Heart and Vessels

, Volume 29, Issue 3, pp 287–299

Effects of rosuvastatin versus atorvastatin on small dense low-density lipoprotein: a meta-analysis of randomized trials

Authors

    • Department of Cardiovascular SurgeryShizuoka Medical Center
  • Masao Niwa
    • Department of Cardiovascular SurgeryShizuoka Medical Center
  • Yusuke Mizuno
    • Department of Cardiovascular SurgeryShizuoka Medical Center
  • Hirotaka Yamamoto
    • Department of Cardiovascular SurgeryShizuoka Medical Center
  • Shin-nosuke Goto
    • Department of Cardiovascular SurgeryShizuoka Medical Center
  • Takuya Umemoto
    • Department of Cardiovascular SurgeryShizuoka Medical Center
Original Article

DOI: 10.1007/s00380-013-0358-6

Cite this article as:
Takagi, H., Niwa, M., Mizuno, Y. et al. Heart Vessels (2014) 29: 287. doi:10.1007/s00380-013-0358-6

Abstract

In addition to their high-intensity effects on the reduction in low-density lipoprotein (LDL) levels, rosuvastatin and atorvastatin would be expected to also reduce small dense LDL (sdLDL) levels. To determine which reduces sdLDL levels more, we performed the first meta-analysis and meta-regression of randomized head-to-head trials of rosuvastatin versus atorvastatin therapy. MEDLINE, EMBASE, and the Cochrane Central Register of Controlled Trials were searched through April 2012. Eligible studies were prospective, randomized controlled trials of rosuvastatin versus atorvastatin therapy reporting final sdLDL (directly measured or calculated) levels as an outcome. For each study, data regarding final sdLDL levels in both the rosuvastatin and atorvastatin groups were used to generate mean differences (MD) and 95 % confidence intervals (CI). Meta-regression analysis was performed to determine whether the effects of rosuvastatin therapy were modulated by the prespecified factors. Of 159 potentially relevant articles screened initially, 28 reports of randomized trials enrolling a total of 7802 patients were included. Pooled analysis suggested a significant reduction in final sdLDL levels among patients randomized to rosuvastatin versus atorvastatin therapy (MD, −1.56 mg/dl; 95 % CI, −2.30 to −0.83 mg/dl; P < 0.0001). The meta-regression coefficients were statistically significant for the baseline LDL/sdLDL level and the difference in LDL changes between the two groups. In conclusion, rosuvastatin rather than atorvastatin therapy is likely more effective in reduction of sdLDL levels. It should be further investigated whether the reduction in sdLDL levels implies overt clinical benefits of rosuvastatin over atorvastatin.

Keywords

AtorvastatinMeta-analysisRandomized trialRosuvastatinSmall dense low-density lipoprotein

Introduction

Low-density lipoprotein (LDL) particles do not comprise a homogeneous population but multiple subclasses with discrete size and density, as well as different physicochemical composition, metabolic behavior, and atherogenicity [1]. Measuring LDL particle size, small dense LDL (sdLDL) content, and LDL particle number provides additional assessment of cardiovascular disease risk [2]. The plasma concentration of sdLDL is likely to be more informative than relative LDL particle size, and although methods are available for quantitation of this subfraction, there is considerable room for improvement [3]. The Québec Cardiovascular Study [4] confirmed a strong association in men of the cholesterol content in sdLDL with risk of ischemic heart disease, especially in the first 7 years of follow-up. The relationship of large LDL to risk, however, was weak and possibly paradoxical. Meanwhile in the Framingham Heart Study [5], small LDL particle concentration increased with the number of metabolic syndrome traits, and those with the syndrome had higher cardiovascular event rates. There was no discernible association, however, between the abundance of small LDL and risk of cardiovascular disease. The discordancy of this result in the Framingham Heart Study [5] with that in the Québec study [4] may be due to the fact that LDL particle size loses its discriminating power in this subset of high-risk subjects [3]. Furthermore, the therapeutic modulation of sdLDL significantly reduces cardiovascular risk [4]. Although weight reduction and increased physical activity may constitute first-line therapy, lipid-lowering drugs are able to favorably alter these particles. The SCRIP (Stanford Coronary Risk Intervention Project) [6], FATS (Familial Atherosclerosis Treatment Study) [7] and the PLAC-I (Pravastatin Limitation of Atherosclerosis in the Coronary Arteries) trial [8] used coronary angiographic changes as outcome variables, and have reported that benefit was concentrated in patients with a predominance of sdLDL who received treatment that tended to lower sdLDL. In addition to their high-intensity effects on the reduction in LDL levels, rosuvastatin and atorvastatin would be expected also to reduce sdLDL levels. Although many randomized head-to-head trials compared the effects of rosuvastatin with those of atorvastatin on lipid profile, to the best of our knowledge there have been very few trials reporting directly measured sdLDL levels. Several methods have been developed for the assessment of sdLDL particles [1013]. These methods, however, can be expensive, time-consuming, and technically demanding, making them too laborious for routine clinical practice or the screening of a large population. Although a recent simple method [14] is easier to implement and has the potential for daily clinical use, the cost of reagents may be prohibitively expensive for general or screening use. Recently, Srisawasdi et al. [15] developed an equation for the estimation of sdLDL by using lipids usually measured in routine clinical laboratories such as total cholesterol (TC), LDL, high-density lipoprotein (HDL), and triglyceride (TG). Using the equation, sdLDL levels can be calculated from TC, LDL, HDL, and TG levels even though directly measured sdLDL levels are not reported. To determine which high-intensity statins reduce sdLDL levels more, we perform herein the first meta-analysis and meta-regression of randomized head-to-head trials of rosuvastatin versus atorvastatin therapy.

