HCC Journal ClubSeptember 2009Statistical Topic: Phase I studies

Selected article:Fong, Boss, Yap, Tutt, Wu, et al. Inhibition of Poly(ADP-Ribose) Polymerase in Tumors from BRCA Mutation CarriersThe New England Journal of MedicineJuly 9, 2009. Vol. 361, No. 2, pp. 123-134

Phase I studies

What are the goals? Dose-finding Safety PK and PD

How are they designed? What is the rationale for dose-

selection?

Fong et al.

Stated goals: Determine the following— Safety adverse-event profile dose-limiting toxicity maximum tolerated dose (MTD) Dose at which PARP is maximally inhibited PK profile PD profile

Study Design

“modified accelerated titration” Not at all! TRUE Accelerated titration design:

Treat 1 person per dose until either one DLT is observed OR, two grade 2 toxicities

Then, treat 3 patients per dose level Dose steps can be doubling or not.

Fong study: uses standard 3+3. probably called modified AT because allows doubling

of dose in the absence of grade 2 or higher. Is NOT accelerated titration in the spirit of the original

paper

Back to basics: Acceptable toxicity

What is acceptable rate of toxicity? 20%? 30%? 50%?

What is toxicity???? Standard in cancer: Grade 4 hematologic or

grade 3/4 non-hematologic toxicity Always? Does it depend on reversibility of toxicity? Does it depend on intensity of treatment?

Tamoxifen? Chemotherapy?

‘3+3’ Design

“Standard” Phase I trials (in oncology) use what is often called the ‘3+3’ design (aka ‘modified Fibonacci’):

Maximum tolerated dose (MTD) is considered highest dose at which 1 or 0 out of six patients experiences DLT.

Doses need to be pre-specified Confidence in MTD is usually poor.

Treat 3 patients at dose K1. If 0 patients experience dose-limiting toxicity (DLT), escalate to dose K+12. If 2 or more patients experience DLT, de-escalate to level K-13. If 1 patient experiences DLT, treat 3 more patients at dose level K

A. If 1 of 6 experiences DLT, escalate to dose level K+1B. If 2 or more of 6 experiences DLT, de-escalate to level K-1

Observed Data in Fong Study

1 2 3 4 5

0.0

0.2

0.4

0.6

0.8

1.0

Dose Level

Pro

po

rtio

n o

f Pa

tien

ts w

ith D

LT

Observed Data: with 95% confidence intervals

1 2 3 4 5

0.0

0.2

0.4

0.6

0.8

1.0

Dose Level

Pro

po

rtio

n o

f Pa

tien

ts w

ith D

LT

This is actually better than most

Most studies treat only 3 or 6 at each dose level

With 0 of 6 DLTs: Estimated DLT rate = 0% 95% CI for DLT rate = [0%, 45%]

With 1 of 6 DLTs: Estimated DLT rate = 17% 95% CI for DLT rate = [0%, 64%]

Why do we use it all the time?

It is terribly imprecise and inaccurate in its estimate of the MTD

Why? MTD is not based on all of the data Algorithm-based method Ignores rate of toxicity!!!

Likely outcomes: Choose a dose that is too high

Find in phase II that agent is too toxic. Abandon further investigation or go back to phase I

Choose a dose that is too low Find in phase II that agent is ineffective Abandon agent

We could use smarter designs!

Phase I is the most critical phase of drug development!

What makes a good design? MTD situation: Accurate selection of MTD

dose close to true MTD dose has DLT rate close to the one specified

Relatively few patients in trial are exposed to toxic doses

What makes a good design? Non-toxic agent situation: Accurate selection of dose (range) which hits target Relatively few patients are treated Relatively few patients are exposed to ineffective

doses

This trial

Is MTD relevant? What is the goal? Should we be looking for hitting the

target? Toxicity ~ Efficacy? PK and PD data presented Although argument made for MTD,

PARP inhibition is relatively constant for the higher doses

Novel Designs

Why not impose a statistical model? What do we “know” that would help?

Monotonicity (often) Desired level of DLT

Statistical models improve: Prediction Efficiency

Accelerated Titration: incorporates model (next slide) Example: CRM (continual reassessment method)

Originally devised by O’Quigley, Pepe and Fisher (1990) dose for next patient was determined based on toxicity

responses of patients previously treated in the trial Others out there (and variations of CRM)

Another Accelerated Titration Feature: Model fit

1 2 3 4 5

0.0

0.2

0.4

0.6

0.8

1.0

Dose Level

Pro

po

rtio

n o

f Pa

tien

ts w

ith D

LT

CRM example

CRM software example

How would CRM have worked in this study?

Would have accelerated quickly Would have iterated at a few doses May not have treated so many patients

at MTD Would likely have been a smaller study Could have used PD data to help dose-

finding.

Discussion Phase 0 trials

PK and PD single dose 6-10 patients

Goals: Define dose range for Phase I Improve chance of success in phase I and II Better planning of phase I

New and exciting! First in man, pre-Phase I Messy though:

Phase 0 vs. Phase 1? How will this change Phase 1 goals?

My humble opinion: the development of Phase 0 strongly suggests that Phase I paradigm needs to be reconsidered