Transcript Lessons from the Ocular Hypertension Treatment Study (OHTS)


Prediction Model Template from
OHTS-EGPS Pooled Analyses

Today’s version is November 14
November 2006
A Prediction Model for Managing
Ocular Hypertensive Patients
Presenter Name
The Ocular Hypertension Treatment Study Group (OHTS)
National Eye Institute, National Center for Minority Healtlh and Health
Disparities, NIH grants EY 09307, EY09341, EY015498, Unrestricted Grant
from Research to Prevent Blindness, Merck Research Laboratories
and Pfizer, Inc.
The European Glaucoma Prevention Study (EGPS)
European Commission BMH4-CT-96-1598 and Merck Research Laboratories
November 2006
Ocular hypertension

Ocular hypertension occurs in 4%-8% of
people in the United States over age 40
(3-6 million people)

The number of affected people will
increase with the aging of the population

Associated with large costs for patient
examinations, tests and treatment
November 2006
Ocular hypertension

Elevated IOP is a leading risk factor for
development of POAG

Only modifiable risk factor for POAG

Patients can lose a substantial proportion
of their nerve fiber layer before POAG is
detected by standard clinical tests
Quigley HA, et al. Arch Ophthal 1981;99:635
November 2006
Why do we need a prediction model?

2002 OHTS publication showed that early treatment
reduces the incidence of POAG by more than 50%

However, only 1% of ocular hypertensive individuals
develop POAG per year

Clear that treating all ocular hypertensive patients is
neither medically nor economically justified
November 2006
Why do we need a prediction model?
Common in the past to base management decisions on
a single predictive factor – usually IOP
What level of IOP do you treat?
– IOP 24 mmHg?
– IOP 26 mmHg?
– IOP 28 mmHg?
– IOP 30 mmHg?
This approach ingores other important predictive factors
November 2006
Why do we need a prediction model?

A prediction model stratifies ocular
hypertensive individuals by level of risk
– To guide the frequency of visits and tests
– To ascertain the benefit of early treatment
November 2006

In 2002, the Ocular Hypertension
Treatment Study (OHTS) published a
prediction model for POAG based on...
– Data from 1,636 ocular hypertensive
participants randomized to either
observation or topical hypotensive
medication
– Median follow-up 6.6 years
.
Gordon et al, Arch Ophthalmol. 2002; 120: 714-720
November 2006
Factors predictive
for the development of POAG
in 2002 OHTS model

5 baseline factors increased the risk of developing
POAG
–
–
–
–
–

Older age
Higher Intraocular pressure
Thinner central cornea
Larger vertical cup/disc ratio by contour
Higher pattern standard deviation
Diabetes decreased the risk of POAG
.
November 2006
2002 OHTS model
needed to be confirmed
in a large, independent sample

2002 prediction model based on data from treated
and untreated ocular hypertensive individuals
– A prediction model should be based solely on untreated
individuals

OHTS sample included 25% African American
participants
– Is the prediction model valid in other groups?

OHTS was 1st study to report central cornea
thickness as a powerful predictor of POAG
– Can this finding be confirmed?
November 2006

A large indepent sample available
through the European Glaucoma
Prevention Study (EGPS)
– EGPS is a randomized clinical trial of 1,077
ocular hypertensive individuals randomized
to either placebo or dorzolamide
– Median follow-up 4.8 years
November 2006
Purpose of collaboration with EGPS

To test the 2002 OHTS prediction
model for the development of
glaucoma in a large, independent
sample

Before undertaking a collaboration
with EGPS, the two study protocols
were compared
November 2006
Comparison of OHTS and EGPS:
Study design
*Similarities between OHTS and EGPS
OHTS
EGPS
Study Design
Unmasked
randomized clinical trial
Double masked
randomized clinical trial
Large Sample
1,636 participants
22 clinics in United States
1,077 participants
18 clinics in 4 countries
Randomization
Groups
Observation
Any commercially available
medication
Placebo
Dorzolamide
POAG
Endpoint
Masked endpoint
ascertainment
Masked endpoint
ascertainment
November 2006

