy switching in forestalling resistance, and epigenetic reprogramming to combat resistance.

Cancer therapy resistance: multi-factorial, heterogeneous and rapidly evolving.

Drug resistance is a barrier to long-term patient survival [1]. Cancers use different routes to escape therapy-induced cell killing and acquire drug resistance, and many of these routes remain unpredictable and incompletely characterized [2, 3]. Improved understanding of the molecular changes that drive tumor progression and drug resistance coupled with the identification of strategies that different cancers use to evade treatment response will aid in designing both cancer subtype-specific treatments and broadly-applicable therapies to combat resistance.

The phenomenon of drug resistance is defined as the inherited ability of cells to survive at clinically-relevant drug concentrations. In the current studies to understand how cancer cells evade drug response two related phenomena, ‘tolerance’ and ‘persistence’, are considered part of a spectrum of therapeutic sensitivity and resistance states [4–6]. While ‘tolerance’ is the ability of cells to survive transient exposure to clinically-relevant concentrations of a drug, ‘persistence’ is the ability of a sub-population of a clonal population of cells to survive such treatment. Although these phenomena are well-documented in the response to anti-microbials [7], their distinction and relationship to the evolution of the fully drug-resistant tumor state remain unclear [8].

Drug resistance in cancer can be classified as intrinsic and acquired (including adaptive) [9], canonical examples of which are recently reviewed elsewhere [10, 11]. Intrinsic resistance arises before therapy and refers to the ability of a population of cells within a treatment-naïve tumor to survive initial therapy due to a pre-existing genetic alteration or cell-state [12–15]. Acquired resistance develops during treatment by therapy-induced selection of pre-existing genetic alterations in the original tumor and/or by acquisition of new mutations or adaptations in the drug target itself, recruitment of another survival factor such as a parallel or downstream pathway protein, metabolic adaptations, and epigenetic changes [16–30]. The tumor microenvironment can also contribute to acquired resistance: growth factors secreted by tumor-cells or tumor-resident stromal cells can enable tumor cell survival during initial treatment [31–33].

Tumor heterogeneity is another challenge that promotes all modes of cancer drug resistance [34–39]. A high degree of biological heterogeneity exists within a tumor (intra-tumor heterogeneity) in individual tumor cells and in cells comprising the tumor microenvironment or within different tumors (inter-tumor heterogeneity) in an individual patient or between patients [40–46]. Large scale studies such as The Cancer Genome Atlas Project (TCGA) provide critical insight into the magnitude of biological heterogeneity present within individual cancer types and across all cancers [47–49]. Such heterogeneity can promote ‘varied’ or ‘no’ response to therapy [50–52].

Tumor heterogeneity evolves temporally over time with tumor progression or with changes in therapy and spatially from primary to metastatic tumors [53–55]. Tumor evolution, which describes the emergence of multiple distinct sub-populations of cancer cells within the same tumor or patient, is a key feature of cancer progression and relapse and plays a key role in drug resistance [56]. Although, cancers usually arise by clonal outgrowth from a single founder cell [57], the expansion of the daughter cells of the founder cell after malignant transformation coupled with the continuous acquisition of alterations/mutations promotes the emergence of divergent cancer cell sub-populations and an increase in heterogeneity (as reviewed elsewhere [58], Figure 1). Alterations that occur early in tumor evolution are clonal whereas those acquired later are subclonal [54], Drug treatment leads to elimination of drug-sensitive sub-populations, tumor microenvironment alteration and positive selection of a drug-resistant sub-population of cells, which can result in a decrease in heterogeneity at least temporarily ([56, 58], Figure 1). Expansion of the resistant sub-population rapidly re-establishes heterogeneity through the acquisition of new mutations that provide survival benefits to the daughter cells of the resistant subclone ([58], Figure 1).

Figure 1. Different mechanisms of drug resistance in cancer.

