Risk, benefit and patient preferences in the treatment of Crohn’s disease Bruce E. Sands, MD, MS Associate Professor of Medicine Harvard Medical School.

Risk, benefit and patient preferences in the treatment of Crohn’s disease

Bruce E. Sands, MD, MSAssociate Professor of Medicine

Harvard Medical SchoolMedical Co-Director, MGH Crohn’s & Colitis Center

Massachusetts General HospitalBoston, USA

Disclosure

• Abbott Immunology• Berlex • BiogenIDEC• Bristol Myers Squibb• Centocor• Elan• Genentech• Millennium Pharmaceuticals• PDL Biopharma• Procter & Gamble• Prometheus Laboratories• Schering Plough• UCB

A question of risks and benefits

• Biologic therapies (indeed, all of the most effective therapies to date) have possible serious adverse effects

• Biologic therapies may also achieve levels of benefit not previously described

• How should risks and benefits be evaluated, and by whom?

Background

• Objective: To estimate the maximum risk of death or disability for specific clinical benefits of biologic therapy that is acceptable to patients with Crohn’s disease

• Conjoint analysis selected as analytical approach• Most valid and reliable technique available for

quantifying patient preferences1,2

1. Johnson FR, et al. Willingness to Pay for Improved Respiratory and Cardiovascular Health: A Multiple-Format Stated-Preference Approach. Health Economics 2000; 9:295–317.

2. Ryan M, et al. Methodological issues in the application of conjoint analysis in health care. Health Economics 1998;7: 373-8.

Conjoint Analysis

• Developed in mathematical psychology, strong theoretical basis

• Gaining widespread use in health care and policy

• Technique based on premises that • any good or service can be described by its

characteristics or attributes, and that• the extent to which an individual values a good or

service depends on these characteristics

• Can be used to• show how people are willing to trade between

characteristics• Estimate relative importance of different characteristics

Ryan and Farrar. BMJ. 2000;320:1530-3.

Questionnaire Content

Designed to approximate treatment decisions that are typical of CD patients

• Demographics

• Elements of the Crohn’s Disease Activity Index (CDAI)

• Short form of Inflammatory Bowel Disease Questionnaire (SIBDQ)

• Description of levels of clinical benefit and treatment attributes

• Description of treatment risks

• Benefit-risk trade-off questions• 10 questions, 6 versions

— 4 questions for internal validity tests— Remaining 6 questions randomized

Treatment Attributes

Attributes Attribute Levels

Serious complications

Prevents all

Reduces some

Has no effect

Time between flare-ups 6 months

2 years

Need to take oral steroids Yes

No

10-year mortality risks due to serious adverse events

• Serious infection

• PML

• Lymphoma

0%

0.5%

2%

5%

Definition of Serious Adverse Events

• Severe InfectionStudies suggest that some medicines for Crohn's disease increase your chance of getting a serious infection. These include sepsis, an infection of the blood system, and tuberculosis and pneumonia, which are infections of the lungs. Despite the availability of the newer antibiotics, infections can and do result in death for some people.

• LymphomaStudies suggest that some medicines for Crohn's disease increase your chance of getting lymphoma, a cancer of the lymph nodes. A person with lymphoma usually becomes aware of a lump in the neck, armpit, or groin. Other symptoms may include fever, night sweats, weight loss, and fatigue. Even with proper treatment lymphoma can result in death.

• PMLStudies suggest that some medicines for CD increase your chance of getting PML (progressive multifocal leukoencephalopathy). PML is a rare brain infection. The symptoms of PML are serious and may include the inability to think clearly, paralysis, blindness, and coma. PML can result in death or serious disability.

Representations of Risk:Absolute Risks

Krupnick A, et al. (2002). Age, health and the willingness to pay for mortality risk reductions: a contingent valuation survey of Ontario residents. Journal of Risk Uncertainty 24(2): 161-86.

Representations of Risk:Comparative Risks

Risk that a 50-year old man will die in 10 years from various causes

Possible risks of serious side effects you will be asked to consider

Accident

CancerHeart Attack All Causes

101,000

51,000

201,000

301,000

401,000

501,000

601,000

701,000

801,000

01,000

0% 0.5% 1% 2% 3% 4% 5% 6% 7% 8%

Risk that a 50-year old man will die in 10 years from various causes

Possible risks of serious side effects you will be asked to consider

Accident

CancerHeart Attack All Causes

101,000

101,000

51,000

51,000

201,000

201,000

301,000

301,000

401,000

401,000

501,000

501,000

601,000

601,000

701,000

701,000

801,000

801,000

01,000

01,000

0% 0.5% 1% 2% 3% 4% 5% 6% 7% 8%

Attributes Treatment A Treatment B

Severity of Crohn's symptoms during treatment

Moderate

• Moderate pain most days or severe pain on some days

• About 8 or more diarrhea stools per day

• Generally feel very poorly

• Considerable problems with work and leisure activities

Remission

• No pain most days

• No more than 1 diarrhea stool per day

• Generally feel well

• No problems with work or leisure activities

Effect on serious complications

(fistulas, abscesses or bowel obstructions)

Prevents some of the serious complications

Time between flare-ups 1 year

Treatment requires taking oral steroids Yes

Chance of dying from a serious infection within 10 years None would die

Chance of dying or having severe disability from PML within 10 years

None would die or have severe disability

20 patients out of 1,000 (2%) would die or have severe disability

Chance of dying from lymphoma within 10 years None would die

Which would you choose if these were the only options available?