Methods

All prospective randomized controlled trials of rosuvastatin versus atorvastatin therapy were identified using a two-level search strategy. First, public domain databases including MEDLINE, EMBASE, and the Cochrane Central Register of Controlled Trials were searched through April 2012 using Web-based search engines (PubMed, OVID). Keywords included rosuvastatin; atorvastatin; and randomized, randomised, randomly, or randomization. Second, relevant studies were identified through a manual search of secondary sources, including references of initially identified articles and a search of reviews and commentaries. All references were downloaded for consolidation, elimination of duplicates, and further analysis.

Studies considered for inclusion met the following criteria: the design was a prospective randomized controlled clinical trial; the study population was unrestricted patients; patients were randomly assigned to rosuvastatin versus atorvastatin therapy; and main outcomes included directly measured sdLDL levels or all of TC, LDL, HDL, and TG levels. Because 10 mg/day rosuvastatin is equivalent to 20 mg/day atorvastatin in terms of its ability to lower LDL levels [16, 17], we excluded studies using less than a twofold dosage of atorvastatin in comparison with rosuvastatin. A QUOROM [18] flow diagram of the study selection process is illustrated in Fig. 1. Data regarding detailed inclusion criteria, duration of follow-up, and directly measured sdLDL, TC, LDL, HDL, and TG levels were abstracted (as available) from each individual study. If directly measured sdLDL levels were not reported, we calculated sdLDL levels using the Srisawasdi equation [15].
https://static-content.springer.com/image/art%3A10.1007%2Fs00380-013-0358-6/MediaObjects/380_2013_358_Fig1_HTML.gif
Fig. 1

QUOROM [13] flow diagram for the meta-analysis. sdLDL small dense low-density lipoprotein, TC total cholesterol, LDL low-density lipoprotein, HDL high-density lipoprotein, TG triglyceride

For each study, data regarding final levels of directly measured sdLDL and calculated sdLDL in both the rosuvastatin and atorvastatin groups were used to generate mean differences (MDs) and 95 % confidence intervals (CIs). When standard deviations (SDs) were unavailable, missing SDs were imputed according to the Cochrane Handbook [19].

The linear regression equation by Srisawasdi et al. [15] is as follows:
$$ {\hbox{sdLDL}} \, = \, 0.580 \times {\hbox{non-HDL-C}} \, + \, 0.407 \times {\hbox{LDL}} \, - \, 0.719 \times {\hbox{cLDL}} \, -12.05 $$
(1)
where, non-HDL-C is non-HDL cholesterol levels calculated by subtracting HDL from TC levels (non-HDL-C = TC − HDL), and cLDL is calculated LDL levels using Friedewald formula (cLDL = TC – HDL − TG/5) [20]. Rearranging the Eq. (1) gives
$$ {\hbox{sdLDL}} \, = \, 0.580 \times \left( {\hbox{TC}} \, - \, {\hbox{HDL}} \right) \, + \, 0.407 \times {\hbox{LDL}} \, - \, 0.719 \times \left( {\hbox{TC}} \, - \, {\hbox{HDL}} \, -{\hbox{TG/5}} \right) \, -12.05 = \, -0.139 \times {\hbox{TC}} \, + 0.139 \times {\hbox{HDL}} \, + \, 0.1438 \times {\hbox{TG}} \, + \, 0.407 \times {\hbox{LDL}} \, - \, 12.05 $$
(2)
According to Eq. (2) and the “variance sum law” on the assumption that TC, LDL, HDL, and TG levels are independent of each other, we calculated the SD of sdLDL levels (SD[sdLDL]) from
$${\text{SD}}\left[ {\text{sdLDL}} \right] = (-0.139^{ 2} \times {\text{SD}}\left[ {\text{TC}} \right]^{ 2} + 0.139^{ 2} \times {\text{SD}}\left[ {\text{HDL}} \right]^{ 2} + 0.1438^{ 2} \times {\text{SD}}\left[ {\text{TG}} \right]^{ 2} { + } 0.407^{ 2} \times {\text{SD}}\left[ {\text{LDL}} \right]^{ 2} )^{1/2} = (0.01932 1\times {\text{SD}}\left[ {\text{TC}} \right]^{ 2} + 0.019321\times {\text{SD}}\left[ {\text{HDL}} \right]^{ 2} + 0.02067844\times {\text{SD}}\left[ {\text{TG}} \right]^{ 2} + 0.165649\times {\text{SD}}\left[ {\text{LDL}} \right]^{ 2} )^{1/2} $$
where SD[TC], SD[HDL], SD[TG], and SD[LDL] are SDs of TC, HDL, TG, and LDL levels.