Collaborative analysis uses data only
from participants not receiving
medication:
– OHTS Observation Group n=819
– EGPS Placebo Group n=500
November 2006
OHTS vs EGPS:
Eligibility criteria
*Similarities between OHTS and EGPS
OHTS
EGPS
Age (years)
40-80 inclusive
> 30
Ocular
eligibility
criteria
Both eyes needed to
meet all criteria
Both eyes required to meet all
criteria except only one eye
needed to meet IOP criterion
21% of EGPS participants had one eye ineligible because of IOP below
entry criterion.
Collaborative analysis was repeated including and excluding participants
enrolled with one eye eligible
November 2006
OHTS vs EGPS:
Eligibility criteria
*Similarities between OHTS and EGPS
OHTS
EGPS
Normal
optic discs
Clinical exam
Review of stereophotos
by masked readers
Similar
Normal and
reliable
visual fields
Humphrey 30-2 Visual
Fields
Humphrey 30-2 Visual Fields
Octopus 32-2 Visual Fields
Masked readers
Masked readers
20% of EGPS participants were tested using Octopus 32-2 visual fields.
Octopus loss variance and mean defect were converted to Humphrey
pattern standard deviation and mean deviation (Anderson et. al., 1999).
November 2006
OHTS vs EGPS:
Exclusion criteria
*Similarities between OHTS and EGPS
Ocular
exclusions
OHTS
EGPS
Excluded pigmentary
dispersion syndrome and
pseudoexfoliation
Included pigmentary
dispersion syndrome and
pseudoexfoliation
Collaborative analysis excluded EGPS participants (19 placebo
participants) with pigmentary dispersion syndrome or pseudoexfoliation.
November 2006
OHTS vs EGPS:
Corneal thickness measurement
*Similarities between OHTS and EGPS
OHTS
Central
DGH 500 Ultrasound
corneal
mean of 5 measurements
thickness
measurements
EGPS
Identical
November 2006
OHTS vs EGPS:
POAG endpoint criteria
*Similarities between OHTS and EGPS
OHTS
EGPS
Definition of
abnormality
3 consecutive VFs with
PSD < 0.05 or GHT < 0.01
Or
2 consecutive
stereophotographs showing
deterioration
3 consecutive VFs with
visual field defects
Or
1 stereophotograph showing
deterioration
Confirmation
of abnormality
Masked readers
Masked readers
Attribution of
abnormality to
POAG
Masked
Endpoint Committee
Masked
Endpoint Committee
November 2006
Collaborative analysis is feasible

OHTS and EGPS protocols are similar
enough to test the validity of the prediction
model after resolution of study differences

Different enough in measures, geographic
distribution and patient characteristics to
test the generalizability of the OHTS
prediction model
November 2006
Results
OHTS vs EGPS control groups:
Baseline characteristics
(Univariate analyses)
OHTS
Observation
Group
n=819
EGPS
Placebo
Group
n=500
58%
52%
55.7 + 9.7
57.7+10.2
Race
African origin
Caucasian/other
25.2%
74.8%
0%
100%
Median follow-up
6.6 yrs
4.8 yrs
Baseline Factors
Female
Mean Age (Years)
November 2006
Results
OHTS vs EGPS control groups:
Definition of baseline IOP (mmHg)
Original
definition of
baseline IOP
(mm Hg)
New
definition of
baseline IOP
(mm Hg)
OHTS Observation Group
EGPS Placebo Group
2-3 IOPs at Randomization Visit
2-3 IOPs at 1 Eligibility Visit
24.9 + 2.7 SD
23.5 + 1.7 SD
4-6 IOPs at 2 Qualifying Visits
plus
2-3 IOPs at Randomization Visit
2-3 IOPs at 1 Eligibility Visit
plus
1 IOP at 6 month visit
Mean of 2 eyes
25.1 + 2.0 SD
Mean of 1 or 2 eyes
22.4 + 2.0 SD
New definition of baseline IOP used data from 2-3 visits and improved estimate
of baseline IOP.
November 2006
OHTS vs EGPS control groups:
Baseline characteristics
OHTS
Observation
EGPS
Placebo
Mean + S.D.
Average of 2 eyes
Mean + S.D.
Average of 2 eyes or
value of one eye
New baseline IOP mmHg
25.1 + 2.0
22.4 + 2.0
Vertical C/D ratio by contour
0.39 + 0.19
0.32 + 0.14
CCT (µm)
574.3 + 37.8
571.6 + 35.9
PSD (dB)
1.90 + 0.21
2.02 + 0.55
Baseline Factors
November 2006
OHTS vs EGPS control groups:
1st eye to develop POAG endpoint
Outcome
Total POAG
(Incidence per year)
OHTS
Observation Group
EGPS
Placebo Group
N=819
N=500
104 POAG of 819
1.9% per year
61 POAG of 500
2.5% per year
Detection Method
Visual field only
33
32%
37
60.7%
Disc only
56
54%
24
39.3%
Visual field & disc
at same visit
15
14%
0
0.0%
November 2006
Why was the incidence of POAG
higher in EGPS than in OHTS?

