Efficacy Evaluation in Acne Clinical Trials

An Outline

  1. Re-visit choice of the primary endpoint for analysis: change score, relative change and lesion counts (statistical viewpoint).
  2. Statistical analysis for efficacy evaluation:
    1. Analysis units: original vs. transformed data
    2. Analysis methods
  1. Gain in power using subjectís repeated assessments in lieu of the final assessment.
  2. Do efficacy findings (whether based on change, counts or IGE) vary by baseline severity?

 

 

I. Re-visit choice of primary endpoint for analysis: change score, relative change and lesion counts (statistical viewpoint):

Efficacy assessment for acne lesions (Inflammatory, Non-inflammatory or Total lesions) can be carried out by analysis of:

Pros and Cons of analysis based on change score:

  1. Easy to interpret and analyze
  2. Attempt to remove the influence of baseline counts on the final counts but may fail since change is negatively correlated with final counts.
  3. Converting baseline and final lesions counts to relative change score may result in highly skewed (asymmetrical) distribution, which violate most parametric statistical tests.

(Figures 1-3: plots of infl., non-infl and total lesion counts, their change and relative changes

by study visits, Drug X, Study 1)

(Figures 4-6: plots of infl., non-infl and total lesion counts, their change and relative changes

by study visits, Drug Y, Study 2)

  1. Statistical analysis for efficacy evaluation:
  1. Analysis units: original vs. transformed data: (pros and cons)
  2. Analysis methods for endpoints (change or final counts):
  1. Simple comparisons of primary endpoints:
  2. ANOVA model with treatment, center and their interaction:
  3. ANCOVA with baseline lesion counts as covariates to account for possible baseline imbalance or regression toward the mean.

 

  1. Gain in power using subjectís repeated assessments in lieu of final assessment.

As successive assessments of a subject lesion counts over the course of the trial are expected to be correlated, one expects that use of these repeated assessments would increase the power for detecting treatment difference. We compare the efficacy based on repeated measurements (GLM, MIXED, MANOVA) approach against that based on the final assessment.

(Table 1: Comparison of efficacy results of various statistical approaches, Drug X, Study 1)

(Table 2: Comparison of efficacy results of various statistical approaches, Drug Y, Study 2)

IV. Do efficacy findings (whether based on change, counts or IGE) vary by baseline severity?

To find out whether efficacy results varies by baseline lesion counts:

(Figures 7-9: Plots of Infl., non-infl, & total lesion counts, their changes and relative changes by baseline category, Drug X, Study 1)

(Figures 10-12: Plots of Infl. non-infl, & total lesion counts, their change and relative changes by baseline category, Drug Y, Study 2)

(Table 3-4: Comparison of efficacy results based on lesion counts & IGE

by baseline category, Drug X, Study 1)

(Table 5-6: Comparison of efficacy results based on lesion counts & IGE

by baseline category, Drug Y, Study 2)

Table 1: Comparison of efficacy results of various statistical methods (Drug X, Study 1)

Infl. lesions

Non-Infl. lesions

Total lesions

Data

Ranks

Data

Ranks

Data

Ranks

Counts

MANOVA

GLM (R)

Week 12

Week 8

Week 4

(18.4, 20.6)*

0.306

0.316

0.130

0.308

0.834

0.069

0.219

0.044

0.222

0.917

(37.4, 46.7) *

<0.001

0.010

0.002

0.003

0.270

<0.001

0.006

<0.001

0.002

0.418

(55.8, 67.2) *

<0.001

0.0129

0.002

0.005

0.325

<0.001

0.003

<0.001

<0.001

0.293

Change

MANOVA

GLM (R)

Week 12

Week 8

Week 4

(9.2, 6.4)*

0.176

0.081

0.031

0.116

0.429

0.158

0.134

0.040

0.132

0.789

(26.0, 15.5)*

<0.001

<0.001

<0.001

<0.001

0.020

<0.001

<0.001

<0.001

0.001

0.021

(35.1, 21.9)*

<0.001

<0.001

<0.001

<0.001

0.027

<0.001

<0.001

<0.001

<0.001

0.030

% Change

MANOVA

GLM (R)

