The Kinetics and (Dys)kinetics of Cancer Chronotherapy

Abstract

Circadian rhythms are the daily cycles that time almost all aspects of physiology, but treatments of the clock or by the clock are rarely tested in the clinic. We develop a framework for identifying interventions that may benefit from administration at the appropriate time of day (chronotherapy). Typically, pharmacokinetics is an important consideration for chronotherapy, with short half-life drugs deemed optimal for such treatments. However, recent data suggest long-lived antibodies can show time-of-day specific effects. Examples include both tumor-targeted antibodies as well as immunotherapies with antibodies that activate T-cells. Clues to the immunotherapy mechanism come from animal vaccination studies, which demonstrate circadian responses of T-cells to a single dose that leads to long-lasting T-cell activation. Conversely, some studies have challenged the efficacy of chronotherapy, underscoring the need to rigorously investigate its application for each drug and tumor type.

Introduction:

Physiological processes in almost all organisms cycle with a circadian (~24h) rhythm. Circadian rhythms are not initiated by exogenous time cues, but instead driven by endogenous circadian clocks, which segregate physiological processes in time and also anticipate daily changes in the environment. Mechanisms that generate a clock were originally identified in Drosophila, but are well-conserved across eukaryotes. The core clock consists of a transcription translation feedback loop (TTFL) in which canonical clock molecules control their own cycling. In mammals, the BMAL1 and CLOCK proteins form an activator complex that cyclically drives transcription of its own repressors, Period (PER)/Cryptochrome (CRY). This core oscillator is complemented by a second stabilizing loop in which periodic expression of BMAL1 is maintained by the REV-ERBα/β repressor and RORα/β/γ activator proteins. As transcription factors, these proteins also drive cyclic expression of numerous target genes to regulate many aspects of physiology. Remarkably, almost every cell has the molecular machinery of the clock, and cellular clocks across the body are synchronized to the day:night cycle by the suprachiasmatic nucleus (SCN) in the brain hypothalamus (1).

Despite an advanced understanding of clock mechanisms, and the pervasiveness of rhythms in physiology, surprisingly few therapies take endogenous timekeeping into account. This is a missed opportunity, since large scale transcriptional profiling found that most drug targets undergo circadian cycling. Timing therapy to these endogenous cycles, such that patients are treated when primary drug targets are most vulnerable, could lead to better therapeutic outcomes. Alternatively, higher tolerance can be achieved when off target drivers of side effects are least active. A prime example of mechanism-based chronotherapy is provided by HMG-CoA reductase inhibitors, commonly known as statins. Pharmacodynamics established that HMG-CoA reductase expression cycles with a peak at night. Administering simvastatin, a statin with a half-life of ~3 hours, at bedtime reduced cholesterol levels significantly more than when it was administered in the morning. The bedtime enhancement was not seen when other statins with longer half-lives or even a timed-release simvastatin was tested. With treatment at night, the peak drug concentration coincided with the peak expression of the target (see figure), and so pharmacokinetics was tuned to pharmacodynamics. The package insert for simvastatin now recommends taking it before bedtime (2).

Unexpected pharmacodynamics of T-cell responses;

(A) Chronotherapy of a short half-life drug. Left, a short half-life drug administered at peak target activity (eg simvastatin) or least toxic time of day (eg oxaliplatin) is more effective than when it is administered at the wrong time as shown on the right. (B) Left, vaccines and checkpoint inhibitors are more effective when administered early in the day when T-cells are more susceptible to activation, compared to treatments later in the day, as illustrated on the right. In this case, the half-life of the treatment does not matter. Activity of T-cells may or may not cycle after treatment (dotted lines)

The example of statins suggests specific properties of a treatment regimen that could benefit from chronotherapy, but other aspects of cycling physiology could also have profound effects on drug efficacy (discussed below). Nevertheless, chronotherapy, or treatment timed to the optimal clock time, has not yet been incorporated into cancer therapeutics. Since chemotherapies, even targeted therapies, often use narrow therapeutic index drugs, any improvement in their use may benefit patients. We discuss below criteria for developing chronotherapy and why immunotherapies may lend themselves to chronotherapy through distinct mechanisms.

Clocks as modulators: A mechanistic approach to Chronotherapy

Clocks create daily cycles that may be amenable to timed treatment, or chronotherapy. In the case of cancer, primary drug targets can cycle, tumors can have daily physiological cycles, drug metabolism can cycle, physiological responses to drugs may cycle, or side effects may cycle. Each of these processes may present opportunities for chronotherapy. Clinical trial protocols that include a component of timing are low risk, since they can be conducted with approved protocols; even doses need not be changed. The only variable is the time of day that a treatment is administered. For clinically administered treatments, retrospective studies regarding efficacy at different times of day have been done from patient scheduling records. Given that several characteristics of the treatments or tumors may predispose them to chronotherapy, prescreening for these characteristics could prioritize tumors most likely to respond.