Drug resistance is multifactorial and heterogenous and poses serious challenges to cancer treatment. The underlying mechanisms of drug resistance are diverse and complex, and often not mutually exclusive. Drug resistance is driven by both genetic and epigenetic mechanisms as well as by several other factors, including drug transporters and adaptive signaling events in tumor cells as well as tumor microenvironment features.

Tumor heterogeneity arises due to both genetic and epigenetic changes in tumor cells and in cells comprising the tumor microenvironment [54, 59]. Genetic changes due to chromosomal instability or genome duplication events as well as a variety of transcriptional states generated by epigenetic changes through altered DNA methylation and histone modifications contribute to tumor heterogeneity [54, 59, 60]. The presence of different lineage-specific gene expression programs within tumor cell sub-populations can also contribute to tumor heterogeneity [11]. Differences in the tumor microenvironment, such as, heterogeneity in the tumor-resident stromal cells from one tissue type to another (like, lung versus liver) is another form of tumor heterogeneity that can also contribute to residual disease and drug resistance [34, 42, 61].

Tumor heterogeneity is well-described in acute leukemias: clonal diseases with high inter-individual genetic variability [62, 63]. Recent whole-genome sequencing and single-cell genotyping studies in acute myeloid leukemia revealed a complex clonal diversity [64, 65]. These studies showed that the leukemic cell population is composed of different clones with high genetic heterogeneity. Further, the size and number of clones follow a complex evolution over the course of the disease [66–69]. Subclonal driver mutations have been shown to contribute to acquisition of drug resistance in chronic myeloid leukemia (CML) [58]. The BCR-ABL translocation is an established driver mutation in CML. The emergence of a subclone of leukemic cells carrying an imatinib resistance mutation in the BCR-ABL kinase domain lead to disease relapse after treatment with the BCR-ABL kinase inhibitor imatinib [70]. By whole exome sequencing and copy number analysis a recent study in chronic lymphocytic leukemia (CLL) identified clonal (MYD88, trisomy 12, and del(13q) or subclonal (Tp53 and SF3BP1) driver mutations corresponding to early or late events in clonal evolution and linked the subclonal driver mutations with rapid disease progression [71]. By genome-scale bisulfite sequencing in CLL another recent study described the contribution of DNA methylation to tumor heterogeneity [72]. Compared to normal B cells, CLL cells exhibited altered DNA methylation patterns. By bulk and single cell transcriptome analysis the DNA methylation disordered state was associated with transcriptional variation in CLL. The disordered DNA methylation state was further associated with adverse clinical outcome [72]. To more accurately identify subclonal driver mutations in CLL, another study developed methods for targeted mutation detection in DNA and RNA samples from single cells and suggested mutated LCP1 and WNK1 as candidate CLL drivers [73].

Tumor heterogeneity is also well-described in lung cancer and contributes to the emergence of EGFR TKI resistance [74, 75]. By whole-exome sequencing in a recent study, more than 75% of early-stage NSCLC (non-small cell lung cancer) tumors were found to harbor a subclonal driver alteration in PIK3CA, NF1, KRAS, TP53, and NOTCH family genes [74]. These subclonal driver mutations appeared clonal in one region but were either absent or subclonal in other regions of a single tumor [74]. EGFR T790M subclones can also co-exist with T790M-negative sub-clones not only after acquisition of resistance to first- and second-generation EGFR inhibitors but even in treatment-naïve lung cancer, although at a lower frequency [75]. A recent study in NSCLC shed new light on the genetic heterogeneity in lung cancer and showed that multiple driver mutations, such as, WNT/β-catenin and cell-cycle gene alterations/ mutations, co-occur and influence clinical outcomes in many EGFR-mutant advanced-stage lung cancers [76].

Mechanisms of drug resistance: multiple and diverse but not mutually exclusive.

