PMA P030016 Specular Microscopy Sub-study

1

PMA P030016 PMA P030016 Specular Microscopy Sub-studySpecular Microscopy Sub-study

Gerry Gray, Ph.D.Gerry Gray, Ph.D.Cardiovascular and Ophthalmic Team Leader, Cardiovascular and Ophthalmic Team Leader,

Division of BiostatisticsDivision of BiostatisticsOffice of Surveillance and BiometricsOffice of Surveillance and Biometrics

Center for Devices and Radiological HealthCenter for Devices and Radiological Health

October 3, 2003October 3, 2003

2

Specular microscopy sub-study Specular microscopy sub-study designdesign

• Endothelial cell countsEndothelial cell counts– Specular microscope photographsSpecular microscope photographs– Multiple images per eyeMultiple images per eye– Images read at a core center (Emory)Images read at a core center (Emory)

• Follow upFollow up– Original design 3 months, 1 and 2 years. Original design 3 months, 1 and 2 years. – Study modified to add 3, 4 year visits.Study modified to add 3, 4 year visits.

• Purpose: investigate effects on endothelial cells Purpose: investigate effects on endothelial cells through timethrough time

• 306 eyes enrolled & have at least one count306 eyes enrolled & have at least one count

3

Eye accountabilityEye accountabilityNo baseline, 1 subsequent visitNo baseline, 1 subsequent visit 2424No baseline, 2+ subsequentNo baseline, 2+ subsequent 7070Baseline, 1 subsequentBaseline, 1 subsequent 66Baseline, 2 subsequentBaseline, 2 subsequent 3434 last visit at 1 yearlast visit at 1 year 13 13

2 years2 years 1010

3 years3 years 1111

Baseline, 3+ subsequentBaseline, 3+ subsequent 172172 last visit at 2 yearslast visit at 2 years 2424

3 years3 years 9191

4 years4 years 5757

TOTALTOTAL 306306

4

Other combinations of visitsOther combinations of visits

3 and 4 year visits3 and 4 year visits 5757Baseline and 4 year visitsBaseline and 4 year visits 5757Baseline, 3 and 4 year visitsBaseline, 3 and 4 year visits 5151All visits up to 3 yearsAll visits up to 3 years 6767All visits up to 4 yearsAll visits up to 4 years 3737

5

ECD results for all eyes and visitsECD results for all eyes and visits

0 1 2 3 4

1000

1500

2000

2500

3000

3500

years

cell

dens

ity /m

m2

Endothelial cell density over time

(time is jittered for clarity)

Blue line segments connect the means

6

Mean cell counts for various cohortsMean cell counts for various cohorts

COHORTCOHORT nn BLBL 3 mo3 mo 1 yr1 yr 2 yr2 yr 3 yr3 yr 4 yr4 yr

All All patientspatients 306306 26572657 25712571 25442544 24762476 24342434 23552355

BL,2+BL,2+ 206206 26542654 25942594 25652565 24982498 24062406 23702370

3,4 yr.3,4 yr. 5757 26082608 25052505 24782478 24282428 23552355 23562356

All All visitsvisits 3737 26142614 25152515 24862486 24082408 23172317 23452345

7

Estimates of cell loss over duration Estimates of cell loss over duration of the studyof the study

• Estimates are fairly stable regardless of Estimates are fairly stable regardless of method of calculationmethod of calculation

• Range of estimates at 3 years: 8.5% to 8.9% Range of estimates at 3 years: 8.5% to 8.9% – Approximately 225 – 235 cells/mm^2Approximately 225 – 235 cells/mm^2– Includes both initial operational loss and normal Includes both initial operational loss and normal

loss due to ageingloss due to ageing• Range of estimates at 4 years: 8.4% to 9.7% Range of estimates at 4 years: 8.4% to 9.7%