Rapid exposure and short half-life drugs

Compounds with short half-lives are optimal for chronotherapy, so this is an important consideration for cancer treatments, which have a wide range of half-lives. At the high end are antibodies, whose half-lives can be measured in weeks, while at the low-end, beam radiation treatments last 10-30 minutes. For daily dosing, drugs with half-lives of 6 hours or less consistently show time of day effects (2). The therapy most amenable to chronotherapy is radiation. Beam radiation is administered in short bursts and timing is easily controlled and recorded since it is administered at an institution. Until recently, antibodies and most other biologics were thought to remain in circulation too long for time of administration to affect outcome. Notably, drug pharmacokinetics can also differ based upon the time of administration. Serum concentration is determined by absorption, distribution, metabolism and excretion, all of which can be under circadian control. For example, pharmacokinetic studies show complex patterns of drug metabolism that are not always predictable (3).

Cycling target: Is there a clock in tumors?

If treatments are to be timed to a protein/pathway that is normally cyclically expressed, one question is whether it continues to cycle within the tumor. Although non-cell autonomous mechanisms can drive molecular cycles, the cycling of most proteins in a tissue is controlled by the local tissue clock. However, the state of the clock in tumors has yet to be resolved. While several tumor cell lines have functional clocks, systematic surveys of clocks in cancer cells have not been done, and even fewer studies have studied the clock within tumors in vivo. Clock genes are frequently mutated or misexpressed in cancers, and mutations in three of the most common oncogenes, Ras, p53 and Myc will perturb the clock, suggesting problems with tumor clocks. Additionally, cancer cells frequently dedifferentiate from their normal counterparts, which may affect clocks, since embryonic cells do not express a robust clock (4).

Clock function is typically measured by expressing luciferase reporter genes under the direction of a clock gene promoter, such as the promoter for the Period2 (Per2) gene. This Per-luc creates a cycling luciferase signal, which is disrupted by genetic perturbation of the clock. Per-luciferase genes cycle in cell culture and in vivo, but this strategy is not a feasible way to measure clocks in humans, as it requires introducing foreign DNA. Instead, tumor specific biomarkers or imaging of activity around the clock could establish if a clock is functional in a tumor. Measurements should be minimally invasive. For example, positron emission scanning, which determines tumor metabolic activation, is likely to be too invasive. Melatonin, which is secreted at night by the pineal gland, could be used as a biomarker in some cases to time chronotherapy. Or perhaps hair follicle samples. But it is important to note that while melatonin and hair may be useful as biomarkers for some types of chronotherapy, they could not be used as biomarkers for tumor clocks because they are not tumor-specific.

In the event that a target within a tumor is expressed cyclically, a drug can be timed to the appropriate point of the molecular cycle. Surveys of antineoplastic agents found that about 50 drugs had an optimal time of administration in rodents. With some drugs there was an optimal time of efficacy, while in other cases the drugs were tolerated better at different times of the day, but for many drugs the optimal times of efficacy coincided with the most tolerated times (3). In another study, many classes of chemotherapy drugs that act through different mechanisms were tested for time-of-day specific action through high throughput screening. Of the 126 drugs tested, 66 showed a circadian effect, and 44 of their primary targets cycled, suggesting that target cycling is common. Although the initial screen was conducted in cultured cells, temporal specificity was also demonstrated for two drugs administered to mice with tumors (5).

Surprisingly, time of day differences can be seen in cancer cells treated with antibodies, though antibodies have very long half-lives. This journal recently reported that resistance in gastric cancer to the monoclonal antibody trastuzumab (Herceptin) can be reversed by disrupting circadian rhythms. Trastuzumab has a half-life of 28-38 days, well beyond the framework half-life of ~6 hours. The underlying mechanism for resistance was facilitated by PER1 regulation of peroxisome proliferator-activated receptor γ (PPARγ) and its downstream target hexokinase 2. Timing treatment with trastuzumab did not overcome resistance, but either timing co-treatment with the glycolysis inhibitor, metformin (half-life 2-6 hrs), or knocking down PER1 both overcame resistance (6). Thus, despite falling outside the framework requirement for short half-life drugs, in some cases, antibody treatments may be improved by exploiting timed treatments.

These results are encouraging suggesting that at least some tumors have a functional clock. However, even when they have functional clocks, it is not known whether tumor cells have altered target cycling, drug transport, drug metabolism, or even drug resistance targets compared to normal tissue. Unfortunately, ways of determining tumor or target cycling in patients are not yet advanced enough for practice.

