Novel phase I study combining G1 phase, S phase, and G2/M phase cell cycle inhibitors in patients with advanced malignancies

Abstract

PURPOSE: Cancer is a manifestation of aberrant cellular proliferation, and the cell cycle is one of the most successfully drugged targets in oncology. No prior study has been reported that simultaneously targets the 3 principal cell cycle phases populated by proliferating cells - G₁, S, and G₂/M.

METHODS: Temsirolimus (G₁ inhibitor), topotecan (S inhibitor), and bortezomib (G₂/M inhibitor) were administered in combination to patients with advanced malignancies using a 3+3 dose escalation schedule to assess the safety and establish the maximum tolerated dose (primary endpoints) of this cell cycle targeting approach. An in silico pharmacodynamic model using established effects of each of these agents on the cell cycle was used to validate the regimen and to guide the dosing regimen.

RESULTS: Sixty-two subjects were enrolled. The most common adverse events and dose-limiting toxicities were cytopenias, consistent with the cell cycle targeting approach employed. All cytopenias resolved to baseline values upon holding study drug administration. The maximum tolerated dose was temsirolimus 15 mg/kg IV D1, 8, 15; topotecan 2.8 mg/m² IV D1, 8; and bortezomib 0.9 mg/m² IV D1, 4, 8, 11 of a 21-day cycle. In silico modeling suggests the regimen induces cell population shifts from G₂/M and S phases to G₁ phase and the quiescent G₀ phase. Eighteen percent of subjects (11/62) achieved partial response (n = 2, serous ovarian and papillary thyroid) or stable disease for > 6 months (n = 9).

CONCLUSION: Combining drugs with inhibitory activity of G₁ phase, S phase, and G₂/M phase is safe and warrants further evaluation.

Introduction

The cell cycle is the collective series of molecular events that a cell undergoes in the process of duplicating itself. In eukaryotic cells, it is comprised of 5 principle phases: G₀, G₁, S, G₂, and M (Fig. 1).¹ In G₁ cells produce proteins necessary for S phase, primarily those needed for chromosomal replication. In S phase cells duplicate their chromosomes and synthesize histones. In G₂ cells produce proteins needed for M phase, most notably microtubules. And in M phase, duplicated chromosomes segregate to 2 pairs (mitosis), and each parent cell physically divides into 2 daughter cells (cytokinesis). Since the M phase is relatively short lived and some G₂ and M phase processes appear to overlap temporally, these 2 phases are often taken together as G₂/M. G₀ refers to the quiescent stage in which a cell is not actively dividing.

Figure 1.

Schematic representation of cell cycle. Location of drug name indicates which phase of the cell cycle the drug inhibits.

Since cancer results from disregulated cellular division, there has been immense interest in understanding the cell cycle and in controlling its component molecular processes. In fact, some of the most commonly used antitumor drugs, including anthracyclines, nucleoside analogs, platinum salts, taxanes, and vinca alkaloids, exert their effects through disrupting various stages of the cell cycle.² Additionally, there are many compounds in clinical development, such as cyclin dependent kinase inhibitors^3,4 and novel microtubule disrupting agents^5,6 that also exert their antitumor effects by disrupting the cell cycle.

Given the importance of the cell cycle in cancer biology, and that it is a well-validated antitumor drug target, exploration of targeting each principle component of the cell cycle simultaneously as a novel antitumor strategy is intriguing. To this end, a phase I clinical trial was designed that combined agents known to have activity in inhibiting the G₁, S, and G₂/M phases of the cell cycle (Fig. 1). Review of the published literature on PubMed, as well as a review of trials registered on ClinicalTrials.gov revealed no current or prior study motivated by such a strategy. A combination of 3 drugs (temsirolimus, topotecan, and bortezomib) with known cell cycle effects and minimally overlapping side effect profiles was chosen, and in silico modeling was used to guide the study's development. Additionally, all drugs were selected to be cytostatic (rather than cytotoxic) in order to minimize potential toxicity of this aggressive cell cycle targeting approach. The primary endpoints of this trial were to determine the maximum tolerated dose (MTD) and dose-limiting toxicities (DLTs) of this 3 drug combination. A secondary endpoint was to provide descriptive analyses of any antitumor efficacy observed with this combination.

