CENTER FOR DRUG EVALUATION AND
`RESEARCH
`
`
`APPLICATION NUMBER:
`
`211192Orig1s000
`
`
`OTHER REVIEW(S)
`
`

`

`IND or NDA
`Brand Name
`Generic Name
`Sponsor
`Indication
`
`Dosage Form
`Drug Class
`
`Interdisciplinary Review Team for QT Studies Consultation:
`QT Study Review
`211192
`TIBSOVO
`Ivosidenib (AG-120)
`Agios Pharmaceuticals, Inc.
`Treatment of patients with relapsed or refractory
`acute myeloid leukemia (AML) with an isocitrate
`dehydrogenase-1 (IDH1) mutation
`Tablet
`Inhibitor of isocitrate dehydrogenase-1 (IDH1)
`enzyme
`500 mg qd
`Chronic
`Has not established in humans
`001 / 12/21/2017
`DHP
`
`Therapeutic Dosing Regimen
`Duration of Therapeutic Use
`Maximum Tolerated Dose
`Submission Number and Date
`Review Division
`
`Note: Any text in the review with a light background should be inferred as copied from the
`sponsor’s document.
`
`SUMMARY
`1
`Ivosidenib inhibits the hERG potassium channel and is associated with concentration-
`dependent QTc interval prolongation in clinical studies. The results of a concentration-
`QTc analysis using data from studies AG120-C-001 and AG120-C-002 are shown by study
`in Table 1. Daily doses of 500 mg caused mean increases in QTcF from baseline of 18 to
`25 ms, respectively. The data from these studies were not pooled because significant
`variation in the slope estimate was detected. Administration of ivosidenib with high-fat
`meal or strong or moderate CYP3A4 inhibitors will further increase exposure.
`Table 1: The Estimated QTc effect based on concentration-QTc modeling
`(FDA Analysis)
`Concentration Mean QTcF (ms) 90% CI (ms)
`
`Treatment
`
`Study
`
`AG120-C-001
`AG120-C-002
`
`500 mg qd
`500 mg qd
`
`6,551 ng/mL
`6,551 ng/mL
`
`17.3
`25.2
`
`(15.1, 19.5)
`(20.8, 29.7)
`
`Reference ID: 4274565
`
`1
`
`

`

`There were patients with substantial QTc prolongation.
` AG120-C-001:
`o 22 (9%) patients had QTc >500 ms
`o 32 (12%) patients had increase from baseline QTc >60 ms. Please note, that
`the label notes 14% of patients had an increase from baseline of greater than
`60 ms – however in the adeg data set provided in sequence 0008, the
`reviewer was only able to identify 32 patients (12%).
`o 12 (5%) patients had QTc >500 ms and change from baseline QTc >60 ms.
` AG120-C-002:
`o 2 (1%) patients had QTc > 500 ms
`o 9 (5%) patients had increase from baseline QTc > 60 ms
`o 2 (1%) patients had QTc > 500 ms and change from baseline > 60 ms
`In study AG120-C-001, there were 5 patents who experienced cardiac AEs within the
`MedDRA SMQ Torsade de Pointes/QTc prolongation. Two subjects
` experienced events of suspected torsade de pointes, however, only one event is
`thought to be drug-related
`. Three subjects
` [cardiopulmonary arrest],
` [ventricular tachycardia], and
` [syncope]) experienced events which were
`not likely to be drug-related. See section 5.4 for more details. Cardiac AEs in AG120-C-
`002 were not evaluated because narratives were not provided.
`
`PROPOSED LABEL
`2
`We have reviewed the most recent label available (Seq 0028, dated 6/5/2018) and overall
`we agree with the language in section 2.3, but we have some suggestions for sections 5.2
`and 12.2. Our suggestions are marked with highlights below and are suggestions only and
`we defer final labeling decisions to the Division. Our suggestion of including only 1 of the
`cardiac AEs is based on Dr. Hicks’ cardiology review of the patient narratives.
`5.2 QTc Interval Prolongation
`Patients treated with TIBSOVO can develop QT (QTc) prolongation [see Clinical
`Pharmacology (12.2)] and ventricular arrhythmias. Of the 258 patients treated with
`TIBSOVO in the clinical trial, 9% were found to have a QTc interval greater than 500 msec
`and 124% of patients had an increase from baseline QTc greater than 60 msec.
`
`
`
`
`ne patient developed ventricular fibrillation
` attributed to TIBSOVO. The clinical trial excluded patients with baseline QTc of
`≥ 450 msec (unless the QTc ≥ 450 msec was due to a pre-existing bundle branch block) or
`with a history of long QT syndrome or uncontrolled or significant cardiovascular disease.
`Concomitant use of TIBSOVO with drugs known to prolong the QTc interval (e.g., anti-
`arrhythmic medicines, fluoroquinolones, triazole anti-fungals, 5-HT3 receptor antagonists)
`and CYP3A4 inhibitors may increase the risk of QTc interval prolongation [see Drug
`Interactions (7.1), Clinical Pharmacology (12.2)]. Conduct monitoring of
`electrocardiograms (ECGs) and electrolytes [see Dosage and Administration (2.3)].
`In patients with congenital long QTc syndrome, congestive heart failure, electrolyte
`abnormalities, or those who are taking medications known to prolong the QTc interval,
`more frequent monitoring may be necessary.
`
`2
`
`Reference ID: 4274565
`
`(b) (4)
`
`(b) (4)
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`