The mechanisms of cancer drug resistance are multiple, diverse and complex (Figure 2). Since cancer is generally thought of as a genetic disease that evolves by clonal expansion and genetic diversification and selection, drug resistance is often considered to arise by genetic alterations [40, 77, 78]. Drug resistance, whether intrinsic or acquired, can be a deterministic and an irreversible phenomenon. Underscoring the genetic basis of drug resistance in cancer, a recent study in BRAF V600E-mutant cancers (melanoma as well as NSCLC) found selection and propagation of distinct BRAF-amplified (BRAF^amp) subclones within the same tumor through parallel evolution shortly after continuous treatment with ERK inhibitor alone that enabled the tumor cells to adapt to their environment while maintaining their intra-tumoral heterogeneity [79]. By contrast, intermittent treatment with a vertical combination of RAF, MEK and ERK inhibitors potently suppressed tumor growth [79]. This effect is presumably due to the inability of tumor cells to adapt well to the changing fitness threshold (or barriers that sub-populations of tumor cells need to overcome to regain fitness in the presence of therapy) imposed by the varying selective pressures exerted by different ERK pathway inhibitors in the intermittent polytherapy.

Figure 2. Changes in tumor heterogeneity during tumor progression and treatment.

Tumor heterogeneity arises due to mutations acquired by daughter cells upon clonal outgrowth of a single founder cell (red) and increases sharply with further development into subclones (different colors reflect different subclones/subpopulations). Some new mutations lead to accelerated growth (for example green and orange subclones). Drug treatment leads to selective survival of a drug resistant clone (orange subclone) and elimination of drug-sensitive subclones (for example green and red subclones) that reduces genetic heterogeneity transiently. Heterogeneity is re-established rapidly through acquisition of mutations by daughter cells of the resistant clone.

Cancer drug resistance is multi-factorial and not driven solely by genetic mechanisms [80, 81]. Non-genetic mechanisms, such as lineage plasticity/switching (a change in cell identity) [82] or epigenetic factors that promote gene expression changes and phenotypic plasticity [83] are increasingly recognized in cancer drug resistance. Lineage plasticity occurs in EGFR-mutant NSCLC treated with EGFR inhibitors and in hormone therapy-resistant prostate cancer with RB or PTEN mutation combined with p53 inactivation [84, 85]. In EGFR-mutant NSCLC, relapse under EGFR inhibition is associated with a transition from NSCLC to a small cell lung cancer-like (SCLC-like) cell state and with sensitivity to the standard chemotherapy used in SCLC [85]. The original EGFR driver mutation is often preserved in the relapsing SCLC-like tumor [85]. The observed lineage plasticity in EGFR-mutant NSCLC is either due to clonal outgrowth of a SCLC-like sub-population pre-existing in the initial EGFR-mutant NSCLC tumor or due to de-novo, adaptive cell-state shift in the NSCLC tumor cells that survive drug exposure.

In anti-androgen therapy-resistant prostate cancer, mutation of RB and p53 leads to overexpression of a pluripotency transcription factor SOX2, which triggers a switch from an anti-androgen therapy responsive luminal prostate cell state to a multi-lineage cell state that does not rely on androgen receptor signaling [86]. Other studies offer evidence for a distinct WNT5A-driven cell state during acquisition of androgen independence [87]. The extent to which these cell-state transitions in EGFR-mutant NSCLC and prostate cancer are mediated by epigenetic alterations remains uncertain but studies in other tumor types are suggestive of this possibility. In T-cell acute lymphoblastic leukemia resistant to NOTCH1 targeting, transition to an epigenetic state upon gamma-secretase inhibition that prevents NOTCH1 activation is associated with a sensitivity to epigenetic therapy with bromodomain and extra-terminal domain (BET) inhibitors [88]. A recent study in BRAF V600E-mutant melanoma found no clear genetic cause for resistance emergence upon continuous RAF inhibitor therapy and raised the possibility of non-genetic mechanisms driving resistance [89]. No genetic mutations were detected in response to drug treatment, but a small population of melanoma cells that displayed profound transcriptional variability at the single-cell level was found and likely to be epigenetically controlled. RAF inhibitor treatment induced epigenetic reprogramming in this cell sub-population, converting the transient transcriptional state to a stable resistant state [89].

