Outcomes in Randomized Controlled Trials in Pain: A Proposed Responder Analysis

1
Outcomes in Randomized Controlled Trials in
PainA Proposed Responder Analysis

• Vibeke Strand, MD
• Biopharmaceutical Consultant
• Clinical Professor, Division of Immunology,
  Stanford University

2
Responder Analyses have Face and Content Validity

• Allow assessment of multiple domains
• Could better categorize analgesics
• Facilitate comparison of efficacy across
• Products
• Heterogeneous populations, and
• Multiple disease indications
• May lead to tiered approach to label indications
• Precedent ACR Responder Index in RA

3
ACR Response Criteria used in RA RCTs

• Require 20 improvement in 5 of 7 measures
• Tender Joint Count and Swollen Joint Count
• and 3 of the following 5 MD
  Global Physical function HAQ Pain by
  VAS Patient Global ESR and/or CRP

4
Strength of Rheumatoid Arthritis GuidanceDocument

• Tricenter agreement Proven track record ? 6
  products!
• Tiered Label Indications
• Improvement in Signs and Symptoms
• By ACR Response Criteria
• At 6 or 12 months
• Inhibition of Radiographic Progression
• Sharp Scores erosions JSN
• At 12 months
• Improvement in physical function and HRQOL
• HAQ and SF-36
• Over 2-5 years
• May be achieved in a single protocol using
  prespecified outcome criteria and Hochberg
  analysis

5
Appropriate Domains for Inclusion in RCTsin
Chronic Pain

• Based on a Breakout Session at the Jointly
• Sponsored FDA / NIH Workshop in May 2002
• Definition for the Workshop
• RCTs of 3 months duration
• In patients with pain of gt 3 months duration,
• regardless of underlying cause

6
Examples of Chronic Pain Indications

• Musculoskeletal
• RA
• OA
• Low Back Pain
• Fibromyalgia
• Neuropathic
• Diabetic Neuropathy
• Post Herpetic Neuralgia
• Trigeminal Neuropathy
• Cancer Pain not necessarily gt3 months
• Rapidly progressive disease
• Adjust intervention as disease progresses

7
Proposed Domains for Chronic Pain

• To Be Selected
• Regardless of clinical indication
• Consider available instruments whether or not
• Validated in pain or specific clinical
  indications
• Previously used in RCTs in Pain
• Disease specific or generic
• Strength of choices based on multiple available
  instruments and prior clinical experience

8
Proposed Domains for Chronic Pain

• PAIN brief, ongoing allodynia Need multiple
  measures of pain Patient global
  assessment Rescue medications Time to
  treatment failure
• Suffering
• Pain relief
• Disease specific measure of improvement /
  physical function
• Health related quality of life HRQOL
• Patient global assessment
• AEs and how perceived by the patient
• Damage irreversible due to disease or its Rx
• Economics

9
Appropriate Domains for Chronic PainFinal Vote

• PAIN unanimous 28
• Physical function / disease specific measure
  27
• HRQOL 24
• Patient Global 15
• AEs 11
• This represents a core set of minimum required
  outcome domains to be assessed. Others, specific
  to the underlying disease process or the clinical
  indication, may be added as secondary endpoints..

10
Goal Define Improvement Multidimensionally

• Separate experience of pain itself, from
• Functional Impairments and Disability
• Which may / may not ensue
• Separate physical impairment from disability
• Utilize individual responder analyses
• Utilize disease specific or relevant,
• as well as generic measures of HRQOL

11
Limitations Do NOT Necessarily Imply Disability

• IMPAIRMENT Due to Pain /or Structural
  Alterations

• Functional Limitations
• ??
• Disability In the eyes of the
  beholder
• Age, Gender Appropriate
• Work v Family / Social

12
Measures for Chronic Pain Patient
GlobalAssessment of Treatment Intervention

• In all the ways your pain affects you, including
  its treatment,how are you doing today?
• Transition question
• Utilities
• EuroQOL or EQ5D
• Health Utility Index HUI

13
Minimum Clinically Important DifferencesMCID

• Degree of improvement in various outcome
  measures
• Perceptible to patients
• Considered clinically important / meaningful
• Defined by patient query, delphi
  technique OMERACT 33-36 improvement 18 gt
  placebo
• Demonstrated by statistical correlations with
  clinical responses in RCTs patient global
  assessments
• Determination of proportion of patients with
  clinically important improvement provides a more
  interpretable result with direct clinical
  implications

