Patient Reported Outcomes as Endpoints in Lung Cancer and Thoracic Malignancies
Richard J. Gralla, MD
New York Lung Cancer Alliance
For the ASCO and FDA Working Group
PATIENT REPORTED OUTCOMES (“PROs”) - Clinical Benefit and Quality of Life -
• Quality of Life– Multidimensional– Includes areas not likely to be affected by chemo
• Clinical Benefit
- Subjective or Palliative Control of Common Problems - Previously Defined to Include such considerations as:
- Pain Control- Weight Loss- Performance Status
QUALITY OF LIFE AND PRO EVALUATION- Is there a Need in Studies of Anticancer Treatments? -
• Highly Symptomatic Disease– Survival and Response data reveal only part of the results that
are important to patients, families, and health care professionals
• Treatments and Agents Vary in their Side- Effects and Risk Profiles
– Balancing patient experienced benefit and risk is needed
• Meaningful Survival Differences are Uncommon
SYMPTOMS OF LUNG CANCER- By Patient Reports (N = 121) -
Ref: Hollen et al. (1993). Eur J Cancer, 29A, S51-S58
84% 79%71% 62%59% 56%57% 60%48%25% 14%
54%
(n = 69) (n = 52)NON-SMALL CELL SMALL CELL
FATIGUE
COUGH
DYSPNEA
ANOREXIAPAINHEMOPTYSIS
NON-SMALL CELL LUNG CANCER- Number of Presenting Symptoms at Baseline -
Percentage
(N = 673 Stage III and IV Patients)
80%
12%
5%
Three or more
Two
One
None 3%
0 20 40 60 80 100
4.96
7.22
7.77
8.66
9.17
8.6
BSC 1995-2001 P alone PE Old Cis + no PE New Cis + Carbo +
+45,5%
+11%
+17,9%
+10,2% -11,4%
NON-SMALL CELL LUNG CANCER- Survival: Supportive Care and Chemotherapy 1991- 2001
(N = 10,995 / 9361) -
718 pts718 pts 783 pts783 pts 509 pts509 pts 1103 pts1103 pts 4648 pts4648 pts 1600 pts1600 pts
Refs: Proc ASCO 2002: Raftopoulos, Bria, Gralla, Eid
PATIENT REPORTED OUTCOMES (“PROs”)- Rationale and Need in Testing Anticancer Agents -
PROs can create an accurate picture of the disease course that is unavailable from the review of other endpoints
Health care professionals are not accurate in evaluating subjective or palliative benefits associated with anti-cancer treatments, when compared with patient self-reports
PROs are often reported by patients as improved with less than major responses to treatment - even with only stable disease…response rates underestimate patient reported benefit
The balance between symptom improvement and toxicity, or the effects of delayed progression summarized in many PRO measures, cannot be consistently predicted by other biomedical endpoints
QUALITY OF LIFE AND PRO’s
- Questions -1) Can we DEFINE quality of life?
2) Can we MEASURE quality of life?
3) Can we agree on how to ANALYZE quality of life results?
4) Can we PRESENT quality of life findings in a clear and useful way?