A growing body of evidence indicates that the genetic and non-genetic/epigenetic resistance-conferring mechanisms are not mutually exclusive but instead co-exist within a given cancer to drive resistance development and therapy failure [79, 89, 90]. A genetic/epigenetic duality model of drug resistance (as described in a recent review, see [8]) may reconcile this complexity. According to this model, pre-existing genetic alterations or transient transcriptional or proteomic variations may allow a minor sub-population of tumor cells to overcome the fitness threshold and survive drug exposure (drug tolerance or persistence) until some cells acquire epigenetic changes and/or secondary genetic mutations that ultimately drive the emergence of drug resistance and tumor progression during therapy.

Other mechanisms like bi-directional interactions between the tumor cells and the tumor microenvironment can also influence drug exposure/response and contribute to drug resistance and are reviewed elsewhere [11, 31–33]. Pharmacokinetic variabilities associated with drug solubility, abnormal expression of drug-uptake and drug-efflux transporters in the tumor cells as well as stromal and physical barriers that restrict drug delivery can reduce drug uptake, intracellular drug concentration and distribution and thereby can lead to incomplete anti-tumor effects, residual disease and drug resistance [91–96]. Several recent studies have linked dysfunctional metabolism of lipid second messengers, such as, ceramide to cancer drug resistance [97, 98]. Ceramide, the basic structural unit of sphingolipids, activates cell death signals induced by cytokines, chemotherapy and radiation therapy. Several anti-cancer agents such as N-(4-hydroxyphenyl) retinamide (4-HPR) have been shown to act, partly, by increasing ceramide synthesis in tumor cells [98, 99]. Impaired ceramide synthesis decreases the cytotoxic effects of chemotherapy and contributes to chemotherapy resistance [97].

Tumor containment versus aggressive therapy to delay resistance emergence.

Tumor cell states are traditionally perceived as dichotomous: drug-sensitive or drug-resistant. Current standard practice based on this binary view of drug sensitivity and resistance is to apply maximum dose-density therapy through multiple cycles to eradicate the tumor as quickly as possible so that resistance does not arise [100, 101]. Such continuous and aggressive treatment regimens eliminate the bulk of the tumor cells which are drug-sensitive but also select for the minor, drug resistant sub-population already present within the tumor which ultimately becomes the primary tumor clone and promotes therapy failure [101–103] (Figure 3). Drug-sensitive and drug-resistant sub-populations are not mutually exclusive and may co-exist within the same tumor, competing against each other for survival. In fact, competition (direct or indirect) for nutrients or other critical resources between drug-resistant and drug-sensitive sub-populations of tumor cells may be the only natural force suppressing resistance emergence [104].

Figure 3. Aggressive therapy versus chronic control of tumors.

There is a fine balance between sustained maximum-dose density therapy and intermittent minimum necessary-dose therapy. Continuous maximum-dose density therapy is based on the principle of eradicating the tumor as swiftly as possible so that drug-resistance does not emerge. However, such rapid and aggressive treatment regimens eliminate the bulk of the tumor cells which are drug sensitive but also select for the minor, drug resistant sub-population already present within the tumor which ultimately takes over the entire tumor and promotes therapy failure. By contrast, in intermittent minimum necessary-dose therapy the idea is to contain the tumors at a fixed tolerable level to allow for expansion of drug sensitive cells at the expense of resistant ones. Although the tumor will increase in size between treatments, it will continue to remain sensitive to therapy overall. Adapted from [8, 102].