Study-specific estimates were combined using in both fixed-effects and random-effects models. Between-study heterogeneity was analyzed by means of standard χ2 tests. Where no significant statistical heterogeneity was identified, the fixed-effect estimate was used preferentially as the summary measure. Sensitivity analyses were performed to assess the contribution of each study to the pooled estimate by excluding individual trials one at a time and recalculating the pooled MD estimates for the remaining studies. Unrestricted maximum likelihood (mixed-effects) meta-regression analysis was performed to determine whether the effects of rosuvastatin therapy were modulated by the prespecified factors: baseline LDL and sdLDL levels; differences in changes from baseline to final LDL (HDL or TG) levels between the rosuvastatin and atorvastatin groups (delta LDL (HDL or TG) change = LDL (HDL or TG) change [final LDL (HDL or TG) − baseline LDL (HDL or TG)] in the rosuvastatin group − LDL (HDL or TG) change in the atorvastatin group); and follow-up length. Meta-regression graphs depict the effect of rosuvastatin therapy on the outcome (plotted as a MD of final sdLDL levels on the y axis) as a function of a given factor (plotted as a mean of that factor on the x axis). Meta-regression coefficients (slopes of meta-regression lines) show the estimated increase in MD per unit increase in the covariate. A negative coefficient would indicate that as a given factor increases, the MD decreases, i.e., rosuvastatin therapy is more beneficial in reducing the outcome of interest. Publication bias was assessed graphically using a funnel plot and mathematically using an adjusted rank-correlation and linear regression test. All analyses were conducted using Review Manager version 5.1 (Nordic Cochrane Centre, Copenhagen, Denmark) and Comprehensive Meta-Analysis version 2 (Biostat, Englewood, NJ, USA).

Results

As outlined in Fig. 1, our search identified 28 prospective randomized controlled clinical trials of rosuvastatin versus atorvastatin therapy. These included 3 measured-sdLDL trials [2123] (reporting directly measured sdLDL levels) and 25 calculate-sdLDL trials [16, 2447] (from which sdLDL levels were calculated). In total, our meta-analysis included data on 7802 patients randomized to therapy with rosuvastatin or atorvastatin. The trial design and lipid profile in the rosuvastatin and atorvastatin groups are summarized in Tables 1, 2, and 3. Pooled analysis of the 3 measured-sdLDL trials (MD, −4.99 mg/dl; 95 % CI, −8.50 to −1.49 mg/dl; P for effect = 0.005; P for heterogeneity = 0.86), the 25 calculate-sdLDL trials (MD, −1.41 mg/dl; 95 % CI, −2.16 to −0.66 mg/dl; P for effect = 0.0002; P for heterogeneity = 0.28), and all the 28 trials (MD, −1.56 mg/dl; 95 % CI, −2.30 to −0.83 mg/dl; P for effect <0.0001; P for heterogeneity = 0.25) demonstrated a statistically significant reduction in final sdLDL levels with rosuvastatin relative to atorvastatin therapy in the fixed-effects model (P for subgroup differences = 0.05; Fig. 2). There was minimal trial heterogeneity and, accordingly, little difference in the pooled result from random-effects modeling. In general, exclusion of any single trial from the analysis did not substantively alter the overall result of our analysis. To assess publication bias, we generated a funnel plot of the effect size versus the reciprocal of standard error for each trial (Fig. 3). There was no evidence of significant publication bias (P = 0.54850 and 0.38076 by the adjusted rank-correlation and linear regression test, respectively). The meta-regression coefficients (slopes of the meta-regression lines) were statistically significant for the baseline LDL level (−0.02465; 95 % CI, −0.04128 to −0.00802; P = 0.00367; Online resource, electronic supplementary material (ESM) Fig. 1), baseline sdLDL level (−0.08313; 95 % CI, −0.14325 to −0.02301; P = 0.00673; ESM Fig. 2), delta LDL change (0.22900; 95 % CI, 0.10374–0.35426; P = 0.00034; ESM Fig. 3), delta HDL change (0.59493; 95 % CI, 0.00596–1.18390; P = 0.04773; ESM Fig. 4), and delta TG change (0.08042; 95 % CI, 0.03538–0.12545; P = 0.00047; ESM Fig. 5), which would indicate that rosuvastatin therapy is more beneficial in reducing final sdLDL levels as baseline LDL and sdLDL levels increase and as delta LDL, HDL, and TG changes decrease. On the other hand, the meta-regression coefficient for the follow-up length was not statistically significant (0.00782; 95 % CI, −0.01092 to 0.02657; P = 0.41327; ESM Fig. 6).
Table 1

Trial design

Trial

Inclusion criteria

Intervention

Patient number

Follow-up

Rosuvastatin (mg/day)

Atorvastatin (mg/day)

