A Phase I Dose-Escalation and Expansion Study of Telaglenastat in Patients with Advanced or Metastatic Solid Tumors

Study design

This first-in-human, open-label, phase I trial (NCT02071862) included dose escalation and dose expansion of telaglenastat as a single agent in patients with advanced and/or treatment-refractory solid tumors (Supplementary Fig. S1). The primary objective was to determine safety, tolerability, and recommended phase II dose (RP2D). Secondary objectives included PK, pharmacodynamics, and antitumor activity. Patients were enrolled according to a traditional 3+3 design in dose escalation. Tumor-specific expansions in solid tumor histologies or histology-agnostic, genomically driven cohorts of interest were then opened to further define safety and explore efficacy. The study was conducted in accordance with the ethics principles of the Declaration of Helsinki and the International Council of Harmonization Guidelines on Good Clinical Practice. All patients provided written informed consent.

Patient selection

Eligible patients had to be ≥18 years of age, ECOG performance status 0–1, measurable disease per RECIST v1.1 criteria (32), adequate organ function, with histologically confirmed diagnosis of locally advanced, inoperable, metastatic, and/or treatment-refractory solid tumors for which no therapies that confer clinical benefit were available. For genomically selected cohorts (KRAS-mutated NSCLC, IDH1/2-mutated tumors, SDH-deficient tumors, FH-deficient tumors, MYC-amplified tumors), oncogenic driver mutations were identified by local standard-of-care genomic testing. MYC amplifications had to be ≥5-fold amplified by the local method. Patients were excluded if receiving concomitant anticancer therapy had untreated/active brain metastases or CNS disease, or had a severe medical illness that would interfere with study participation (see Supplementary Data for details).

Study treatment

Telaglenastat was administered orally in 21-day cycles until disease progression or intolerable toxicity. Initial dose-escalations levels were 100, 150, 250, 400, 600, and 800 mg for the 3 times daily (TID) “fasted” cohort (8 hours without food for morning dose, on empty stomach for midday and evening doses). Starting dose was based on preclinical toxicology studies and PK modeling that suggested that sustained plasma exposures of telaglenastat over 200 ng/mL would maximize target inhibition (Calithera, data on file). During the conduct of the study, increased exposure of telaglenastat was observed when dosed with food, and twice daily (BID) dosing with food (“BID-fed”) was explored with 3 dose-escalating cohorts: 600, 800, and 1,000 mg BID. Starting dose in the BID-fed cohorts was based on the highest previously cleared dose level in TID escalation cohort. Intrapatient dose escalation was permitted after completion of cycle 1 once the next higher dose level had been demonstrated to be well tolerated. For dose expansion, patients received either 600 or 800 mg per oral route (PO) BID.

Safety and efficacy evaluations

All patients were evaluated with medical history, physical examinations, and clinical laboratory results weekly for the first cycle, then on day 1 of each subsequent cycle. Adverse events (AE) were recorded by Common Toxicity Criteria for Adverse Events (CTCAE) v.4 from first dose of study drug up to 28 days after the last dose. Disease assessments with CT scans or MRI were performed at baseline and on day 1 of every third cycle (every 9 weeks). Tumor-response assessments were completed by RECIST v 1.1 criteria (32) for all solid tumors except pleural mesothelioma that used the modified RECIST criteria (33).

Pharmacokinetics

PK samples were collected on days 1 and 15 of cycle 1 at the following times: Predose, 0.5, 1, 2, 4, 6, and 8 hours postdose. Plasma concentrations of telaglenastat and glutamine were determined using HPLC/MS/MS. Telaglenastat PK parameters were estimated by non-compartmental analysis method. C_max and T_max were determined directly from the observed data, and AUC was calculated for days 1 and 15 of cycle 1.

