in Rare Patients " Some day risk will be zero My, oh my Some - - PowerPoint PPT Presentation

6/22/2013 1

Ninewells Hospital & Medical School

"Assessing Therapies

in Rare Patients"

Tom MacDonald

Some day drugs will be perfect If we try Some day drugs will be perfect And no one will ever die Some day risk will be zero My, oh my Some day pills will be magic And they’ll taste of apple pie

How do we decide?

Hierarchy of Evidence

1a Systematic review of RCTs with homogeneity 1b Individual RCT with narrow confidence intervals 1c All or none (i.e. fatal before Rx or ‘Lazarus’) 2a Systematic review of cohorts with homogeneity 2b Individual cohorts (including low quality RCTs) 2c Outcomes research, ecological studies 3a Systematic review of case-control studies 3b Individual case-control studies 4 Case series 5 Expert opinion, bench research, first principles

www.cebm.net/levels_of_evidence.asp

6/22/2013 2

Types of Rx to be evaluated

• Rapid onset + rapid loss of effect

– i.e. short acting symptomatic relief

• Rapid onset with carry-over effects

– i.e. short acting symptomatic relief with disease modifying effects

• Slow onset + slow loss of effect

– disease modifying drugs

• Pre-licensing
• Post-licensing

More Tests = Improved Precision

Test Frequency Value Obtained

Mean of Multiple Measures the target

6/22/2013 3

SDD

The Standard Deviation of the Differences between repeated measures in the same subjects

SDD values of 8.3 for ABPM and 11.0 for Clinic SBP

Am J Hypertens 2002;15:101-104

For a parallel trial to detect

∆ of 10 mmHg SBP,

51 subjects using Clinic SBP v 29 using ABPM

140 135 130 125 120 115 110

MASBP (mmHg) Lumiracoxib 100 mg qd Ibuprofen 600 mg tid

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Hour of the day using 24-hour clock

24h BP 5mmHg Lower Lumiracoxib v Ibuprofen

MacDonald et al J Hyp 2008;26:1695-1702

6/22/2013 4

Outcome Measures

• Symptom scores
• Exercise capacity

– 6 min walking – Treadmill, bicycle, etc (+/- VO2 max)

• Activity

– 24h activity monitors – 24h O2 saturation monitors

• Something else?

– Pulmonary artery pressure? – Biomarkers?

Pros + Cons

• Symptom scores / QOL

– Pro: Easy – Con: Soft

• Exercise Tests

– Pro: Harder endpoint – Con: Tiring, training effect

• Activity / Saturation Monitors

– Pro: Measure important variables – Con: Less ‘hard’ than testing

• Other endpoints

– Surrogates, ? relevance

Cohort Studies

• Sample size issues
• Accurate baseline important
• Serial measures of outcome
• Good for drugs with slow onset

& disease modifying effects

(not randomisation)

BMJ 2005;330;843

6/22/2013 5

Rx New Treatment Rx Placebo or Standard Rx

Assess suitability & randomize

Placebo

Expensive placebo more effective than cheap placebo

JAMA 2008;299:1016-7

6/22/2013 6

Where possible (pre licensing) compare to placebo

On top of best Rx

Repeated ‘measures’ tiring for patients

Need to embrace new technology iPhone Accelerometer App

http://www.jawbone.com/up

Programmable to Buzz to stimulate exercise!

Use Technology to capture large amounts of outcome data over time

6/22/2013 7

Patients are Precious

Multiple Measurements improve precision, reduce variance and increase power

Phenotypes

Severe disease: few symptoms Mild disease: many symptoms

Disease Severity

It is hard to improve the nearly normal

Example

Study severe hypertension: big falls in BP Study mild hypertension: small falls in BP Study low (normal) BP no fall in BP

6/22/2013 8 N Engl J Med 2010;362:1575-85.

14mmHg Lower BP: No benefit on APTC events

Reduced stroke

ACCORD-BP: Not a Mysterious Result

BP RISK

154 144 UKPDS 134 119 ACCORD

Study More Severe Disease

If Possible

6/22/2013 9

Given the between patient and within patient variability How can we study the effects of drug treatments?

How many patients are needed to show the beneficial effects

• f a drug?