Week 12

Week 8

Week 4

(32.1, 21.4)*

0.067

0.079

0.012

0.201

0.498

0.046

0.0614

0.008

0.108

0.712

(41.1, 24.8)*

<0.001

<0.001

<0.001

<0.001

0.050

<0.001

<0.001

<0.001

0.001

0.037

(38.8, 24.8)*

<0.001

<0.001

<0.001

<0.001

0.023

<0.001

<0.001

<0.001

<0.001

0.034

(active, vehicle)*

 

Table 2: Comparison of efficacy results of various statistical methods (Drug Y, Study 2)

 

Infl. lesions

Non-Infl. lesions

Total lesions

 

Data

Ranks

Data

Ranks

Data

Ranks

Counts

GLM (R)

Cycle 6

Cycle 5

Cycle 4

ANCOVA(R)

ANCOVA

Simple test

(13.7, 16.2)*

0.020

0.020

0.031

0.160

0.015

0.014

0.068

0.031

0.018

0.024

0.375

0.016

0.032

0.035

(43.6, 49.2)*

0.415

0.294

0.636

0.803

0.072

0.044

0.428

0.120

0.071

0.231

0.099

0.030

0.017

0.176

(57.3, 65.3)*

0.211

0.161

0.327

0.561

0.021

0.015

0.288

 

0.023

0.022

0.072

0.098

0.001

< 0.001

0.077

Change

GLM (R)

Cycle 6

Cycle 5

Cycle 4

ANCOVA(R)

ANCOVA

Simple test

(8.1, 5.7)*

0.037

0.060

0.016

0.117

0.015

0.014

0.093

 

0.017

0.036

0.007

0.124

0.009

0.023

0.050

(6.6, -2.4)*

0.043

0.027

0.204

0.240

0.072

0.044

0.055

 

0.008

0.001

0.024

0.014

0.003

< 0.001

0.004

(14.7, 3.3)*

0.013

0.010

0.063

0.118

0.021

0.015

0.028

 

0.001

< 0.001

0.002

0.001

< 0.001

< 0.001

< 0.001

% Change

GLM (R)

Cycle 6

Cycle 5

Cycle 4

ANCOVA(R)

ANCOVA

Simple test

(31.0, 22.0)*

0.076

0.097

0.078

0.249

0.074

0.096

0.102

0.009

0.017

0.010

0.188

0.008

0.016

0.020

(13.2, -5.2)*

0.035

0.021

0.078

0.039

0.061

0.037

0.040

0.005

0.005

0.020

0.012

0.005

0.004

0.013

(22.6, 8.5)*

0.003

0.002

0.010

0.021

0.005

0.004

0.010

< 0.001

< 0.001

0.004

0.004

< 0.001

< 0.001

< 0.001

(active, vehicle)*

 

 

Table 3: Comparison of efficacy results by baseline category (Drug X, Study 1)

Lesion Type

Data

Baseline Category, mean (s.d.)

1

2

3

4

Inflammatory

Counts

Active

Placebo

13.3 (9.8)

13.2 (10.0)

17.0 (10.3)

20.0 (10.5)

19.4 (17.8)

22.9 (15.2)

 

23.2 (18.4)

26.8 (21.0)

Change

Active

Placebo

6.2 (9.2)

4.3 (8.8)

7.6 (9.6)

6.0 (11.2)

8.3 (14.4)

6.9 (13.5)

13.6 (19.0)

8.9 (15.7)

%Change

Active

Placebo

30.9 (45.3)

24.4 (51.7)

28.4 (43.9)

18.4 (48.9)

32.2 (40.5)

17.7 (67.8)

35.9 (42.5)

26.8 (37.0)

Non-inflammatory

Counts

Active

Placebo

18.1 (11.7)

23.5 (12.1)

25.7 (15.9)

36.7 (18.4)

41.3 (23.7)

47.0 (23.5)

59.7 (38.1)

85.6 (43.7)

Change

Active

Placebo

14.7 (12.4)

10.9 (12.1)

19.1 (16.9)

7.8 (17.9)

24.1 (23.4)

17.5 (20.8)