Differences in entry criteria

Differences in POAG endpoint criteria

Differences in risk characteristics of
participants
November 2006
Steps in testing the
validity of the OHTS prediction model
1.
Perform separate analyses of OHTS
Observation Group and EGPS
Placebo Group
(Multivariate Cox proportional hazards models)
2.
Compare results of the two analyses
November 2006
Results of independent multivariate analyses
OHTS vs EGPS:

Separate predictive models in OHTS and
in EGPS identified the same 5 predictors
for POAG
Age
IOP
CCT
PSD
Vertical cup/disc ratio by contour

The predictive factors in the OHTS model
and the EGPS model have similar hazard
ratios
All comparisons of hazard ratios by t-test, p values > 0.05
D’Agostino et al., JAMA;2001: 180-187
November 2006
Multivariate Hazard Ratios for
OHTS Observation group and EGPS Placebo group
HR 95% CI
Age Decade
1.37 (1.00, 1.88)
EGPS
1.16 (0.94, 1.43)
OHTS
IOP (mm Hg)
1.11 (0.98,1.27)
EGPS
1.21 (1.11, 1.31)
OHTS
CCT (40 µm decrease)
EGPS
2.07 (1.49, 2.87)
OHTS
Vertical CD ratio
by contour
PSD (per 0.2 dB increase)
2.00 (1.59, 2.50)
1.27 (1.04,1.54)
EGPS
1.26 (1.12, 1.41)
OHTS
1.05 (0.95, 1.16)
EGPS
1.16 (0.95,1.41)
OHTS
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
November 2006
OHTS prediction model for POAG is
confirmed in EGPS

Prediction model is validated...
– In an independent European study population
– In ocular hypertensive individuals
not on treatment

Thinner central corneal measurement is
confirmed as a predictive factor for POAG
November 2006
Next step was to pool OHTS and EGPS
data in the same prediction model

To increase the sample size to 1,319
participants (165 POAG endpoints)

To tighten 95% confidence intervals
for estimates of hazard ratios for
POAG
November 2006
Multivariate Hazard Ratios
OHTS Observation Group, the EGPS Placebo Group
Pooled OHTS and EGPS dataset
Age Decade
EGPS
OHTS
Pooled
IOP (mm Hg)
EGPS
OHTS
Pooled
CCT (40 µm decrease)
EGPS
OHTS
Pooled
Vertical CD Ratio (per 0.1 increase)
EGPS
OHTS
Pooled
PSD (per 0.2 dB increase)
EGPS
OHTS
Pooled
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
November 2006
Factors not in the prediction model:
Heart disease

In univariate analyses, history of heart
disease was a significant predictive
factor in OHTS but not in EGPS

In multivariate analyses, heart
disease was not a significant
predictive factor in OHTS, EGPS or
the pooled sample
November 2006
Factors not in the prediction model:
Diabetes

History of diabetes reduced the risk of developing
POAG in the 2002 OHTS prediction model

The effect of diabetes was difficult to estimate in
current OHTS models – data based solely on selfreport

Diabetes was not significant in univariate or
multivariate EGPS prediction models

Because of poor statistical estimation, diabetes
was not included in the final prediction models
November 2006
Which model performs best?

A model averaging data from both eyes?

A model using data from the worst eye?

A model using data from both eyes
including asymmetry between the eyes?
These models all perform similarly and
correlation coefficients ranging from
0.94 – 0.98.
November 2006
The OHTS and EGPS pooled data were
reanalyzed using tree analyses to look for
predictive factors that might be missed in
Cox model

Results from tree analyses
– Identified the same 5 predictive factors
for POAG (Age, IOP, CCT, Vertical C/D, PSD)
– Confirmed that heart disease, diabetes,
hypertension, myopia and self-identified
race had no detectable effect on risk of
developing POAG
November 2006
How accurate is the OHTS-EGPS
prediction model for POAG?

The accuracy of prediction models in discriminating
between patients who do and do not develop a
disease is measured using the C statistic

C statistic ranges from 0.50 (random agreement) to
1.00 (perfect agreement)
November 2006
Accuracy of prediction models for POAG
compared to Framingham Heart Study*
Prediction Models
C-statistic
*Framingham Heart Study prediction
model applied to different studies
0.63 - 0.83
OHTS observation group
0.76
EGPS placebo group
0.73
Pooled OHTS EGPS sample
0.74
D’Agostino et al. JAMA, 2001.
November 2006
Comparision of observed vs. predicted 5 year incidence
of POAG for the OHTS-EGPS pooled sample
0.36
Observed
Predicted
0.32
Probability
0.28
0.24
0.20
0.16
0.12
0.08
0.04
0.00
1
2
3
4
5
6
7
8
9
Decile of Predicted Risk (112 participants per decile)
10
November 2006
Using the prediction model
Available on web free of charge
 https://ohts.wustl.edu/risk