Results

Drug selection

A database of all antitumor drugs with at least one FDA granted marketing indication which contained each drug's mechanism of action and its most common side effects was compiled. From this database, 3 drugs were selected that were cytostatic, were known to interfere with cell cycle progression at G₁, S, and G₂/M, and which had minimally overlapping side effect profiles.

The G₁ inhibitor selected was temsirolimus, an inhibitor of the mammalian target of rapamycin (mTOR), which is a kinase that initiates translation of proteins necessary for G₁.^7-9 Some of temsirolimus’ commonest side effects are weakness, mouth sores, hyperglycemia, and hypertriglyceridemia.

The S phase inhibitor selected was topotecan. Topoisomerases create transient breaks in the DNA to allow for strand separation during replication.¹⁰ Topotecan inhibits religation of these breaks, resulting in cell stall during DNA replication (S phase).^11,12 Some of topotecan's commonest side effects are nausea, vomiting, and diarrhea.

The G₂/M inhibitor selected was bortezomib. Cell cycle progression is controlled by the ordered action of cyclins and cyclin-dependent kinases (CDKs). After a cyclin-CDK complex has performed its function, the cyclin is quickly destroyed by proteins known as proteasomes.^13,14 The transient existence of the correct cyclins provides directionality of the cell cycle. The proteasome inhibitor bortezomib stalls cells in G₂/M by inhibiting degradation of the specific cyclin (cyclin B) required for exit from mitosis.^14,15 Some of bortezomib's commonest side effects are peripheral neuropathy, thrombocytopenia, and nausea.

In silico modeling of the 3 drug combination's effect on distribution of cells in the cell cycle

All 2-way drug combinations of the 3 drugs included in this study have been previously investigated (temsirolimus, bortezomib¹⁰; temsirolimus, topotecan¹⁷; topotecan, bortezomib^18-20), and demonstrate cooperative drug effects. Given the 3-way combination has not been reported, the effect of the 3-way combination was examined in silico by constructing a computational model (Methods; Fig. 2A). The model is based on established pharmacodynamics and known effects of each of the agents^21-29 on the cell cycle for 10 tumor types^31-38 and for erythroblasts³⁹ (erythroblasts have the highest proliferative index of marrow cells). The model's output is the change in the distribution of cells in the various phases of the cell cycle. The a priori hypothesis was that inhibition of cell cycle progression should result in an increased proportion of cells in earlier phases of the cell cycle and decreased proportions in later phases. The model shows that for the 10 included tumor types, after 7 days of treatment the mean changes in cellular distribution are +17.2 ± 4.8% for cells in G₀/G₁, −10.8 ± 3.5% for cells in S phase, and −6.8 ± 3.5% for cells in G₂/M (Fig. 2B). In the model, low doses result in similar final cellular distributions as high doses by 7 days of treatment, but with different kinetics (Fig. 2C). The sequence of drug administration was found to be relevant, with best efficacy obtained upon G₁, S, and then G₂/M inhibition.

Figure 2.

Effect of temsirolimus (TS), topotecan (TP), and bortezomib (BZ) on the cell cycle. (A) Change in populations of proliferating (dP/dt) and quiescent (dQ/dt) cells is a function of proliferation (γ) and decay (δ) rates, transition rates between proliferating (α) and quiescent (β) states, and function of drug action, f(t), which is based on the effects of each agent. (B)The model described by the equation is implemented in MATLAB and evaluated by Euler's method for 10 tumor types as well as erythroblasts. Plotted is the absolute change in percent of cells in each cell cycle phase after 7 days of treatment. (C) Representative curves for time course of relative change in cell cycle distribution for erythroblasts. Black = low dose (5mg, 0.8mg/m², 0.3 mg/m², TS/TP/BZ). Red = high dose (25 mg, 4 mg/m², 1.3 mg/m², TS/TP/BZ).

