Department of Health and Human Services

Food and Drug Administration

Center for Drug Evaluation and Research

Office of Biostatistics

 

 

Statistical Review and Evaluation

 clinical studies

 

NDA:

21-366

Name of drug:

Crestor (rosuvastatin calcium)

Applicant:

AstraZeneca Pharmaceuticals

Indication:

Treatment of hyperlipidemia

Documents reviewed:

\\CDSESUB1\N21366\N_000\2001-06-26\clinstat

Dates:

 

Statistical reviewer:

Joy Mele, M.S. (HFD-715)

Biometrics team leader:

Todd Sahlroot, Ph.D. (HFD-715)

Biometrics division director:

Ed Nevius, Ph.D. (HFD-715)

Project manager:

Bill Koch, R.Ph. (HFD-510)

Clinical reviewer:

Bill Lubas, M.D. (HFD-510)

 

Keywords:

NDA review, clinical studies, active control, dose response


 

 

Introduction................................................................................................ 3

Reviewer’s Methods.................................................................................... 5

Clinical trials in patients with IIa/IIb dyslipidemia.............................. 6

Multiple fixed-dose studies of Rosuvastatin....................................................................................................... 6

Statistical Methods for Studies 8 and 23................................................................................................................. 6

Study 8 (conducted 8/98 to 1/99)........................................................................................................................... 6

Study 23 (conducted 4/99 to 12/99)..................................................................................................................... 12

Statistical Methods for Study 33........................................................................................................................... 15

Study 33 (conducted  10/99 to 6/00).................................................................................................................... 15

Rosuvastatin 5 mg and 10 mg compared to pravastatin and simvastatin................................................. 22

Statistical Methods for Studies 27 and 28............................................................................................................. 22

Study 27 (conducted 6/99 to 4/00)....................................................................................................................... 22

Study 28 (conducted 4/99 to 10/00)..................................................................................................................... 28

Clinical trials in patients with IIb/IV dyslipidemia............................. 34

Study 35 (conducted 12/99 to 8/00)..................................................................................................................... 34

Study 29 (conducted 10/99 to 10/00)................................................................................................................... 39

Study 36 (conducted 9/99 to 10/00)..................................................................................................................... 44

Clinical trials in patients with other dyslipidemias....................... 49

Study 54 (conducted 4/00 to 2/01)....................................................................................................................... 49

Study 30 (conducted 7/99 to 6/00)....................................................................................................................... 53

Study 31 (conducted 10//99 to 9/00).................................................................................................................... 57

Assessment of treatment effect........................................................ 60

40 mg versus 80 mg............................................................................................................................................. 60

Dose response for rosuvastatin............................................................................................................................. 61

5 and 10 mg rosuvastatin doses............................................................................................................................ 63

Findings in Subgroup Populations........................................................ 64

Dyslipidemia Type................................................................................................................................................ 64

Gender................................................................................................................................................................. 66

Age, Race and Baseline LDL................................................................................................................................ 66

Rosuvastatin versus other statins.................................................... 67

Summary and Conclusions...................................................................... 70


Introduction

 

            The sponsor has submitted the results of 14 controlled clinical trials to establish the efficacy and safety of rosuvastatin for the treatment of dyslipidemia. These trials are briefly described in Tables 1, 2 and 3 below. Studies 24, 25 and 26 are reviewed in a separate document by FDA statistical reviewer Cynthia Liu. The results for those studies are included in summaries in this review. The remaining 11 trials are reviewed here. 

            No studies had the same overall design, therefore each study provides additional insight regarding the efficacy of rosuvastatin.

            Eight trials were conducted in Type IIa and Type IIb patients (Table 1).  Doses for rosuvastatin ranged from 1 mg to 80 mg in these trials; Studies 8, 23 and 33 were specifically designed to examine the dose response relationship of rosuvastatin.

            Atorvastatin was an active control in five of the eight trials while pravastatin and simvastatin were active controls in two trials.  In Study 33, doses from 5 to 80 mg for rosuvastatin and doses from 10 to 80 mg for atorvastatin were examined; this trial, by design, was the best comparative trial submitted since it contained a full range of doses for both active drugs.

 

 

 

Table 1. Clinical trials in patients with IIa/IIb dyslipidemia and LDL primary endpoint

Study #

Design

Rosuvastatin doses (mg)

Pla?

Active Control (dose)

Duration of treatment

Multiple fixed doses of Rosuvastatin compared to multiple fixed doses of Atorvastatin

8

 

23

 

33

Fixed-dose

 

Fixed-dose

 

Fixed-dose

1, 2.5, 5, 10, 20, 40

 

40, 80

 

5, 10, 20, 40, 80

Yes

 

Yes

 

No

Ator (10, 80 OL)

 

 

 

Ator (10, 20,40, 80)

6 weeks

 

6 weeks

 

6 weeks

Rosuvastatin compared to Atorvastatin

24

 

25

 

 

26

Fixed-dose

 

Forced titration

 

 

Titration to NCEP goal

5, 10

 

5÷20÷80

10÷40÷80

 

5 ÷max 80

10 ÷max 80

Yes

 

No

 

 

No

Ator (10)

 

Ator 10÷40÷80

 

 

Ator 10÷max 80

12 weeks

 

12 wks at 1st dose

6 wks at each higher dose

12 wks at 1st dose

52 weeks total

Rosuvastatin compared to Pravastatin and Simvastatin

27

 

28

Fixed-dose

 

Titration to NCEP goal

5, 10

 

5 ÷max 80

10 ÷max 80

No

 

No

Prav (20), Sim (20)

 

Prav 20÷max 40

Sim 20÷max 80

12 weeks

 

12 wks at 1st dose

52 weeks total

 

    


     Patients with Type IIb or Type IV dyslipidemia were enrolled in Studies 35, 29 and 36 (Table 2).  Study 35 was a fixed dose placebo-controlled study while the other two studies examined rosuvastatin in combination with niaspan or fenofibrate

 

Table 2. Clinical trials in patients with IIb/IV dyslipidemia

Study #

Design

Rosuvastatin doses (mg)

Pla?

Active Control (dose)

Duration of treatment

35

Fixed-dose

 (TG EP)

 5, 10, 20, 40, 80

Yes

None

6 weeks

29

Forced titration

 (LDL EP)

10÷20÷40 OL

No

Niaspan alone and combined

12 wks at 1st dose

6 wks at each higher dose

36

(Type II

Diabetes)

Forced titration

 (TG EP)

10÷20÷40

Yes

Fenofibrate alone and combined

6 wks at 1st dose

6 wks at each higher dose

 

            Patients with homozygous or heterozygous familial dyslipidemia or severe hypercholesterolemia were enrolled in Studies 54, 30 and 31, respectively (Table 3).  In these high risk populations, rosuvastatin doses of 20, 40 and 80 mg were used. In Study 31, the combination of rosuvastatin plus cholestyramine (Questran) was examined.

 

Table 3. Clinical trials in patients with other dyslipidemias

Study #

Design

Rosuvastatin doses (mg)

Active Control (dose)

Duration of treatment

54

(Homozygous Familial)

Forced titration/ crossover

 (LDL EP)

20÷40÷80

Ator 80

6 wks at each rosuvastatin dose

followed by 2-period crossover (6 wks/period)

30

(Heterozygous Familial)

Forced titration

 (LDL EP)

20÷40÷80

Ator 20÷40÷80

6 wks at each dose

31

(Severe HC)

Combination

 (LDL EP)

40÷80

Questran 16 mg

6 wks at rosuvastatin 40; randomized to rosuvastatin 80 or rosuvastatin 80 +Questran

for 6 weeks

 


Reviewer’s Methods

 

     The sponsor provided datasets for each of the 11 studies reviewed here.  All statistical results, tables and figures in this review were created by this reviewer unless otherwise noted.

The protocol-defined primary analysis population in all studies except Study 8 was the ITT population where all patients with baseline and at least one post-baseline measurement were included. For patients with missing data at the primary endpoint, the last observation for that patient was used (LOCF analysis).

Baseline was computed as the average of Week –2, -1 and 0 unless otherwise noted.

                        For most studies (exceptions are noted in the review), an analysis of variance with treatment and region as fixed effects was used to analyze the response variable. Tests for interactions of treatment with subgroup and region were performed and the results are noted when significant. [Note that  the protocols specified that center would be included in the ANOVA model; however, from the sponsor’s output, it was clear that country or region was included as a term instead. This is acceptable since many centers had a small number of patients overall or were missing patients in 1 or more treatment groups. It seems logical to this reviewer to group small centers based on country or US region.] In some models, baseline was included as a covariate. This reviewer generally only performed analyses of the primary efficacy variable. Few important differences between the results of the sponsor and those of the reviewer were found, though,  there are many differences in interpretation and presentation of the results.