– Approximately 220 – 260 cells/mm^2Approximately 220 – 260 cells/mm^2

8

““Steady state” long term lossSteady state” long term loss• Estimate depends largely on:Estimate depends largely on:

– Cohort that is usedCohort that is used– Whether we use all the data or only the 3 & 4 year dataWhether we use all the data or only the 3 & 4 year data

• Sponsor’s analysis: Sponsor’s analysis: – Percent change between 3 & 4 yearsPercent change between 3 & 4 years– Using only 3 & 4 year observations from cohort with both 3 & 4 year Using only 3 & 4 year observations from cohort with both 3 & 4 year

visitsvisits• Recall from the previous Table that this 57-patient cohort has a relatively Recall from the previous Table that this 57-patient cohort has a relatively

low 3-year count.low 3-year count.– Estimated percent change = 0.07% (i.e. a slight gain)Estimated percent change = 0.07% (i.e. a slight gain)

• 95% CI [-1.44%, 1.58%]95% CI [-1.44%, 1.58%]• Other cohorts (e.g. BL, 2+ visits) have relatively higher 3-Other cohorts (e.g. BL, 2+ visits) have relatively higher 3-

year count.year count.– Various analyses using these cohorts produce a loss of about 2% per Various analyses using these cohorts produce a loss of about 2% per

year.year.

9

Test for change in slopeTest for change in slope• In all cohorts, the loss does appear to “level off” after 3 In all cohorts, the loss does appear to “level off” after 3

years.years.• But there’s no strong statistical evidence that the “leveling But there’s no strong statistical evidence that the “leveling

off” is real (versus random chance and/or a small 4-year off” is real (versus random chance and/or a small 4-year sample).sample).

• Piecewise linear model:Piecewise linear model:– Initial (operative) loss from baseline to 3 monthsInitial (operative) loss from baseline to 3 months– Linear loss from 3 months to 3 yearsLinear loss from 3 months to 3 years– Linear loss (possibly different slope) from 3 years to 4 yearsLinear loss (possibly different slope) from 3 years to 4 years– Test for different 3-4 year slope: p = 0.37Test for different 3-4 year slope: p = 0.37

• Implication: “steady state” loss should be estimated using Implication: “steady state” loss should be estimated using all data after 3 months.all data after 3 months.

10

Two different fitsTwo different fits

0 1 2 3 4

1000

1500

2000

2500

3000

3500

years

cell

dens

ity /m

m2

Endothelial cell density over time

time is jittered for clarity

meansregression model3-4 year percent change

11

Extrapolation from different fitsExtrapolation from different fits

0 5 10 15 20 25 30

1000

1500

2000

2500

3000

3500

years

cell

dens

ity /m

m2

Endothelial cell density over time

time is jittered for clarity

regression model3-4 year percent change

12

Extrapolation caveatsExtrapolation caveats

• ALWAYSALWAYS a questionable exercise to extrapolate a questionable exercise to extrapolate beyond the range of available data, beyond the range of available data, especiallyespecially to to the degree we want here.the degree we want here.

• Highly dependent on the model we use & the Highly dependent on the model we use & the assumptions we make.assumptions we make.– BOTHBOTH of the previous extrapolations assume that loss of the previous extrapolations assume that loss

will continue linearly for 30 years.will continue linearly for 30 years.• Probably much more important to think about:Probably much more important to think about:

– If it’s necessary to obtain good long-term data.If it’s necessary to obtain good long-term data.– If so, how to go about it.If so, how to go about it.

13

Phase IV study possibilitiesPhase IV study possibilities• Continuation of phase III studyContinuation of phase III study

– Higher quality dataHigher quality data– Fewer patientsFewer patients– More costlyMore costly

• ““Registry” approachRegistry” approach– Simpler & cheaperSimpler & cheaper– More patientsMore patients– Less information (specular microscopy not generally Less information (specular microscopy not generally

available)available)• Choice depends on goalsChoice depends on goals

14

How do individual patients fare?How do individual patients fare?