QUALITY OF LIFE INSTRUMENTS- Dimensions -
Physical
Functional
Psychological
Social
Spiritual
- Conceptual Model for Clinical Trials: THE “LCSS” -
PHYSICALDIMENSION*
Symptoms
Symptomatic Distress
distress from
FUNCTIONALDIMENSION*
ActivityStatus
QUALITY OF LIFEFOR THE LUNG
CANCER EXPERIENCE
Qualityof Life
Global
QUALITY OF LIFE AND PRO’S IN LUNG CANCER
Global
Global symptomatic
lung cancer
DimensionsCaptured: Dimensions
Captured:
OVERALL
•Cognitive•Physical
•Social(Role)
•Cognitive•Psychological
•Spiritual•All others
•Appetite•Fatigue•Cough•Dyspnea•Hemoptysis•Pain
•Social
* PRO Dimensions
QUALITY OF LIFE INSTRUMENTS
- Instrument Focus -
DISEASE-SPECIFIC:
SITE-SPECIFIC:
TREATMENT-SPECIFIC:
GENERAL HEALTH: All Populations
Cancer Diabetes Arthritis
LymphomaLungCancer
Clinical Trials Post - Op
Clinical TrialsBMT
QUALITY OF LIFE INSTRUMENTS- Lung Cancer Specific -
1. Lung Cancer Symptom Scale (LCSS) - Patient (9 items) & Observer (6 items) Forms
- Developed Specifically for Clinical Trials
2. EORTC - General and Lung Cancer Modules (30-40 items)
- Developed for General Use
3. FACT-L - General and Lung Cancer Modules (30-40 items)
- Developed for General Use
LUNG CANCER SPECIFIC INSTRUMENTS
- Psychometrics (1) -PSYCHOMETRICS CHARACTERISTICS
FEASIBILITY: Short administration time Low reading level required Easily understood Multi-center utility
CONTENT VALIDITY: Oncology expert agreement Patient agreement
RELIABILITY: Items internally consistent Intra / interrater agreement Patient reproducibility
QUALITY OF LIFE INSTRUMENTS
- Good reliability features include: -
• Internal consistency = Cronbach’s alpha > 0.70
for new measures
• Stability = Reliability coefficient > 0.70
• Equivalence = Kappa statistic > 0.61
Ref: Nunnally & Bernstein, 1994; Landis & Koch, 1977
QOL MEASURES FOR LUNG CANCER
- Example: Reliability Coefficients -FACT-L
Total core measure
(alpha, 0.89) for 116 patients
Lung cancer module (alpha 0.68) for 116 patients
LCSS
Total patient scale
(alpha 0.82) for 207 patients
Observer scale (alpha 0.75) for 21 observers
Cronbach’s alpha of 0.70 for new measures
LUNG CANCER SPECIFIC INSTRUMENTS
- Psychometrics (2) -PSYCHOMETRICS CHARACTERISTICS
Based on conceptual model Valid for LC patients with different extents of disease
Compares well to "gold standards"
673 LC patients from two North American cancer trials (30 centers)
CONSTRUCT VALIDITY:
CRITERION-RELATED(CONCURRENT) VALIDITY:
NORMATIVE DATA:
CLINICAL SIGNIFICANCE: KPS and LCSS Observer scales used as anchors
QUALITY OF LIFE AND PRO EVALUATION
- Additional Information -
• Clinically “meaningful” difference
– Often subject to “risk-benefit” considerations
– Not clearly defined for survival or response endpoints too
• Normative data for subgroups
Ref: Mayo Proceedings, 2002
NON-SMALL CELL LUNG CANCER
- Clinical Benefit and Quality of Life –
Assessment in Patients
In Phase II Trials
RANDOMIZED PHASE II TRIAL OF GEFITINIB AT TWO DOSE LEVELS – “IDEAL 2”
Quality of Life / Clinical Benefit: ASCO 2002 Abstract #1167
• A subscale of the FACT-L instrument was used (the LCS)
• Palliation was noted rapidly when it occurred: generally within 7 to 10 days
• Responding patients had greater symptom relief than those with stable disease or progressive NSCLC– 43% with symptom improvement
– 34% with quality of life improvement
QUALITY OF LIFE AND PRO EVALUATION - Difficulties with Analysis: Phase II Trials -
Analysis Problem – as with Surivial Analysis – relates to the lack of a Control Group for Judging Context
Appropriate Standard Palliation Confounds Analysis:
– Complicates benefit assessment when there is no control group
– Leads to overestimate of benefit with study agent when patients are receiving standard approaches as well