Rosuvastatin

Atorvastatin

Measured-sdLDL trials

Bahadir 2009

Hypercholesterolemic metabolic syndrome

10

20

17

12

8 weeks

Mori 2009

Hypercholesterolemia with concomitant type 2 diabetes

5

10

50

48

12 weeks

STELLAR 2008

Hyperlipidemia

40

80

135

136

6 weeks

Calculated-sdLDL trials

Anagnostis 2011

Non-diabetes with dyslipidemia

10

20

18

18

12 weeks

ARIES 2006

Hypercholesterolemic African Americans

10

20

186

178

6 weeks

ARTMAP 2012

Mild coronary atherosclerotic plaques

10

20

128

143

6 months

ATOROS 2006

Cardiovascular disease-free subjects with primary hyperlipidemia

10 (6 weeks) → 10/20 (18 weeks)

20 (6 weeks) → 20/40 (18 weeks)

60

60

24 weeks

Brunetti 2007

Diabetes

10

20

11

11

3 months

Davidson 2002

Hypercholesterolemia

5

10

128

127

12 weeks

ECLIPSE 2008

High risk of cardiovascular disease with hypercholesterolemia

10 (6 weeks) → 20 (6 weeks) → 40 (12 weeks)

10 (6 weeks) → 20 (6 weeks) → 40 (6 weeks) → 80 (6 weeks)

464

476

24 weeks

Her 2010

Hypercholesterolemia

10

20

25

25

8 weeks

Hong 2011

Intermediate coronary stenosis

20

40

65

63

11 months

IN-PRACTICE 2010

Cardiovascular disease (CVD), diabetes, or high risk of CVD

5/10

40

258

259

6 weeks

Liu 2011

Stable atherosclerosis

10

20

18

18

4 weeks

Mazza 2008

Primary hypercholesterolemia

10

20

52

54

48 weeks

MERCURY I 2004

Hypercholesterolemia with coronary heart disease, atherosclerosis, or type 2 diabetes

10

20

521

299

16 weeks

Okada 2011

Coronary artery disease

5

20

38

35

12 weeks

Olsson 2001

Moderate hypercholesterolemia

10/20/40

80

63

10

6 weeks

2.5/5

10

30

13

Olsson 2002

Primary hypercholesterolemia

5

10

135

139

12 weeks

POLARIS 2007

High-risk patients with hypercholesterolemia

20 (2 weeks) → 40 (24 weeks)

40 (2 weeks) → 80 (24 weeks)

428

432

26 weeks

Puccetti 2011

Hypercholesterolemia

10

20

30

30

8 weeks

RADAR 2005

Cardiovascular disease and low HDL

10 (6 weeks) → 20 (6 weeks) → 40 (6 weeks)

20 (6 weeks) → 40 (6 weeks) → 80 (6 weeks)

230

231

18 weeks

Rawlings 2009

Stable atherosclerosis

10

40

15

15

28 days

SATURN 2011

Coronary disease

40

80

520

519

104 weeks

Schneck 2003

Hypercholesterolemia

10

20

45

39

6 weeks

20

40

28

42

40

80

44

41

5

10

38

43

Sindhu 2011

Obese type 2 diabetes mellitus

10–40

40–80

20

20

6 months

SUBARU 2008a

Hypercholesterolemia

5

10

198

205

8 weeks

Tsutamoto 2011

Stable chronic heart failure with dilated cardiomyopathy

2.5

5

31

32

6 months

Total

   

4029

3773

 

HDL high-density lipoprotein, sdLDL small dense low-density lipoprotein

aCalculated sdLDL was abstracted because sdLDL was measured by relative mobility calculated from a densitogram

Table 2

Baseline and final lipid profile in the rosuvastatin group

Trial

Rosuvastatin

Dosage (mg/day)

Baseline (mg/dl)

Final (mg/dl)

TC

LDL

HDL

TG

sdLDL

TC

LDL

HDL

TG

sdLDL

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Measured-sdLDL trials

Bahadir 2009

10

272.1

55.7

179.3

50.7

47.1

11.1

223.1

80.4

29.6

24.8

174.5

33.3

96.2

34.9

45.6

9.6

161.0

42.5

8.9

8.5

Mori 2009

5

NA

172.7

21.5

61.0

16.4

160.2

81.0

44.5

21.7

NA

83

20.6

66.2

15.4

114.4

54.3

21.8

10.9

STELLAR 2008

40

284

27

204

27

50

13

182

69

71

30

167

46

96

45

54

15

127

88

28

38

Calculated-sdLDL trials

Anagnostis 2011

10

278

35

187

24

62

17

149

79

55

16

176

33

95

26

60

14

104

46

25

13

ARIES 2006

10

270.7

31.9

191.8

27.2

51.5

11.2

136.5

49.7

55.2

14.0

198.7

48.8

120.6

43.5

55.1

12.9

114.7

61.0

33.6

21.0

ARTMAP 2012

10

186

34

109

31

40

9

182

121

38

22

126

25

53

18

47

11

125

65

17

12

ATOROS 2006

10 (6 weeks) → 10/20 (18 weeks)