`

`Interrupt TIBSOVO if QTc increases to greater than 480 msec and less than 500 msec.
`Interrupt and reduce TIBSOVO if QTc increases to greater than 500 msec. Permanently
`discontinue TIBSOVO in patients who develop QTc interval prolongation with signs or
`symptoms of life-threatening arrhythmia [see Dosage and Administration (2.3)].
`12.2
`Pharmacodynamics
`Cardiac Electrophysiology
`A concentration-dependent QTc interval prolongation of approximately 16.1 msec (90%
`CI: 13.3, 18.9) was observed at the steady-state Cmax following a 500-mg daily dose based
`on an analysis of 171 patients with an IDH1 mutation, including 136 patients with relapsed
`or refractory AML, who received TIBSOVO 500 mg daily [see Warnings and Precautions
`(5.1)]. Co-administration with moderate or strong CYP3A inhibitors is expected to further
`increase QTc interval prolongation from baseline.
`
`3 BACKGROUND
`
`3.1 PRODUCT INFORMATION
`AG-120 is a selective, potent inhibitor of the isocitrate dehydrogenase-1 (IDH1) mutant
`protein, being developed for the treatment of patients with cancers that harbor IDH1
`mutations, including those with acute myelogenous leukemia (AML). The proposed clinical
`dosing regimen is 500 mg daily (QD).
`
`3.2 MARKET APPROVAL STATUS
`Ivosidenib is not approved for marketing in any country.
`
`3.3 PRECLINICAL INFORMATION
`In the manual patch clamp assay, AG-120 inhibited IKr (hERG) with an IC20 and IC50
`value of 3 and 12.6 μM, respectively. Individual animals with possible (≥30 msec) and
`probable (≥60 msec) test article-related QTcB prolongation (Morganroth 2001) have been
`noted in both the 28-day and 3-month GLP cynomolgus monkey studies at free Cmax
`values ≥0.7-fold the C2D1 500 mg free Cmax human exposure. In addition, prolonged
`QTcB was observed at the 45 and 135 mg/kg dose levels in a non-GLP single dose monkey
`CV safety pharmacology study, in which group mean Cmax values were similar to that of
`individual animals in the 28-day and 3-month repeat-dose studies. Reversibility of the
`prolonged QTcB effect was assessed in a single high-dose male in the 3-month monkey
`study; the QTcB levels returned to baseline following the 28-day recovery period (AG120-
`N-059-R1). Prolonged QTc interval is known to occur at drug concentrations below its
`hERG IC50 value and can occur at inhibitory concentrations as low as 12-30% (Redfern, et
`al. 2003). For AG-120, the hERG IC20 is approximately 3 μM, thus hERG inhibition is a
`plausible explanation for the prolonged QTcB intervals observed in the cynomolgus
`monkeys.
`Reviewer’s Comment: Based on the results of the manual patch clamp assay, the hERG
`safety margin is ~11x, which suggests that the observed QT prolongation in animals, and
`in the clinical study, could be mediated via direct blockade of the hERG potassium
`channel.
`
`3
`
`Reference ID: 4274565
`
`

`

`3.4 PREVIOUS CLINICAL EXPERIENCE
`Appendix 6.1 summarizes the previous clinical experience.
`
`3.5 CLINICAL PHARMACOLOGY
`Appendix 6.1 summarizes the key features of ivosidenib’s clinical pharmacology.
`
`4
`
`SPONSOR’S SUBMISSION
`
`4.1 OVERVIEW
`The QT-IRT did not review the protocols for any of the studies included in the proposed
`concentration-QT analysis, but did provide input on the proposed concentration-QTc
`approach under IND 119,341 (DARRTS 07/05/2016). The QT-IRT agreed that the
`proposed concentration-QTc analysis would be adequate to detect large mean changes in
`the QTc interval, which is acceptable based on the proposed indication.
`The proposed concentration-QTc analysis includes three studies: AG120-C-001 (dose
`escalation/expansion in R/R AML patients, CSR), AG120-C-002 (dose
`escalation/expansion in patients with solid tumors, CSR) and AG120-C-004 (food effect
`study in healthy volunteers, CSR). There are differences in dosing and ECG/PK sampling
`schedule between the two patient studies and the healthy volunteer study and we will
`therefore focus on our analysis on the two patient studies in our review, except for section
`4, which describes the sponsor’s pooled concentration-QTc analysis.
`The sponsor submitted the “Concentration-QTc Report AG120-C-META-CQT” (CSR) for
`ivosidenib, including electronic datasets and waveforms to the ECG warehouse.
`
`4.2 CONCENTRATION-QT REPORT
`
`4.2.1 Title
`Modeling the Relationship Between AG-120 Concentration and Electrocardiogram QTc in
`Subjects with Advanced Malignancies and Healthy Subjects
`
`4.2.2 Protocol Number
`Concentration-QTc Report: AG120-C-META-CQT
`
`4.2.3 Study Dates
`Date first subject enrolled:
` AG120-C-001: 12 March 2014
` AG120-C-002: 14 March 2014
` AG120-C-004: 29 September 2015
`Date last subject completed:
` AG120-C-001: 12 May 2017 (cutoff for the provided CSR)
` AG120-C-002: Ongoing (cutoff of 12 May 2017 in the interim CSR)
` AG120-C-004: 31 December 2015
`
`4.2.4 Objectives
`The objectives of this population concentration-QTc analysis were to:
`
`4
`
`Reference ID: 4274565
`
`