Missing data/dropouts was not an issue in these trials since generally over 90% of the patients completed treatment. Therefore, no analyses to assess bias due to missing data were performed by this reviewer.

A statistical methods section is included with those studies where additional description of the methods is needed beyond what is given here.

                         


Clinical trials in patients with IIa/IIb dyslipidemia

Multiple fixed-dose studies of Rosuvastatin

 

Statistical Methods for Studies 8 and 23

 

For both Studies 8 and 23, the protocol stipulated as the primary analysis population patients with baseline and Week 6 data (i.e. patients completing the randomized treatment phase). The intent-to-treat analysis of all randomized patients using LOCF for missing data was proposed as a secondary analysis. Analysis of covariance of % change from baseline of LDL with baseline as a covariate and treatment group as a term in the model was performed. William’s test was planned to identify the minimum effective dose and Dunnett’s test was proposed to compare each rosuvastatin dose to placebo. Only the results of William’s test are included in the NDA study report.  A regression analysis was done to assess the dose response relationship.

The sponsor suggested that dose response could best be assessed using only patients that complete 6 weeks of therapy. Since the bulk of the response to treatment occurs during the first 2 weeks of therapy, this reviewer thinks that an ITT LOCF analysis will not unduly bias against the drug or cloud interpretation of the dose response relationship. Also, this reviewer felt that a Week 6 LOCF analysis would produce estimates consistent with estimates from the other studies in this submission. For these reasons, only ITT analyses are presented here. Note that there were very few dropouts in these studies so there are no important differences between the sponsor’s results and this reviewer’s results.

Both studies were multicenter studies conducted outside the USA.  The sponsor did not present results by center or country or perform analyses with center or country as terms in the model. This reviewer did these analyses for the primary efficacy variable, LDL and found that inclusion of center or country as a fixed effect  had no effect on assessment of efficacy  and there were no positive interactions with treatment.

 

Study 8 (conducted 8/98 to 1/99)

 

Design

 

Study 8 is a double-blind, multicenter, randomized Phase II/III trial designed to compare multiple doses of rosuvastatin  to placebo.  Doses of 1, 2.5, 5, 10, 20 and 40 mg of rosuvastatin were studied. In addition, atorvastatin (10 and 80 mg) was studied with the objective of only estimating the treatment effect to provide information for planning future trials. Note that atorvastatin was administered open-label to patients; the sponsor states that investigators remained blinded to atorvastatin assignment.  So there were a total of 9 treatment arms. After a 6-week dietary run-in period, patients were randomized to treatment and followed for 6 weeks.

The primary outcome variable was percent change from baseline in LDL at Week 6.  Secondary endpoints named in the protocol were the following:

·        % change at 6 weeks for TC, HDL,  TG, ApoB, ApoA-1, Apo-A-2, fibrinogen and Lp(a)

 

Upon advice from the medical reviewer, the following secondary endpoints are reviewed here:

·        TC, HDL, ApoB, TG  (non-HDL was not reported)

 

Inclusion criteria included the following:

1.     160 mg/dL#LDL<220 mg/dL at Visits 2 (Week –2) and 3  (Week –1)

2.     TG<300 mg/dL at Visits 2 (Week –1) and 3  (Week –1)

3.     males 18-70 years and post-menopausal females 50-70 years

 

Fasting lipids were measured at Weeks –6, -2, -1, 0, 1, 2, 4, 6, 8 and 10. (Weeks 8 and 10 were follow-up after withdrawal of therapy)

Patients were to be withdrawn from the trial if CK>10xULN with pain or ALT or AST>3xULN.

 

Patient Disposition

 

     A total of 142 patients completed screening and were randomized to treatment  in 4 countries (Table 4).

Table 4. Study  8 Patient Disposition by Country and Treatment

Country

(# centers)

Placebo

ROSU

1

ROSU 2.5

ROSU

5

ROSU

10

ROSU

20

ROSU

40

ATOR

10

ATOR

80

Norway

(3)

 

6

 

7

 

7

 

7

 

7

 

7

 

8

 

7

 

7

Netherlands

(3)

 

2

 

2

 

2

 

3

 

2

 

2

 

2

 

2

 

2

Finland

(3)

 

3

 

3

 

3

 

3

 

4

 

4

 

4

 

3

 

2

Sweden

(5)

 

2

 

2

 

3

 

4

 

4

 

4

 

4

 

3

 

2

 

Only 6 patients did not complete the 6-week treatment period; one patient did not continue into the follow-up period (Table 5). Three patients (2 not treated and 1 rosu 5mg patient) had no post-baseline data.  The difference, then, between a completer analysis population (all patients with 6-week data; the sponsor’s analysis population) and an LOCF analysis population (this reviewer’s analysis population) is the data of 3 patients (1 rosu 20, 1 ator 10 and 1 ator 80).

 

Table 5. Study  8 Patient Disposition by Week on Study

 

Placebo

ROSU

1

ROSU 2.5

ROSU

5

ROSU

10

ROSU

20

ROSU

40

ATOR

10

ATOR

80

Randomized

14

15

15

18

17

17

18

15

13

(Not treated)

(1)

(1)

 

 

 

 

 

 

 

  Wk 2

  Wk 6

13

13

14

14

15

15

17

17

17

17

17

16

18

18

15

14

12

12

Completers

13 (93%)

14

(93%)

15

(100%)

17

(94%)

17

(100%)

16

(94%)

18

(100%)

14

(93%)

12

(92%)

ITT

13 (93%)

14

(93%)

15

(100%)

17

(94%)

17

(100%)

17

(100%)

18

(100%)

15

(100%)

13

(100%)

 

            The reasons for trial discontinuation for the 6 dropouts were ADE (rosu 20 and ator 10), patient request and protocol violation.

           

Baseline Demographics

 

            Three patients were Asian and the remainder were Caucasian. About one-third of the patients were female (Table 6). The average age was 55 years (range of 24 to 70). About 15% of the patients were 65 years or older. Treatment groups were not well-balanced regarding gender or patients 65 or older[1].

 

            Table 6. Study  8  Patient Demographics for ITT Patients

 

Placebo

 

(n=13)

ROSU

1

(n=14)

ROSU 2.5

(n=15)

ROSU

5

(n=17)

ROSU

10

(n=17)

ROSU

20

(n=17)

ROSU

40

(n=18)

ATOR

10

(n=15)

ATOR

80

(n=13)

Age

  Mean (SD)

  Range

  %$65years

 

56 (7)

45-64

0%

 

59 (7)

48-69

29%

 

54 (10)

31-67

13%

 

55 (7)

42-65

6%

 

56 (11)

24-69

18%

 

52 (11)

29-66

6%

 

55 (11)

38-70

28%

 

56 (10)

30-67

13%

 

56 (11)

36-69

23%

Gender

  % female

 

38%

 

36%

 

33%

 

24%

 

53%

 

24%

 

33%

 

40%

 

15%

Race

  % white

 

92%

 

93%

 

100%

 

100%

 

100%

 

100%

 

100%

 

100%

 

92%

 

Efficacy

 

            The Week 6 LOCF results for the primary endpoint LDL and four secondary endpoints are shown in Table 7 below for all nine treatment groups. All doses of rosuvastatin showed a significant decrease in LDL, total cholesterol (TC) and ApoB (p<.001). The rosuvastatin results for HDL and TG are variable and do not appear to be dose-related.