Example of a Stated Preference Trade-Off Question

Patients

• Inclusion criteria• Diagnosis of Crohn’s disease• Age > 18• US resident• On-line informed consent

• Patient identification and recruitment• Natalizumab naïve internet panel (n=342)• Natalizumab naïve clinical panel (clinical research and clinical

practice sites) (n=140)• Natalizumab experienced patient panel (clinical research sites)

(n=98)

• Patients required to access survey via the internet• All patients issued web URL and unique password

• IRB review and HIPAA compliant data collection process

Primary Endpoints

• The relative contribution of the treatment efficacy and risk attributes to patient preferences

• The mean maximum acceptable annual risk (MAR) for clinically relevant treatment efficacy attributes

• The percent of patients accepting various levels of risk for clinically relevant treatment efficacy attributes

Results: Relative Importance of Treatment Attributes

0102030405060708090

100

SymptomSeverity

LymphomaRisk

SeriousInfection

Risk

PreventComp.

Time toFlare-up

Natalizumab Naïve Internet PanelNatalizumab Naïve Clinical Panel

Imp

ort

an

ce

0102030405060708090

100

PMLRisk

SteroidUse -

Natalizumab Patient Panel

Imp

ort

an

ce

Results: Maximum Acceptable Risk for Clinically Relevant Benefits

0.00.1

0.20.3

0.40.50.60.7

0.80.91.01.1

Moderate toMild

Moderate toRemission

Severe toModerate

Severe toMild

Severe toRemission

MA

R (

An

nua

l % R

isk)

PMLSerious InfectionLymphoma

EstimatedPML Risk(95 % CI)

4,611Estimated Risk: 3 cases of PML in 3,116 Natalizumab patients treated for 17.9 months (4,648 person years) (Yousri NEJM, 2006)

3= 0.065%

EstimatedPML Risk(95 % CI)

4,611

Results: Maximum Acceptable Risk for Clinically Relevant Benefits

Estimated Risk: 3 cases of PML in 3,116 Natalizumab patients treated for 17.9 months (4,648 person years) (Yousri NEJM, 2006)

3

MA

R (

An

nua

l % R

isk)

0

0.1%

0.2%

0.3%

0.4%

0.5%

0.6%

0.7%

0.8%

Moderate to Mild

Moderate to Remission

Severe to Mild

Severe to Remission

Severe to Moderate

Natalizumab Naïve Internet PanelNatalizumab Naïve Clinical PanelNatalizumab Patient Panel

Moderate to Mild

Moderate to Remission

Severe to Mild

Severe to Remission

Severe to Moderate

= 0.065%

Results: % of Patients Accepting Risk for Clinically Relevant Benefits

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0% 0.2% 0.3% 0.4% 0.5%Percent Risk

Moderate to Remission

Severe to Remission

Pe

rce

nt A

cce

ptin

g R

isk

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0% 0.1%Percent Risk

Severe to MildSevere to Moderate

Moderate to Mild

EstimatedPML Risk(95 % CI)

Discussion• MAR findings are similar across patient panels

• Treatment decisions were primarily driven by the desire for improvements in daily symptom severity

• CD patients are willing to accept significant risks for clinically relevant therapeutic benefits

• Great majority of patients would accept the observed risks in order to achieve clinically relevant therapeutic benefits

Acknowledgements

• RTI• Reed Johnson• Semra Ozdemir

• Elan• Steve Hass• Jeff White• David Miller

• Dartmouth-Hitchcock• Corey Siegel

Treatment AttributesSymptom Severity and Activity Limitations

Remission Mild

CD

Moderate

CD

Severe

CD

No pain most days

Mild pain most days

Moderate pain most days or severe pain some days

Severe pain most days

No more than one loose stool

per day

About 3 diarrhea stools per day

About 8 or more diarrhea stools

per day

More than 12 diarrhea stools

per day

Generally feel well

Generally feel below par

Generally feel poorly

Generally feel terrible

No problems with work or

leisure activities

Some problems with work or

leisure activities

Considerable problems with work or leisure

activities

Unable to engage in work

or leisure activities

Results: Disposition of Patients in Samples

1,007 persons accessed the survey

(8%)

650 patients failed entry criteria or discontinued

(65%)

357 patients completed the questionnaire

(35%)

25 patients failed data quality tests

(4%)

342 patients passed data quality tests

(96%)