`

` Explore and characterize the relationship of change from baseline QTc (ΔQTc) with
`AG-120 plasma concentration in Study AG120-C-001.
` Explore and characterize the relationship of ΔQTc with AG-120 plasma
`concentration in pooled data sets of AG120-C-001 together with AG120-C-002 and
`healthy subjects in AG120-C-004.
` Predict AG-120 concentration-related QTc prolongation with associated confidence
`intervals at relevant exposures.
`
`4.2.5 Study Description
`
`4.2.5.1 Design
`The concentration-QTc report for AG-120 consists three phase 1 studies: AG120-C-001,
`AG120-C-002, and AG120-C-004. The detail of design for the three studies (001, 002 and
`004) that were evaluated in this report are as follows:
`AG120-C-001 (“001”): This is an ongoing Phase 1, multicenter, open-label, dose-escalation
`and expansion, safety, pharmacokinetic, pharmacodynamic, and clinical activity study of
`orally administered AG-120 in subjects with advanced hematologic malignancies with an
`IDH1 mutation, including R/R AML. The study includes a dose escalation portion to
`determine the maximum tolerated dose and/or the recommended Phase 2 dose and an
`expansion portion to further evaluate the safety, tolerability, and clinical activity of AG-
`120 in select populations.
`During the dose escalation phase, consented eligible subjects were enrolled into sequential
`cohorts of increasing doses of AG-120. AG-120 was administered in continuous 28-day
`cycles at doses of 100 mg twice daily (BID), 300 mg once daily (QD), 500 mg QD, 800
`mg QD, and 1,200 mg QD. The first 3 subjects in each cohort received a single dose of
`AG-120 on Day -3, three days before starting continuous daily dosing, in order to assess
`the PK/pharmacodynamics profile. On the basis of the available safety, PK/PD, and
`clinical activity data observed during dose escalation, 500 mg QD was selected as the
`recommended dose for expansion.
`On the basis of the available safety, PK/PD, and clinical activity data observed during dose
`escalation, 500 mg QD was selected as the recommended dose for expansion. Subjects
`were enrolled into 1 of 4 arms: (1) R/R AML (n=126 enrolled and received at least 1 dose
`of study drug), (2) untreated AML (n=25), (3) non-AML IDH1-mutated R/R advanced
`hematologic malignancies (n=11), and (4) relapsed AML not eligible for Arm 1 (n=18).
`Subjects continued treatment with AG-120 until disease progression, development of other
`unacceptable toxicity, confirmed pregnancy, undergoing a hematopoietic stem cell
`transplant (HSCT), death, withdrawal of consent, loss to follow-up, or Sponsor ending the
`study, whichever occurred first.
`In the dose escalation portion, time-matched AG-120 concentration and single 12-lead
`ECG assessments were made on Day -3 (pre-dose and 4 hours post-dose). For all
`subjects in the dose escalation portion, additional single 12-lead ECG assessments were
`made on Cycle 1 Day 1, 8, and 15, on Cycle 2 Day 1 and 15, on Day 1 of each
`treatment cycle thereafter, and at the end of treatment (EOT). The ECG assessment times
`
`5
`
`Reference ID: 4274565
`
`

`

`were pre-dose (Cycle 1 Day 1 only), four hours post-dose (Cycle 1 only), or anytime (after
`Cycle 1).
`In the expansion portion, time-matched AG-120 concentration and triplicate
`(approximately 2 minutes apart) 12-lead ECG assessments were made on Day 1 (pre-dose
`and 3, 4, and 8 hours post- dose) of Cycles 1 and 2, and at EOT. In addition, a single 12-
`lead ECG was obtained on Cycle 1 Day 8 and 15 (4 hours post-dose). For comparison
`with these ECG assessment times, the median Tmax of AG-120 after multiple doses was
`2.98 hours [15].
`AG120-C-002 (“002”): This is an ongoing multicenter, open-label, dose-escalation study
`with an expansion portion to evaluate safety/tolerability, MTD, PK/PD relationships,
`and clinical activity in subjects with advanced solid tumors with an IDH1 mutation.
`Sampling and design were similar to AG120-C-001.
`AG120-C-004 (“004”): This was a two-part study in healthy subjects. Part 1 was an open-
`label randomized study, with two periods and a crossover design, to determine the effect
`of food on the PK of a single 500-mg dose. Part 2 was an open-label study to determine
`safety and PK parameters following a single 1000-mg dose of AG-120. In both parts,
`time-matched AG-120 concentration and triplicate ECG assessments were made on Day 1.
`pre-dose, and 1, 2, 4, 24, and 48 hours post-dose. All data with and without food were
`included in this analysis.
`Overall design for the three studies are presented below together with another study that
`was not included in the pooled analysis (003)
`
`Source: DARRTS 01-07-16, Table 12.
`
`Reference ID: 4274565
`
`6
`
`