 

Table 7. Study  8  Lipoprotein results (mg/dL) at Week 6 LOCF

 

Placebo

 

(n=13)

ROSU

1

(n=14)

ROSU 2.5

(n=15)

ROSU

5

(n=17)

ROSU

10

(n=17)

ROSU

20

(n=17)

ROSU

40

(n=18)

ATOR

10

(n=15)

ATOR

80

(n=13)

LDL

Baseline

% change

 

 

197 (14)

-7% (7)

 

 

191 (18)

-35% (9)

 

 

190 (15)

-42% (9)

 

 

193 (16)

-45% (7)

 

 

190 (16)

-52% (9)

 

 

191 (22)

-56% (13)

 

 

184 (19)

-63% (9)

 

 

189 (9)

-44% (9)

 

 

192 (18)

-54% (15)

P-value

vs. pla

 

 

 

<.001

 

<.001

 

<.001

 

<.001

 

<.001

 

<.001

 

<.001

 

<.001

TC

Baseline

% change

 

 

272 (12)

-5% (6)

 

 

267 (21)

-24% (7)

 

 

264 (24)

-30% (8)

 

 

269 (20)

-33% (6)

 

 

267 (16)

-36% (7)

 

 

267 (21)

-41% (10)

 

 

257 (27)

-46% (8)

 

 

264 (16)

-32% (7)

 

 

265 (16)

-42% (13)

P-value

vs. pla

 

 

<.001

 

<.001

 

<.001

 

<.001

 

<.001

 

<.001

 

<.001

 

<.001

HDL

Baseline

% change

 

 

49 (12)

+3% (10)

 

 

55 (14)

+8.5% (10)

 

 

48 (10)

+8.8% (10)

 

 

52 (9)

+13% (13)

 

 

50 (15)

+14% (12)

 

 

50 (13)

+7.5% (9)

 

 

52 (13)

+9.4% (8)

 

 

50 (16)

+7% (10)

 

 

49 (9)

-3% (13)

P-value

vs. pla

 

 

.4

 

.5

 

.04

 

.04

 

.4

 

.3

 

>.3

 

>.3

TG

Baseline

% change

 

 

130 (41)

-3% (23)

 

 

116 (49)

-16% (18)

 

 

132 (45)

-14% (33)

 

 

123 (51)

-35% (16)

 

 

135 (52)

-12% (35)

 

 

134 (52)

-27% (18)

 

 

107 (48)

-25% (23)

 

 

126 (52)

-15% (17)

 

 

119 (41)

-25% (24)

P-value

vs. pla

 

 

.2

 

.7

 

.001

 

.8

 

.07

 

.009

 

>.3

 

>.1

Apo-B

Baseline

% change

 

 

140 (16)

-3% (10)

 

 

132 (14)

-27% (11)

 

 

135 (12)

-34%

 

 

139 (18)

-38% (9)

 

 

143 (18)

-42% (8)

 

 

136 (20)

-46% (11)

 

 

130 (15)

-55% (6)

 

 

138 (11)

-36% (11)

 

 

135 (12)

-46% (15)

P-value

vs. pla

 

 

<.001

 

<.001

 

<.001

 

<.001

 

<.001

 

<.001

 

<.001

 

<.001

 

Statistically significant changes in LDL for rosuvastatin compared to placebo are seen as early as Week 1 with large part of the response achieved by Week 2  and essentially complete by Week 4 (Figure 1). 

 

Figure 1.  Study 8 LDL-C (mg/dL)  and % change from baseline by week on study for all observed cases

 

 

 

 

 

 


            The boxplots in Figure 2 show the distribution of the LDL % change from baseline data at endpoint (Week 6 LOCF) for all treatment groups.  There is a clear dose response for rosuvastatin. The overlap of the 95% confidence intervals between adjacent doses of rosuvastatin would be expected especially considering the small number of patients in each treatment group (14-18).

            A regression analysis of the five rosuvastatin doses by the sponsor yielded a significant slope and the following regression equation (r2=.49):

 

            LDL response at Week 6= -35.1 – 7.5* Ln(dose)

 

According to this linear model, a doubling of the dose up to a maximum of 40 mg results in a further decrease of about 5% (Ln2*7.5). See Appendix 1 for a graph of the dose response and the fitted line.

 

 

Figure 2  Study 8  Boxplots[2] of LDL % change from baseline at Week 6 LOCF


            According to the protocol, the sponsor did not intend to compare the atorvastatin responses to the rosuvastatin responses. For a subsequent Phase 3 trial, Study 33, this comparison was made. This reviewer thinks that it is useful to examine the relationship between atorvastatin and rosuvastatin in this study for further comparison to Study 33. The boxplots suggest that the response for atorvastatin 10 is similar to rosuvastatin 5 and atorvastatin 80 to rosuvastatin 20. Also from Figure 1 we saw that the lines for rosuvastatin 20 mg and atorvastatin 80 mg are superimposed. Confidence intervals on the treatment differences were computed by this reviewer and are shown in Table 8 below. Since no standards for comparability were named in this trial and also atorvastatin was administered open-label, one cannot draw any definitive conclusions from this data. Nevertheless we can use the clinical standard of  6% as a clinically significant difference as a guide for interpreting the limits of the confidence intervals.  In doing so, we can not draw any conclusions regarding the comparability of Rosu 5 to Ator 10; the confidence interval for Rosu 20/Ator 80 suggests comparability with an upper limit of 5%.  The data does show that Rosu 40 is statistically better than Ator 80.

 

Table 8. 95% confidence intervals on the treatment difference between atorvastatin and rosuvastatin for LDL % change from baseline at the primary endpoint (Week 6 LOCF)

 

95% confidence interval

(neg. favors rosuvastatin)

ATOR10 versus

    ROSU 2.5

    ROSU 5.0

 

-5%, 9.5%

-8%,  6%

ATOR 80 versus

    ROSU 10

    ROSU 20

    ROSU 40

 

-6%, 9%

-10%, 5%

-17%, -2%

 

 

Reviewer’s Comments on Study 8

 

     In conclusion, Study 8 shows that rosuvastatin  significantly decreases LDL in a dose-related manner for doses from 1 mg to 40 mg. Doubling the dose results in an additional mean decrease of about 5%.  Similar dose-related responses are seen for total cholesterol and Apo-B but not for HDL and TG.  For HDL, only the 5 mg and 10 mg doses showed significant increases compared to placebo. For TG, only the 5 mg and 40 mg doses showed significant decreases compared to placebo.  Also,  for atorvastatin, neither dose showed significantly different changes in HDL or TG compared to placebo.  The lack of effects seen for HDL and TG in this study is most likely due to the fact that 93% of the patients had Type IIa dyslipidemia, a population characterized by high LDL; the mean HDL at baseline was about 50 mg/dL and the mean TG about 125 mg/dL.  Comparisons of rosuvastatin to atorvastatin suggest that at most half of the dose of rosuvastatin is needed to get a response similar to atorvastatin; however, due to sample size and design (open-label administration of limited doses of atorvastatin), these comparisons are inconclusive.


Study 23 (conducted 4/99 to 12/99)

 

Study 23 is a double-blind, multicenter, randomized trial designed to compare the 40 mg and 80 mg doses of rosuvastatin to placebo. After a 6-week dietary run-in period, patients were randomized in a 1:1:2 ratio to placebo, Rosu 40 mg and Rosu 80 mg and followed for 6 weeks.

The primary outcome variable was percent change from baseline in LDL at Week 6.  Secondary endpoints named in the protocol were the following:

·        % change at 6 weeks for TC, HDL,  TG, ApoB, ApoA-1, Apo-A-2, fibrinogen and Lp(a)

 

Upon advice from the medical reviewer, the following secondary endpoints are reviewed here:

·        TC, HDL, ApoB, TG  (non-HDL was not reported)

 

Inclusion criteria included the following:

1.     160 mg/dL#LDL<220 mg/dL at Visits 2 (Week –2) and 3  (Week –1)

2.     TG<300 mg/dL at Visits 2 (Week –1) and 3  (Week –1)

3.     males 18-70 years and post-menopausal females 50-70 years

 

Fasting lipids were measured at Weeks –6, -2, -1, 0, 1, 2, 4, 6, 8 and 10. (Weeks 8 and 10 were follow-up after withdrawal of therapy).

Patients should have been withdrawn from the trial if CK>10xULN with pain or ALT or AST>3xULN.

 

Design

           

A total of 65 patients (Table 9) were randomized to treatment at 9 centers (all 9 centers were used in Study 8 also).  Only one patient (rosu 40 mg) discontinued treatment; this patient is included in this reviewer’s analyses.

 

Table 9. Study  23 ITT Patient Disposition

 

Placebo

ROSU 40

ROSU 80

Randomized

17

16

31

Dropouts

0

1 at Week 5

0

ITT

17

16

31

 

Baseline Demographics

           

The average age of patients was about 58 years with about ¼ of the patients 65 or older (Table 10). About half of the patients were female and all patients were Caucasian.

 

Table 10. Study  23 ITT Patient Disposition

 

Placebo

(n=17)

ROSU 40

(n=16)

ROSU 80

(n=31)

Age

  Mean (SD)

  Range

  %$65years

 

57 (7)

43-69

24%

 

59 (9)

41-69

31%

 

57 (8)

39-69

16%

Gender

  % female

 

47%

 

50%

 

42%

Race

  % white

 

100%

 

100%

 

100%

 


Efficacy

 

     Both doses (40 and 80 mg) of rosuvastatin show significant decreases in LDL, TC and ApoB (Table 11). Results for HDL and TG are borderline significant. Overall the results show small differences between the 40 and 80 mg rosuvastatin doses suggesting no important advantages to increasing the dose from 40 to 80 mg.