14
Consistency of MCID Values

• Changes in disease specific or relevant
  measures of function / HRQOL related to much
  or very much improvement in patient global
  assessment
• Changes in generic measures of HRQOL related to
  improvements in patient global assessment
• Consistent high correlations between disease
  specific and generic measures of HRQOL
• MCID values consistent. Examples OA
  RA FM

15
Measures of Chronic Pain
16
Measures of Chronic Pain PAIN

• Numeric Pain Intensity Scale
• VAS or Faces or graphic scales anchor VAS
• Verbal rating scales Likert or VAS
• Brief Pain Inventory - BPI
• McGill Pain Questionnaire MPQ
• Chronic Pain Coping Inventory - CPCI
• West Haven-Yale Multidimensional Pain Inventory
  MPI
• ASA Nine Outcomes Measures
• Treatment Outcomes in Pain Survey - TOPS

17
Faces Rating Scale
18
MCID Pain Intensity Numerical Rating Scalev.
Patient Global Assessment of Change

• 10 placebo RCTs of Pregabalin in Diabetic
  neuropathy, Postherpetic neuralgia, LBP,
  Fibromyalgia and OA
• Relationship of much, very much improved in
  PGIC LIkert 5 to PI-NRS 10 points
• Reduction of 30 or 2 points in Pain Intensity
  PI-NRS MCID
• Regardless of Baseline pain or disease state

Farrar et al Pain 2001 94149-158
19
Measures of Physical Function /or HRQOL Chronic
PAIN

• Brief Pain Inventory BPI cancer HRQOL not
  extensively validated in non malignant pain
• McGill Pain Questionnaire MPQ intensity and
  subjective experience of pain
• Chronic Pain Coping Inventory CPCI coping
  strategies and level of adjustment
• Multidimensional Pain Inventory MPI
  HRQOL how psychosocial role functioning
  affected omits work-related activity
• Treatment Outcomes in Pain Survey TOPS HRQOL
  measured longitudinally in individuals

20
Generic HRQOL Instruments

• Sickness Impact Profile SIP implies illness
• Nottingham Health Profile - NHP
• Medical Outcomes Survey MOS - SF-12 and SF-36
• HRQOL in large groups across disease states
• Limited assessment of UE, facial pain
• Poor differentiation of LBP v upper body pain
• WHO Quality of Life Instrument - WHOQOL-100
  Newer instrument
• EuroQOL or EQ5D widely used in EU
• Quality of Well Being - QWB

21
Disease Specific RelevantMeasures of
Physical Function /or HRQOL
22
Disease Specific Measures Physical Function /or
HRQOL

• Rheumatoid Arthritis HAQ, MHAQ, MDHAQ
• AIMS, AIMS-2
• MACTAR, PET
• Osteoarthritis WOMAC
• Harris Hip Score
• HSS Knee Score and Knee Society Score
• Ankle OA Score
• AOFAS hind-, mid- and fore-foot scores
• LBP Roland-Morris Oswestry Disability
  Questionnaire
• Geriatrics Katz Index of Independence in ADLs
• HAQ
• Cancer Functional Living Index, Cancer FLIC
• Functional Assessment of Cancer Therapy
  FACT

23
EXAMPLERheumatoid Arthritis Disease Specific
Measures of Physical Function /or HRQOL
24
Health Assessment Questionnaire - HAQ

• Widely accepted, validated, rheumatology-specific
  instrument to assess physical function in RA
• Gold Standard OMERACT/FDA Guidance
• 20 questions covering 8 types of activities
  Dressing Grooming Arising Eating Walking
  Hygiene Reaching Gripping, ADLs
• HAQ Disability Index (HAQ DI)
• Scores the worst items within each scale
• Based on use of aids and devices
• Other, shorter versions equally useful MHAQ,
  MDHAQ

25
SF-36 Short Form 36 Health Survey

• Validated, widely used generic measure of HRQOL
• 8 Domains scored 0 - 100 age, gender adjusted
  norms
• Designed to evaluate health status of large
  groups
• Doesnt separate work limitations v everyday
  activity
• Summary Scores Normative based (Mean 50, SD
  10)
• Physical Component PCS
• Impact of physical fxn impairment / disability
• Mental Component MCS
• Impact of mental affect, pain symptoms
• Change evident within 4 weeks