42.5 (34.1)

26.6 (35.6)

%Change

Active

Placebo

44.0 (35.3)

31.2 (34.5)

41.3 (36.9)

17.0 (38.2)

35.8 (35.8)

27.2 (34.8)

42.6 (30.9)

21.9 (34.4)

Total

Counts

Active

Placebo

31.4 (17.7)

36.7 (16.3)

42.7 (21.6)

56.6 (21.7)

60.8 (33.3)

69.8 (27.8)

82.9 (47.1)

112.5 (50.6)

Change

Active

Placebo

20.9 (17.1)

15.3 (15.8)

26.7 (21.8)

13.9 (22.6)

32.3 (31.0)

24.4 (25.9)

56.1 (45.7)

35.4 (45.1)

%Change

Active

Placebo

40.2 (32.0)

29.4 (30.7)

38.3 (30.5)

19.1 (32.0)

35.3 (33.0)

26.2 (28.2)

40.8 (30.6)

23.5 (30.2)

 

 

Table 4: Comparison of efficacy results based on IGE by baseline category (Drug X, Study 1)

Treatment

Baseline Category

Overall

1

2

3

4

Active

18/51 (35%)

11/53 (21%)

4/46 (9%)

7/68 (10%)

40/218 (18%)

Placebo

13/49 (27%)

6/56 (11%)

6/61 (10%)

0/52 (0%)

25/218 (11%)

 

 

 

 

 

Table 5: Comparison of efficacy results by baseline category (Drug Y, Study 2)

Lesion Type

Data

Baseline Category, mean (s.d.)

1

2

3

4

Inflammatory

Counts

Active

Placebo

9.55 (8.6)

10.17 (5.9)

11.76 (8.9)

12.03 (8.0)

14.49 (9.4)

17.37 (9.7)

18.88(17.1)

24.36 (23.5)

Change

Active

Placebo

1.90 (8.9)

1.46 (5.9)

3.94 (8.9)

3.51 (8.3)

7.42 (9.2)

4.22 (9.6)

19.71 (17.7)

13.91 (23.1)

%Change

Active

Placebo

14.83 (76.9)

12.30 (50.0)

25.05 (56.9)

21.44 (54.9)

33.93 (42.2)

19.42 (44.1)

49.87 (36.7)

36.19 (52.4)

Non-inflammatory

Counts

Active

Placebo

10.31 (8.2)

13.17 (13.7)

17.66 (10.6)

22.47 (20.6)

38.86 (37.1)

44.02 (40.8)

110.7 (105.3)

112.3 (96.6)

Change

Active

Placebo

0.98 (8.2)

-1.76 (13.5)

4.57 (10.0)

-0.51 (21.1)

7.33 (37.0)

0.24 (38.7)

13.70 (86.8)

-7.10 (71.7)

%Change

Active

Placebo

5.44 (80.3)

-17.7 (126.7)

20.19 (46.4)

-6.7 (109.4)

14.80 (89.4)

3.11 (85.5)

11.79 (69.5)

0.31 (56.8)

Total

Counts

Active

Placebo

21.70 (12.4)

25.84 (20.3)

33.52 (23.4)

35.75 (18.9)

50.22 (37.8)

60.00 (31.4)

127.6 (117.1)

139.5 (105.5)

Change

Active

Placebo

5.35 (12.4)

0.76 (19.9)

9.37 (22.0)

7.48 (17.9)

19.48 (36.9)

11.20 (30.5)

25.05 (95.2)

-5.62 (84.2)

%Change

Active

Placebo

19.20 (45.1)

2.76 (67.8)

23.00 (47.9)

17.71 (40.3)

28.25 (50.4)

15.51 (42.4)

19.48 (57.9)

-0.84 (60.9)

 

Table 6: Comparison of efficacy results based on IGE by baseline category (Drug Y,

Study 2)

Treatment

Baseline Category

Overall

1

2

3

4

Active

28/43 (65%)

25/51 (49%)

21/46 (46%)

15/44 (34%)

89/184 (48%)

Placebo

28/49 (57%)

17/42 (40%)

12/46 (26%)

12/47 (26%)

69/184 (38%