 

Table 11. Study  23  Lipoprotein responses at  Week 6 LOCF

 

Placebo

(n=17)

ROSU 40

(n=16)

ROSU 80

(n=31)

LDL

Baseline

% change

p vs. PLA

 

190 (15)

-0.8% (11)

 

186 (16)

-61% (7)

<.001

 

188 (13)

-63% (8)

<.001

TC

Baseline

% change

p vs. PLA

 

269 (21)

-0.2% (7)

 

264 (24)

-44% (8)

<.001

 

263 (20)

-45% (6)

<.001

HDL

Baseline

% change

p vs. PLA

 

56 (9)

+2.6% (11)

 

53 (13)

+11% (13)

.10

 

51 (14)

+15% (15)

.04

TG

Baseline

% change

p vs. PLA

 

114 (47)

-0.1% (39)

 

127 (60)

-27% (35)

.05

 

119 (46)

-23% (25)

.06

Apo-B

Baseline

% change

p vs. PLA

 

139 (16)

-1.8% (11)

 

138 (21)

-52% (8)

<.001

 

139 (14)

-54% (6)

<.001

 

The lack of a difference between 40 and 80 is further illustrated by a graph of LDL overtime.

 

Figure 3. Study 23 LDL-C (mg/dL) by week on study for all observed cases


     The boxplots below (Figure 4) show the range of % change in LDL seen in both Study 8 and 23 at Week 6. The overlap of the confidence intervals (the gray area) for the 40 mg doses demonstrate  the consistency of response across the studies. The 40 mg placebo-subtracted effects are –56% and –60% for Studies 8 and 23, respectively. 

 

Figure 4.  LDL % change from baseline at Week 6 LOCF for Studies 8 and 23

 

 

Reviewer’s Comments on Study 23

 

              One of the objectives of Study 23 was to further characterize the dose response of rosuvastatin to 80mg. The intention was to combine the results of Study 8 and Study 23 if the 40 mg doses showed similar responses.  Though similarity of response was seen, there appears to be little utility to combining the studies since it seems clear that the 80 mg dose offers no clinically significant advantage over 40 mg (this was also seen in Study 33 which is reviewed in the following section).


 

Statistical Methods for Study 33

 

            The sponsor’s initial model was an analysis of covariance model defined as follows:

 

y = β01 Baseline2 Tx3 log(Dose)4 Tx*log(Dose)5 Center6 Tx* Center

 

The treatment by center interaction term (β6) was dropped if it was found to be non-significant . (In the sponsor’s analysis, region instead of centers was used in the model with region defined as shown in Table 12.) The treatment by log(dose)[3] interaction tested whether the slopes for atorvastatin and rosuvastatin were parallel; if they were found to be parallel (non-significant β4) then this interaction term was dropped from the model and the remaining terms re-estimated. The results of this model and an alternative model defined by this reviewer are discussed with the LDL results.

 

 

Study 33 (conducted  10/99 to 6/00)

 

Design

 

Study 33 was a double-blind, multicenter, randomized trial designed to compare multiple doses of rosuvastatin (5, 10, 20, 40 and 80 mg) to multiple doses of atorvastatin (10, 20, 40 and 80 mg). After a 6-week dietary run-in period, patients were randomized to treatment and followed for 6 weeks.

The primary outcome variable was percent change from baseline in LDL at Week 6.  In addition to LDL, the following lipoprotein data was collected:

·        TC, HDL,  TG, ApoB, ApoA-1 and non-HDL

 

Upon advice from the medical reviewer, the following secondary endpoints are reviewed here:

·        TC, HDL, ApoB, TG,  non-HDL

 

Inclusion criteria included the following (differences from Studies 8 and 23 are bolded):

4.     160 mg/dL#LDL<250 mg/dL at Visits 2 (Week –2) and 3  (Week –1)

5.     TG<400 mg/dL at Visits 2 (Week –1) and 3  (Week –1)

6.     males and females 18 years or older

 

Fasting lipids were measured at Weeks –6, -2, -1, 0, 1, 2, 4, and 6.

 

Patients may be withdrawn from the trial if CK>10xULN with pain or ALT or AST>3xULN.

 

Patient Disposition

 

             A total of  374 patients were randomized to treatment at 35 centers in the USA and 4 centers in Canada (Table 12).  For analysis purposes, centers were pooled into regions as shown in the table below.

           

            Table 12. Study  33 Patient Disposition by Region and Treatment

Region

(# centers)

ROSU

5

ROSU

10

ROSU

20

ROSU

40

ROSU

80

ATOR

10

ATOR

20

ATOR

40

ATOR

80

Northeast

(9)

 

9

 

12

 

8

 

11

 

10

 

10

 

9

 

11

 

9

Southeast

(7)

 

6

 

8

 

8

 

9

 

6

 

9

 

9

 

7

 

10

Central

(12)

 

15

 

16

 

14

 

16

 

15

 

18

 

16

 

13

 

15

West (7)

Canada (4)

 

8

 

9

 

7

 

9

 

11

 

6

 

7

 

8

 

10

 

            Over 95% of the randomized patients completed the study (Table 13); only 2 patients did not have data on study and are excluded from the ITT analysis.

 

Table 13. Study  33 Patient Disposition by Week on Study

 

ROSU

5

ROSU

10

ROSU

20

ROSU

40

ROSU

80

ATOR

10

ATOR

20

ATOR

40

ATOR

80

Randomized

38

45

39

45

42

43

39

42

41

(Not treated)

 

 

(1)

(1)

 

 

 

 

 

  Wk 2

  Wk 6

38

38

45

42

38

37

44

43

42

41

43

41

39

37

41

41

40

40

Completers

38

(100%)

42

(93%)

37

(95%)

43

(96%)

41

(98%)

41

(95%)

37

(95%)

39

(93%)

39

(95%)

ITT

38

(100%)

45

(100%)

38

(97%)

44

(98%)

42

(100%)

43

(100%)

39

(100%)

42

(100%)

41

(100%)

Note that some patients dropped during Week 6 and had Week 6 data but were not completers.

 

            The reasons for trial discontinuation show that ADE is the most common reason for trial discontinuation for both rosuvastatin and atorvastatin patients (Table 14).

 

            Table 14. Study  33 Reasons for treatment discontinuation

 

ROSU

5

ROSU

10

ROSU

20

ROSU

40

ROSU

80

ATOR

10

ATOR

20

ATOR

40

ATOR

80

Randomized

38

45

39

45

42

43

39

42

41

ADE

Pt request

Prot. Viol.

Other

Lost-to-FU

0

0

0

0

0

2

0

0

0

1

1

1

0

0

0

1

0

0

0

1

1

0

0

0

0

1

1

0

0

0

0

2

0

0

0

1

0

0

2

0

2

0

0

0

0

 

 

 

 

 

Baseline Demographics

 

            The treatment groups were fairly well-balanced regarding age, gender and race (Table 15 on following page). About 88% of the patients were Caucasian; another 8% were Black. About half of the patients were female. The average age was 57 years (range of 25 to 81). About 25% of the patients were 65 years or older.

 

            Table 15. Study  33  Patient Demographics for ITT Patients

 

ROSU

5

(n=38)

ROSU

10

(n=45)

ROSU

20

(n=38)

ROSU

40

(n=44)

ROSU

80

(n=42)

ATOR

10

(n=43)

ATOR

20

(n=39)

ATOR

40

(n=42)

ATOR

80

(n=41)

Age

  Mean (SD)

  Range

  %$65years

 

55 (12)

30-78

21%

 

58 (10)

36-79

29%

 

56 (12)

33-81

26%

 

57 (9)

38-80

18%

 

57 (9)

39-75

26%

 

59 (11)

38-80

33%

 

56 (12)

25-78

21%

 

57 (11)

34-74

26%

 

54 (12)

27-74

22%

Gender

  % female

 

53%

 

49%

 

61%

 

50%

 

45%

 

53%

 

49%

 

43%

 

32%

Race

  % white

 

82%

 

82%

 

89%

 

89%

 

88%

 

88%

 

90%

 

88%

 

95%

 

Efficacy

 

            The LDL (Figure 5) overtime shows a pattern of response akin to what was seen in Study 8; the bulk of response occurs by Week 2 with some further lowering observed primarily in the higher doses. 