285

30

205

42

48

6

159

51

61

19

182

32

105

21

50

6

113

35

29

11

Brunetti 2007

10

191.6

43.6

105.7

32.3

47.2

13.0

200.4

128.3

39.7

23.5

131.1

34.7

48.9

20.1

40.0

7.3

269.0

248.0

33.9

36.9

Davidson 2002

5

278

24

188

19

52

12

186

77

60

14

200

40

113

34

59

13

154

92

36

20

ECLIPSE 2008

10 (6 weeks) → 20 (6 weeks) → 40 (12 weeks)

276.6

25.3

189.2

21.0

52.0

13.6

176.8

64.8

59.2

13.3

162.4

46.4

80.8

49.5

56.4

21.0

133.3

80.0

25.3

24.3

Her 2010

10

242

27

163

21

50.3

10.3

166

73

52

14

157

28

81

21

51.7

11.5

137

39

26

11

Hong 2011

20

185

42

122

37

47

10

125

94

36

21

124

25

62

20

47

12

94

56

16

12

IN-PRACTICE 2010

5/10

181.7

23.2

96.7

15.5

54.1

11.6

141.7

79.7

29.9

13.6

177.2

36.0

93.8

27.7

54.1

13.2

135.6

89.5

28.5

17.9

Liu 2011

10

235.6

129.8

162.3

31.0

50.1

22.5

182.1

107.3

54.4

27.1

164.3

96.7

96.6

29.3

56.2

19.9

146.1

95.0

33.2

22.7

Mazza 2008

10

313.25

51.15

217.74

60.50

56.55

13.94

195.00

191.45

68.93

37.66

201.19

34.31

121.49

29.69

55.19

13.12

124.00

77.70

34.93

17.23

MERCURY I 2004

10

246.6

34.5

164.9

31.0

48.6

10.9

165.5

64.7

51.3

16.5

214.0

269.3

117.4

348.3

58.9

358.5

147.1

640.1

35.3

180.2

Okada 2011

5

198.0

25.0

120.3

18.4

49.6

10.2

134.0

50.0

35.6

11.0

180.4

29.2

101.5

22.5

51.0

12.8

130.0

46.1

30.0

12.1

Olsson 2001

10/20/40

261.5

22.6

185.4

17.5

52.5

12.2

125.0

53.1

52.3

11.1

151.1

31.9

76.1

25.7

58.1

13.7

96.3

63.9

19.9

14.7

2.5/5

267.4

21.1

191.7

15.4

50.3

9.4

124.0

52.2

53.6

10.3

186.5

29.3

111.7

23.8

56.3

11.5

94.0

64.7

28.8

14.1

Olsson 2002

5

273.5

24.5

188.0

19.3

54.5

14.1

155.8

67.0

56.4

13.0

186.0

40.1

101.5

34.3

57.8

16.3

132.4

80.9

30.5

19.1

POLARIS 2007

20 (2 weeks) → 40 (24 weeks)

275.2

26.4

189.3

21.2

48.0

12.1

190.3

71.9

60.8

14.1

161.8

42.6

81.4

36.7

53.3

15.5

145.2

94.5

26.9

21.2

Puccetti 2011

10

288.5

8.9

223.1

12.6

48.0

5.4

90.3

10.5

58.3

5.5

190.3

6.6

122.2

7.2

49.9

6.0

85.0

10.5

30.4

3.5

RADAR 2005

10 (6 weeks) → 20 (6 weeks) → 40 (6 weeks)

224.3

50.3

139.2

46.4

30.9

3.9

248.0

132.9

53.4

27.8

124.0

58.9

62.2

52.9

32.4

6.4

160.2

149.1

23.6

31.5

Rawlings 2009

10

232

39

146

35

48

15

184

108

48

22

165

35

80

27

49

15

174

124

29

22

SATURN 2011

40

193.9

34.1

120.0

27.3

45.3

11.8

128.0

66.7

34.5

15.5

139.4

27.4

62.6

22.8

50.4

11.4

120.0

50.4

18.3

12.5

Schneck 2003

10

276

25

190

18

51

14

180

62

60

12

184

29

101

23

54

15

140

81

31

16

20

270

25

188

24

50

10

164

52

57

13

170

27

91

26

55

11

134

66

28

15

40

276

26

188

20

53

14

176

67

59

13

163

30

81

24

60

15

131

84

25

16

5

281

27

193

22

53

14

180

89

61

16

198

33

113

28

57

15

138

103

34

19

Sindhu 2011

10–40

245.93

45.15

173.72

27.80

44.11

8.31

182.07

38.33

56.78

14.11

158.40

40.80

74.66

18.48

52.23

8.88

164.16

32.36

27.19

10.58

SUBARU 2008

5

186.1

28.8

102.9

25.1

60.9

17.6

128.5

67.4

30.9

14.8

178.5

28.5

95.3

24.2

60.7

17.7

136.7

80.4

30.0

15.9

Tsutamoto 2011

2.5

189

34

111

28

43

10

192

80

40

17

143

221

71

17

46

10

130

49

22

32

HDL high-density lipoprotein, LDL low-density lipoprotein, NA not available, SD standard deviation, sdLDL small dense low-density lipoprotein, TC total cholesterol, TG triglyceride