Pharmacodynamics

GLS inhibition was measured in circulating platelets. Platelets were sampled predose and 4 hours after dosing on cycle 1, day 1. Paired fresh tumor samples were also assessed by similar methodology in patients who underwent optional pretreatment and on-treatment biopsies (cycle 2 and day 1). Platelet pellets from −80°C freezer were thawed on ice and resuspended in 230 μL of ice-cold lysis buffer. Samples were subjected to two rounds of homogenization on a Bioruptor for 5 minutes per cycle and filtered over a gel-filtration spin column (Zeba Spin Desalting Column, 7K MWCO). Spin columns were centrifuged at 2,500 RPM, 4–5 minutes, 4°C. Spin-filtered samples were serially diluted in a 384-well microplate for assay. Homogenates were quantified for total protein amount (BCA protein assay). Enzymatic assays were assembled in low-volume black 384-well HE microplates (Molecular Devices) by mixing (i) 5 μL platelet or tissue homogenates, (ii) 5 μL glutamate dehydrogenase (GDH), and (iii) 5 μL substrate (NADP⁺ and glutamine). Each component was separately prepared as a 3× solution and kept at room temperature. A 3×-GDH-containing solution was prepared by diluting GDH into assay buffer to reach a final concentration of 18 U/mL GDH. A 3×-substrate-containing solution was prepared by diluting glutamine and NADP⁺ stock solutions into assay buffer to reach final concentrations of 30 and 6 mmol/L, respectively. Fully assembled reactions were thus 15 μL in volume and contained 6 U/mL GDH, 10 mmol/L glutamine, 2 mmol/L NADP⁺, and various amounts of platelet or tissue homogenate. A second sample of each serial dilution was incubated without glutamine added. Generation of NADPH was monitored by fluorescence (Ex340/Em460 nm) every minute for 15 minutes on a SpectraMax M5e plate reader (Molecular Devices at 25°C). Relative fluorescence units (RFU) were converted to units of NADPH concentration (μmol/L) using a standard curve of NADPH. Under these assay conditions, up to 75 μmol/L glutamate is stoichiometrically converted to α-ketoglutarate/NADPH by GDH.

For the on-treatment biopsies, the GLS assay was specifically developed to take advantage of the slow dissociation between telaglenastat and GLS when cell/tissue lysates are prepared in the presence of high salt and low phosphate concentrations, keeping the enzyme in an inhibited state; inhibited GLS activity was then measured by separating the enzyme–inhibitor complex from free substrate and inhibitor and determining the activity of the residual uninhibited enzyme.

RP2D determination and DLT definition

Dose escalation followed a standard 3+3 design with a minimum of 3 patients assigned per dose level. If 0 of 3 or 1 of 6 patients in a cohort experienced a dose-limiting toxicity (DLT), dose escalation would continue to enroll per protocol. MTD was defined as the highest dose level with either no DLTs reported in 3 DLT-evaluable patients, or ≤1 DLT in 6 DLT-evaluable patients. MTD was exceeded if >1 of 3 or ≥2 of 6 patients experienced a DLT.

DLTs were evaluated only in patients receiving at least 75% of planned doses in the first treatment cycle and defined as any AE that could not be determined to be unrelated to study treatment, occurs within the first treatment cycle, and meets at least one of the following criteria: (i) Any grade ≥3 clinically significant non-hematologic toxicity per the CTCAE v.4, except nausea/vomiting/diarrhea lasting <48 hours and controlled with antiemetic/antidiarrheal therapy, grade 3 hyperglycemia lasting <72 hours with standard antidiabetic therapy, laboratory abnormalities reversible to grade ≤1 or baseline status within 72 hours with outpatient care and/or monitoring, or that are considered not clinically significant by the investigator; (ii) grade 4 neutropenia (ANC < 0.5 × 10⁹/mL); (iii) grade 3 febrile neutropenia (ANC < 1.0 × 10⁹/L with temperature ≥38.3°C); (iv) grade 4 thrombocytopenia (platelet count < 25.0 × 10⁹/L) lasting >4 days or requiring platelet transfusion; (v) grade ≥3 thrombocytopenia (platelet count <50.0 × 10⁹/L) associated with grade ≥3 bleeding; (vi) any other AE that is felt to be treatment-limiting in the medical opinion of the principal investigator and medical monitor. After each cohort was cleared for safety, an additional 4 patients with biopsy-amenable disease for mandatory pre- and on-treatment biopsies could be enrolled.

Per protocol, RP2D was to be selected based on MTD or, if MTD was not reached, other considerations, including: (i) Proportionality of dose-exposure relationship; (ii) magnitude of systemic GLS inhibition in platelets (goal: ≥90% inhibition at C_min); and (iii) overall safety/efficacy observations, including concurrent clinical trials.

Statistical analysis

Descriptive statistics were used to analyze the data, and summary statistics for continuous variables. Categorical variables are presented as frequency counts and percentages. All patients who received at least one telaglenastat dose were analyzed for safety. Results were tabulated by frequency, organ systems affected, and relationship to study treatment.