Intermittent and minimum necessary-dose treatment regimens to contain the tumors at a fixed tolerable level could allow expansion of drug-sensitive cells at the expense of resistant ones and may be more effective in delaying the emergence of resistance [8] (Figure 3). However, this treatment strategy should be practiced with caution as it could potentially be detrimental for the patient, for instance when the drug-sensitive cells acquire de-novo alterations and become drug-resistant [8, 104]. Mathematical models proposed in recent studies analyze the fitness thresholds to guide treatment decisions, such as aggressive therapy versus chronic control of tumors [104]. For example, when the patient cannot tolerate any tumor burden it is advisable to use a swift and attack-all aggressive therapy [104]. When tumors are rapidly mutating, it is advantageous to enforce a stable tumor burden by allowing the drug-sensitive, less aggressive cells to survive, which in turn will competitively suppress the aggressive drug-resistant cells to achieve chronic control of tumor evolution [8, 102–104]. Drug scheduling strategies based on these principles may be useful in halting and/or delaying resistance emergence by harnessing tumor heterogeneity and improving clinical outcomes and should be increasingly incorporated into clinical trials.

Polytherapy versus monotherapy in forestalling resistance development.

Drug resistance emergence in cancer may be suppressed by combination therapies that target multiple cancer dependencies [80]. In comparison to monotherapy, rational polytherapies that simultaneously target distinct mechanisms of resistance are less likely to fail, particularly when non-cross resistant treatments are applied together. Despite their promise, several challenges face the use of polytherapy such as limited drug options (which can be overcome partially with drug repurposing, as reviewed elsewhere [105, 106]) or drug access and treatment-related toxicity.

Several resistance mechanisms are ‘pathway-based’ mechanisms and frequently driven by pathway reactivation of a drug-inhibited oncoprotein [3, 80]. One way to achieve downstream pathway reactivation is by rendering a protein drug-target in-sensitive to therapy through acquisition of second-site or third-site mutations (i.e. by drug-target alterations) with the original oncogenic driver mutation. A rational polytherapy strategy to suppress drug-target alteration and pathway reactivation is to inhibit the target with two non-overlapping drugs simultaneously that bind to the same target but at different sites or with different modes of action. Such a dual-blockade approach has been applied clinically in HER2-amplified breast cancer, where HER2 function is blocked by both pertuzumab and trastuzumab [107, 108] and by lapatinib and trastuzumab [109] combinations. In chronic myeloid leukemia (CML) driven by BCR-ABL1, combining ATP-competitive/catalytic and allosteric inhibitors of the ABL kinase helped to overcome acquired resistance in pre-clinical studies [110].

Alternatively, primary oncogenic pathway reactivation can be achieved by activating effector proteins acting either upstream or parallel or downstream of the primary drug target. In these cases, multiple nodes in the same pathway should be targeted to block pathway reactivation and prevent resistance. For example, in BRAF-mutant melanoma, vertical combination of RAF and MEK inhibitors is clinically superior to RAF-directed monotherapy in preventing MAPK pathway reactivation and combating resistance [111, 112]. In prostate cancer, combination of ADT (androgen deprivation therapy) and the second-generation anti-androgen drug abiraterone suppresses AR signaling more potently than ADT alone [113, 114]. Despite their initial promise to improve clinical outcomes in many settings, these approaches are often not durable and resistance eventually emerges due to the limitations of single-pathway inhibition in constraining cancer evolution. On the other hand, targeting parallel pathways/dependencies with horizontal cancer drug combinations holds promise. But these regimens are often toxic, because the pathways targeted are often essential for the survival of normal cells, and hence are not always clinically-feasible [115, 116]. Alternatively, other convergent molecular targets that can compensate for single-pathway inhibition, for instance activation of the Hippo pathway effector YAP1, or its key gene targets, in mutant BRAF and RAS-driven cancers may need to be blocked in a non-cross resistant manner to achieve durable clinical remissions [117–123].

Along these lines, many oncogenic signal transduction cascades alter the function of downstream transcriptional programs to implement gene expression changes that drive cell transformation. Targeting an overactive, cancer specific transcription factor/regulator (e.g. YAP1) may combat several upstream oncogenes more effectively. If an oncogenic transcription factor is a broad integrator of expression of many different genes, like YAP1 or the master regulatory transcription factor MYC, its function may not be easily replaced by any other pathway or mechanism, decreasing the risk of resistance development.