Table 3

Baseline and final lipid profile in the atorvastatin group

Trial

Atorvastatin

Dosage (mg/day)

Baseline (mg/dl)

Final (mg/dl)

TC

LDL

HDL

TG

sdLDL

TC

LDL

HDL

TG

sdLDL

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Mean

SD

Measured-sdLDL trials

Bahadir 2009

20

245.5

27.5

161.2

26.3

45.2

11.0

210.1

66.9

26.2

15.0

167.0

26.8

86.8

22.2

43.6

11.9

184.0

76.3

14.8

9.6

Mori 2009

10

NA

173.6

32.1

58.2

13.1

156.4

81.2

49.5

22.9

NA

94.9

22.7

60.2

12.5

136.3

71.0

27.0

12.8

STELLAR 2008

80

278

31

198

27

52

14

180

77

63

25

168

48

98

43

53

16

124

100

31

31

Calculated-sdLDL trials

Anagnostis 2011

20

272

41

183

34

58

8

167

90

57

20

182

25

106

18

55

10

115

44

30

10

ARIES 2006

20

269.4

28.1

189.1

24.0

49.9

12.8

146.3

49.7

55.4

12.8

192.1

45.6

116.3

40.6

51.7

14.4

117.6

62.0

32.7

19.9

ARTMAP 2012

20

183

36

110

31

40

13

165

93

37

19

128

23

56

18

47

12

122

67

17

13

ATOROS 2006

20 (6 weeks) → 20/40 (18 weeks)

285

43

204

40

48

8

157

55

61

19

180

27

113

38

47

7

113

49

32

17

Brunetti 2007

20

200.3

29.4

99.1

26.1

40.4

4.9

363.0

184.9

58.3

28.9

146.5

32.6

52.7

25.3

38.5

4.0

293.3

123.1

36.6

21.0

Davidson 2002

10

272

25

186

20

50

12

183

72

59

14

204

40

106

34

56

13

218

87

42

20

ECLIPSE 2008

10 (6 weeks) → 20 (6 weeks) → 40 (6 weeks) → 80 (6 weeks)

274.9

25.0

188.3

20.4

51.4

11.8

176.4

67.0

58.9

13.3

166.3

46.0

90.0

49.0

52.3

19.7

127.0

81.8

27.0

24.2

Her 2010

20

251

24

168

15

50.4

9.1

166

73

52

13

170

34

92

24

51.3

13.2

132

63

28

14

Hong 2011

40

181

43

117

38

48

15

124

117

35

24

133

32

70

24

47

12

103

193

19

30

IN-PRACTICE 2010

40

181.7

23.2

100.5

15.5

54.1

11.6

132.9

73.1

30.2

12.8

166.7

36.1

89.4

28.5

52.9

13.2

122.1

87.7

26.1

17.9

Liu 2011

20

229.6

132.8

149.7

28.0

56.0

22.1

167.4

151.0

48.8

30.9

173.0

106.5

87.5

25.0

57.3

19.5

133.5

85.7

26.7

22.0

Mazza 2008

20

318.11

67.79

232.57

65.17

54.00

15.40

157.42

93.00

68.53

31.24

249.31

44.87

162.78

41.62

56.50

13.77

150.42

85.77

49.03

21.95

MERCURY I 2004

20

248.3

37.9

166.7

30.1

49.9

11.8

158.6

62.1

51.0

16.1

217.5

40.3

122.7

34.7

55.6

20.9

141.4

69.5

35.7

18.4

Okada 2011

20

191.2

20.4

114.1

14.7

50.9

9.9

126.0

74.1

33.0

12.6

168.4

19.4

94.5

16.8

47.7

9.8

127.0

62.2

27.9

11.7

Olsson 2001

80

266.8

15.5

193.4

11.6

46.4

7.7

124.0

35.4

53.9

7.3

143.8

24.2

80.1

21.2

44.9

10.8

85.9

46.1

19.1

11.5

10

263.0

15.5

189.4

7.7

50.3

15.5

124.0

53.1

53.3

8.8

178.8

25.2

105.7

18.7

53.7

16.4

107.1

57.3

29.0

12.0

Olsson 2002

10

273.8

23.1

188.1

18.1

53.9

12.2

160.6

68.7

57.0

12.8

197.1

37.1

114.7

32.2

57.1

14.4

134.9

82.4

34.6

18.5

POLARIS 2007

40 (2 weeks) → 80 (24 weeks)

274.8

27.3

189.0

22.1

47.6

10.9

191.3

63.9

60.8

13.5

166.8

43.2

89.8

35.7

50.6

13.4

138.3

86.2

28.2

20.1

Puccetti 2011

20

288.5

8.9

218.5

13.2

49.1

2.9

95.7

11.8

57.4

5.8

190.3

6.6

126.1

7.2

49.9

3.2

90.3

8.5

32.7

3.3

RADAR 2005

20 (6 weeks) → 40 (6 weeks) → 80 (6 weeks)

