BIAS IN CONTROLLED
CLINICAL TRIALS
Ljubljana 19 March 2012
SILVIO GARATTINI
METHODOLOGICAL REQUIREMENTS
FOR CLINICAL TRIALS
Ask important questions…
…answer them reliably
The objective is the patient,
the goal is his benefit
Yusuf S, Collins R, Peto R.
Why do we need some large, simple randomized trials? Stat Med 1984; 3: 409-420
PATIENTS CAN BE INVOLVED IN CLINICAL
TRIALS ONLY IF THERE IS A REASONABLE
POTENTIAL ADVANTAGE. FOR THEM OR
FOR FUTURE PATIENTS THE ADVANTAGE
COULD BE INCREASED EFFICACY,
DECREASED TOXICITY, DIFFERENT TOXIC
PROFILES, BETTER COMPLIANCE.
LONGER DURATION OF ACTION, etc.
MEDIANA 0,097 QALY
51% < 0,1 QALY
12% > 1,0 QALY
MIGLIORAMENTI INDOTTI
DA 281 NUOVI FARMACI
Basic Clin Pharmacol
2009, 105, Suppl. 1, 032
Walker et al., 2009
• THE EXCESSIVE USE OF PLACEBO
PLACEBO SHOULD BE USED ONLY WHEN
THERE ARE NO EFFECTIVE DRUGS FOR
A GIVEN INDICATION INAPPROPIATE USE
OF PLACEBO MAY LEAD TO AN OPTIMISTIC
EVALUATION OF A NEW DRUG
DENOSUMAB (ANTI-RANK LIGAND MAB)
HAS BEEN TESTED AGAINST PLACEBO IN
POST-MENOPAUSAL WOMEN WHEN THERE ARE
AVAILABLE SEVERAL ANTI-OSTEOPOROTIC AGENTS
Scrip 19/9/08
Giovannoni et al., 2010
Kappos et al., 2010
ADD-ON DESIGN
NEW DRUG PLACEBO
BASIC TREATMENT
+ +
Zinman et al., 2007
A FAIR COMPARISON WOULD HAVE
BEEN TO USE INSTEAD OF PLACEBO
ONE OF THE MANY ANTIADIBETIC
AGENTS AVAILABLE ON THE MARKET
Pratley et al., 2010
Effect of Noninsulin Antidiabetic Drugs Added
to Metformin Therapy on Glycemic Control,
Weight Gain, and Hypoglycemia
in Type 2 Diabetes Olivia J. Phung, PharmD
Jennifer M. Scholle, PharmD
Mehak Talwar, BS
Craig I. Coleman, PharmD
Conclusion When added to maximal metformin therapy, all noninsulin
antidiabetic drugs were associated with similar HbA1c reductions but
differed in their associations with weight gain and risk of hypoglycemia.
JAMA. 2010;303(14):1410-1418
Phung et al., 2010
• DESIGN OF CLINICAL TRIALS
CLINICAL TRIALS MAY BE
DESIGNED TO DEMONSTRATE
• SUPERIORITY
• EQUIVALENCE
• NON INFERIORITY
IN THE NON-INFERIOR TYPE OF TRIAL WITH AN ACTIVE
CONTROL, INVESTIGATORS ARE TESTING THE NULL
HYPOTHESIS THAT A NEW DRUG IS WORSE THAN THE
ACTIVE CONTROL (STANDARD) AND WHEN THEY CAN
REJECT THE NULL HYPOTHESIS THEY ACCEPT THE
ALTERNATIVE, THAT THE NEW DRUG IS NOT WORSE
THAN THE ACTIVE CONTROL.
ARAS, 2001 DRUG INFORMATION J.35,1157
SUPERIORITY
EQUIVALENCE
NON
INFERIORITY
CONTROL BETTER NEW AGENT BETTER 0 - +
Are there specific reasons for allowing a
non-inferiority approach?
• There may be non-responders to current treatments
and products with comparable activity may offer a
useful alternative.
If the target is non-responders to current
treatments, why not test their superiority over
drugs with little effect in this subset of patients?
18 ANTICANCER AGENTS
21 INDICATIONS
12 ONLY PHASE II
9 PHASE III
6 EQUIVALENCE OR NON INFERIORITY
3 SUPERIORITY
Fueglistaler et al., 2007
A B C Therapy
Eff
icacy %
OUT OF 383 CLINICAL TRIALS
64 % COULD DETECT A DIFFERENCE > 50 %
84 % COULD DETECT A DIFFERENCE > 25 %
MOHER et al., 1994
Between 2002 and 2009, of 43 NDAs
containing data from at least one noninferiority
trial, the FDA approved 18 on the basis of
pivotal evidence from noninferiority trials but
found nine NDAs had poorly designed
noninferiority trials that could not be used to
establish the efficacy of the new drug.