November 2006
Home Page
Benefits of risk stratification
to clinicians and patients

Decide on frequency of visits and tests

Ascertain the benefit of early treatment

Potentially reduce medical costs
November 2006
Cost Utility Analysis

Kymes et. al.*, reported that it was
cost effective to treat ocular
hypertensive individuals with > 2%
per year risk of developing POAG
*Kymes et al., AJO, 2006;141: 997-1008.
November 2006
Benefits of risk stratification

Approximately 30%-40% of the participants
in the pooled sample have <1% per year
risk of developing POAG

Many of these individuals could be seen
and tested once a year

Most of these individuals do not require
treatment

Potential cost savings
November 2006
LIMITATIONS AND CAUTIONS

There is no guarantee that the predicted risk is accurate for
a specific patient.

The predictions are more likely to be accurate for patients
who are similar to the patients studied in the OHTS and the
EGPS, and if your testing protocols for your patients
resemble those used in the studies.

The model predicts the development of early POAG. It is
not clear whether the model also predicts progression of
established disease or the development of visual disability.

The model is based on baseline parameters. Changes
November 2006
during follow-up will alter the risk of developing POAG.
Limitations and Cautions:
Application of prediction models to individual patients
must include information outside the model


THE PREDICTIONS ARE DESIGNED TO AID
BUT NOT TO REPLACE CLINICAL JUDGMENT.
Need to consider factors such as health status, life
expectancy and patient preferences
– An 18 year old ocular hypertensive with a low
5-year risk of developing POAG might be a
candidate for treatment
– A seriously ill 63 year old ocular hypertensive
with a high 5-year risk of developing POAG
might not be a candidate for treatment
November 2006
Summary

5 baseline factors accurately stratify
ocular hypertensive individuals by
their risk for developing POAG:
– Age
– IOP
– Central corneal thickness
– PSD
– Vertical cup/disc ratio by contour
November 2006
Summary

OHTS prediction model for POAG has
demonstrated high external validity
– OHTS model validated in EGPS sample and
Diagnostic Innovations in Glaucoma Study
sample (Medeiros FA, et al., Archives of Ophthalmology,
2005.)
– Model accurately predicts development of
POAG in ocular hypertensive individuals not on
treatment.
– Predictive model is accurate in self-identified
whites and African Americans
November 2006
Next Steps

Clarify the effects of diabetes, cardiovascular disease, ethnic
origin, myopia and family history of glaucoma on the risk of
developing POAG

Test the generalizability of the predictive model in other
populations

Add new diagnostic technology
– Quantitative assessments of disc and nerve fiber layer parameters
– Psychophysical tests

Identify new predictive factors
– Diet
– Environmental exposures
– Genetic factors
Predictive models will evolve with new information
November 2006
Collaborative Group
Ocular Hypertension
Treatment Study






Mae Gordon
Michael Kass
Phil Miller
Julie Beiser
Feng Gao
Ralph D’Agostino
European Glaucoma
Prevention Study





Valter Torri
Stefano Miglior
Irene Floriani
Davide Poli
Ingrid Adamsons
– Consulting Statistician,
Boston University
November 2006
OHTS Clinical Centers

















Bascom Palmer Eye Institute
Eye Consultants of Atlanta
Eye Physicians and Surgeons
Cullen Eye Institute
Devers Eye Institute
Emory Eye Institute
Henry Ford Hospitals
Johns Hopkins University
Krieger Eye Institute
Howard University
University of Maryland
University of California, Los
Angeles
Charles Drew University
Kellogg Eye Center
Kresge Eye Institute
Great Lakes Eye Institute
University of Louisville
















Mayo Clinic
New York Eye & Ear Infirmary
Ohio State University
Ophthalmic Surgeons & Consultants
Pennsylvania College of Optometry
MCP/Hahnemann University
Scheie Eye Institute
Keystone Eye Associates
University of California, Davis
University of California, San Diego
University of California, San
Francisco
University Suburban Health Center
University of Ophthalmic
Consultants
Washington Eye Physicians &
Surgeons
Eye Associates of Washington, DC
Washington University, St. Louis
EGPS Clinical Centers
Belgium



University of Antwerpen
University of Buxelles
University of Gent
Germany





University of Leuven
University of Mainz
University of Freiburg
University of Heidelberg
University of Wuerzburg
Italy







University of Milan, S. Paolo
Hospital
University of Milan, L. Sacco
Hospital
University of Verona
University of Parma
Oftalmico Hospital, Rome
S. Giovanni Hospital, Rome
Fatebenefratelli Hospital, Rome
Portugal



Coimbra, AIBILI
Viseu, S. Teotonio Hospital
Lisbon, S. Jose’ Hospital
November 2006