- Email: [email protected]
Cholesterol lowering and stroke: no longer room for pleiotropic effects of statins – confirmation from PCSK9 inhibitor studies
Accepted Manuscript Cholesterol lowering and stroke: no longer room for pleiotropic effects of statins – confirmation from PCSK9 inhibitor studies
Tanya Salvatore, Riccardo Morganti, Roberto Marchioli, Raffaele De Caterina PII: DOI: Reference:
S0002-9343(19)30564-9 https://doi.org/10.1016/j.amjmed.2019.06.029 AJM 15267
To appear in:
The American Journal of Medicine
Please cite this article as: T. Salvatore, R. Morganti, R. Marchioli, et al., Cholesterol lowering and stroke: no longer room for pleiotropic effects of statins – confirmation from PCSK9 inhibitor studies, The American Journal of Medicine, https://doi.org/10.1016/ j.amjmed.2019.06.029
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Cholesterol lowering and stroke: no longer room for pleiotropic effects of statins – confirmation from PCSK9 inhibitor studies Short title: Cholesterol lowering and stroke Tanya Salvatorea, MD, Riccardo Morganti, PhDb, Roberto Marchioli, MDc,
PT
and Raffaele De Caterina, MD, PhDb, d Institute of Cardiology, “G. d’Annunzio” University, Chieti, Italy
b
Section of Statistics, Azienda Ospedaliero-Universitaria Pisana, Pisa University Hospital,
RI
a
Pisa, Italy
CV, Metabolic & Renal Disease, Medical & Scientific Services, IQVIA, Milan, Italy
d
Chair of Cardiology, University of Pisa, Pisa, Italy
NU
SC
c
MA
Word count: 3261 – 1 Table – 1 Figure – 1 Online Supplement
Funding sources related to this manuscript: none Conflict of interest statement for all authors: none
Correspondence:
CE PT E
D
All authors had access to the data and a role in writing the manuscript.
Prof. Raffaele De Caterina, MD, PhD
Chair of Cardiology, University of Pisa, and Cardiovascular Division, Pisa University Hospital
AC
Via Paradisa 2 56124 Pisa, Italy
Tel.: +39-050-996-751
E-mail: [email protected]
ACCEPTED MANUSCRIPT ABSTRACT
Background: The relationship between cholesterol levels and stroke has been much less clear than between cholesterol levels and coronary heart disease. This is likely mostly due to the inadequate power of older studies and the low intensity of cholesterol-lowering interventions at that time available. Because a reduction in stroke has been, conversely, clearly observed in trials with statins, for long “pleiotropic” effects of such drugs, unrelated
PT
to cholesterol lowering, have been invoked. In a previous analysis of all randomized trials of cholesterol-lowering treatments reporting on stroke we had, however, reached the
RI
conclusion that any cholesterol lowering is related to a significant reduction of stroke, in a
SC
relationship that appeared to exist for both statin and non-statin cholesterol-lowering interventions. Outcome results of the FOURIER trial with evolocumab, SPIRE-1 and -2
NU
with bococizumab, and ODISSEY OUTCOMES trial with alirocumab now offer the opportunity of clearly confirming or confuting this concept.
MA
Methods: We here report on an updated meta-regression of the relationship of total cholesterol changes occurred with various drugs or treatments and changes in the risk of
D
stroke compared with control.
CE PT E
Results: Figures of the RR here found in FOURIER, SPIRE-1/2 and ODISSEY OUTCOMES (0.79, 0.60 and 0.79) are extremely close to the RR of 0.79, 0.79 and 0.84 predicted by our new meta-regression, respectively. Conclusions: These findings offer a definitive proof that the pure total (and low-density lipoprotein) cholesterol lowering, with any available lipid-lowering intervention, reduce
AC
stroke risk proportional to the extent of cholesterol reduction, without the need of invoking “pleiotropic” effects of any such treatment.
Graphical abstract
Key words: cholesterol; stroke; atherosclerosis; lipid-lowering; statins; PCSK9 inhibitors; ezetimibe.