26
SF-36 Two-Component Model
Physicalcomponent
27
Minimum Clinically Important Differences MCID
Score Direction MCID Range of
Scoring Literature
HAQ DI 1-4 0 - 3 0.22
SF-36 2, 5-7 0 - 100 5 - 10 points
PCS/MCS mean 50 10 2.5 - 5 points
1 Guzman et al. Arth Rheum. 1996 395208 2
Kosinski et al. Arth Rheum. 2000 431478-873
Redelmeier et al. Arch Intern Med. 1993
1531337-42 4 Wells et al. J Rheumatol. 1993
20557-605 Kosinski et al. Arth Rheum. 2000
43S140 6 Samsa et al. Pharmacoeconomics. 1999
15141-155 7 Thumboo et al. J Rheumatol. 1999
2697-102.
28
Mean Improvement in HAQ Disability Index Year-2
Cohort at 24 Months
LEF
MTX
SSZ
Worsening
US301
MN302/304
MN301/303/305
(248)
(51)
(46)
(101)
(273)
(97)
0
-0.22
Mean Change from Baseline
-0.37
-0.5
-0.48
-0.56
-0.56
-0.6

-0.73
Improvement
LEF vs MTX p0.01
-1
29
ATTRACT HAQ Disability IndexMean Improvement
through Week 102
0.5
0.5
0.5
0.45
0.4
0.4
0.4
0.3
0.2
Mean improvement
0.2
0.1
0
3 mg/kg q8w
3 mg/kg q4w
10 mg/kg q8w
10 mg/kg q4w
All infliximab
MTX Placebo
lt 0.001
lt 0.001
lt 0.001
lt 0.001
p-value vs. MTX Placebo
30
ERA Improvement in HAQ Disability Indexat 24
months
Baseline 1.44 1.49
31
US 301 Baseline SF-36 Scores US Norms vs US301
Population
Study US301 Population
US Norms (A/S Adjusted)
100
90
80
70
60
50
40
30
20
10
0
PhysicalFunction
RolePhysical
BodilyPain
GeneralHealthPerception
Vitality
Social Function
RoleEmotion
MentalHealth
32
US301 Mean Improvement in SF-36 DomainsITT
Cohort at 12 Months
PBO (n101)
LEF (n157)
MTX (n162)
24

Improvement

20

16


12

Mean Change from Baseline
8

4
0
Worsening
Physical Role Bodily General Vitality Social Role
Mental Function Physical Pain Health Function E
motion Health Perception
-4
LEF vs PBO plt0.05LEF vs MTX plt0.05
33
US301 Mean Improvement in SF-36 Year-2 Cohorts
Leflunomide and Methotrexate
US Norms (A/S Adjusted)
Baseline Year-2 Cohort
90
80
70
60
50
Mean Scores
40
30
20
10
0
Physical Role Bodily General Vitality Social Role
Mental Function Physical Pain Health Function Em
otion Health Perception
34
US301 Mean Improvement in SF-36 Year-2 Cohorts
Leflunomide and Methotrexate
US Norms (A/S Adjusted)
LEF 24 Months (n 93)
Baseline Year-2 Cohort
MTX 24 Months (n 89)
90
80
70
60
50
Mean Scores
40
30
20
10
0
Physical Role Bodily General Vitality Social Role
Mental Function Physical Pain Health Function Em
otion Health Perception
35
ATTRACT Mean Improvement in SF-36 Week 54
Physical Domains
MTX Control
3 mg/kg q8 Wks
3 mg/kg q4 Wks
10 mg/kg q8 Wks
10 mg/kg q4 Wks
MTX Control
3 mg/kg q8 Wks
3 mg/kg q4 Wks
10 mg/kg q8 Wks
10 mg/kg q4 Wks
0.035
lt0.001
lt0.001
0.002
0.083
lt0.001
lt0.001
0.002
p-value vs. MTX control
p-value vs. MTX control
36
US301 Mean Improvement in PCS and MCS
Leflunomide Year-2 Cohort at 12 24 Months
Baseline
Month 12
Month 24
(n93)
(n93)
(n93)
60
53.2
52.2
48.5
US Norm
50
42.7
41.7
40
2 SDs below US Norm
30.9
30
20
10
0
PCS
MCS
PCS Physical Component Summary Score MCSMental
Component Summary Score
37
ATTRACT Median Improvement in SF-36 PCS Week 102
Baseline 23.9 30.8
38
ERA Mean Improvement in SF-36 PCS 12 Months
US Norm