Targeting transcriptional dependencies in cancer is challenging, as transcription factors often lack structural features that can be readily and selectively targeted with small molecule inhibitors. This limitation can be overcome indirectly by targeting the chromatin regulatory factors that influence transcription factor expression, activity, and stability and this approach is being tested in the clinic using small molecule inhibitors. For example, histone deacetylase (HDAC) inhibitors, which indirectly inhibit certain oncoproteins such as the melanocyte lineage-specific transcription factor MITF, are combined with MAPK pathway inhibitors to overcome MITF-driven resistance in melanoma [124]. The bromodomain inhibitors, targeting the BET family chromatin regulators and transcriptional co-activators (BRD4, 3, 2 and T), when administered as monotherapy produce less durable response in several cancers and are being combined with other agents, such as kinase inhibitors (e.g., RAF [125], MEK [126] and PI3K [127] inhibitors), cell-cycle inhibitors (e.g., CDK9 inhibitor [128]), DNA-damage repair inhibitors (e.g., ATR inhibitor, AZ20 [129]), immune-checkpoint inhibitors (e.g., PD-1 and PD-L1 inhibitors) [130], cytotoxic chemotherapeutic agents and other epigenetic drugs (e.g., HDAC [131] inhibitors) to enhance therapy response and forestall resistance. These combinations either induce a greater cytotoxic effect where merely a cytostatic effect was observed with either agent alone or prevent/delay tumor progression in-vivo suggesting that they may be able to overcome resistance to a single agent. Such combinations may be effective with reduced doses of each drug, potentially limiting toxicity issues. Yet, in some cases the combined toxicity of the combination regimens may be significant, and remains a major focus when designing combinatorial therapies. One potential solution is to consider intermittent or alternating drug administration schedules in the clinic.

Reversion to drug sensitivity by epigenetic reprogramming.

Another pertinent question in cancer treatment is how to tackle drug resistance after it emerges. Following emergence of drug resistance, drug discontinuation can lead to drug-sensitive populations outcompeting drug-resistant populations, restoration of drug-sensitivity, and clinical response upon drug re-exposure. For example, after a drug-free interval, drug-resistant lung cancers can re-respond to EGFR TKIs [132, 133]. In lung cancer patients whose tumors were assessed at multiple points along their treatment course, it was observed that genetic resistance mechanisms were lost without continued TKI treatment, providing a mechanistic basis for the re-treatment clinical responses [36]. Drug scheduling strategies based on these principles may be helpful in achieving reacquisition of drug sensitivity and response to therapy after a drug holiday to achieve improved chronic cancer control.

Cancer cells can use epigenetic mechanisms to control cell state. Emerging evidence shows the potential to reverse cancer-associated epigenetic abnormalities to reprogram neoplastic cells. While genetic alterations are deterministic and irreversible, epigenetic modifications are not caused by changes in the DNA sequence and are inherited at cell division. Epigenetic modifications are frequently reversible. Because of this inherent plasticity, there is an opportunity to revert back to a heritable drug-sensitive state by epigenetic reprogramming after resistance emerges. In a recent study in EGFR-mutant NSCLC PC9 cells, exposure to high concentrations of the EGFR inhibitor erlotinib killed nearly all parental PC9 cells but spared a small population of non-dividing cells [6]. These drug-tolerant persister cells re-initiated growth in the presence of erlotinib and displayed global chromatin alterations rather than genetic changes (such as EGFR-T790M or MET amplification [26]) known to confer erlotinib resistance. The drug-tolerant state could be reversed, converting the persister cells to drug-sensitive, either by continuous culture of these cells in the absence of erlotinib or by concomitantly treating them with an epigenetic drug (HDAC inhibitor) [6]. A recent study in BRAF-mutant melanoma showed that epigenetic therapy with BET bromodomain inhibitors could re-sensitize cells that acquired resistance to the RAF inhibitor vemurafenib in a YAP/TAZ-dependent manner after chronic exposure to vemurafenib [117]. In prostate cancer where regulation of GR signaling via a BET-dependent tissue-specific enhancer drives enzalutamide resistance, BET bromodomain inhibition re-sensitized drug-resistant tumors to enzalutamide by selectively impairing the GR signaling axis via this enhancer [134]. These studies show that cancer cells can become recalcitrant to therapy by acquiring a reversible drug-tolerant state during treatment that is epigenetically programmed and can be reversed by epigenetic re-programming, providing a rationale for innovative clinical trials.