Figure 5. Study 33 LDL (mg/dL) by week on study and treatment group

            The Week 6 LOCF results for the primary endpoint LDL are shown in Table 16 below for all nine treatment groups. (See Appendix 2 for a plot of the LDL % change from baseline data for the duration of the trial.)

 

 

Table 16. Study  33 LDL results (mg/dL) at Week 6 LOCF

 

ROSU

5

(n=38)

ROSU

10

(n=45)

ROSU

20

(n=38)

ROSU

40

(n=44)

ROSU

80

(n=42)

Baseline mean  (SD)

% change mean (SD)

193 (22)

-42% (10)

190 (18)

-48% (13)

188 (24)

-50% (19)

188 (20)

-58% (12)

198 (22)

-61% (14)

 

 

 

 

ATOR

10

(n=43)

ATOR

20

(n=39)

ATOR

40

(n=42)

ATOR

80

(n=41)

Baseline mean  (SD)

% change mean (SD)

 

190 (24)

-37% (13)

185 (19)

-46% (10)

188 (22)

-45.5% (14)

190 (18)

-55% (10)

p-value (Rosu vs. Ator)

Sponsor’s model

Alternative model

 

 

<.0001

.0001

 

<.0001

.13

 

<.0001

<.0001

 

<.0001

.03

 

 

 

This reviewer considered two models to analyze this data. The first model is the sponsor’s model:

 

y = β01 Baseline2 Tx3 log(Dose)4 Tx*log(Dose)5 Center6 Tx* Center

 

 The treatment by region and treatment by dose interactions were non-significant. The latter indicates that the difference between the slopes for the two treatment groups is not statistically significant indicating parallel slopes for the two treatment groups.  The sponsor then dropped both terms from the model.  In their final model, then, response was regressed on both baseline LDL[4] and dose. This is essentially akin to separate regressions for each treatment group (see Appendix 3 for a plot of the regression lines).  The fit of these lines is not good with an r2#.2. The sponsor’s model, then, does not well-characterize the dose response for each treatment group. Comparisons between like doses based on this model and the linearity of the dose response are not acceptable to this reviewer.

Instead of regressing on dose, dose may be treated as a classification variable. This second alternative model yields the results shown at the bottom of Table 16. Using the sponsor’s model,  rosuvastatin is significantly different from atorvastatin at each dose (recall that the differences will all be the same since the comparisons are based on regressing on dose) while this reviewer’s model shows no statistically significant difference between the 20 mg doses. Given the mean results by dose (particularly the similarity of the responses for the atorvastatin 20 and 40 mg doses) it is not surprising that the results of the two models would differ.

In addition to being interested in how the two treatment groups compare at like doses, we are also interested in seeing which rosuvastatin doses and atorvastatin doses are comparable. The results in Table 17  on the following page are from this reviewer’s model described above (rosuvastatin 80 mg beats all doses of atorvastatin and is not included in the table, p<.03).  The bolded numbers are for the doses where rosuvastatin is either better or comparable  to atorvastatin, using an upper limit of the CI of +6% as the acceptable margin. According to these LDL results, rosuvastatin is as good as twice the dose of atorvastatin

 

Table 17. Study  33 Comparison of Rosu versus Ator for LDL results (mg/dL) at Week 6 LOCF

 

ATOR

10

ATOR

20

ATOR

40

ATOR

80

ROSU 5 vs.  ATOR

    LS means difference

    p-value

    95% CI

 

-4.3%

.13

-10, 1

 

+4.7%

.13

-1, 10

 

+4.1%

.16

–1.7, 9.8

 

+14%

.0001

8, 20

ROSU 10 vs.  ATOR

   LS means difference

   p-value

   95% CI

 

-10.9%

.0001

-16, -5

 

-2.0%

.48

-7.6, 3.6

 

-2.4%

.38

-7.9, 3.1

 

+7.3%

.01

1.8, 13

ROSU 20 vs.  ATOR

   LS means difference

   p-value

   95% CI

 

-13%

<.0001

-19, -8

 

-4.4%

.13

-10.3, 1.4

 

-4.8%

.10

-10.6, 1

 

+4.9%

.09

-0.8, 11

ROSU 40 vs.  ATOR

   LS means difference

   p-value

   95% CI

 

-21%

<.0001

-26, -15

 

-12%

<.0001

-17, -6

 

-12%

<.0001

-18, -7

 

-2.4%

.39

-8, 3

Note that negative values favor rosuvastatin.

 

The overlap of boxes in Figure 6 illustrates the variation in response which was also reflected in the low r2 ‘s observed in the regression analyses.

 

Figure 6. Boxplots for LDL % change from baseline at endpoint by dose and treatment group

 

 

           


            The rosuvastatin results for TC, non-HDL and ApoB show a clear dose response (Table 18). The comparisons to atorvastatin look similar to those seen for LDL (note due to time constraints, analyses comparing rosuvastatin and atorvastatin were not done on secondary endpoints by this reviewer). As in Study 8, no dose response is seen for TG and HDL.  

 

 

Table 18. Study  33  Lipoprotein results (mg/dL) at Week 6 LOCF

 

ROSU

5

(n=38)

ROSU

10

(n=45)

ROSU

20

(n=38)

ROSU

40

(n=44)

ROSU

80

(n=42)

ATOR

10

(n=43)

ATOR

20

(n=39)

ATOR

40

(n=42)

ATOR

80

(n=41)

 

TC

Baseline

% change

 

 

281 (27)

-30% (8)

 

 

276 (25)

-35% (10)

 

 

270 (25)

-36% (14)

 

 

276 (27)

-41% (9)

 

 

286 (27)

-44% (11)

 

 

280 (29)

-27% (11)

 

 

271 (23)

-34% (7)

 

 

274 (24)

-35% (11)

 

 

278 (23)

-43% (8)

 

HDL

Baseline

% change

 

 

53 (14)

+7% (10)

 

 

51 (15)

+6% (11)

 

 

50 (10)

+9% (13)

 

 

53 (14)

+12% (11)

 

 

52 (10)

+10% (14)

 

 

54 (15)

+4% (11)

 

 

49 (11)

+7% (12)

 

 

49 (10)

+4% (9)

 

 

48 (11)

+2% (10)

 

non-HDL

Baseline

% change

 

 

228 (29)

-39% (9)

 

 

225 (23)

-44% (11)

 

 

221 (27)

-46% (17)

 

 

223 (24)

-53% (11)

 

 

233 (26)

-56% (14)

 

 

226 (9)

-34% (13)

 

 

222 (23)

-43% (8)

 

 

225 (25)

-43% (14)

 

 

229 (25)

-52% (9)

 

TG

Baseline

% change

 

 

180 (89)

-23% (15)

 

 

180 (62)

-22% (24)

 

 

164 (52)

-18% (29)

 

 

176 (67)

-26% (18)

 

 

177 (72)

-20% (44)

 

 

179 (71)

-17% (28)

 

 

188 (90)

-25% (26)

 

 

181 (66)

-27% (22)

 

 

193 (68)

-34% (29)

 

Apo-B

Baseline

% change

 

 

183 (24)

-35% (9)

 

 

182 (19)

-41% (10)

 

 

181 (21)

-43% (17)

 

 

178 (21)

-48% (10)

 

 

189 (19)

-51% (12)

 

 

184 (24)

-32% (11)

 

 

181 (20)

-38% (10)

 

 

183 (20)

-39% (13)

 

 

184 (22)

-48% (9)

 

 

 

Reviewer’s Comments on Study 33

 

            The results for Study 33 show a rosuvastatin dose-response relationship for LDL, TC, non-HDL and Apo-B. As in Study 8 no dose–response is seen for HDL or TG. Comparisons to atorvastatin show that comparable responses are seen when the dose of atorvastatin is double the dose of rosuvastatin.

           


Rosuvastatin 5 mg and 10 mg compared to pravastatin and simvastatin

 

Statistical Methods for Studies 27 and 28

 

The sponsor planned to perform pairwise t-tests to compare each dose of rosuvastatin to each comparator using the following steps:

 

 

1.     Test each dose of rosuvastatin to pravastatin and simvastatin for non-inferiority using a 6% margin. If the rosuvastatin is found to be non-inferior to simvastatin or pravastatin, then a test of superiority will be performed.

2.     Tests of superiority will be performed for the 10 mg dose of rosuvastatin versus each comparator. If the results of a test are significant at the .05 level, then a test of the 5 mg dose will be performed.