Using a Simon 2-Stage design, the sample size of 11 patients per expansion cohort was selected to allow adequate confirmation of safety and tolerability of telaglenastat and to identify evidence of preliminary clinical activity worthy of further investigation. In cases of clinical activity in any of the expansion cohorts based on response rate (i.e., at least 1 responder from the initial 11 patients), the protocol allowed an additional 26 patients for a total of 37 patients per cohort. With this design, the null hypothesis was an objective response rate (ORR) of ≤2% versus the alternate hypotheses of an ORR ≥15%, which would be reasonable for treatment-refractory solid tumors. A sample size of 37 patients per tumor type cohort would maintain an alpha level of 0.05 and 0.80 power. ORR was defined as the proportion of patients with complete response (CR) or partial response (PR) and calculated with a 95% confidence interval. Disease control rate (DCR) was defined as overall response rate + stable disease. The efficacy analysis set comprised all patients who completed 1 post-baseline tumor assessment. Patients who discontinued study medication early due to treatment-related toxicity or due to disease-related death or symptomatic deterioration (clinical progression) must have had received at least 48 doses TID or 32 doses BID to be considered evaluable for efficacy.

Patients

From February 2014 to October 2016, 56 patients (32 TID-fasted; 24 BID-fed) enrolled in dose-escalation and 64 in dose-expansion parts of the trial (Table 1). In dose escalation, 20 patients (36%) had TNBC, 9 (16%) RCC, 5 (9%) NSCLC, 5 (9%) mesothelioma, and the remaining with cancers of the bone, liver, colorectal, soft tissue, stomach, uterus, and gall bladder. In dose expansion, 22 patients enrolled in the RCC cohort, 9 NSCLC, 7 TNBC, 1 mesothelioma, and the remaining 25 across the 5 mutation-specific cohorts. Median age of patients was 60 and 59.5 years for dose escalation and dose expansion, respectively. Patients were heavily pretreated; >50% received ≥4 prior lines of systemic therapy (combined adjuvant/metastatic). High levels of GLS, based on H-score, were seen with immunohistochemical staining of archival tumor samples (n = 36) from patients with TNBC, NSCLC, RCC, mesothelioma, and SDH-deficient GIST (Supplementary Fig. S2). H-score was >100 for all but two patients (1 TNBC, H-score = 0; 1 TCA cycle mutant, H-score = 10).

Safety

Of 56 patients treated in dose escalation, 3 DLTs occurred in 2 patients in the TID-fasted cohort: 1 had creatinine increase at the 250 mg TID dose; 1 had alanine transaminase (ALT) increase and aspartate transaminase (AST) increase at 400 mg TID. No DLTs occurred on the BID-fed schedule. MTD was therefore not established.

Overall, across doses and schedules, treatment-related AEs of any grade were reported in 87/120 (72.5%) patients (Table 2): Fatigue (23%), nausea (19%), ALT increased (17%), AST increased (13%), and photophobia (11%). Patients receiving telaglenastat BID-fed had a lower incidence of treatment-related grade 3/4 AEs than those on the TID-fasted schedule [3/88 (3%) BID-fed vs. 7/32 (22%) TID-fasted]. In particular, grade 3/4 AST or ALT increase was significantly less frequent on the BID-fed schedule (1% AST, 2% ALT) than the TID-fasted schedule (16% ALT, 16% AST). Concurrent bilirubin elevation with AST/ALT elevation was rare; no patients on either schedule met criteria for the Hy's Law. There were no grade 5 AEs. Overall, 50 of 120 patients (42%) had dose interruptions, primarily due to AEs, 17 (14%) had dose reductions, and 4 (3%) had dose increased.

Serious AEs (SAE) occurred in 27 (22.5%) patients in dose-escalation and -expansion cohorts, most of which were considered unrelated to treatment. Four treatment-related SAEs occurred in 3 patients: ALT and AST increase (1 patient, 400 mg TID), blood creatinine increase (250 mg TID), and a grade 2 convulsion deemed possibly related to treatment (1,000 mg BID), which occurred after the first dose in a patient with metastatic TNBC who had 3 previously undiagnosed MRI-documented brain metastases. Seven patients died during the study-reporting period, all due to progressive disease (PD).

Pharmacokinetics

Following multiple administrations of 100 to 800 mg telaglenastat under fasted conditions, median t_max ranged from 1.0 to 2.0 hours over the dose range (Table 3). For the 100, 150, 250, 400, 600, and 800 mg TID dose levels, mean AUC_last were 2,150, 7,603, 4,626, 3,850, 5,162, and 5,832 ng·h/mL, respectively. Mean C_max values for the same dose levels were 395, 1,444, 914, 818, 1,142, and 1,058, respectively. Because blood samples were collected only for the first 8 hours before subsequent treatment with telaglenastat, the terminal phase could not be characterized for most patients, except one each in the 250 and 600 mg dose levels (3.73 hours). On the TID schedule, patients received telaglenastat without food throughout dosing, with the exception of cycle 2, day 1, when telaglenastat was administered with food. AUC_0–8h and C_max values were approximately 1.5-fold higher under fed conditions when compared with fasted conditions (Table 3). The average t_max with food was delayed by 2–3 hours compared with fasted conditions. On the basis of these observations, telaglenastat was administered on a BID schedule with food in all subsequent treatment cohorts.