`

`4.2.6 Treatment Regimen
`
`4.2.6.1 Treatment Arms
`AG120-C-001 (“001”):
` Escalation doses at 100 mg BID, 300 mg QD, 500 mg QD, 800 mg QD and 1200
`mg QD
` Expansion dose selected 500 mg QD. Subjects were enrolled into 1 of 4 arms:
`Arm 1: R/R AML (n=126 enrolled and received at least 1 dose of study drug),
`Arm 2: untreated AML (n=25),
`Arm 3: non-AML IDH1-mutated R/R advanced hematologic malignancies (n=11), and
`Arm 4: relapsed AML not eligible for Arm 1 (n=18).
`
`AG120-C-002 (“002”):
` Escalation doses at 100 mg BID, 300 mg QD, 400 mg QD, 500 mg QD, 600 mg
`QD, 800 mg QD, 900 mg QD and 1200 mg QD
` Expansion dose at 500 mg QD
`
`AG120-C-004 (“004”):
`500 mg Fasted, 500 mg Fed and 1000 mg Fasted
`
`
`4.2.6.2 Sponsor’s Justification for Doses
`The expansion cohorts in AG120-C001 and AG120-C002 evaluated the proposed
`therapeutic dose.
`
`4.2.6.3 Instructions with Regard to Meals
`Ivosidenib was administered without regard to food in both AG120-C001 and AG120-
`C002 and study AG120-C004 included administration under both fed and fasting
`conditions.
`Reviewer’s Comment: An increase in Cmax and AUC of 25% and 98% respectively was
`observed when ivosidenib is taken with a high-fat meal.
`
`4.2.6.4 ECG and PK Assessments
`See above Table.
`Reviewer’s Comment: Acceptable, the ECG/PK sampling in the two expansion phases is
`adequate to capture time of peak effect (~3 h) and allow for detection of delayed effects.
`
`4.2.7 Sponsor’s Results
`
`4.2.7.1 Study Subjects
`AG120-C-001:
`A total of 258 subjects with advanced hematologic malignancies have received at least 1
`dose of ivosidenib: 78 subjects were treated in dose escalation (dose range 100 mg BID
`- 1200 mg QD) and 180 subjects were treated in the expansion phase (500 mg QD).
`
`Reference ID: 4274565
`
`7
`
`

`

`AG120-C-002:
`A total of 168 subjects enrolled and received at least 1 dose of AG-120 across 8 groups at
`doses of 100 mg BID, 300 mg QD, 400 mg QD, 500 mg QD, 600 mg QD, 800 mg QD, 900
`mg QD, and 1,200 mg QD. One-hundred and thirty subjects received 500 mg QD.
`
`AG120-C-004:
`A total of 30 subjects entered and 27 completed the study for Part 1. A total of 6 subjects
`entered and completed the study for Part 2.
`
`4.2.7.2 Statistical Analyses
`
`4.2.7.2.1 Primary Analysis
`This is a concentration-QTc study.
`
`4.2.7.2.2 Assay Sensitivity
`Not applicable.
`
`4.2.7.2.3 Categorical Analysis
`Table 2 presented sponsor’s categorical analyses for full QT dataset and triplicate dataset
`from Studies 001 and 002. Table presented counts of records with elevated QTcF (>450,
`>480, and >500 msec) or ΔQTcF (>30, >60, and >100 msec) by study, visit/time, and
`dataset.
`Table 2: Sponsor’s Counts of Elevated Mean QTcF and ΔQTcF Instances by
`Cycle/Day/Time in Studies AG120-C-001 and AG120-C-002
`
`Source: Study Report of ag120-c-meta-cqt.pdf, Table 6-3, page 22.
`
`8
`
`Reference ID: 4274565
`
`