ACCEPTED MANUSCRIPT INTRODUCTION Hypercholesterolemia is a powerful risk factor for coronary heart disease 1. Cholesterollowering drugs and non-pharmacological treatments significantly reduce morbidity from coronary heart disease, thus proving a causal role for cholesterol in coronary events 1. Conversely, the relationship between cholesterol levels and stroke has been much less clear. Trials with statins have shown decreased stroke incidence in treated populations 1, 2,
PT
but this had not been observed with non-statin drugs or treatments 2, supporting the concept that such effects are attributable to statin-related “pleiotropism” – cholesterol-
RI
independent (neuro)protective properties related to the interference by statins with the
SC
mevalonate pathway 3.
As of today, results from the recently published FOURIER 4, SPIRE 1/2
5
and ODYSSEY
NU
OUTCOMES 6 trials allow a testing of this hypothesis because the three drugs here used – evolocumab, bococizumab and alirocumab – reduce total and low-density lipoprotein (LDL) cholesterol by a mechanism totally independent of the mevalonate pathway and to a
MA
much more substantial extent compared with previous non-statin treatments. We here therefore now present a new comprehensive meta-regression of all reported
D
cholesterol-lowering trials, (a) providing an updated meta-regression of such relationship; and (b) reporting an updated equation allowing to predict the extent of total stroke
METHODS
AC
Literature search
CE PT E
reduction per each definite change in total cholesterol.
To retrieve published randomized controlled trials (RCT), we used the database of studies developed for our reviews on cholesterol-lowering interventions, stroke and secondary prevention of coronary heart disease
2, 7-12
. The database was updated using the
commonly adopted approach of computer-aided literature search of the Pubmed database. The new search was limited to randomized controlled trials performed in humans published between January 2000 to April 2019, using the same key terms adopted for our previous publications plus those for new lipid-lowering drugs (i.e., (Random allocation) OR (clinical trial)) AND ((3-hydroxy-3-methylglutaryl-coenzyme-A reductase inhibitor) OR (HMG CoA reductase inhibitor) OR (statin) OR (simvastatin) OR
ACCEPTED MANUSCRIPT (fluvastatin) OR (atorvastatin) OR (pravastatin) OR (rosuvastatin) OR (lovastatin) OR (colestipol) OR (niacin) OR (gemfibrozil) OR (bezafibrate) OR (probucol) OR (fibrates) OR (diet) OR (ciprofibrate) OR (etofibrate) OR (nicotinic acid OR (acipimox) OR (clofibrate) OR (fenofibrate) OR (colestyramine) OR (ezetimibe) OR (CETP inhibitor) OR (cholesterol ester transfer protein inhibitor) OR (PCSK9 inhibitor). Then titles and abstracts were screened to check if the publications met the following inclusion criteria: a) outcome including the incidence of death; b) randomized study design comparing cholesterol-
PT
lowering treatment with placebo or no treatment; c) inclusion of at least 100 patients; d) follow-up duration of at least 6 months. If more than one publication was found referring to
RI
a single study, all publications were used to compile the most complete dataset. We also
SC
performed a manual search of references cited by the published original studies and recent, relevant review articles.
NU
Potentially relevant studies were reviewed independently by 2 investigators (T.S. and R.D.C.) in duplicate to determine whether they met eligibility criteria for inclusion. Where discrepancies between investigators occurred for inclusion or exclusion, a third
MA
investigator (R.M) was involved to perform an additional evaluation, and discrepancies
D
were resolved by discussion.
CE PT E
Criteria for trial selection and data abstraction
Studies were eligible for inclusion if: (1) the study design was a RCT; (2) the intervention consisted of a single treatment compared with placebo or no treatment; (3) the number of patients included in the study was greater than 100; (4) the intervention duration was at least 6 months; and (5) the number of events occurring during the study were reported by
AC
intervention and control groups. We excluded studies with a crossover design. Although PRISMA guidelines
13
were not formally followed and declared at the time of our former
meta-analyses, criteria used for checking the search accuracy and data reporting were substantially similar. Placebo-controlled trials are less prone to bias than open-conducted studies, but dietary interventions, typical and frequent side effects of cholesterol-lowering drugs, and blinding of laboratory records often make the adoption of a placebo-controlled design unfeasible or at least problematic. Both open and blinded trials were therefore accepted. In multipleintervention trials and in the case of prolongation of follow-up after the end of the study, it is not possible to clearly attribute differences in outcome between the intervention and
ACCEPTED MANUSCRIPT control groups to the actions of any specific treatment. Therefore multi-factorial trials and studies that reported data that were gathered beyond the planned trial duration were excluded. Whenever possible, intention-to-treat analysis data were used. Some RCT had randomized 2 active treatments and a single control group 14-20. For these cases, each arm was considered as a separate one, and the data of control groups were used twice. All data were independently abstracted in duplicate by 2 investigators (T.S. and R.D.C.) using a standardized data collection form. Discrepancies were resolved by discussion with
PT
a third investigator (R.M.) and by referencing the original report. We did not contact
RI
authors to request additional information.