220.4

50.3

143.1

50.3

30.9

3.9

239.1

124.0

54.2

28.0

133.4

58.6

74.3

56.6

31.8

6.4

163.6

140.1

27.6

31.7

Rawlings 2009

40

217

27

135

122

44

60

193

229

47

60

144

56

77

77

45

36

116

63

22

34

SATURN 2011

80

193.5

34.2

119.9

28.9

44.7

10.7

130.0

59.3

34.8

15.4

144.1

27.3

70.2

22.8

48.6

11.4

110.0

46.7

19.1

12.2

Schneck 2003

20

272

23

185

19

49

11

189

90

59

15

184

27

105

23

53

12

141

105

33

18

40

274

24

188

22

49

10

182

66

59

14

175

28

97

26

51

11

132

84

29

17

80

278

23

190

18

48

11

193

68

61

13

166

29

88

24

49

12

126

88

26

17

10

280

30

190

24

54

15

180

70

60

15

200

35

117

29

57

16

149

87

37

18

Sindhu 2011

40–80

243.56

45.15

163.98

25.88

42.73

7.95

173.16

52.00

51.67

14.40

194.77

40.80

91.05

19.39

51.92

9.10

162.00

55.26

28.45

12.62

SUBARU 2008

10

192.3

34.8

109.3

30.6

60.1

15.3

130.9

72.2

32.9

17.1

187.4

32.9

106.7

28.7

58.8

14.6

129.7

89.5

32.2

18.1

Tsutamoto 2011

5

203

34

115

32

42.6

11

190

108

40

21

146

27

72

29

43.4

10

131

73

22

16

HDL high-density lipoprotein, LDL low-density lipoprotein, NA not available, SD standard deviation, sdLDL small dense low-density lipoprotein, TC total cholesterol, TG triglyceride

https://static-content.springer.com/image/art%3A10.1007%2Fs00380-013-0358-6/MediaObjects/380_2013_358_Fig2_HTML.gif
Fig. 2

Forest plot for final small dense low-density lipoprotein (sdLDL). CI confidence interval, IV inverse variance, SD standard deviation

https://static-content.springer.com/image/art%3A10.1007%2Fs00380-013-0358-6/MediaObjects/380_2013_358_Fig3_HTML.gif
Fig. 3

Publication bias funnel plot of the effect size versus the reciprocal of standard error for each trial

Discussion

The results of our analysis suggest that rosuvastatin therapy may reduce final sdLDL levels by 1.56 mg/dl over atorvastatin therapy, depending on baseline LDL and sdLDL levels. This result was robust in sensitivity analyses, even eliminating any single trial from the analysis. Only 4 [22, 28, 34, 40] of the 28 individual trials included in the present meta-analysis, however, were able to demonstrate a statistically significant benefit to rosuvastatin therapy, likely attributable to systematic underpowering of the other trials in the design phase. The results from calculated-sdLDL trials (from which sdLDL levels were calculated) are the most compelling, with data from 7404 patients randomized in 25 different trials demonstrating a 1.41-mg/dl reduction in final sdLDL levels with rosuvastatin therapy relative to atorvastatin therapy. The data from only three measured-sdLDL trials [2123] (reporting directly measured sdLDL levels) are less robust, primarily because of the small number of enrolled patients (n = 398). Though underpowered as reflected in the wide CIs (−8.50 to −1.49 mg/dl), the observed 4.99-mg/dl reduction in final sdLDL levels in the patients receiving rosuvastatin is statistically significant (P = 0.005). Although few patients have been randomized in the measured-sdLDL trials, the acceptable reduction in final sdLDL levels associated with rosuvastatin therapy in these trials would likely make design of additional randomized controlled trials in which sdLDL levels are directly measured.

A recent meta-analysis [48] of 75 head-to-head randomized controlled trials has indicated that ≥20 mg/day rosuvastatin and atorvastatin could reduce LDL by >40 %; 10 mg/day atorvastatin, 80 mg/day fluvastatin, 40–80 mg/day lovastatin, and 20 mg/day simvastatin could decrease LDL by 30 %–40 %; and 40 mg/day fluvastatin, 10–20 mg/day lovastatin, 20–40 mg/day pravastatin, and 10 mg/day simvastatin could decrease LDL by 20 %–30 %. Another recent meta-analysis [49] of 37 studies from the VOYAGER (Individual Patient Meta-Analysis of Statin Therapy in At Risk Groups: Effects of Rosuvastatin, Atorvastatin and Simvastatin) trial has shown that increasing doses of rosuvastatin, atorvastatin, and simvastatin predictably resulted in an incremental benefit in terms of TG reduction. Despite differences in the potency of each individual agent, the incremental impact of dose doubling was comparable, with a 2 %–4 % increase in TG lowering, which was less than the incremental effect of increasing dose on LDL lowering. Another analysis [50] of the VOYAGER database has revealed that the HDL-raising ability of rosuvastatin and simvastatin was comparable, with both being superior to atorvastatin. Increases in HDL were positively related to dose with rosuvastatin (from 5.5 % to 7.9 % over the dose range of 5–40 mg/day) but inversely related to dose with atorvastatin (being 4.5 % at the 10-mg/day dose and falling to 2.3 % at a dose of 80 mg/day).