`

`Reviewer’s Comments: We conducted independent categorical analyses based on escalation
`and expansion cohorts from Studies 001 and 002 at 500 mg QD in Section 5.2 by using
`“adeg.xpt” dataset.
`
`4.2.7.3 Safety Analysis
`The most common adverse reactions (≥20%) of any grade were leukocytosis, diarrhea,
`nausea, electrocardiogram QT prolonged, and rash.
`Serious adverse reactions (≥5%) were IDH differentiation syndrome (10%), leukocytosis
`(10%), and electrocardiogram QT prolonged (7%). No patient had a Grade 4 IDH
`differentiation syndrome or electrocardiogram QT prolonged event and none of the events
`of IDH differentiation syndrome, leukocytosis, or electrocardiogram QT prolonged were
`fatal.
`The most common adverse reactions leading to dose interruption were electrocardiogram
`QT prolonged (7%), IDH differentiation syndrome (3%), and leukocytosis (3%). Three out
`of 179 patients (2%) required a dose reduction due to an adverse reaction. The most
`common adverse reaction leading to a dose reduction was electrocardiogram QT prolonged
`(1%). No patients permanently discontinued TIBSOVO due to the adverse reaction of
`electrocardiogram QT prolonged. No patient permanently discontinued or required a dose
`reduction of TIBSOVO due to IDH differentiation syndrome or leukocytosis adverse
`reactions. One of 179 patients (0.6%) permanently discontinued TIBSOVO due to rash.
`A summary of adverse events by MedDRA preferred term in the broad SMQ for torsade de
`pointes/QT prolongation is shown below in Table 3.
`Table 3: Summary of Grade ≥3 and All-Grade Adverse Events by MedDRA
`Preferred Term in the SMQ (Broad) Torsade de Pointes/QT Prolongation (Safety
`analysis set) for study AG120-C-001
`
`Source: Summary of clinical safety, Table 21, Page 58
`
`9
`
`Reference ID: 4274565
`
`

`

`Reviewer’s comments: There are two cardiac adverse events that are suspected torsade
`events as described in section 5.4.1.
`
`4.2.7.4 Clinical Pharmacology
`
`4.2.7.4.1 Pharmacokinetic Analysis
`The PK results after multiple dosing in study AG120-C-001 is shown below in Table 4.
`
`Table 4: Summary of Ivosidenib Plasma Pharmacokinetic Parameters After Multiple
`Oral Administration (C2D1) of 500 mg QD Ivosidenib (AG120-C-001)
`
`Source: Summary of Clinical Pharmacology Studies, Table 14, Page 38
`
`4.2.7.4.2 Exposure-Response Analysis
`Consistent with the reviewer’s initial assessment, the sponsors assessment showed an
`absence of a delay between changes in QTcF and ivosidenib concentration and that a linear
`model would be appropriate to describe the relationship. Afterwards, the sponsor evaluated
`the model on the full data set and triplicate only dataset as well as by study and the results
`of the sponsor’s assessment is provided below.
`
`10
`
`Reference ID: 4274565
`
`

`

`Table 6-6 compares slopes in concentration across models for each dataset, combined and
`by individual study, and for the R/R AML subpopulation of AG120-C-001 (see Table 10-5
`through Table 10-12 for parameter estimates). Slopes in the primary final model were
`similar to those in the corresponding base model. Note that the Study AG120-C-001 slope
`is sensitive to the single highest outlier (with ΔQTcF = 186.7 msec). The effect of this
`outlier on Study AG120-C-001 predictions is shown in Table 6-7, discussed below.
`Table 5: Comparison of Primary and Supporting Models
`
`Source: Concentration-QTc report, Table 6-6, Page 26
`Reviewer’s Analysis: The reviewer’s independent analysis of the data is shown in section
`5.2 and is consistent with the sponsor’s analysis.
`
`5 REVIEWERS’ ASSESSMENT
`
`5.1 EVALUATION OF THE QT/RR CORRECTION METHOD
`The sponsor used QTcF for the primary analysis, which is acceptable as no significant
`increases or decreases in heart rate were observed (see Section 5.3).
`
`Reference ID: 4274565
`
`11
`
`

`

`5.2
`
`STATISTICAL ASSESSMENTS
`
`5.2.1 QTc Analysis
`
`Study ID
`AG120-C-001
`
`C1D8
`C1D15
`C2D1
`
`5.2.1.1 Central Tendency Analysis for Ivosidenib 500 mg QD
`All analyses are based on data from Studies 001 and 002. Table 6 presents the summary
`statistics (mean and standard deviation) and the 90% confidence interval for mean changes
`of AG-120 at 500 mg daily for Cycles 1 and 2 for the expansion phase. The reason for
`omitting Cycle 3 or later ECG values was lack of triplicate measurements (see report,
`Section 4.4.2, page 12).
`Using central tendency analysis, for Study 001, the largest upper bound of the 2-side 90%
`CI on the mean change from baseline in QTcF is 25.3 ms (on Cycle 2 Day 1 at 3-hour post-
`dose) which is >20 ms. For Study 002, the largest upper bound on the mean change from
`baseline is 19.8 ms (on Cycle 2 Day 1 at 4-hour post-dose).
`Table 6: Analysis Results of QTcF for Ivosidenib 500 mg QD
` by Study, Cycle, Day and Time (Expansion Cohort)
`Visit
`Time
`Std
`90% CI for
`N Mean
`Cycle/Day
`(H)
`Dev
`Mean
`174
`8.3
`C1D1
`3
`13.1
`(6.7, 10.0)
`177
`9.0
`4
`13.8
`(7.3, 10.8)
`168
`6.4
`8
`12.7
`(4.8, 8.0)
`165
`16.6
`4
`20.0
`(14.0, 19.2)
`154
`13.3
`4
`29.0
`(9.5, 17.2)
`137
`22.0
`3
`23.6
`(18.6, 25.3)
`137
`20.5
`4
`22.0
`(17.4, 23.6)
`136
`16.4
`8
`24.9
`(12.9, 20.0)
`107
`10.0
`3
`10.5
`(8.3, 11.7)
`107
`9.3
`4
`11.0
`(7.5, 11.1)
`107
`7.0
`8
`11.0
`(5.2, 8.7)
`102
`13.8
`4
`17.7
`(10.9, 16.7)
`100
`14.7
`4
`17.4
`(11.8, 17.6)
`97
`15.9
`3
`16.4
`(13.2, 18.7)
`96
`17.2
`4
`15.6
`(14.5, 19.8)
`95
`12.4
`8
`15.3
`(9.8, 15.0)
`Notes: Dataset “qtpk_fda.xpt” is used in the analysis.
`
`AG120-C-002
`
`C1D1
`
`C1D8
`C1D15
`C2D1
`
`Reference ID: 4274565
`
`12
`
`