 

 

 

Study 27 (conducted 6/99 to 4/00)

 

Design

 

Study 27 is a double-blind, multicenter, randomized trial designed to compare two low doses of rosuvastatin (5 mg and 10 mg) to the 20 mg dose of pravastatin and simvastatin.  After a 6-week dietary run-in period, patients were randomized to one of 4 treatment arms and followed for 12 weeks.

The primary outcome variable was percent change from baseline in LDL at Week 12.  Secondary endpoints named in the protocol were the following:

·        % change at 12 weeks for TC, HDL, LDL/HDL, TC/HDL, non-HDL/HDL, TG, ApoB, ApoB/ApoA-1, ApoA-1 and Lp(a)

·        % change at 2, 6, and 10 weeks for LDL, TC, HDL, LDL/HDL, TC/HDL, non-HDL/HDL, and TG

·        % of patients within NCEP and EAS guidelines at 12 weeks

 

Upon advice from the medical reviewer, the following secondary endpoints are reviewed here:

·        TC, HDL, non-HDL, ApoB, TG

 

Inclusion criteria included the following:

1.     160 mg/dL#LDL<250 mg/dL at Visits 2 (Week –1) and 3  (Week –1)

2.     TG<400 mg/dL at Visits 2 (Week –1) and 3  (Week –1)

3.     Men and women$18 years

 

Fasting lipids were measured at Weeks –6, -2, -1, 0, 2, 6, 10 and 12.

 

Patients could be withdrawn from the trial if CK>10xULN with pain or ALT or AST>3xULN.

 

 

Patient Disposition

 

            A total of 502 patients were randomized to treatment at 63 centers in 7 countries (Table 19 on following page). About of the patients were from the United Kingdom.

 

            Table 19. Study 27 Distribution of centers and patients by country

Country

# centers

ROS 5

ROS 10

PRAV 20

SIM 20

France

10

15

13

21

19

Germany

7

6

6

9

9

Holland

4

5

6

6

6

Italy

9

12

15

18

17

Poland

7

22

20

20

20

Spain

7

14

15

14

16

United Kingdom

19

46

40

49

42

 

Table 20 shows the number of patients on study by study week; overall only 6% of the patients did not complete the 12 weeks of the study. Dropout rates were similar across treatment groups. Seven patients were not included in the ITT analysis due to missing data at baseline or on study.

 

Table 20. Study  27 Patient Disposition by Week on Study

 

ROSU 5

ROSU 10

PRAV 20

SIM 20

Randomized

120

115

137

130

  Wk 2

  Wk 6

  Wk 10

  Wk 12

120

117

114

114

114

109

108

106

137

133

132

131

129

126

122

122

Completers

114 (95%)

106 (92%)

131 (96%)

122 (94%)

ITT

119 (99%)

111 (97%)

136 (99%)

129 (99%)

 

           

 

            In the rosuvastatin groups and the pravastatin group, the major reason for dropout was adverse event (ADE) while in the simvastatin group the major reason was patient request (Table 21).

 

           

Table 21. Study 27   Reasons for discontinuation

 

ROSU 5

(n=120)

ROSU 10

(n=115)

PRAV 20

(n=137)

SIM 20

(n=130)

ADE

Pt request

Prot. Viol.

Other

Lost-to-FU

2 (1.7%)

0 (0%)

2 (1.7%)

2 (1.7%)

0 (0%)

6 (5.2%)

1 (0.9%)

1 (0.9%)

1 (0.9%)

0 (0%)

3 (2.2%)

2 (1.5%)

0 (0%)

0 (0%)

1 (0.7%)

1 (0.8%)

3 (2.3%)

2 (1.5%)

0 (0%)

1 (0.8%)

 


Baseline Demographics

 

            The treatment groups were well-balanced regarding baseline demographics (Table 22) and medical history including CHD risk factors. The average age of patients in this study was 58 years with about of the patients 65 years or older.  Except for 2 patients, all patients were Caucasian.

 

 

Table 22.  Study 27 Patient Demographics for All Randomized Patients

 

ROSU 5

(n=120)

ROSU 10

(n=115)

PRAV 20

(n=137)

SIM 20

(n=130)

Age

  Mean (SD)

  Range

  %$65years

 

57 (12)

28-79

30%

 

60 (10)

28-84

32%

 

59 (11)

20-78

37%

 

59 (11)

22-81

32%

Gender

  % female

 

49%

 

57%

 

54%

 

50%

Race

  % white

 

100%

 

100%

 

99%

 

100%

 

 

Efficacy Results

 

Primary Endpoint LDL-C

 

The primary endpoint in this trial is LDL-C measured at Week 12 with the last observation carried forward for missing data. Results of an ANOVA of  LDL-C for the ITT population (Table 23) clearly show that each dose of rosuvastatin is superior to pravastatin and simvastatin at the doses tested.  The p-values and 95% confidence intervals clearly show that the results meet the standards for both non-inferiority and superiority set in the protocol.

 

 Table 23.  Study 27 LDL-C Week 12 LOCF ITT Results (mg/dL)

 

ROSU 5

(n=120)

Mean (SD)

ROSU 10

(n=115)

Mean (SD)

PRAV 20

(n=137)

Mean (SD)

SIM 20

(n=130)

Mean (SD)

Baseline

% Change from Baseline

190 (20)

-42% (12)

186 (19)

-49% (15)

189 (19)

-28% (12)

188 (22)

-37% (13)

Results vs. Prav 20

   p-value

   95% Confidence Interval

 

.0001

-17.5%, -11.2%

 

.0001

-24.3%, -17.8%

 

 

Results vs. Sim 20

   p-value

   95% Confidence Interval

 

.03

-8.2%, -1.8%

 

.0001

-15.0%, -8.5%

 

 

        For the 95% confidence intervals, negative values favor rosuvastatin.

 

 

In addition to testing for treatment effect, the protocol stated that a test for interaction would be done; these results are not included in the sponsor’s report.  This reviewer’s test of treatment by country interaction yielded a p-value of .057 suggesting that further examination of the country results is warranted.  This reviewer found that  removing  Poland from the analysis increased the  interaction p-value to .42.  A comparison of the results for each country showed that the interaction was quantitative not qualitative with the relationship between doses for each country similar to what is illustrated in the boxplots below (Figure 7).

 

 

 

Figure 7.  Study 27  Boxplots of % change from baseline for LDL-C at Week 12 LOCF

 


            A graph of LDL over the duration of the trial (Figure 8) illustrates two points; 1) the response is quite stable from Week –2  to 0 so averaging of these three values to obtain a baseline is acceptable and 2) it appears that most of the response occurs during the first 2 weeks of treatment. Further examination of the latter point by this reviewer showed that in the rosu 5 mg group, the average additional decrease from Week 2 to endpoint was about 2% and in the rosu 10 mg group, about 4%.  About 56% of the rosu 5 mg patients and 70% of rosu 10 mg patients had a smaller LDL at endpoint than at Week 2 but the bulk of their response occurred by Week 2.

 

 

Figure 8. Study 27 LDL-C by week on study (observed cases)

 

 

 

 


Secondary Endpoints

 

            The results for five secondary endpoints (selected by the medical reviewer from the 10 named by the sponsor) are shown in Table 24 below. Significantly larger decreases in TC, non-HDL, and Apo-B were seen for both doses of rosuvastatin compared to pravastatin and simvastatin. Similar changes in HDL and TG were seen in all groups. No criteria for non-inferiority were named in the protocol for secondary endpoints.

