Clinical Trial Design

Definition

“Clinical trials are a special kind of cohort study in which interventions are specifically introduced by the investigators in ways that improve the possibility of observing effects that are free from bias”

Retrospective vs Prospective• Retrospective

– Gather data on historical cases. Limited by data collection, lead-time bias, temporal bias, yadda yadda

– Generates questions, NOT answers (way too much bias)

– Results should only rarely alter standard of care

– Unfortunately, much of our S-of-C in Vet Med based on retrospective analysis

• Prospective– Compare 1 population to another by entering patients as they are seen and using predetermined allocation procedures to control bias

Basic Clinical Trial Structure

Experimental treatment

Control treatment

PatientSelection

Allocation Intervention OutcomeMeasurement

Purpose of clinical trials

• Ultimate goal: The need to improve “standard-of-care”

• Determine dose and toxicity profile of a new drug/procedure– “Phase I trial”

• Determine “efficacy/activity” of new drug/procedure– “Phase II trial”

• Is it better than current treatment – (“comparative trial” or “Phase III trial”)

• Gather ancillary data (prognostic and predictive factors)

• Inform human clinical trials

Bringing a new anticancer drug to market … several

phases

Table 1: Goals of Phase I Š III Clinical Trials

Phase of Clinical Trial Characteristic Phase I (Dose Finding) Phase II (Activity/Efficacy) Phase III (Comparative) Primary Goals Determine MTD

Define DLT Elucidate parameters of

toxicity

Determine activity/efficacy in defined populations

Inform the decision to move to a Phase III trial

Compare a new drug or combination to therapy currently regarded as standard of care

Secondary Goals PK/PD issues Scheduling issues Target modulation effects Preliminary efficacy data

Estimate therapeutic index

Expand toxicity data Evaluate additional

dosing groups Expand target

modulation data Quality of life measures

Quality of life comparisons

Comparative costs

MTD, maximally tolerated dose; DLT, dose limiting toxicity; PK/PD, pharmacokinetic/pharmacodynamic.

Phase I Trial = “Toxicity Trial”

• Goal is to determine the MTD (maximally tolerated dose) for future phase II

• Elucidate parameters of toxicity

• Schedule and route• Secondary goals

– PK/PD issues (ADME)– Target modulation– Efficacy is NOT a primary

goal, but nice to see. Average response rate is <10% in phase I trials.

• Increasing dose in Cohorts of 3 is typical (3 + 3 design)

Who enters phase I• Refractory to standard-of-care• Often heavily pretreated and poor performance scores

• In veterinary trials, who enters a phase I?– No standard-of-care exists– standard-of-care sucks– Financial constraints on standard-of-care•Offset treatment costs •Additional incentive for alternate therapy at failure

– Altruistic?

Setting the Initial dose level

• Weigh toxicity probability against efficacy– Too low starting dose results

in prolonged trial, poor utilization of resources, and ethics of using sub-efficacious doses

– Normal volunteers - 1/10 NOAEL prediction

– Not the case in cancer trials, rather 1/3 NOAEL

– Patient advocates - patient can choose their starting dose

• If normal dog tox known, then 50% of MTD– beagles are not cancer-bearing

dogs• Typically, if no normal dog,

than 10% of the LD10 of rodent or other species

Table 2: Standard Phase I dose escalation scheme*

Number of Patients with DLT

at a Given Dose Level

Escalation Decision Rule

0 out of 3

Enter 3 patients at the next dose level

> 2

Dose escalation will be stopped. This dose level will be declared the maximally administered dose (highest dose administered). Three (3) additional patients will be entered at the next lowest dose level if only 3 patients were treated previously at that dose.

1 out of 3

Enter at least 3 more patients at this dose level. If 0 of these 3 patients experience DLT, proceed to the next

dose level. If 1 or more of this group suffer DLT, then dose escalation

is stopped, and this dose is declared the maximally administered dose. Three (3) additional patients will be entered at the next lowest dose level if only 3 patients were treated previously at that dose.

<1 out of 6 at highest dose level

below the maximally administered dose

This is generally the recommended phase 2 dose (MTD). At least 6 patients must be entered at the recommended phase 2 dose.

*From NCI CTEP Phase 1 protocol template http://ctep.cancer.gov/guidelines/templates.html). MTD, Maximum tolerated dose; DLT, dose-limiting toxicity

Dose Limiting Toxicity

– a dose-limiting toxicity (DLT) is defined as ≥ Grade III toxicity in any category except hematologic

– Grade IV used for hematologic toxicity

– use Veterinary Cooperative Oncology Group Common Terminology Criteria for Adverse Events (VCOG-CTCAE) v1.0.