Patients characteristics

Sixty-two patients were enrolled in this study between September 2008 and November 2011 (Table 1). Median patient age was 57 years (range 13-90 years), and 66% of patients were women. The most common malignancies were ovarian cancer (n = 13; 21%), colorectal cancer (n = 9; 15%), and thyroid cancer (n = 6; 10%). The study enrolled one pediatric patient (age 13, hepatocellular carcinoma).

Table 1.

Baseline Patient Characteristics

	Number of Patients (%)
Total	62 (100)
Sex
Female	41 (66)
Male	21 (33)
Age in years
Median	57
Range	13-90
Race
Asian	5 (8)
Black	10 (16)
Hispanic	13 (21)
White	34 (55)
Cancer Diagnosis
Breast	1
Cancer of unknown primary	2
Gasterointestinal
Anal	1
Colorectal	9
Esophageal	1
Hepatocellular	1
Genitourinary
Bladder	1
Renal cell	2
Gynecological
Cervical	3
Endometrial	1
Ovarian	13
Lymphoma	2
Thoracic / Head and Neck
Mesothelioma	1
Nasopharyngeal	1
NSCLC	2
SCC Head and Neck	3
SCLC	4
Melanoma	2
Sarcoma	4
Thyroid	6
Other
Ganglioma	1
Neuroendocrine	1

Abbreviations: NSCLC = non-small cell lung cancer, SCC = squamous cell cancer, SCLC = small cell lung cancer.

Dose escalation and toxicity

Side effects in each cohort are shown in Tables 2–3. One patient had a DLT of confusion that was attributed to study treatment at dose level 6 (temsirolimus 15 mg, topotecan 2.8 mg/m2, bortezomib 0.6 mg/m2), and 2 patients had grade 4 dose-limiting thrombocytopenia at dose level 7 (temsirolimus 15 mg, topotecan 2.8 mg/m2, bortezomib 0.9 mg/m2). While dose level 6 was the MTD per the protocol's definition of MTD, dose level 5 was selected for expansion given the participating investigators felt dose level 6 may be too heme-toxic to allow for a significant number of cycles to be administered. This was approved by the investigators’ IRB. At dose level 5, no DLTs were observed.

Table 2.

Dose-limiting Toxicities Observed in Each Cohort (Drug-Related, First Course)

Dose Level	Temsirolimus (mg)	Topotecan (mg/m2)	Bortezomib (mg/m2)	Number of Patients	Dose-Limiting Toxicities
1	5	0.8	0.3	5
2	10	0.8	0.3	3
3	10	1.5	0.3	4
4	10	1.5	0.6	4
5	15	1.5	0.6	34*
6	15	2.8	0.6	5	Confusion (n=1)
7	15	2.8	0.9	7	Thrombocytopenia (n=2)

^*Dose level 5 was chosen for expansion.

Table 3.

Non-DLT Adverse Events ≥ Grade 2 Observed During Treatment (Irrespective of Relation to Study Treatment)

Toxicity	Grade 2 (# of Patients)	Grade 3 (# of Patients)	Grade 4 (# of Patients)
Hypercholesterolemia	1
Anemia	3	1
Neutropenia	1	9
Thrombocytopenia	1	10	1
Hypertriglyceridemia	1
Elevated transaminases	1	1
Diarrhea	1
Pyrexia	1
Nausea	1	1
Vomiting	1	1
Febrile Neutropenia		2
Herpes Zoster	1
Cellulitis	1
Edema	1
Sepsis			1
Pulmonary Embolism		1
Hyponatremia		1	1