 

Table 24.  Study 27 LDL-C Results at Week 12 LOCF (mg/dL)

Sponsor’s Results

 

ROSU 5

(n=120)

Mean (SD[5])

ROSU 10

(n=115)

Mean (SD)

PRAV 20

(n=137)

Mean (SD)

SIM 20

(n=130)

Mean (SD)

TC

Baseline

% Change

Results vs. Prav 20

   p-value (95% CI)

Results vs. Sim 20

   p-value (95% CI)

 

274 (24)

-30% (1)

 

.001 (-12%, -8%)

 

.004 (-6%, -1%)

 

271 (23)

-34% (1)

 

.001 (-17%, -12%)

 

.001 (-10%, -5%)

 

275 (23)

-20% (1)

 

 

 

274 (25)

-26% (1)

non-HDL

Baseline

% Change

Results vs. Prav 20

   p-value (95% CI)

Results vs. Sim 20

   p-value (95% CI)

 

224 (24)

-38% (1)

 

.0001 (-16%, -10%)

 

.004 (-7%, -1%)

 

218 (24)

-44% (1)

 

.0001 (-22%, -16%)

 

.0001 (-14%, -8%)

 

221 (23)

-26% (1)

 

219 (25)

-34% (1)

HDL

Baseline

% Change

Results vs. Prav 20

   p-value (95% CI)

Results vs. Sim 20

   p-value (95% CI)

 

51 (13)

+6.2% (1.2)

 

.26 (-1%, +5%)

 

.15 (-1%, +5.5%)

 

53 (12)

+6.8% (1.3)

 

.14 (-1%, +6%)

 

.07 (-0.2%, +6%)

 

54 (13)

+4.4% (1.2)

 

55 (14)

+3.9% (1.2)

TG

Baseline

% Change

Results vs. Prav 20

   p-value (95% CI)

Results vs. Sim 20

   p-value (95% CI)

 

168 (62)

-12%  (3)

 

.80 (-6%, +8%)

 

.62 (-6%, +9%)

 

160 (59)

-18% (3)

 

.15 (-13%, +2%)

 

.23 (-12%, +3%)

 

161 (64)

-13% (3)

 

 

 

156 (62)

-14% (3%)

Apo-B

Baseline

% Change

Results vs. Prav 20

   p-value (95% CI)

Results vs. Sim 20

   p-value (95% CI)

 

183 (24)

-34% (1)

 

.001 (-15%, -9%)

 

.011 (-7%, -1%)

 

177 (22)

-39% (1)

 

.001 (-21%, -14%)

 

.001 (-12%, -6%)

 

178 (24)

-21% (1)

 

178 (24)

-29% (1)

 


Study 28 (conducted 4/99 to 10/00)

 

Design

 

Study 28 was a double-blind, multicenter, randomized trial designed to compare rosuvastatin to pravastatin and simvastatin.  The primary objective was to compare doses of 5 mg and 10 mg of rosuvastatin to the 20 mg dose of pravastatin and simvastatin after 12 weeks of treatment. After the 12-week fixed dose period, patients were followed for an additional 40 weeks during which the dose could be titrated to achieve NCEP target LDL-C.  

The primary outcome variable was percent change from baseline in LDL at Week 12.  Secondary endpoints were the following:

·        % change at 52 weeks for LDL

·        % change for TC, HDL, LDL/HDL, TC/HDL, non-HDL/HDL, TG, ApoB, ApoB/ApoA-1, ApoA-1 and Lp(a) at 12 weeks and 52 weeks

·        % of patients within NCEP and EAS guidelines at 12 weeks and 52 weeks

·        number of titration steps

·        % of patients on each titrated dose at 52 weeks

 

Upon advice from the medical reviewer, the following secondary endpoints are reviewed here:

·        TC, HDL, non-HDL, ApoB, TG

 

The Eating Pattern Assessment Tool was administered during the 6-week dietary lead-in period at Weeks –6 and –2 and during double-blind treatment at Weeks 12, 20, 28, 36, 44 and 52. The results of this test were analyzed by the sponsor to assess adherence to the NCEP Step-I diet and they are not included here.

Inclusion criteria included the following:

·        160 mg/dL#LDL<250 mg/dL at Visits 2 (Week –1) and 3  (Week –1)

·        TG<400 mg/dL at Visits 2 (Week –1) and 3  (Week –1)

·        men and women $18 years

 

Fasting lipids were measured at Weeks –6, -2, -1, 0, 2, 6, 10, 12, 20, 28, 36, 44, 50 and 52. The dose could be increased to the next dose level at Weeks 12, 20, 28, 36, 44 or 50 based on whether the NCEP target LDL had been met and at the discretion of the investigator.

Patients should have been withdrawn from the trial if CK>10xULN with pain or ALT or AST>3xULN on 2 or more occasions.

 

Patient Disposition

 

            A total of 477 patients were randomized to treatment at  44 centers in the United States (Table 25). About 40% of the patients were from Central United States. The number of patients in each center ranged from 1 to 36 patients;  3 centers were missing patients in 1 or more treatment groups.

 

                        Table 25. Study 28 Distribution of centers and patients by USA region

 

# centers

ROS 5

ROS 10

PRAV 20

SIM 20

Northeast

11

27

28

28

28

Southeast

10

30

27

28

30

Central

14

50

45

48

48

West

8

14

16

13

14

 

Table 26 shows the number of patients on study by study week.  At Week 12 about 95% of the patients on rosuvastatin 10 mg, pravastatin 20 mg or simvastatin 20mg remained on study while 89% remained on study in the rosuvastatin 5 mg group. At the end of the study, over 80% of the patients were still on study.  Only 3 patients were not included in the ITT analysis due to missing data at baseline or on study.

 

Table 26. Study  27 ITT Patient Disposition by Week on Study

 

ROSU 5

ROSU 10

PRAV 20

SIM 20

Randomized

123

116

118

120

  Wk 2

  Wk 6

  Wk 10

  Wk 12

  Wk 20

  Wk 28

  Wk 36

  Wk 44

  Wk 50

  Wk 52

121

117

113

109 (89%)

109

107

105

105

102

101

116

114

112

109 (94%)

107

104

101

100

98

98

117

114

113

112 (95%)

109

108

105

100

97

95

120

120

116

115 (96%)

114

110

107

105

103

102

Completers

101 (82%)

98 (84%)

95 (81%)

102 (85%)

ITT

121 (98%)

116 (100%)

117(99%)

120 (100%)

 

            The major reason for dropout was adverse event (ADE) in all treatment groups; patient request was the second most frequent reason (Table 27).

           

Table 27. Study 27   Reasons for discontinuation

 

ROSU 5

(n=123)

ROSU 10

(n=116)

PRAV 20

(n=118)

SIM 20

(n=120)

ADE

Pt request

Prot. Viol.

Other

Lost-to-FU

12 (9.8%)

4 (3.3%)

1 (0.8%)

3 (2.4%)

2 (1.6%)

10 (8.6%)

3 (2.6%)

1 (0.9%)

1 (0.9%)

3 (2.6%)

11 (9.3%)

5 (4.2%)

3 (2.5%)

3 (2.5%)

1 (0.8%)

9 (7.5%)

5 (4.2%)

3 (2.5%)

0 (0%)

1 (0.8%)

 

Baseline Demographics

 

            The treatment groups were well-balanced regarding baseline demographics (Table 28) and medical history including CHD risk factors. The average age of patients in this study was about 58 years with about of the patients 65 years or older. More elderly patients were seen in the pravastatin group than each of the other groups. Majority of the patients were Caucasian with fewer than 10% in each of the other racial groups.

 

Table 28.  Study 28 Patient Demographics for All Randomized Patients

 

ROSU 5

(n=123)

ROSU 10

(n=116)

PRAV 20

(n=118)

SIM 20

(n=120)

Age

  Mean (SD)

  Range

  %$65years

 

57 (10)

30-79

24%

 

58 (10)

34-81

33%

 

60 (11)

28-82

40%

 

59 (12)

19-86

38%

Gender

  % female

 

60%

 

63%

 

58%

 

63%

Race

  % white

  % black

  % Hispanic

  % Asian

 

81%

11%

4%

3%

 

84%

5%

7%

3%

 

86%

7%

4%

3%

 

85%

8%

3%

3%

 

 

Efficacy Results

 

Primary Endpoint LDL-C

The primary endpoint in this trial was percent change from baseline in LDL at Week 12 when all patients were still on their starting doses.  The results (Table 29) are consistent with what was observed in Study 27 (Table 23); each dose of rosuvastatin was found to be superior to pravastatin and simvastatin.

    

Table 29.  Study 28 LDL-C Week 12 LOCF ITT Results

 

ROSU 5

(n=121)

Mean (SD)

ROSU 10

(n=116)

Mean (SD)

PRAV 20

(n=117)

Mean (SD)

SIM 20

(n=120)

Mean (SD)

Baseline (mg/dL)

% Change from Baseline

187 (18)

-39% (14)

187 (21)

-47% (15)

189 (19)

-27% (10)

188 (19)

-35% (16)

Results vs. Prav 20

   p-value

   95% Confidence Interval

 

.0001

-16%, -8.8%

 

.0001

-23.8%, -16.5%

 

 

Results vs. Sim 20

   p-value

   95% Confidence Interval

 

.01

-8.1%, -1%

 

.0001

-16.0%, -8.7%

 

 

            For the 95% confidence intervals, negative values favor rosuvastatin.

 

            LDL was also measured from Week 12 to Week 52. During this time period the dose could be titrated to meet NCEP goals. The LDL response for the full duration of the trial (Figure 9 ) shows no important changes in LDL after Week 12 in any of the treatment groups (a graph of just completers looks the same).