`

`5.2.1.2 Sensitivity analysis
`Not applicable.
`
`5.2.1.3 Categorical Analysis
` lists the number of subjects whose QTcF values are ≤ 450 ms, between 450 ms and 480 ms,
`between 450 ms and 480 ms, between 450 ms and 480 ms, and >500 ms. Twenty-two
`subjects from Study 001 and two subjects from Study 002 had QTcF >500 ms.
`Table 7: Categorical Analysis of ΔQTcF
` (Escalation and Expansion Cohort at 500 mg QD)
`Total N QTcF<=450 ms 450<QTcF<=480 ms 480<QTcF<=500 ms QTcF>500 ms
`
`Study ID
`
`AG120-C-001
`
`258
`
`91 (35.3%)
`
`112 (43.4%)
`
`AG120-C-002
`168
`95 (56.5%)
`68 (40.5%)
`Notes: Dataset “adeg.xpt” is used in the analysis.
`
`33 (12.8%)
`
`3 (1.8%)
`
`22 (8.5%)
`
`2 (1.2%)
`
`Table 8 lists the number of subjects’ changes from baseline ΔQTc ≤30 ms, between 30 and
`60 ms, and >60 ms. Thirty-two subjects from Study 001 and nine subjects from Study 002
`had ΔQTcF>60 ms.
`
`Table 8: Categorical Analysis of ΔQTcF
` (Escalation and Expansion Cohort at 500 mg QD)
`30
`ms<Value<=60
`ms
`
`Study ID
`
`Total
`N
`
`Value<=30 ms
`
`Value>60
`ms
`
`AG120-C-001
`
`258
`
`117 (45.3%)
`
`109 (42.2%)
`
`32 (12.4%)
`
`67 (39.9%)
`92 (54.8%)
`168
`AG120-C-002
`Notes: Dataset “adeg.xpt” is used in the analysis.
`
`9 (5.4%)
`
`5.2.2 HR Analysis
`The descriptive statistics in HR for central tendency analysis are listed in Table 9. Using
`central tendency analysis, the largest upper bounds of the 2-sided 90% CI for mean changes
`from baseline in HR are 4.2 bpm (on Cycle 2 Day 1 at 8-hour post-dose) and 4.5 bpm (on
`Cycle 1 Day 15 at 4-hour post-dose) from studies 001 and 002, respectively.
`
`Table 9: Analysis Results of HR for Ivosidenib 500 mg QD by Study, Cycle, Day and
`Time (Expansion Cohort)
`Time
`(H)
`3
`4
`8
`4
`
`N Mean
`178
`-0.7
`179
`-1.0
`169
`-0.1
`171
`-0.4
`
`Std
`Dev
`7.3
`7.4
`8.7
`10.9
`
`90% CI
`for Mean
`(-1.6, 0.3)
`(-1.9, -0.1)
`(-1.2, 1.0)
`(-1.8, 1.0)
`
`Visit
`Study ID
`AG120-C-001 C1D1
`
`C1D8
`
`Reference ID: 4274565
`
`13
`
`