Concluding Remarks

Tumor heterogeneity and cancer evolution underlie the emergence of therapeutic resistance and eventually disease relapse. Therefore, understanding how tumor heterogeneity evolves during the course of the disease and how to measure tumor evolution are essential to design more effective and durable rational polytherapies as well as to determine the timing of specific therapy deployment, therapy interruption, dose alteration or therapy switch in order to tackle the multifaceted problem of drug resistance in cancer (see Outstanding Questions). The analysis of the tumor composition (clones, subclones and the tumor microenvironment) in so-called “super-responder” patients who exhibit a dramatic anti-tumor response to a molecular therapeutic may help yield insight into the factors influencing exceptional response (or resistance) [135].

Current treatment protocols apply continuous maximum dose-density therapy to achieve maximum cell kill and eradicate the tumor as quickly as possible so that resistance does not arise. In the context of tumor evolution, however, it will be beneficial to employ intermittent or adaptive therapy [56, 136] to maintain a fine balance between drug-sensitive and drug-resistant tumor cell sub-populations (Figure 3), while avoiding excessive toxicity. Consistent with this prediction, in a recent study in breast cancer by employing adaptive therapy tumor progression was acutely controlled with an intensive chemotherapy followed by tumor maintenance with progressively smaller drug doses, in contrast, to rapid re-growth of tumor after the completion of maximum dose-density therapy [46, 56]. Similarly, in targeted therapy-based BRAF-V600E mutant melanoma and KRAS-mutant colorectal cancer models the onset of resistance was also stalled by intermittent BRAF inhibition or EGFR inhibition. In the chemotherapy-based breast cancer model, however, intermittent dosing strategy was less successful than variable dosing schedules [46, 56]. Observations from these studies suggest that intermittent and adaptive therapy and chronic control of tumors hold promise for combating resistance and should be tested more often in clinical trials in the future. These trials should be guided by future research that capitalizes on emerging preclinical models such as patient-derived organoids (PDOs) and patient-derived xenografts (PDXs), which can offer more clinically-relevant systems to facilitate high fidelity clinical translation [137–139].

Further, repeated molecular assessment will be critical to measure changes in tumor heterogeneity and cancer evolution during therapy to determine or adjust drug combinations or scheduling. Tumor profiling at multiple metastatic sites or at multiple regions within a single tumor can provide a comprehensive but only a static overview of genetic alterations. Single-site or even multi-site single biopsies are sub-optimal and will not detect all the molecular alterations within a tumor. Monitoring clonal dynamics and capturing changes in subclonal alterations as the tumor evolves (either temporally or spatially or both) through the analysis of circulating tumor DNA in the blood is becoming an increasingly useful tool. Techniques to perform tumor-sampling (such as, serial biopsies) and blood sampling (such as, analysis of circulating tumor DNA) should be further developed to allow for a more comprehensive analysis of the tumor heterogeneity and clonal evolution. The adjustment of the treatment strategy by incorporating the genetic heterogeneity factors, which is currently lacking in the standard protocols of cancer therapy, will be critical and should help prevent or delay the emergence of drug resistance to transform more cancers into chronic diseases.

Highlights.

The emergence of drug resistance is a barrier to effective cancer treatment. Resistance develops during chemotherapy, radiotherapy, molecularly-targeted therapy, and immunotherapy in most cancer patients and prevents their long-term survival.