Response and Palliation:
– Major response underestimates benefit: Lesser responses may give symptom relief
– Benefit in patients with stable disease may be due to either the study agent or to standard palliation: can lead to overestimation
NON-SMALL CELL LUNG CANCER
- Clinical Benefit and Quality of Life –
Assessment in Patients
In Phase III Trials
PATIENT REPORTED OUTCOMES IN CLINICAL TRIALS
- Problems in Evaluation and Analysis -
1) Cumbersome instruments
2) Patient deterioration
3) Lack of investigator commitment
PROSPECTIVE CLINICAL TRIAL IN NSCLC
- Causes of Patient Attrition -
Causes for attrition
Death
Disease progression
Unknown
Patients entered
Remaining on studyafter 3 cycles
673
97
131
14
431
14%
19%
2%
64%
100%
QUALITY OF LIFE AND PRO EVALUATION- Baseline Values for Age and LCSS -
7972
76
60
(p = 0.0001)(p = 0.0002)
Percent of Patients
60 62
Age Average Symptom Burden
QL Item(p = NS)
Patients remaining on study (n=431); attrition group (n=242)
(N = 673 Patients with NSCLC)
0
10
20
30
40
50
60
70
80
90
On StudyAttrition Group
PATIENT REPORTED OUTCOMES IN CLINICAL TRIALS
- Prospective Emphasis on PRO: A Recent Study * -
1) A brief training session for all investigative and data management personnel on the methods and role of PRO evaluation
2) Inclusion of baseline QoL data as part of eligibility for randomization
3) Continued emphasis during the trial for vigilance in assessing PRO endpoints
4) As a result, more than 90% of the planned weekly assessments occurred over the initial 6 cycles of the trial
* Vogelzang et al, J Clin Oncol 2003; ** Gralla et al, Proc ASCO 2003.
Quality of LifeSurvival
Tumor Response &
Side Effects
Treatment
Malignancy
ENDPOINTS AND TREATMENTRelationships and Role of Patient Reported Outcomes (“PROs”)
NON-SMALL CELL LUNG CANCER- Quality of Life at Baseline: Influence on Survival -
- Prospective Analysis of 673 Patients at 30 Centers -
* p = 0.0001, using the LCSS quality of life instrument
0
20
40
60
80
100
0 2 4 6 8 10 12 14 16 18 20 22 24
MONTHS
PE
RC
EN
T S
UR
VIV
ING
*
LOWER QL HIGHER QL
QUALITY OF LIFE EVALUATION IN CLINICAL TRIALS
- Difficulties with Results Analysis: Phase III Trials -
Standards for statistical approaches remain controversial:
– Simply evaluating averages of scores at subsequent time points is problematic:• In Single Arm evaluation: Overestimates QoL and Clinical Benefit
• In Comparison trials: Underestimates QoL differences between study arms IF survival differences also are found
Survival differences complicate QoL analysis– Patient attrition (due to death or progression) is not random
• The most symptomatic patients drop out of the analysis first
• Patients with the poorer prognostic factors drop out first
• Thus, a regimen with poorer survival loses more lower QoL patients earlier and paradoxically - but incorrectly - appears to gain in QoL
Results from ALL patients on trial need to be Analyzed
Response and PRO Outcomes in a Random Assignment Trial: Added Value from Patient Determined Data
- Using Pain Scores within Major Response as an Example* -
-2
0
2
4
6
8
10
12
14
16
18
20
CR/PR SD PD/other
Pem+Cis
Cis
N=92 N=77N=94
N=41N=87N=37
Change from
baseline (mm)
Improvement
Worsening
Note: y-axis error bars
represent SE of the means
* Greater benefit reported by patients in 8 of 8 PRO parameters (p <0.05), validated LCSS-meso
(Model-based means.)
Survival and PRO Outcomes in a Random Assignment Trial: Added Value from Patient Determined Outcome Data
Week 12 Week 18
Pem+cis Cis Pem+cis Cis
% Surviving *
Quality of Life **
Symptom Distress **
92%92%
82%86%
AUC
AUC
47% 43% 45% 38%
53% 50% 51% 44%
p = 0.167
p = 0.162
p = 0.012
p = 0.009
p = 0.797 p = 0.247
*Vogelzang et al, J Clin Oncol 2003; ** Gralla et al, Proc ASCO 2003.