The most common non-dose limiting toxicities ≥ grade 2 observed during the course of this study were hematological, with 12 subjects (19%) experiencing grade 2-4 thrombocytopenia (1 grade 2, 10 grade 3, and 1 grade 4), and 10 subjects (16%) experiencing grade 2-3 neutropenia (1 grade 2, 9 grade 3). Anemia was less common (1 grade 3 event and 3 grade 2 events). In the majority of cases, cytopenias resolved to baseline with cessation of study drugs. The other grade 4 adverse events observed were one episode of sepsis (reported by investigator to be related to study drugs) and one episode of hyponatremia (reported as unrelated to study drugs), both of which resolved.

Subject disposition

Of the 62 subjects enrolled on this study, 53 had at least one restaging evaluation. Of the 9 who did not undergo restaging, 2 had early deaths (after 15 and 35 days, respectively); 3 came off study prior to restaging for non-treatment related side-effects; 3 withdrew consent; and one withdrew for non-compliance. Median time from enrollment to last follow-up date before data cutoff (30-Apr-2012) was 6.7 months. At the time of the data cutoff all patients had discontinued study therapy, and 13 patients were still alive. The most common reason for discontinuing study therapy was progressive disease (76%; 66% radiographic, 10% clinical). Other reasons included consent withdrawal (11%), adverse event (5%), lost to follow up (3%), death (2%), protocol deviation (2%), and administrative (2%).

Tumor response

As of the data cutoff date, 18% (11/62) patients had a partial response (PR, n = 2) or stable disease for ≥ 6 months (SD, n = 9). The partial responses were seen in a 59-year old female with papillary thyroid carcinoma whose best RECIST response was −49% at month 13, and who eventually had progressive disease from this nadir after a total of 15.0 months on treatment; and a 46-year old female with serous ovarian cancer whose best RECIST response was −34% at month 2, and who eventually had progressive disease after 6 months. The nine cases of prolonged stable disease were seen in 2 patients with renal cell carcinoma (time to progressive disese (PD) = 9.0 and 8.8 months), one with choroidal melanoma (7.2 months until PD), one with follicular thyroid carcinoma (withdrew consent after 13.2 months without PD), 2 with ovarian cancer (8.8 and 9.6 months until PD), one with colorectal cancer (6.3 months until PD), one with nasopharyngeal carcinoma (6.6 months until PD), and one with carcinoma of unknown primary (7.2 months until consent withdrawal). Both patients with PR were initially treated at dose level 7. The patient with papillary thyroid carcinoma was dose decreased to level 5 starting with cycle 2 and the patient with ovarian cancer was dose decreased to level 6 starting with cycle 3. The nine cases of prolonged SD were seen at a variety of dose levels (3 at level 1, one at level 2, one at level 3, and 3 at level 5). The best response for the 42 subjects with tumor changes quantifiable by RECIST is shown by waterfall plots (Fig. 3A). Of the patients not included in the waterfall plot, 11 subjects had at least one new lesion at the first restaging and were therefore not quantifiable by RECIST, and 9 subjects did not have a restaging evaluation (for the reasons given in Subject Disposition above).

Figure 3.

Waterfall plot of best response colored by dose level.

Survival outcomes

The median PFS in the 62 patients who participated in this study is 2.8 months (80% CI, 2.1 to 3.4). Median OS is 7.3 months (80% CI, 5.4 to 9.0). As of the time of the data cutoff (April 30, 2012), 13 patients are still alive including some who are censored before the median OS. All subjects who were censored for PFS are permanently censored.