SC
The key data elements sought from each trial were categorized as study characteristics, patients characteristics, treatment characteristics, and clinical outcomes. Specifically, the
NU
characteristics considered for each study were: the name of the first author, journal and year of publication, the duration of follow-up, the number of patients enrolled, the type of intervention, and drug dosages. For patients, the following variables were collected: sex;
MA
mean age; total, fatal, and non-fatal strokes; the means of total cholesterol, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, and triglycerides at
D
baseline, at the end of study and during the follow-up. When the final value of serum lipids was not reported, an estimate was calculated as the mean of the observed values during
CE PT E
the follow-up. As to co-morbidities, information on gender, age, and main cardiovascular risk factors (e.g., history of smoking, diabetes mellitus, and hypertension) was collected. Previous cardiovascular events and pharmacological treatment were also recorded. The clinical outcomes considered were the incidence of total, fatal, and non-fatal stroke, as
AC
defined in the original reports of the trials. Statistical methods
We calculated estimates of the average effect and 95% confidence intervals (95% CI) of statins, fibrates and other cholesterol-lowering interventions on serum lipids estimating the relative risk (RR) to assess the effect of the interventions on the risk of stroke of the population recruited in each trial. To explore the relationship of cholesterol reduction and total stroke, we performed a metaregression by using inverse variance-weighted linear regression
21
. We used the logarithm
RR for total stroke as the dependent variable in the model against the aforementioned
ACCEPTED MANUSCRIPT variables, and weights in each study were the reciprocals of the variances for the logarithm RR for stroke
11
. In each trial, we calculated the percent reduction in total serum
cholesterol levels (%ΔTC) by subtracting end-study (or mean in-study) TC from baseline TC in treated and control groups:
%TC
TCf TCb x100 TCb
where TCf was end-study (or mean in-study) total serum cholesterol, and TCb was
PT
baseline total serum cholesterol.
SC
RI
We conducted statistical testing for efficacy at a 2-tailed α-level of 0.05.
NU
RESULTS
MA
Data on total stroke from the FOURIER, SPIRE 1/2 and ODISSEY OUTCOMES are presented in Table 1. A new meta-regression of all cholesterol-lowering trials vs the observed risk reduction of stroke is presented in Figure 1, The new equation to predict the
CE PT E
D
saving of strokes as a function of total cholesterol lowering is now: LnRR = - 0.061 – 0.005*(% cholesterol reduction)
According to this equation, one should expect a RR of 0.851, 0.810 and 0.770 for a 20%,
AC
30% and 40% reduction in total cholesterol, respectively.
DISCUSSION
We here show that the reduction in stroke by cholesterol lowering fits a log-linear relationship with the reduction in cholesterol levels. These data quite explicitly indicate that stroke reduction is a most likely consequence of cholesterol lowering. They would thus now offer a quite definitive proof that no special property of any cholesterol-lowering intervention has to be invoked to explain stroke reduction previously observed only in statin-intervention trials. These data also allow a precise estimate of the expected results on stroke in future trials affecting total cholesterol.