The 13-year follow-up data from the Québec Cardiovascular Study [4] indicated that estimated cholesterol levels in the large LDL subfraction were not associated with an increased risk of ischemic heart disease in men and that the cardiovascular risk attributable to variations in the LDL size phenotype was largely related to markers of a preferential accumulation of sdLDL particles. In participants in cycle 6 of the Framingham Offspring Study [51], women, and not men, with coronary heart disease had higher sdLDL concentrations than controls. Furthermore, sdLDL levels are more powerful than LDL levels for the determination of severe stable coronary heart disease [52]. In addition, sdLDL is independently associated with the incidence and extent of coronary artery disease, and can be a risk factor for the development of acute coronary syndrome [53]. Even in subjects with the metabolic syndrome and without overt coronary artery disease, sdLDL shows an independent predictive role for future cardiovascular and cerebrovascular events [54]. No cardiovascular events, however, were reported in all the trials included in the present meta-analysis except for the largest (>1000 patients) trial with the longest follow-up (2 years), the SATURN (Study of Coronary Atheroma by Intravascular Ultrasound: Effect of Rosuvastatin versus Atorvastatin) [43] trial. In the SATURN trial, the final sdLDL level was nonsignificantly lower in the rosuvastatin group than in the atorvastatin group (18.3 ± 12.5 vs 19.1 ± 12.2 mg/dl; MD, −0.80 mg/dl; 95 % CI, −2.30 to 0.70 mg/dl; P = 0.30), and the frequency of the first major adverse cardiovascular events was similar in the two groups (7.5 % vs 7.1 %). It is unclear whether a little reduction in final sdLDL levels (−1.56 mg/dl demonstrated in the present meta-analysis) with rosuvastatin relative to atorvastatin therapy implies overt clinical benefit of rosuvastatin over atorvastatin. Further investigation, such as large randomized head-to-head trials with long follow-up, reporting clinical events as main outcomes, would be needed.

One of the present meta-regression analyses indicates that rosuvastatin therapy is more beneficial in reducing final sdLDL levels as a delta LDL change (=LDL change in the rosuvastatin group − LDL change in the atorvastatin group) decreases. The vast majority of the trials included in the present meta-analysis used twofold dosage of atorvastatin compared with rosuvastatin (e.g., 40 mg/day atorvastatin vs 20 mg/day rosuvastatin or 80 mg/day atorvastatin vs 40 mg/day rosuvastatin). It has been also shown that 10 mg/day rosuvastatin is equivalent to 40 mg/day atorvastatin in LDL-lowering ability [55]. If fourfold dosage of atorvastatin compared with rosuvastatin (e.g., 40 mg/day atorvastatin vs 10 mg/day rosuvastatin or 80 mg/day atorvastatin vs 20 mg/day rosuvastatin) reduces LDL levels equally, sdLDL levels also may be equivalently decreased.

Our analysis must be viewed in the context of its limitations. First, we used only data from randomized controlled trials. Patients enrolled in randomized trials may not be representative of patients typically seen in clinical practice. However, because randomized trials balance both known and unknown confounders across treatment groups, this is the study design least vulnerable to bias. Second, our results may be influenced by a publication bias favoring rosuvastatin (introduced to the market more recently than atorvastatin) therapy. This risk was minimized through an exhaustive search of the available literature and through the statistical tests not indicating publication bias (P = 0.54850 and 0.38076 by the adjusted rank-correlation and linear regression test, respectively). Third, only three trials [31, 34, 43] included in the present meta-analysis had a duration of follow-up >6 months. Despite no association of MD of final sdLDL levels with follow-up length in meta-regression, additional trials with a longer follow-up would be needed. Fourth, we abstracted “calculated” instead of “measured” sdLDL levels from 25 of all 28 trials included in the present meta-analysis, because no “measured” sdLDL levels were reported in these trials. Using the equation developed by Srisawasdi et al. [15], sdLDL levels were calculated from TC, LDL, HDL, and TG levels. Despite a strong linear relationship between measured and calculated sdLDL values, with an R2 of 0.88 [15], calculated sdLDL-C values may be biased because the subgroup difference between pooled MD estimates in the measured-sdLDL and calculated-sdLDL trials tends to be statistically significant (P for subgroup differences = 0.05).

In conclusion, despite acknowledged limitations, we found that, based on a meta-analysis, rosuvastatin rather than atorvastatin therapy is likely more effective in reduction of sdLDL levels. It should be further investigated whether the reduction in sdLDL levels implies overt clinical benefits of rosuvastatin over atorvastatin therapy.

Conflict of interest

No author has any conflict of interest associated with this article.

Supplementary material

380_2013_358_MOESM1_ESM.pdf (837 kb)
Supplementary material 1 (PDF 836 kb)

Copyright information

© Springer Japan 2013