CONSTITUTIONAL SYMPTOMSGrade

Adverse Event 1 2 3 4 5Lethargy/fatigue Mild

lethargyoverbaseline

ModeratelethargycausingsomedifficultywithperformingADL

Compromised,severelyrestricted inADL,ambulatoryonly to thepoint ofperformingADL

Disabled, mustbe force fed andhelped toperform ADL

Death

Fever (in theabsence of grade3 or 4neutropenia)

1o C aboveULN

1 - 2o CaboveULN

2 - 3o C aboveULN

>3o C aboveULN

Hypothermia --- 36 Š >33o

C33 - >28o C <28o C or life-

threateningconsequences(e.g., coma,hypotension,pulmonaryedema)

Death

Weight loss 5 Š 10%frombaseline;interventionnotindicated

>10 - 20%frombaseline;nutritionalsupportindicated

> 20% ofbaseline

---

Other (Specify,_________ )

Mild Moderate Severe Life-threatening;disabling

Death

ADL = activities of daily living (eating, sleeping, defecating and urinating)

BLOOD/BONE MARROWGrade

Adverse Event 1 2 3 4 5Bone marrowcellularity

Mildlyhypocellular;< 25%reductionfrom normalcellularityfor age

Moderatelyhypocellular;>25 - < 50%reductionfrom normalcellularityfor age

Severelyhypocellular;>50reduction ofcellularityfrom normalfor age

--- ---

Hemoglobin Dog:<LLN Š 10g/dlCat:<LLN Š 8.0g/dl

Dog:<10 Š 8.0g/dlCat:<8.0 Š 6.5g/dl

Dog:< 8.0 Š 6.5g/dlCat:<6.5 Š 5.0

Dog:< 6.5 g/dlCat:<5.0

---

Neutropenia 1,500 Š2,000/µL

1,000 -1,499/µL

500 -999/µL

< 500/µL ---

Thrombocytopenia 100,000 Š149,999/µL

50,000 Š99,000/µL

25,000 Š49,000/µL

<25,000 ---

Other (Specify,_________ )

Mild Moderate Severe Life-threatening;disabling

Death

LLN = lower limit of normal

Dose escalation strategies

• Increments:– Fixed-dose modified Fibonacci– Accelerated titration - single patients, dose

increased by factor of 2 till a grade 2 toxicity occurs, then more typical acceleration

– These < # of patients treated at doses far from the MTD

– If myelosuppression prime toxicity, then can base dose escalation on degree

– PK-based strategies– Target modulation strategies– Continuous reassessment

methods • Bayesian methods

– Always a trade off of risk versus benefit

MTD vs BOD

• MTD may be irrelevant for new so-called static drugs or targeted drugs

• The dose that maximizes the target modulation would be the end-point

• Could use validated surrogate assays

Phase II = “Efficacy trial”

• Goal– Using the MTD/BOD from phase I, identify the clinical or biologic activity (i.e., which tumor types respond)

– Standard single-arm study• Treat a minimum of 9 of each tumor histology or target population (this keeps the likelihood of a false negative result to 80% [type II error] assuming a 25% RR)

• If 1 responder, then > to 31 patients to get accurate response rate

• Treat more if expecting lower response rate

– Estimate therapeutic index– Expand toxicity data (e.g.,

cat doxil and renal toxicity)

Are all errors created equal?

• Old school: Consequence of type I error (false positive) is less deleterious than type II error (false negative) as false positive trials are likely to be repeated versus false negative trials may lead to the abandonment of a particular treatment– e.g., Cisplatin

• New School: False positives are just as serious due to– >>> # of new agents to investigate

– High cost $$ and patient resource cost is so high

Clinical Endpoints• Other endpoints:

– Q of L measures (gemzar)– Temporal measures for newer static agents• Time to progression• Progression free rate

– Molecular endpoints• Dephosphorylation of growth factor receptor

• Vessel density• Functional imaging*

*or new lesion

Phase III trials (Comparative Trials

•Phase II are “learning” trials and phase III are “confirming” trials

•Larger, randomized blinded trials

0 50 100 150 200 250 300 350 4000

10

20

30

40

50

60

70

80

90

100SPI-77Median = 219 d

CarboplatinMedian = 204 d

Days

P = 0.31/0.25

Only 14% of phase II trials in Oncology are

positive

“companies should consider phase II trial designs that are more predictive of phase III success.”

Booth, B, et al. Oncology trials. Nature Reviews Drug Discovery 1:609, 2003.