ACCEPTED MANUSCRIPT
We had already reported such a significant relationship, with no significant association with changes of high-density lipoprotein cholesterol levels, and inconsistent associations with reduction of triglycerides
11
. We had subsequently shown
12
that IMPROVE IT, a trial
assessing the incremental cardiovascular benefit of adding the non-statin agent ezetimibe to statin therapy 22, fitted very well the regression line from our previous analysis, offering a first proof of our thesis with a non-statin treatment. Several new lipid-lowering trials have
PT
appeared since that report (see Online Supplement). Most recent trials with PCSK9 inhibitors have made use of drugs with pure effects on LDL (and total) cholesterol, and
RI
with an unprecedented extent of LDL reduction. When examined in their totality, these
SC
results (a) confirm that no special property of any cholesterol lowering intervention has to be invoked to explain the reduction in stroke, fitting a log-linear relationship; (b) allow a
NU
precise estimate of the expected results on stroke in future intervention trials affecting total cholesterol.
MA
We recognize two limits of this report. First, we have run this analysis using total cholesterol and not LDL cholesterol, while the latter that should more directly relate to stroke. We used total cholesterol because this was the only cholesterol-related parameter
D
reported in older trials, and their exclusion would have grossly flawed the entire construct
CE PT E
by excluding earlier interventions with a weaker effect on cholesterol, but also with more varied mechanisms of action. A second limitation is the use of total stroke as an endpoint: again due to the necessity of not excluding older trials, which did not report on adjudicated distinction of ischemic and hemorrhagic strokes, with only the former positively related to total or LDL cholesterol. It has been recently reported that cholesterol lowering might be 23
. If anything, such occurrence
AC
related to an increase in hemorrhagic stroke in women
would have attenuated the direct relationship of total cholesterol with total stroke here reported. An inference from our data is that the favorable effects of cholesterol lowering on ischemic stroke by far outweighs any possible detrimental effect on hemorrhagic stroke overall for most achieved cholesterol levels. Disclosures None.
ACCEPTED MANUSCRIPT REFERENCES
1.
Mihaylova B, Emberson J, Blackwell L, Keech A, Simes J, Barnes EH, Voysey M, Gray A, Collins R, Baigent C. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet. 2012;380:581-590 Di Napoli P, Taccardi AA, Oliver M, De Caterina R. Statins and stroke: evidence for
PT
2.
cholesterol-independent effects. Eur Heart J. 2002;23:1908-1921 Amarenco P, Lavallee P, Touboul PJ. Stroke prevention, blood cholesterol, and
RI
3.
statins. Lancet Neurol. 2004;3:271-278
Sabatine MS, Giugliano RP, Keech AC, Honarpour N, Wiviott SD, Murphy SA,
SC
4.
Kuder JF, Wang H, Liu T, Wasserman SM, Sever PS, Pedersen TR, for the
NU
FOURIER Steering Committee Investigators. Evolocumab and Clinical Outcomes in Patients with Cardiovascular Disease. N Engl J Med. 2017;376:1713-1722 Ridker PM, Revkin J, Amarenco P, Brunell R, Curto M, Civeira F, Flather M, Glynn
MA
5.
RJ, Gregoire J, Jukema JW, Karpov Y, Kastelein JJP, Koenig W, Lorenzatti A, Manga P, Masiukiewicz U, Miller M, Mosterd A, Murin J, Nicolau JC, Nissen S,
D
Ponikowski P, Santos RD, Schwartz PF, Soran H, White H, Wright RS, Vrablik M,
CE PT E
Yunis C, Shear CL, Tardif J-C. Cardiovascular Efficacy and Safety of Bococizumab in High-Risk Patients. N Engl J Med. 2017;376:1527-1539 6.
Schwartz GG, Steg PG, Szarek M, Bhatt DL, Bittner VA, Diaz R, Edelberg JM, Goodman SG, Hanotin C, Harrington RA, Jukema JW, Lecorps G, Mahaffey KW, Moryusef A, Pordy R, Quintero K, Roe MT, Sasiela WJ, Tamby J-F, Tricoci P, White
AC
HD, Zeiher AM. Alirocumab and Cardiovascular Outcomes after Acute Coronary Syndrome. N Engl J Med. 2018;379:2097-2107 7.
Marchioli R, Marfisi RM, Carinci F, Tognoni G. Meta-analysis, clinical trials, and transferability of research results into practice. The case of cholesterol-lowering interventions in the secondary prevention of coronary heart disease. Arch Intern Med. 1996;156:1158-1172
8.