Under PK evaluation under the BID-fed schedule showed that after approximately 2 weeks of dosing (cycle 1, day 15), median t_max ranged from 4.0 to 6.0 hours, suggesting a delayed absorption due to food effect. Exposure appeared to increase from 600 (n = 41) to 800 mg (n = 10), with mean AUC_0–8h of 7,611 and 9,660 ng*h/mL, respectively, and C_max of 1,455 and 1,774, respectively. For patients receiving 1,000 mg BID-fed (n = 4), mean AUC_0–8, AUC_0–8h, C_max, and median C_min were 915 ng·h/mL, 5,073 ng·h/mL, 5,005 ng/mL, and 147 ng/mL, respectively. Mean t_max for the 600, 800, and 1,000 mg BID-fed groups was 4.00, 4.00, and 6.08 hours, respectively (Table 3).

To further evaluate the effect of food on telaglenastat exposure, PK parameters were determined after the first administration of telaglenastat on cycle 1, day 1 under both fasted and fed regimens. Patients receiving 600 mg BID with food experienced a >3-fold increase in mean AUC_0–8h and C_max, compared with those receiving 600 mg BID while fasted. At the 800 mg BID dose, AUC_0–8h and C_max was 1.3-fold higher with fed than fasted dosing. Because of the associated delay in t_max and the flattening of the PK curve, these data supported the switch from TID dosing without food to BID dosing with food (Supplementary Fig. S3; Supplementary Table S1).

Pharmacodynamics

Measurement of GLS activity in peripheral platelets demonstrated >90% inhibition of GLS activity at plasma telaglenastat exposures >250 ng/mL (Fig. 1A). In patients with paired biopsies (n = 5) tumor GLS was inhibited by at least 75% (Cycle 2, Day 1), with exception of 1 patient who had low systemic exposure, likely due to radical gastrectomy (Fig. 1B). Plasma glutamine increased by approximately 150% on cycle 1, day 15 relative to predose on cycle 1, day 1 (P = 0.0073; n = 6; Fig. 1C).

RP2D selection

The initial dose selected for dose expansion was 600 mg BID based on available safety and PK data. Over the course of study, emerging PK/PD data demonstrated greater GLS inhibition with higher exposure at the 800-mg dose. In addition, the 800 mg dose led to higher exposure with equivalent safety. Thus, the final RP2D that was selected was 800 mg BID with food. Because RP2D was not selected until after initiation of the dose expansion phase, more patients received 600 mg BID (n = 55) than the final RP2D of 800 mg BID (n = 11).

Clinical activity

Of 101 patients evaluable for efficacy, 27 were in TID dose escalation, 16 in BID dose escalation, and 58 in the expansion cohorts (BID dosing); 4, 8, and 3 patients, respectively, were not evaluable. No objective responses occurred in either BID or TID dose escalation cohorts. Best response of stable disease (SD) occurred in 7/27 (25.9%) patients in TID and 10/16 (62.5%) patients in BID dose escalation. In dose expansion, 1/58 patients had a PR (in an RCC patient), 24/58 (41.4%) had SD, 33/58 (56.9%) had PD (Table 4). SDs occurred across different tumor types, with time on study extending to ≥6 months for 12 patients (Fig. 2A and B).

Patients with RCC exhibited more prominent clinical activity than other tumor types and were prioritized for further study. One of 9 RCC dose-expansion patients attained a PR, prompting expansion of the RCC cohort. Patient characteristics from the RCC expansion cohort (n = 22) were similar to the overall population (median age 65 years, 45% female, heavily pretreated; 65% with >3 prior systemic therapies). Data were not evaluable for 1 and 2 patients with RCC in dose escalation and dose expansion, respectively. Among 28 evaluable patients with RCC in both dose escalation and dose expansion, DCR was 14/28 (50%; Table 4).

The patients with RCC who achieved a PR had been diagnosed with metastatic clear cell RCC, receiving 3 prior lines of systemic therapy (sunitinib, pazopanib, everolimus; immediate prior treatment duration: 4.6 months; best response: PD). Treatment with telaglenastat led to a 32% reduction in target lesions and complete resolution of lymphadenopathy, whereas an adrenal target lesion remained unchanged (Fig. 2C). Duration of response was 7.4 months. PK data from the patient showed high telaglenastat exposure (AUC: 19,900 ng•hr/mL, C_max 2,960 ng/mL; C_min 1,490 ng/mL). Four patients with RCC remained on study for more than 10 months, including one clear cell patient with RCC who remained on for >2.4 years.