Cycling toxicity (or physiology)

Cancer therapeutics presents a unique situation relative to therapies for other diseases. As so many chemotherapy drugs are narrow therapeutic index drugs, the most important application of chronotherapy may lie in timing doses to minimize toxicity. Nevertheless, the only known case of chronotherapy to minimize toxicity is the use of oxaliplatin to treat colorectal cancer. Oxaliplatin, a platinum compound, was initially developed by a company that found it too toxic. They sold oxaliplatin to a just-created pharmaceutical group, Debiopharm, which redesigned clinical trials around rodent studies that found a peak tolerance at 16:00. In support of the rodent data, subsequent phase II and phase III chronotherapy trials found oxaliplatin better tolerated when it was delivered at a specific time of day, leading to approval of the drug. Hence, chronotherapy rescued a drug abandoned due to toxicity (2). In another example, significant progress has been made in modeling toxicity by the topoisomerase inhibitor irinotecan, which shows time and sex dependent effects. Although the underlying mechanisms have yet to be defined, toxic effects correlate with cyclic expression of clock genes in the liver, or alternatively with cycling of glutathione (antioxidant) and DNA repair pathways. In addition to chemotherapy, radiation treatment can produce time-of-day specific toxicity (4). Unfortunately, off target effects are not as specific as primary targets, thereby limiting efforts to minimize toxicity in this fashion.

Time trials: surprise from cancer immunotherapy

Based on rhythmic toxicity determined from animal studies, the EORTC 05011 trial addressed if a peak time could be established in treating metastatic colorectal cancer by varying the treatment times for irinotecan (7). Irinotecan was chronomodulated with a peak delivery rate at 1 of 6 clock hours staggered by 4 hours on day 1. 199 patients (130 males and 63 females) were treated in six groups of randomized patients. Then fixed-time Fluorouracil-Leucovorin-Oxaliplatin was administered for 4 days and repeated after 3 weeks. Irinotecan administration in the morning benefitted males while administration in the afternoon benefitted females without impairing efficacy. This study established that the tolerance for some chemotherapies could be enhanced through chronomodulation and that optimal times in patients were sex dependent, as previously found in mice. However, response rates, progression-free survival or overall survival did not differ by sex or by the timing of peak irinotecan delivery. Several other clinical studies that tested effects of time of day, all designed from pharmacokinetic modeling, have yielded mixed results (4). Furthermore, though chronotherapy trials played a central role in the approval of oxaliplatin, current clinical guidelines do not yet recommend chronotherapy regimens for its use.

Contrary to the framework for chronotherapy established by the short half-lives of small molecule drugs, long-acting immunotherapies may also be responsive to chronotherapy. Recently, a retrospective study was done to assess the impact of time of day on treating melanoma with the monoclonal antibody checkpoint inhibitors pembrolizumab (anti-PD-1), nivolumab (anti-PD-1), or ipilimumab (anti- CTLA-4) (8). These three antibodies activate T-cells, which then attack the tumors. 299 patients with stage IV melanoma at a tertiary cancer center in the USA melanoma patients were classified as being treated in the morning or afternoon. Surprisingly a significant effect on survival was noted. The overall survival was significantly longer among patients who received fewer than 20% of their infusions of immune checkpoint inhibitor in the evening than among patients who received 20% or more of these infusions in the evening: OS 68% vs. 48% (95% CI 2.04). The effect was seen even after addressing a number of variables including sex, BRAF^V600 status, brain metastasis and other treatments. Another study reported at a conference found that treating lung cancer metastasis in the morning produced a four-fold greater overall survival (OS) than treating in the afternoon (OS 34.2 mo. vs. 9.6 mo.) (9).

The striking observations of the two studies above was that a circadian effect was observed with monoclonal antibody therapy. The half-life of ipilimumab is about 2 weeks, of nivolumab is about 3.5 weeks, and of pembrolizumab is about 3 weeks. With such long half-lives any initial benefit of time-of-day administration would be diluted out by many weeks of high circulating drug concentrations. The underlying mechanisms cannot be explained by pharmacokinetic models seen with small molecule drugs that have short half-lives.

The authors of these studies did not test mechanisms, but speculated that T-cells, which are targeted by all of these antibodies, may be more responsive to initial injections at earlier times of day, and that furthermore they maintain the activation. These immunotherapy results are not the only cases where T-cell stimulation showed a circadian response. A mouse study found that vaccination in the daytime was more effective than vaccination at night at stimulating CD8-T cell responses and that this difference was reduced in a CD8 T cell-specific BMAL1 deletion. Retrospective studies of human subjects vaccinated at different times of day have yielded mixed results as to whether vaccination at different times of day lead to better antibody responses (10).

Conclusions:

Circadian rhythms are universal and can be exploited through time-of-day specific treatments to improve responses with better tolerance. The most effective applications were developed for statins and oxaliplatin based on the pharmacokinetics of drugs and pharmacodynamics of markers and toxicity in animal and clinical models. Unexpectedly, immunotherapies and antibody drugs may respond to chronotherapy through mechanisms that cannot be explained by their pharmacokinetic/pharmacodynamic modeling. At the same time, the efficacy of timed therapy has been challenged by a few studies, underscoring the need to rigorously investigate its application to each tumor and drug type.