• Perhaps more important than “average” cell Perhaps more important than “average” cell loss through time.loss through time.– What proportion of patients will have major What proportion of patients will have major

operative loss?operative loss?– What proportion of the patients will show cell What proportion of the patients will show cell

loss greater than some critical amount?loss greater than some critical amount?

15

Regression on individual eyesRegression on individual eyesEstimated postoperative ECD change

(baseline to 3 months)

cells per mm^2

Freq

uenc

y

-1000 -800 -600 -400 -200 0 200

010

2030

Estimated annual ECD change(after three months)

cells per mm^2 per year

Freq

uenc

y

-300 -200 -100 0 100 200 300

020

4060

Piecewise linear fits to 206 individual eyes (BL, 2+ cohort).

16

Regression on individual eyesRegression on individual eyes• Mean baseline = 2654 cells/mm^2Mean baseline = 2654 cells/mm^2• Mean initial loss = 204 cells/mm^2/yr Mean initial loss = 204 cells/mm^2/yr (absolute loss 51 cells/mm^2, or 1.9%)(absolute loss 51 cells/mm^2, or 1.9%)• Mean rate after 3 months = 53 cells/mm^2/yr, or 2%/yrMean rate after 3 months = 53 cells/mm^2/yr, or 2%/yr• Tolerance interval for long term rate:Tolerance interval for long term rate:

– 95% confident that 95% confident that • 60% of patients have a loss no worse than 82 cells/mm^2^yr (~3.1%)60% of patients have a loss no worse than 82 cells/mm^2^yr (~3.1%)• 90% of patients have a loss no worse than 163 cells/mm^2/yr (~6.1%)90% of patients have a loss no worse than 163 cells/mm^2/yr (~6.1%)• 99% of patients have a loss no worse than 247 cells/mm^2/yr (~9.3%)99% of patients have a loss no worse than 247 cells/mm^2/yr (~9.3%)

17

Predictors of endothelial cell lossPredictors of endothelial cell loss• There appear to be several statistically significant There appear to be several statistically significant

predictors of endothelial cell loss (baseline measurements):predictors of endothelial cell loss (baseline measurements):• This includes anterior chamber depth, MRSE, diopter, This includes anterior chamber depth, MRSE, diopter,

ECD coefficient of variationECD coefficient of variation– Dependent variable (annual loss) taken from per-eye regression.Dependent variable (annual loss) taken from per-eye regression.– Using continuous measurements for covariates.Using continuous measurements for covariates.– Sponsor presented analyses using “binned” data (cut at 15.0D or at Sponsor presented analyses using “binned” data (cut at 15.0D or at

7D & 10D) that do not show MRSE effect.7D & 10D) that do not show MRSE effect.• Age, gender, IOP, cylinder, axis, etc. did not appear to be Age, gender, IOP, cylinder, axis, etc. did not appear to be

significant predictors of cell loss.significant predictors of cell loss.

18

Clinically significant ACD effect?Clinically significant ACD effect?

Annual cell loss estimated using per-eye regressions.

3.0 3.5 4.0 4.5

-300

-200

-100

010

020

030

0

ACD

EC

D c

hang

e (c

ells

/mm

^2/y

r)

Estimated slope = 46.5

Estimated annual cell loss versus ACD

19

Sponsor’s ACD summarySponsor’s ACD summaryParameter Estimate P-Value

Intercept -0.9495 0.1017

Visit interval

1 Year to 2 Years -0.2328 0.71252 Years to 3 Years -0.9465 0.18863 Months to 1 Year 1.2483 0.0650Pre-op to 3 Months 0

ACD group

2.80 to 3.00 mm -2.6544 0.0015>3.00 to 3.50 mm -0.9729 0.0263>3.50 to 4.00 mm -1.4923 0.0026>4.00 mm 0

Estimates are percent loss relative to index group.

20

Questions for panelQuestions for panel

21