Nature Medicine 2010, 16, 1049
THE DESIGN OF N.I. TRAILS REQUIRES
ALWAYS THE USE OF STANDARD AND
PLACEBO WILE A SUPERIORITY DESIGN
REQUIRES ONLY THE USE OF STANDARD.
Garattini S, Bertele’ V.
How can research ethics committees protect patients
better?
BMJ 2003; 326:1199-201
Draft informed consent
“Let us treat you with something
that at best is the same as what
you would have had before, but
might also reduce - though this is
unlikely - most of the advantages
previously attained in your
condition. It might even benefit
you more than any current
therapy but, should that actually
happen, we shall not be able to
prove it. Nor have we enough
chance to let you know whether
the new treatment may somehow
bother or even harm you more
than the standard one”.
NON-INFERIORITY TRIALS
FEW PATIENTS WOULD AGREE TO
PARTICIPATE IF THIS MESSAGE IS
CLEAR IN THE INFORMED CONSENT FORM.
THREE ARM TRIAL
RANDOMIZATION
PLACEBO COMPARATOR NEW DRUG
PLACEBO IS NOT NECESSARY IN THE DESIGN OF SUPERIORITY
• SURROGATE END POINTS
QUALITY OF LIFE, MORBILITY, MORTALITY
SHOULD BE THE FINAL OBJECTIVES
OF CLINICAL TRIALS
• ENCAINIDE AND FLECAINIDE DECREASE ARRHYTHMIAS
BUT INCREASE MORTALITY
• OESTROGENS INCREASE HDL-CHOLESTEROL
BUT DO NOT PREVENT CARDIOVASCULAR EVENTS
• TORCETRAPIB INCREASES HDL-CHOLESTEROL
BUT INCREASES MORTALITY
HbA1c IS NOT A SURROGATE END-POINT FOR
CARDIOVASCULAR DISEASES IN TYPE 2 DIABETES
SULFONYLUREAS HbA1C MYOCARDIAL INFARCTION
ROSIGLITAZONE HbA1C HEART FAILURE
NEED TO EVALUATE PARAMETERS IMPORTANT FOR PATIENTS
Arguedas et al., 2009
RIMONABANT
antagonist of cannabinoid-1 receptor
• DECREASE OF BODY WEIGHT
• INCREASE OF HDL, ADIPONECTIN
• DECREASE OF TRIGLYCERIDES, BLOOD
GLUCOSE, FASTING INSULIN, LEPTIN
• DECREASE OF TOTAL VOLUME ATHEROMA
Topol et al., 2010
CRESCENDO TRIAL
• UTILIZATION OF COMPOSITE END-POINTS
Lapostolle et al., 2007
Ferreira-Gonzàlez et al., 2007
• SELECTIVE PUBLICATIONS
SELECTIVE PUBLICATIONS
OF 5 SSRI
• 21 STUDIES SHOW DRUG BETTER THAN PLACEBO
• 19 PRIMARY PUBLICATIONS
IN TOTAL 42 STUDIES
Sweedish Drug Regulatory Authority Melander et al., BMJ 2003
SELECTIVE PUBLICATIONS
OF 5 SSRI
• 21 STUDIES SHOW DRUG BETTER THAN PLACEBO
• 19 PRIMARY PUBLICATIONS
IN TOTAL 42 STUDIES
• 21 STUDIES SHOW NEGATIVE RESULTS
• 6 PRIMARY PUBLICATIONS
Sweedish Drug Regulatory Authority Melander et al., BMJ 2003
SELECTIVE PUBLICATIONS
OF 5 SSRI
• 21 STUDIES SHOW DRUG BETTER THAN PLACEBO
• 19 PRIMARY PUBLICATIONS
EVIDENCE
B(I)ASED
MEDICINE
IN TOTAL 42 STUDIES
• 21 STUDIES SHOW NEGATIVE RESULTS
• 6 PRIMARY PUBLICATIONS
Sweedish Drug Regulatory Authority Melander et al., BMJ 2003
Kirsch et al., 2008
Eyding et al., 2010
Eyding et al., 2010
SELECTIVE REPORTING OF CLINICAL TRIALS MAY
BIAS META-ANALYSES AND MISDIRECT GUIDE-LINE.
IN A RETROSPECTIVE REVIEW OF 130 TRIALS,
THOSE WITH POSITIVE RESULTS WERE THREE
TIMES MORE LIKELY TO BE PUBLISHED WITH A
SIGNIFICANT SHORTER TIME TO PUBLICATION THAN
THOSE WITH NEGATIVE RESULTS.
STERN, SIMES, 1997