31 patients failed entry criteria or discontinued

(11%)

9 patients failed data quality tests

(4%)

98 received Natalizumab in a

clinical trial

278 patients enrolled from clinical practice sites accessed the

survey

247 patients completed

questionnaire(89%)

238 patients passed data quality tests

(96%)

140 patients had not participated in a

clinical trial

Natalizumab Naïve

ClinicalPanel

(N= 130)

Natalizumab Patient

Panel(N= 98)

NatalizumabNaïve Internet

Panel(N= 342)

Results: Patient Characteristics

Variable Natalizumab Naïve Internet Panel

Natalizumab Naïve Clinical

PanelNatalizumab Patient Panel

Sample Size 342 140 98

Age*

mean (SD)45 39 43

Gender*

% Female73 66 53

Years since diagnosis*

mean (SD)

7.2

(3)

6.8

(3)

8.3

(3)

# diarrhea stools in past 7 days

mean (SD)

15

(21)

13

(17)

26

(26)

IBDQ score

mean (SD)

144

(38)

155

(39)

132

(39)

* P < 0.05

Rationale for the Selection of a 10-Year Risk Horizon• The chronic nature of CD and the clinical benefits of its

treatments are easily described in relation to a time horizon of several years

• A 10-year period enhances the plausibility of the risk exposure for a medication taken on a continuing basis for a chronic condition

• Increases the magnitude of risks to a range that patients can evaluate. Previous studies indicate that most people have difficulty conceptualizing probabilities less than 1/1,000 (Krupnick et al., 2002)

• Pretests of the instrument confirmed that 10 years is a reasonable planning horizon for MS patients

Validity Tests in Crohn’s Survey

Validity Test

Natalizumab Naïve

Internet Panel

(n=342)

Natalizumab Naïve

Clinical Panel

(n=140)

Natalizumab Patient

Panel

(n=98)

Stability 93% 89% 93%

Rational Choice 93 95 96

Monotonicity 96 89 84

Transitivity 89 96 95

PERCENT OF PATIENTS WHO PASSED TEST

Distribution of IBDQ Scores: ENACT-1 Baseline Versus Risk-Benefit Study

0

5

10

15

20

25

30

<69 70-94 95-119 120-144 145-169 > 170

ENACT-1 Baseline

CD Internet Patient Panel

Pe

rce

nt

RemissionImproving QOL

•Panel had somewhat better QOL scores than ENACT-1 patients at baseline

Internet Patient Panel CD Treatment Experience

Treatment Current TX Past TX Never Used

5-ASA 51% 44% 5%

Immunomodulators 41% 30% 29%

Infliximab 27% 30% 43%

Oral steroid 14% 74% 12%

Anti-infectives 10% 71% 19%

Budesonide 10% 37% 53%

Adalimumab 3% 2% 95%

Natalizumab 0.4% 3% 97%

Possible Disadvantages of Stated-Choice Approach

• Potential for hypothetical bias• Not same clinical, financial, emotional consequences as

real treatment choices• Realistic treatment options

• Cognitively challenging for subjects• Need for careful subject preparation• Definition of conditions• Probability concepts• Practice with tradeoff task• Internal validity tests

• Requires expertise in • Qualitative and quantitative survey methods• Experimental design• Advanced statistical analysis

Results: Maximum Acceptable Risk by IBDQ Score

IBDQ <= median (146) IBDQ > median (146)

0%

0.2%

0.4%

0.6%

0.8%

Moderate to Mild

Moderate to

Remission

Severe to Mild

Severe to

Remission

Estimated PML Risk

Remission

MA

R (

An

nua

l % R

isk)

0.1%

0.3%

0.5%

0.7%

Multiple Sclerosis StudyMaximum Acceptable Risk Estimates

0%

0.2%

0.4%

0.6%

0.8%

0.1%

0.3%

0.5%

0.7%

MA

R (

An

nua

l % R

isk)

PMLLiver FailureLeukemia

EstimatedPML Risk(95 % CI)

BenefitRelapses: 4 1 during a 5 year period

Progression: 3 yrs 5 yrs

Questionnaire Development

• Questionnaire was designed to:• Approximate treatment decisions that are typical of CD patients• Map the patients’ CD clinical status to standard measures of

severity

• Content drawn from:• Published literature• Established CD instrumentation• Collaboration of gastroenterologists & researchers experienced in

questionnaire design

• Pre-testing:• Assessed the patients’ ability to understand survey content• Assessed their willingness to tradeoff treatment efficacy against

serious side effect risks — Cognitive interviews (n=10)— Pilot data collection (n=51)

Traditional View

Reject

Accept

Minimum acceptable

efficacy

Benefit

Risk

Maximum acceptable

risk

More Realistic View

Risk

Benefit

Maximum acceptable

risk for patients

Minimum acceptable efficacy for

patients

Reject

Accept

Minimum acceptable efficacy for physician

Maximum acceptable

risk for physician