 

Figure 9.  LDL-C (mg/dL) by week on study for all observed case

 

            As for Study 27,  most of the response is seen after 2 weeks of treatment. Small additional decreases averaging from 1-4% between Week 2 and Week 12 were seen in all treatment groups.

            After 12 weeks of the therapy at the randomized dose, patients could be titrated to a higher dose to achieve NCEP goals.  In the rosuvastatin groups, 26% of the rosuvastatin 5 mg patients and 15% of rosuvastatin 10 mg patients were titrated to a higher dose (Table 30 ).  About 59% of pravastatin patients were titrated to 40 mg and 40% of simvastatin patients were titrated to either 40 mg or 80 mg.

 

 

          Table 30. Study 28 Percentage of patients by highest dose received during the trial

Highest Dose

ROSU 5

(n=123)

ROSU 10

(n=116)

PRAV 20

(n=118)

SIM 20

(n=120)

5

10

20

40

80

74%

16%

5%

3%

2%

NA

85%

10%

2%

3%

NA

NA

41%

59%

NA

NA

NA

60%

21%

19%

 

 

            From Figure  10 it can be seen that the LDL continues to decrease in patients titrated to a higher dose (right graph) while for patients maintained on the same dose, on average, the response is largely unchanged.

 

Figure 10.   Study 28 LDL by week on study for patients remaining on the same dose throughout the trial (left graph) and for patients titrated after Week 12 (right graph)

 

Secondary Endpoints

 

 

            The results for five secondary endpoints (selected by the medical reviewer from the 10 named by the sponsor) are shown in Table 31 below. Significantly larger decreases in TC, non-HDL, and Apo-B were seen for both doses of rosuvastatin compared to pravastatin and simvastatin. Significant TG results for rosuvastatin 10 mg compared to pravastatin and simvastatin were observed with borderline results for the 5 mg dose.  Similar changes in HDL were seen for rosuvastatin 5 and the comparators; significantly greater increases were seen for rosuvastatin 10 mg. No criteria for non-inferiority were named in the protocol for secondary endpoints.

 

Table 31.  Study 28 LDL-C Results at Week 12 LOCF (mg/dL)

Sponsor’s Results

 

ROSU 5

(n=120)

Mean (SD[6])

ROSU 10

(n=115)

Mean (SD)

PRAV 20

(n=137)

Mean (SD)

SIM 20

(n=130)

Mean (SD)

TC

Baseline

% Change

Results vs. Prav 20

   p-value (95% CI)

Results vs. Sim 20

   p-value (95% CI)

 

276 (24)

-28% (1)

 

.001 (-12%, -7%)

 

.002 (-7%, -1.5%)

 

273 (23)

-33% (1)

 

.001 (-17%, -12%)

 

.001 (-12%, -7%)

 

274 (24)

-19% (1)

 

 

 

274 (25)

-24% (1)

non-HDL

Baseline

% Change

Results vs. Prav 20

   p-value (95% CI)

Results vs. Sim 20

   p-value (95% CI)

 

226 (22)

-36% (1)

 

.0001 (-14%, -8%)

 

.005 (-8%, -1%)

 

223 (22)

-43% (1)

 

.0001 (-22%, -15%)

 

.0001 (-15%, -9%)

 

224 (24)

-25% (1)

 

223 (23)

-31% (1)

HDL

Baseline

% Change

Results vs. Prav 20

   p-value (95% CI)

Results vs. Sim 20

   p-value (95% CI)

 

51 (11)

+8.2% (1.2)

 

.95 (-3%, +3%)

 

.69 (-4%, +2.5%)

 

50 (11)

+12% (1.2)

 

.03 (+0.4%, +6.8%)

 

.06 (-0.1%, +6%)

 

50 (11)

+8.3% (1.2)

 

51 (11)

+8.8% (1.2)

TG

Baseline

% Change

Results vs. Prav 20

   p-value (95% CI)

Results vs. Sim 20

   p-value (95% CI)

 

193 (72)

-18%  (3)

 

.07 (-13%, +0.6%)

 

.03 (-14%, -0.7%)

 

180 (62)

-22% (3)

 

.004 (-17%, -3%)

 

.001 (-18%, -5%)

 

178 (67)

-11% (3)

 

 

 

176 (63)

-10% (2.5%)

Apo-B

Baseline

% Change

Results vs. Prav 20

   p-value (95% CI)

Results vs. Sim 20

   p-value (95% CI)

 

178 (20)

-31% (1)

 

.001 (-14%, -8%)

 

.006 (-8%, -1%)

 

174 (21)

-37% (1)

 

.001 (-20%, -14%)

 

.001 (-14%, -7%)

 

175 (21)

-20% (1)

 

175 (21)

-27% (1)

 

 

Reviewer’s comments on Studies 27 and 28

 

            Both the 5 mg and 10 mg doses of rosuvastatin showed significantly larger drops in LDL than the 20 mg dose of pravastatin and simvastatin in Studies 27 and 28. The magnitude of the LDL % change from baseline for both doses of rosuvastatin was consistent with what was observed in Studies 8 and 33 even though the duration of treatment was twice as long in Studies 27 and 28. In Study 28,  patients were treated up to 52 weeks with no further lowering of LDL observed.

            In Study 28, patients were titrated after 12 weeks of therapy to achieve NCEP goals. About 87% of the rosuvastatin patients achieved goal with 5 or 10 mg of rosuvastatin. Only 5% of the patients were titrated to the high  doses of 40 or 80 mg.  It appears that for most patients doses of 10 mg or less are sufficient to reach LDL goals.

            The HDL results show favorable results (not always significant) for rosuvastatin 10 mg over pravastatin and simvastatin in both studies; the comparisons to rosuvastatin 5 mg were non-significant but comparable.

            The TG results for rosuvastatin show no difference from pravastatin and simvastatin in Study 27 but favor rosuvastatin significantly in Study 28.


Clinical trials in patients with IIb/IV dyslipidemia

 

Study 35 (conducted 12/99 to 8/00)

 

Design

 

Study 35 is a double-blind, multicenter, randomized Phase III trial designed to compare multiple doses of rosuvastatin  to placebo in Type IIb or IV patients.  Doses of 5, 10, 20, 40  and 80 mg of rosuvastatin were studied. After a 6-week dietary run-in period, patients were randomized to treatment and followed for 6 weeks.

The primary outcome variable was percent change from baseline in triglycerides at Week 6.  The sponsor measured about 25 secondary endpoints. Upon advice from the medical reviewer, the following secondary endpoints are reviewed here:

·        TC, HDL-C, ApoB, LDL-C, LDL-TG, VLDL-C, VLDL-TG

 

Inclusion criteria included the following:

·        300#TG<800 mg/dL Visits 2 (Week –2) and 3  (Week –1)

·        males and females >18 years

 

Fasting lipids were measured at Weeks –6, -2, -1, 0,  2, 4, and 6.

 

Patients were to be withdrawn from the trial if CK>10xULN with pain or ALT or AST>3xULN.

 

 

Patient Disposition

 

     A total of 156 patients completed screening and were randomized to treatment  at 31 centers in the USA (Table 32).

            Only 7 randomized patients did not complete the 6-week treatment period; only 2 patients are excluded from the ITT analysis due to missing data.

 

Table 32. Study  35 Patient Disposition by Week on Study

 

Placebo

ROSU

5

ROSU

10

ROSU

20

ROSU

40

ROSU

80

Randomized

26

26

23

28

26

27

(Not treated)

 

(1)

 

 

 

 

  Wk 2

  Wk 4

26

25

26

25

23

23

28

27

26

24

27

27

Completers

24 (92%)

25

(96%)

23

(100%)

27

(96%)

24

(92%)

26

(96%)

ITT

26 (100%)

25

(96%)

23

(100%)

27

(100%)

25

(96%)

27

(100%)

 

            The primary reasons for trial discontinuation for the 7 dropouts were ADE and patient request.

 


Baseline Demographics

 

            More than 90% of the patients were Caucasian. About 40% of the patients were female (Table 33). The average age was about 56 years (range of 28 to 82). About 20% of the patients were 65 years or older.  About 59% of the patients had Type IV dyslipidemia and 41% had Type IIb.

            Table 33. Study  35  Patient Demographics for ITT Patients

(Extracted from sponsor’s Table 11)

 

Placebo

 

(n=26)

ROSU

5

(n=26)

ROSU

10

(n=23)

ROSU

20

(n=28)