Modifications of Standard Trial Designs• Phase 0 Trials and Phase IV trials

• Adaptive Trial Designs– Stopping rules– Bayesian (continuous learning) methods

• Randomized phase II– Biologically effective dose (BED) may be well below the MTD (i.e., TK inhibitors)

– Scheduling issues– Multiple agents with enough power to make “inferences” as to which is best drug to take forward (but not enough power for direct comparison to standard-of-care = phase III)

• Randomized discontinuation design• Phase Combinations• Enrichment• Non inferiority trials• Crossover Trials

Stopping Rules

• Terminate trial within a predetermined adaptive trial design

• Protect patients from unsafe drugs and/or hasten availability of superior drugs

• 3 reasons to stop:1. Investigational drug

clearly better than control

2. It is clearly worse than control (< activity, > toxicity)

3. It is not likely to be better = “stopping for futility”

Stopping Rules

• Interim analysis by a blinded individual

• Predetermined rules• “conditional power”

– The probability that the final result will demonstrate statistical significance conditional on the current data and assumptions on future data

• “Stochastic curtailing”

Bayesian (continuous learning) Designs

• Adaptive designs:– Can stop trials early– Can change randomization weight to better performing arms

– Can add new arms– Extend accrual beyond target

• Frequentist approach– Traditional method uses fixed parameters - inflexible

• Bayesian - makes statistical inferences using:– Accumulated data– Historical data– Data from other trials

Bayesian Example

• EGFR2 +ve breast cancer study

• Initial design = 164 patients

• Bayesian approach after 34 patients enrolled– 67% CR in study arm– 25% in control– Bayesian predictive probability of 95% if 164 enrolled so trial stopped and Phase III initiated early

Buzdar AU, et al. J Clin Oncol 2005;23

Enrichment• Intent is to select a subset of patients to enroll that are relatively homogenous with respect to prognostic or predictive factors and randomize only these patients

• e.g:– All start, then randomize compliers/no-tox

– “drugable” target identification

– Exclude poor prognostic group

– Randomized discontinuation

Enrichment example:• Trastuzumab Her-2 mAb

+ chemo• 469 patients needed in study when entering only Her-2 over-expressers (1 yr. OS 78% vs. 67%)

• If unselected patients - 23,586 would have been required to show the difference!

• Problem: what if your wrong about the “target”?

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Slamon DJ, et al. NEJM 2001;344.

Phase II Trials: Paradigm Shift

• Cytotoxic phase II design: end point is tumor shrinkage (i.e.,= RECIST)

• Newer putative cytostatic agents slow tumor growth or induce stasis (induction of stable disease)– Heterogenous growth rates in patient population lead to problems

– Some naturally grow very slowly– Must distinguish antiproliferative activity of drug from indolent disease process (“no treatment effect”)

– e.g., small molecule TK inhibitors, antiangiogenic agents, metronomic protocols

16 weeks of Treatment:All patients!

StableDisease?

N(PD)

Y

RANDOM IZE

Treatment

Placebo*

Out oftrial

According to simulation, expect 70% of patients will progress during the 16 weeks, 0% will regress spontaneously and 30 % will be stable. * If placebo patients progress off drug; code broken and they are switched back to drug

Stage I (“run in”) Stage II(accrual in stage I is halted when sufficient numbers areRandomized into stage II [power calculation])

Rosner GL, et al. JCO 20:2002

Clinical

Response(RECIST)

Continue Treatment

Note, if high grade toxicity, also removed from trial!

Copyright ©2007 American Association for Cancer Research

Stadler, W. M. Mol Cancer Ther 2007;6:1180-1185

Figure 1">

Sorafenib +ve trial

• RAF kinase inhibitor

• 34% shrinkage of RCC by 12 wks

• 69 patients stable and randomized– 50% stayed stable in treatment group

– 18% in placebo– p = 0.0077

Crossover studies

Randomization

Control Tx

Control TxExperimental Tx

Experimental Tx

Wash outperiod

- comparison within groups: Patients serve as own control- appropriate for chronic, slowly progressive disease like allergies, arthritis, diabetes, incontinence

Table 4: Suggested "Elements of Consent" to include in

Informed Client Consent documents

1. Purpose of Research

2. Expected Duration of Participation

3. Description of Procedures

4. Possible Discomforts and Risks

5. Possible Benefits

6. Alternative Treatment (or Alternative to Participation)

7. Extent of Confidentiality of Records

8. Compensation or Therapy for Injuries

9. Contact Person for the Study

10. Voluntary Participation and Right to Withdraw

11. Te rmination of Participation by the Principal Investigator

12. Unforeseen Risks

13. Financial Obligations

14. Hospital Review Committee Contact Person

With permission from Morris Animal Foundation:

www.morrisanimalfoundation.org/reports/grants/full_proposal/Established_Investi

gator_guidelines.pdf