`

`Study ID
`
`Visit
`C1D15
`C2D1
`
`AG120-C-002 C1D1
`
`Time
`90% CI
`Std
`for Mean
`Dev
`N Mean
`(H)
`(0.8, 4.1)
`12.7
`159
`2.5
`4
`(-1.1, 2.5)
`12.7
`139
`0.7
`3
`(-0.9, 2.4)
`11.9
`141
`0.7
`4
`(0.4, 4.2)
`13.4
`136
`2.3
`8
`(-0.6, 1.5)
`6.4
`107
`0.5
`3
`(-1.1, 0.9)
`6.4
`107
`-0.1
`4
`(-0.2, 1.8)
`6.4
`107
`0.8
`8
`(1.1, 3.7)
`7.9
`104
`2.4
`4
`(1.5, 4.5)
`9.0
`100
`3.0
`4
`(0.3, 3.1)
`8.2
`97
`1.7
`3
`(-0.4, 2.2)
`7.8
`97
`0.9
`4
`(1.4, 4.2)
`8.4
`96
`2.8
`8
`Notes: Dataset “qtpk_fda.xpt” is used in the analysis.
`
`C1D8
`C1D15
`C2D1
`
`5.2.3 PR Analysis
`The descriptive statistics in PR for central tendency analysis are listed in Table 10. Using
`central tendency analysis, the largest upper bounds of the 2-sided 90% CI for mean changes
`from baselines in PR are 2.0 ms (on Cycle 1 Day 1 at 3-hour post-dose) and 4.0 ms (on Cycle
`2 Day 1 at 4-hour post-dose) from studies 001 and 002, respectively.
`
`Visit
`Study ID
`AG120-C-001 C1D1
`
`N Mean
`172
`0.9
`173
`-0.1
`164
`0.2
`166
`-0.5
`153
`-0.4
`135
`-1.7
`136
`-1.7
`132
`-0.8
`107
`0.7
`107
`0.9
`107
`0.2
`
`Std
`Dev
`9.0
`9.3
`10.0
`13.5
`14.7
`13.7
`14.9
`13.4
`8.0
`7.6
`8.7
`
`90% CI
`for Mean
`(-0.2, 2.0)
`(-1.3, 1.0)
`(-1.1, 1.5)
`(-2.2, 1.2)
`(-2.4, 1.5)
`(-3.7, 0.2)
`(-3.8, 0.4)
`(-2.7, 1.2)
`(-0.6, 2.0)
`(-0.3, 2.1)
`(-1.2, 1.5)
`
`Table 10: Analysis Results of PR Ivosidenib 500 mg QD by Study, Cycle, Day and
`Time (Expansion Cohort)
`Time
`(H)
`3
`4
`8
`4
`4
`3
`4
`8
`3
`4
`8
`
`C1D8
`C1D15
`C2D1
`
`AG120-C-002 C1D1
`
`Reference ID: 4274565
`
`14
`
`

`

`Study ID
`
`Visit
`C1D8
`C1D15
`C2D1
`
`Time
`90% CI
`Std
`for Mean
`Dev
`N Mean
`(H)
`(-2.2, 1.4)
`11.2
`104
`-0.4
`4
`(-3.2, 0.4)
`11.0
`100
`-1.4
`4
`(-0.9, 2.4)
`10.0
`97
`0.8
`3
`(0.6, 4.0)
`10.1
`97
`2.3
`4
`(-2.0, 1.4)
`10.1
`95
`-0.3
`8
`Notes: Dataset “qtpk_fda.xpt” is used in the analysis.
`, AG120-C-
`Subjects AG120-C-001
`, AG120-C-001
`, and AG120-C-001
`have missing PR intervals.
`
`C1D8
`C1D15
`C2D1
`
`5.2.4 QRS Analysis
`The descriptive statistics in QRS for central tendency analysis are listed in
`Table 11. Using central tendency analysis, the largest upper bounds of the 2-sided 90% CI
`for mean changes from baselines in QRS are 0.8 ms (on Cycle 2 Day 1 at 3-hour post-dose)
`and 0.7 ms (on Cycle 1 Day 1 at 3-hour post-dose) from studies 001 and 002, respectively.
`Table 11: Analysis Results of QRS for Ivosidenib 500 mg QD by Study, Cycle, Day
`and Time (Expansion Cohort)
`Visit Time(H)
`N Mean Std Dev 90% CI for Mean
`Study ID
`AG120-C-001 C1D1
`3
`178
`0.3
`4.0
`(-0.2, 0.7)
`4
`179
`-0.1
`4.0
`(-0.6, 0.4)
`8
`169
`-0.1
`4.6
`(-0.6, 0.5)
`4
`171
`-0.3
`6.2
`(-1.1, 0.5)
`4
`159
`-0.1
`5.9
`(-0.9, 0.7)
`3
`139
`-0.1
`6.0
`(-0.9, 0.8)
`4
`140
`-0.2
`5.9
`(-1.0, 0.6)
`8
`136
`-0.7
`5.4
`(-1.5, 0.1)
`3
`107
`0.0
`3.8
`(-0.6, 0.7)
`4
`107
`-0.2
`3.0
`(-0.7, 0.2)
`8
`107
`-0.6
`3.8
`(-1.3, -0.0)
`4
`104
`-0.6
`4.9
`(-1.4, 0.2)
`4
`100
`-0.9
`4.7
`(-1.7, -0.1)
`3
`97
`-0.1
`4.9
`(-0.9, 0.7)
`4
`97
`-0.4
`5.0
`(-1.3, 0.4)
`8
`96
`-1.7
`5.0
`(-2.5, -0.8)
`Notes: Dataset “qtpk_fda.xpt” is used in the analysis.
`
`AG120-C-002 C1D1
`
`C1D8
`C1D15
`C2D1
`
`Reference ID: 4274565
`
`15
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`