Discussion

The primary endpoints of this study were to determine the dose-limiting toxicities and maximum tolerated dose of the regimen described in patients with advanced malignancies. Given the cell cycle targeting strategy employed, a priori there was concern for potential significant toxicity. For this reason, the first dose cohort enrolled at approximately 20% of each agent's usually prescribed dose, and all drugs were selected to be cytostatic rather than cytotoxic. The regimen was fairly well tolerated through dose escalation, with the final MTD of 15 mg temsirolimus, 2.8 mg/m² topotecan, and 0.6 mg/m² in the schedule described above representing 50-70% of each agent's commonly prescribed dose. It is not surprising that the most common toxicities of a cell-cyle targeting approach were cytopenias (76% of all grade 3 and grade 4 toxicities) or directly cytopenia-related (7% of all grade 3 and grade 4 toxicities were neutropenic fever). Given this study's treatment regimen was generally well tolerated, use of at least one cytotoxic drug in the combination (such as replacing bortezomib with a taxane) should be considered in a future study employing the paradigm of complete cell cycle, as this could enhance antitumor effect.

A secondary objective of this study was to report descriptive findings regarding antitumor activity observed with this regimen. As seen in Figure 3, the largest decreases in tumor burden were achieved at the higher dose levels, which is not unexpected given the known effective doses of these agents is 5-fold greater than that administered in this study's initial dose cohorts. The two partial responses observed in this study occurred in the highest dose cohort (level 7), and all 3 subjects enrolled in this dose cohort had significant tumor responses (−19% to −49% change in tumor burden by RECIST). Further development of this regimen or regimens based on a similar mechanism of action should consider these data when selecting the optimal dose, and could consider incorporation of haematopoietic growth factors into study protocols to allow for mitigation of cytopenic adverse events.

Limitations of this study include the heterogeneous population, including multiple tumor types and patients who have been treated with varying prior regimens. Additionally, the 3 agents selected have additional anti-tumor effects beyond their cell cycle inhibitory effects.

In summary, the regimen reported here is the first to our knowledge to evaluate a combination of cell cycle inhibitors. Similar approaches which simultaneously target multiple critical points in a cellular or viral network have yielded clinical benefit in other disease settings, most notably infectious diseases. The approach has provided incremental improvement in oncology as well, such as that seen with the combination chemotherapies regimens used to treat leukemias.⁴⁰ In all such cases, regimens have improved with iterative experimentation, with each subsequent regimen being informed by its predecessor. The concept of combination cell cycle blockade should be further evaluated in future studies as this paradigm should too improve with further refinement.

Patients and Methods/Materials

Eligibility

Patients were required to have unresectable or metastatic cancer, with no standard therapy available that increased survival by at least 3 months. Other criteria were adequate performance status (ECOG ≤ 2); no chemotherapy, antibody, or biological therapy within 4 weeks or 5 half-lives; and adequate bone marrow, liver and kidney function (absolute neutrophil count ≥ 1000/μL, platelets ≥ 75,000/μL, total bilirubin ≤ 2 × upper limit of normal (ULN), creatinine ≤ 2 × ULN). Pediatric patients and patients with CNS involvement were not excluded. The study was conducted by at the University of Texas MD Anderson Cancer Center and is in accordance with the Helsinki Declaration of 1975. All patients gave written informed consent in accordance with local Institutional Review Board (IRB) requirements.

Interventions

Drugs were administered intravenously in 21 day cycles on an outpatient basis. Temsirolimus was administered on days 1, 8, and 15, topotecan on days 1 and 8, and bortezomib on days 1, 4, 8, and 11. On days that more than one agent was administered, the sequence was temsirolimus, topotecan, and then bortezomib, as guided by the in silico modeling. The starting dose was 5 mg for temsirolimus, 0.8 mg/m² for topotecan, and 0.3 mg/m² for bortezomib, and doses were escalated using a 3+3 design. The protocol allowed a potential maximum dose of 25 mg of temsirolimus, 4.0 mg/m² of topotecan, and 1.3 mg/m² of bortezomib over 10 dose levels (Table S1). In order to further evaluate the drug combination, additional patients were enrolled onto a dose-expansion cohort after MTD determination (this cohort was dosed at one dose level below the MTD). Patients continued on treatment until progressive disease, unacceptable toxicity, withdrawal of consent, protocol deviation, or death.