One

6/22/2013 10

N of One Crossover Studies

Plac Rx A

• -
• +

+ + + + +

6 + v 6 - = P<0.05 even with a non-parametric test

Preferred Rx

50 subjects to detect 30% preference with 85% Power N of 1 studies carried out in relatively small numbers can be very informative Provides one solution to the pre-licensing assessment of drugs in rare conditions

How many Crossovers of what duration?

• Effect size expected

– Binary: 6 & 6 minimum – Small effects: many crossovers – Large effects: fewer crossovers

• Pharmacodynamics determines

Duration of treatments

– Short acting: weekly crossover – Medium acting: 1 month crossover

6/22/2013 11

How Drugs are Licensed

Disease Activity Proportion of subjects Placebo

Average of 8% benefit

New Rx

Individual Effects

Benefit No Benefit

Benefit : Risk Analysis

6/22/2013 12

Trial of Therapy Always Useful

Lots of benefit: accept some risk No benefit: accept no risk

N-of-One Studies

Pick the Winners post licensing

Streamlined Studies

Treatments Mailed to Patients

Research Pharmacy

6/22/2013 13

New Rx Licensed

GP Rx n-of-one pack (free)

Patient enters data each week

Enters unique pack code at and of crossover

Long term e-follow up

Only Significant Benefit Rx

Cost- effective threshold

Downside

• r possible advantage

Carry-over effects

Washout Period

Plac Rx A

Helps reduce carry-over effects

6/22/2013 14

Washout Period (fixed or random) with re-measured baseline

Test: is ∆ Rx A > ∆ Plac ?

Plac Rx A

∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆ ∆

1 2 4 3 6 5

1 2 3 5 4 6

Also: 6 time-dependent baselines following placebo and 6 following active treatment

Vary block duration to measure carry over effects

Plac Rx A

Random Variable Washout Periods

Plac Rx A

Analyse the duration of carry over effects

Measure Carry-over

R e s p e n s e t

• t

r e a t m e n t 1 2 3 4 5 6 7 8 9 1 D a y s 7 1 4 2 1 2 8 3 5 4 2 4 9 5 6

6/22/2013 15

Carry-over effects help

Mitigate missed dose effects etc

Ethical Issues

• Everyone gets active Rx v placebo

– Pre licensing

• Or active Rx v alternative Rx

– Post licensing

• Added to ‘best current Rx’
• Patients completing at least 1

crossover informative

• Multiple long term crossovers feasible

N-of-one studies

Pre & post licensing

• Good for:

– Symptoms – Patient measured outcomes – ‘Activity’ / exercise capacity – Measureable variables

• Not so good for:

– Long term outcomes.

Catch-22

• Cannot get medicine used

until cost-effective & safe

• Cannot get cost-

effectiveness (or safety) data until medicine is used

6/22/2013 16

Designs for licensed drugs

Random introduction of ‘New Rx’ with ‘Designed delay’ in comparators 50% Usual Care 50% New Rx 100% New Rx

1 year?

Designed Delay, Stepped Wedge, or Similar description

Patients Not Rx Patients Treated/Time Randomised if Possible

100 100

10% Rx 90% not Rx

Designed Delay, Stepped Wedge, or Similar description

Patients Not Rx Patients Treated/Time

50%

Randomised if Possible

100 100

50% Rx 50% not Rx

6/22/2013 17

Designed Delay, Stepped Wedge, or Similar description

Patients Not Rx Patients Treated/Time

100 100

100% Rx 0% not Rx

Post Licensing

Stepped wedge / designed delay studies likely to be approved by ethics committees

BMJ 2008;337:1085-7

Cost and staff resources a major problem

6/22/2013 18

MEMO

Sign up Patients

To:

Engage Patients

Can we do internet

• nly studies?

Never or rarely see patients?

6/22/2013 19

Is this a crazy idea?

BJCP 2011 Nov 15. doi: 10.1111/j.1365-2125.2011.04142.x. [Epub ahead of print]

eContact

• Dedicated website

https://www.safetyswineflu.co.uk

• Online / paper registration

– Patient information sheets – Consent

–Monthly follow-up

• Automated email or SMS

6/22/2013 20

6/22/2013 21

MEMO MEMO

“Investigators should develop and improve methods to help decision makers appraise the evidence”

Harveian oration at the Royal College of Physicians, London www.rcplondon.ac.uk/pubs/brochure.aspx?e=262

6/22/2013 22

Balancing Benefit & Risks

My Advisors

6/22/2013 23