`

`5.3 CLINICAL PHARMACOLOGY ASSESSMENTS
`The objective of the clinical pharmacology analysis is to assess the relationship between
`ivosidenib concentration and ΔQTcF. Of note, the reviewer used a different dataset with
`the sponsor, however, similar slope and predicted mean effect was observed to that of the
`sponsor’s analysis. Prior to evaluating the relationship using a linear model, the three key
`assumptions of the model were evaluated using exploratory analysis: 1) absence of
`significant changes in heart rate (more than a 10 bpm increase or decrease in mean HR); 2)
`delay between plasma concentration and ΔQTcF and 3) presence of non-linear relationship.
`An evaluation of the time-course of ivosidenib pharmacokinetics and changes in ΔHR and
`ΔQTcF is shown in Figure 1, which shows an absence of significant changes in HR and do
`not appear to show significant hysteresis.
`Figure 1: Time course of ivosidenib concentration (top), heart rate (middle) and QTcF
`(bottom) for AG120-C-001 (solid) and AG120-C-002 (dashed) for cycle 1 day 1 and
`cycle 2 day 1 (triplicate ECGs)
`
`Reference ID: 4274565
`
`16
`
`

`

`After confirming the absence of significant heart rate changes or delayed QTc changes, the
`relationship between ivosidenib concentration and ΔQTcF was evaluated to determine if a
`linear model would be appropriate. Figure 2 shows the relationship between ivosidenib
`concentration and ΔQTcF and supports the appropriateness of a linear model by study
`(AG120-C-001 and AG120-C-002). While, the relationship appears similar between the
`two studies, a significant difference in the slope of the relationship was observed and it is
`unknown if this difference is due to a difference in patient population.
`Figure 2: Assessment of linearity of concentration-QTc relationship
`
`Finally, the linear model was applied to the data and the goodness-of-fit plot is shown in
`Figure 3, which shows a linear relationship between ivosidenib concentration and ∆QTc by
`study. A similar conclusion was reached by the sponsor.
`Figure 3: Goodness-of-fit plot for QTc
`
`Reference ID: 4274565
`
`17
`
`

`

`5.4 CLINICAL ASSESSMENTS
`
`5.4.1 Safety assessments
`Table 3 summarizes the adverse events associated with the MeDRA SMQ Torsade/QTc
`Prolongation in Study AG120-C-001, an open-label, single-arm, multicenter clinical trial in
`125 patients with relapsed or refractory acute myeloid leukemia (AML) with an IDH1
`mutation who were assigned to receive AG-120 (ivosidenib) 500 mg daily. Dr. Karen
`Hicks (cardiologist in DCRP) reviewed the narratives for five subjects with cardiac adverse
`events. In brief, two subjects
` experienced events of suspected
`torsade de pointes of which one is thought to be drug-related
`, and three subjects
` [cardiopulmonary arrest],
` [ventricular tachycardia], and
`
`[syncope]) experienced events which were not likely to be drug-related.
`
` are
`The two events of suspected torsade de pointes for Subjects
`described below. In addition to reviewing the narratives, Dr. Hicks reviewed the available
`ECG waveforms.
`
`1. Subject
`
` (syncope and ventricular fibrillation arrest)
`
` is an 83-year-old woman who on Study Day 20 experienced events of
`Subject
`syncope and subsequent ventricular fibrillation arrest due to suspected torsade de pointes.
`The torsade de pointes (TdP) arrhythmia is multifactorial in etiology, and drug effect from
`ivosidenib cannot be excluded. The supratherapeutic plasma concentration of ivosidenib,
`electrolyte abnormalities (hypokalemia and hypomagnesemia), and concomitant
`medications known to cause QT prolongation (fluconazole, amiodarone, and ondansetron)
`that the subject was receiving in the setting of renal impairment (which further increased
`the plasma concentration of fluconazole) and heart failure contributed to the events of
`suspected torsade de pointes.
`
`
`received her first dose of ivosidenib 500 mg qd on
`In brief, Subject
`(Study Day 1) and experienced a syncopal event followed by a ventricular fibrillation arrest
`on
` (Study Day 20). Study drug was interrupted on
` and
`permanently discontinued on
` after the subject withdrew consent. On
` two days after the last dose of study drug, the subject died due to a ventricular
`arrhythmia. No autopsy was performed.
`
`
`
`had received 3 anti-cancer regimens previously but no radiotherapy or
`Subject
`bone marrow transplantation for the underlying malignancy. Her past medical history was
`remarkable for hypokalemia, hypomagnesemia, hypouricemia, and increased alkaline
`phosphatase. She also had a history of hypotension, uterine brachytherapy, cardiac
`murmur, mastitis, non-Hodgkin’s lymphoma, pleural effusion, pneumonia, radiotherapy,
`and scarlet fever. Her past surgical history included a colectomy, colostomy, tubal ligation,
`and tonsillectomy.
`
`Pertinent medical issues at the time of study entry included electrolyte imbalance,
`hyperglycemia, anemia, epistaxis, nausea, malaise, decreased appetite, asthenia,
`
`18
`
`Reference ID: 4274565
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`(b) (6)
`
`

`

`hypoalbuminemia, cervical carcinoma, diverticulum, neutropenia, thrombocytopenia, easy
`bruising, back pain/spinal compression fracture, osteoporosis, and weight loss.
`
`An echocardiogram performed at Screening on
`ventricular ejection fraction