Monitoring and treatment assessment

Physical exam and routine laboratory evaluations (complete blood count, serum chemistries, liver function studies) were conducted weekly during cycle 1 and at week 1 of each subsequent cycle. Radiological evaluations were repeated after every 2 cycles unless clinical course prompted earlier evaluation, and were evaluated using Response Evaluation Criteria in Solid Tumors (RECIST 1.0)⁴¹ for all solid malignancies except for lymphoma, which was evaluated with World Health Organization (WHO) criteria.⁴²

Maximum Tolerated Dose (MTD) and Dose-Limiting Toxicity (DLT)

MTD was defined as the highest dose studied in which less than 33% of patients experienced a DLT. Patients were enrolled in a 3+3 fashion and followed for 21 days for evaluation of DLTs. If 1 of 3 subjects developed a DLT then 3 additional subjects were added. If no DLTs were observed in the first 3 subjects or ≤ 1 DLT in the expanded cohort of 6, the subsequent dose level was opened for enrollment. DLT was defined as any grade 4 hematologic toxicity lasting 2 weeks or longer, any grade 4 nausea or vomiting greater than 5 days, or any other grade 3 non-hematologic toxicity. Toxicity grades are as defined by CTCAE v3.0, and toxicities are reported using descriptive statistics.

In silico modeling of the 3-drug combination's effect on distribution of cells in the cell cycle

A test population initialized with 1000 cells was modeled for one of 11 different cases (10 different tumor types and erythroblasts). The rate of change in the populations of actively proliferating and quiescent cells within a tumor can be described by Formula 1, as has been previously described:⁴³

$${\displaystyle \frac{\text{d}P}{\text{d}t}=\left(\gamma - \alpha - \delta -f\left(\text{t}\right)\right)P+\text{β}Q}$$

$${\displaystyle \frac{\text{d}Q}{\text{d}t}= \alpha P- \beta Q}$$

Briefly, the rate of change in the percentage of proliferating cells, P, and quiescent cells, Q, is based on the proliferation rate γ, the transition rates between proliferating and quiescent states α and β, the cell decay rate δ, and the function of drug action, f(t). f(t) is a probability of effect based on the individual^16-20 or combined^21-29 effects of temsirolimus, topotecan, and bortezomib which have been previously described.

Every 4 hours the model was updated, with each cell in P represented as being in G₁, S, or G₂/M phase of the cell cycle. Cells affected by one of the 3 treatments were halted in the respective phase (in G₁ for temsirolimus, S for topotecan, and G₂/M for bortezomib), becoming part of the quiescent Q population. The model described was implemented in MATLAB, and the differential equations representing the changes in P and Q were evaluated using the Euler method of integration. For each of the 11 cases, the populations were examined to determine changes to the baseline distributions of cells in G₁, S, and G₂/M. Different doses of each treatment were tested by changing values for f(t).

Statistical methods and considerations

All patients were enrolled by November 2011, and follow-up data through April 2012 are included in all analyses. Median progression-free and overall survival (PFS, OS) were estimated using the Kaplan-Meier method.³⁰ PFS is defined as the time from enrollment to either disease progression per RECIST 1.0 or death. If a subject discontinued study prior to documentation of progression per RECIST or death, then the last date that a RECIST-specified assessment was performed is taken as a censored end date for PFS calculations. OS is defined as the time from enrollment to time of death. Patients without a reported death date were censored at the day of last contact.

Disclosure of Potential Conflicts of Interest

Rajul Jain is employed by Kite Pharma. Yash Gad is the founder of SimVivo. The content of this manuscript is not of material interest to either Kite Pharma or SimVivo.

Acknowledgments

We would like to thank M. Kistler and S. Lockless for invaluable discussion on the study concept and phase I trial design.

Supplemental Material

Supplemental data for this article can be accessed on the publisher's website.