29.20
Mean Normal Scores
28.01
25 mg (198)
MTX (217)
10 mg (208)
25 mg (207)
MTX (204)
10 mg (194)
MTX (199)
10 mg (188)
25 mg (193)
Baseline
6 Months
12 Months
p lt 0.01, 25 mg vs 10 mg
39
MCID Values Consistent in RCTs in RA

• Improvements in HAQ DI and SF-36 in RA with
  newly approved therapies are statistically
  significant more importantly, CLINICALLY
  MEANINGFUL
• MCID values are consistent across agents and
  patient populations
• Disease specific relevant measure HAQ
• Generic measure SF-36
• Improvements in disease specific highly
  correlated with generic measures

40
EXAMPLEOsteoarthritis Disease Specific Measures
of Physical Function /or HRQOL
41
Western Ontario and McMaster Universities (WOMAC)
Osteoarthritis Index

• Self-administered questionnaire
• Developed querying patients with hip or knee OA
• Reflects physical activities most affected by
  symptoms, disease manifestations
• Composite score based on 24 questions subscores
• Pain (5 questions)
• Joint stiffness (2 questions)
• Physical function (17 questions)
• Scored by 0 - 4 Likert or 0 - 10 cm VAS scales
• Improvement negative change

42
WOMAC Scores in Osteoarthritis MCID

• MCID in WOMAC composite score, Likert scale
• 12 wk pivotal OA RCTs with Celecoxib 10.1 0
  89
• Pain, Stiffness, Physical Fxn 2.1, 1.2, 6.50
  20 0 8 0 61
• MCID in WOMAC VAS
• Anchoring to Patient Response to Rx 0-4 Likert
  scale
• 6 wk RCTs OA hip, knee Rofecoxib v Ibuprofen v
  PL
• Pain, Stiffness, Physical Fxn 9.7, 10, 9.3 mm,
  VAS
• 11 mm VAS for Patient Global Assessment

Zhao et al. Pharmacother 1999191269-78
Ehrich et al JRheum 200027 2635-2641
43
Improvement in WOMAC Composite ScoresWeek 12
from Baseline Pivotal OA Trials
Improved Scores
Zhao et al Pharmacother 1999191269-78
44
WOMAC Physical Function Subscale12 months
Pivotal RCT, OA knee or hip
Rofecoxib 12.5 mg Rofecoxib 25 mg Diclofenac 150
mg
0
-5
-10
-15
Mean Change (mm)
-20
-25
-30
Mean baseline 69.6 mm
-35
R
2
4
8
12
26
39
52
Week
R randomization P lt 0.05 for all groups
treatment response compared with baseline Cannon
GW, et al. Arthritis Rheum. 200043978987.
45
Mean Improvement in SF-36 All Rofecoxib v
Normative Data US Population
Difference between ages 45-54 and 55-64 US
population. Ware et al 1993
PF RP PAIN GHP VITAL SOC RE
MH
46
Change in SF-36 Scores at Week 12 from
Baseline Pivotal Trial in OA of knee
p lt .05 v placebo
47
SF-36 Scores at Week 12 in OA of kneev US
Normative Data in 55-65 year olds
48
MCID Values Consistent in RCTs in OA

• Improvements in WOMAC and SF-36 in OA with newly
  approved therapies are statistically
  significant more importantly, CLINICALLY
  MEANINGFUL
• MCID values are consistent across agents and
  patient populations
• Disease specific measure WOMAC
• Generic measure SF-36
• Improvements in disease specific highly
  correlated with generic measures

49
EXAMPLEFibromyalgia Disease Specific Measures
of Physical Function /or HRQOL
50
Fibromyalgia Pain, Sleep Disturbance and Fatigue
Correlated

• Consistent relationships between patient
  reported
• Pain Diary NRS Pain by MPQ VAS Sleep
  Quality Diary NRS Multidimensional
  Assessment of Fatigue MAF
• Numerical rating NRS recorded daily visual
  analog VAS scales reported weekly
• High baseline scores impaired sleep and much
  fatigue
• Low scores in SF-36 RP, BP and Vitality
• Poor sleep quantity and quality MOS-Sleep
• High scores in MAF Fatigue
• Higher levels of anxiety than depression HADS