Multiple mechanisms can promote the emergence of drug resistance and therapy failure. While gene mutations pre-existing in the tumor or acquired during therapy can drive drug resistance, cancers can also evade drug response through non-genetic and epigenetic mechanisms. Other mechanisms contributing to therapy resistance include tumor microenvironment influences, which can modify both drug exposure and response, and tumor heterogeneity within or between patients that underlies varied or minimal therapy response.

Cancer drug resistance is multi-factorial and evolves dynamically. Tumors evolve rapidly and studying how tumors adapt or change during therapy could be the key to tackling drug resistance. There is a need to recognize that drug sensitivity and resistance in cancer, and the various resistance mechanisms are not mutually exclusive but can operate together within the same tumor, or across different metastatic tumors in an individual patient. This reality must be considered in designing therapies and determining therapy duration, drug-dosing and timing for deployment of monotherapy or combination therapy. An up-front combination therapy that prevents or delays the evolution of tumors and/or dynamic switching of polytherapies during initial tumor response before resistance fully emerges holds promise for more effectively attenuating drug resistance to increase patient survival.

Outstanding questions.

1. How does tumor heterogeneity impact drug response and resistance?

2. How do genetic and epigenetic factors cooperate to promote drug resistance?

3. What are the most appropriate treatment strategies to forestall the evolution of drug resistance?

4. Can transcriptional nodes be effectively targeted to blunt resistance-associated cell state programs?

5. What are the most appropriate polytherapy strategies for clinical use to achieve chronic cancer control: co-administration therapy, alternating therapy, and/or intermittent and therapy switching regimens?

Glossary

Drug resistance

The phenomenon of drug resistance is defined as the inherited ability of cells to survive at high drug concentrations.

Tolerance and persistence

Two other related phenomena, ‘tolerance’ and ‘persistence’, are considered part of a spectrum of therapeutic sensitivity and resistance states [4–6]. While tolerance is the ability of cells to survive transient exposure to clinically-relevant concentrations of a drug, persistence is the ability of a sub-population of a clonal population of cells to survive such treatment. Drug resistance, tolerance and persistence are well documented in the response to anti-microbials in microbiology [7].

Intrinsic and acquired resistance

Drug resistance in cancer can be classified as intrinsic and acquired (including adaptive) [9], for canonical examples refer to [10, 11]. Intrinsic resistance arises before therapy and refers to the ability of a population of cells within a treatment-naïve tumor to survive initial therapy due to a pre-existing genetic alteration or cell-state [12–15]. Acquired resistance develops during treatment by therapy-induced selection of pre-existing genetic alterations in the original tumor and/or by acquisition of new mutations or adaptations in the drug target itself, recruitment of another survival factor such as a parallel or downstream pathway protein, metabolic adaptations, and epigenetic changes [16–30].

Lineage plasticity/switching

It is a change in cell identity as reviewed elsewhere [82]. Many cancers evade targeted therapies through this mechanism known as lineage plasticity, whereby tumor cells acquire phenotypic characteristics of a cell lineage whose survival no longer depends on the drug target.

Phenotypic plasticity

It is the ability of an organism to alter its phenotype in response to environmental influences, without altering its genome as reviewed elsewhere [83].

Maximum dose-density therapy

This therapeutic regimen is applied through multiple cycles to eradicate the tumor as quickly as possible so that resistance does not arise [100, 101].

Intermittent and minimum necessary-dose treatment

regimens are employed to contain the tumors at a fixed tolerable level could allow expansion of drug-sensitive cells at the expense of resistant ones and may be more effective in delaying the emergence of resistance [8]

Tumor heterogeneity

another challenge that promotes all modes of cancer drug resistance [34–39]. A high degree of molecular and genetic heterogeneity exists within a tumor (intra-tumor heterogeneity) in individual tumor cells and in cells comprising the tumor microenvironment or within different tumors (inter-tumor heterogeneity) in an individual patient or between patients [40–46].