Corbin et al Arth Rheum 2001 44S66-67, S212
51
EXAMPLECancer Pain Disease Specific Measures of
Physical Function /or HRQOL
52
Measures of Physical Function /or HRQOLCANCER

• Functional Assessment of Cancer Therapy FACT
• Linear Analog Self Assessment LASA
• Functional Living Index Cancer FLIC
• Quality of Life Index QLI
• European Organization for Research and Treatment
  of Cancer Questionnaire EORTC
• Cancer Rehabilitation Evaluation System CARES
• Treatment Outcomes in Pain Survey TOPS
• Missoula-Vitas Quality of Life Index MVQLI

53
Measures of Physical Fxn, HRQOL - CANCER

• Treatment Outcomes in Pain Survey TOPS
  Designed as extension of SF-36 HRQOL Tracks
  responses in individual patients over time Valid
  in multiple models of chronic pain /
  multidisciplinary treatment of pain
• Functional Living Index, Cancer FLIC Scales
  for specific patient populations, diseases
• Functional Assessment of Cancer Therapy FACT
  Scales for specific patient populations,
  diseases 5 subscales generally relevant
  Likert scales
• Linear Analog Self Assessment Scales LASA
  Scales for specific patient populations,
  diseases VAS scales readily comprehensible,
  convenient

54
Appropriate Domains Responder Analysis for
Chronic Pain

• PAIN multiple instruments NRS, VAS or
  Face Scales
• Disease specific relevant physical fxn /
  HRQOL Many instruments specific to disease
  state or TOPS when relevant, may add measures
  of sleep, depression, etc.
• Generic measure of HRQOL such as SF-36
  facilitate comparisons across treatments,
  populations, diseases
• Patient Global Assessment of Risk/Benefit Specifi
  c Question, or Health Utility Measure HUI, EQ5D
• AEs

55
Appropriate Domains Responder Analysis for Acute
Pain 24-48 hours duration

• PAIN
• Brief Pain Inventory BPI
• VAS or Faces Rating Scale
• Numerical Rating Scale NRS
• Time to Treatment Failure
• Rescue Medications
• Patient Global Assessment of Risk/Benefit
• AEs
• ? Necessity for HRQOL Measure in Acute Pain ?

56
Appropriate Domains Responder Analysis for Acute
Pain of 2 Weeks Duration

• PAIN Brief Pain Inventory BPI VAS or
  Faces Rating Scale Numerical Rating
  Scale NRS Time to Treatment Failure
  Rescue Medications
• Physical function / disease relevant
  measure According to disease population
  indication or TOPS
• Generic HRQOL SF-36 well validated
• Patient Global Assessment of Risk/Benefit Specif
  ic question HUI or EQ5D
• AEs

57
MCID Acute Pain Cancer Related Breakthrough Pain

• Titration phase multiple cross-over RCT of
  oral transmucosal fentanyl citrate
• 130 treatment not opioid naïve patients 1268
  episodes of pain
• Differences in pain scores between episodes which
  did, did not yield adequate pain relief
• MCID Pain Intensity Difference PID 0-10
  33Maximum Total Pain Relief TOTPAR 60 mins
  33
• Absolute PID, Pain Relief PR, Sum of PID over
  60 mins SPID 2 of 5 points in Likert scale

Farrar et al Pain 2000 88287-94
58
Conclusions Responder Analyses in Pain RCTs

• Domains consistent for chronic pain and 2 weeks
• Minimum number of required domains
• Assessed by a variety of validated instruments
• Add other domains as secondary endpoints
• PAIN include time to Rx failure use of rescue
  meds
• Chronic Pain TOPS or Disease specific
  relevant HRQOL measures available validated
• Addition of generic measure of HRQOL important

59
Conclusions Responder Analyses in Pain RCTs

• As with other Responder Analyses could require
• Improvement in a majority of domains but NOT ALL
• Without deterioration in the others
• Degree of improvement required could be based on
  MCID values
• Improvement across multiple domains not closely
  correlated
• Reflects a ROBUST clinical response
• May add statistical power decrease sample sizes

60
That these decisions be evidence based!!!
61
Universal Quality of Life Scale