Marchioli R, Marfisi RM. [Primary and secondary prevention of ischemic cardiopathy. A meta-analysis with meta-regression as a biologic mechanism: reflections on clinical practice based on scientific evidence]. G Ital Cardiol. 1998;28:434-459
ACCEPTED MANUSCRIPT 9.
Di Mascio R, Marchioli R, Tognoni G. Cholesterol reduction and stroke occurrence: an overview of randomized clinical trials. Cerebrovasc Dis. 2000;10:85-92
10.
Lewington S, Whitlock G, Clarke R, Sherliker P, Emberson J, Halsey J, Qizilbash N, Peto R, Collins R. Blood cholesterol and vascular mortality by age, sex, and blood pressure: a meta-analysis of individual data from 61 prospective studies with 55,000 vascular deaths. Lancet. 2007;370:1829-1839
11.
De Caterina R, Scarano M, Marfisi R, Lucisano G, Palma F, Tatasciore A, Marchioli
PT
R. Cholesterol-lowering interventions and stroke: insights from a meta-analysis of randomized controlled trials. J Am Coll Cardiol. 2010;55:198-211 De Caterina R, Salvatore T, Marchioli R. Cholesterol-lowering interventions and
RI
12.
13.
SC
stroke: Insights from IMPROVE-IT. Atherosclerosis. 2016;248:216-218 Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic
14.
NU
reviews and meta-analyses: the PRISMA statement. PLoS Med. 2009;6:e1000097 The Coronary Drug Project Research Group. The Coronary Drug Project. Initial findings leading to modifications of its research protocol. JAMA. 1970;214:1303-
15.
MA
1313
The Coronary Drug Project Research Group. Clofibrate and niacin in coronary heart
16.
D
disease. JAMA. 1975;231:360-381
The Coronary Drug Project Research Group. A co-operative trial in the primary
CE PT E
prevention of ischaemic heart disease using clofibrate. Report from the Committee of Principal Investigators. Br Heart J. 1978;40:1069-1118 17.
Canner PL, Berge KG, Wenger NK, Stamler J, Friedman L, Prineas RJ, Friedewald W. Fifteen year mortality in Coronary Drug Project patients: long-term benefit with niacin. J Am Coll Cardiol. 1986;8:1245-1255 Brown G, Albers JJ, Fisher LD, Schaefer SM, Lin JT, Kaplan C, Zhao XQ, Bisson
AC
18.
BD, Fitzpatrick VF, Dodge HT. Regression of coronary artery disease as a result of intensive lipid-lowering therapy in men with high levels of apolipoprotein B. N Engl J Med. 1990;323:1289-1298 19.
Sawayama Y, Shimizu C, Maeda N, Tatsukawa M, Kinukawa N, Koyanagi S, Kashiwagi S, Hayashi J. Effects of probucol and pravastatin on common carotid atherosclerosis in patients with asymptomatic hypercholesterolemia. Fukuoka Atherosclerosis Trial (FAST). J Am Coll Cardiol. 2002;39:610-616
ACCEPTED MANUSCRIPT 20.
Mohler ER, 3rd, Hiatt WR, Creager MA. Cholesterol reduction with atorvastatin improves walking distance in patients with peripheral arterial disease. Circulation. 2003;108:1481-1486
21.
Thompson SG, Higgins JP. How should meta-regression analyses be undertaken and interpreted? Stat Med. 2002;21:1559-1573
22.
Cannon CP, Blazing MA, Giugliano RP, McCagg A, White JA, Theroux P, Darius H, Lewis BS, Ophuis TO, Jukema JW, De Ferrari GM, Ruzyllo W, De Lucca P, Im K,
PT
Bohula EA, Reist C, Wiviott SD, Tershakovec AM, Musliner TA, Braunwald E, Califf RM. Ezetimibe Added to Statin Therapy after Acute Coronary Syndromes. N Engl J
SC
Rist PM, Buring JE, Ridker PM, Kase CS, Kurth T, Rexrode KM. Lipid levels and
CE PT E
D
MA
NU
the risk of hemorrhagic stroke among women. Neurology. 2019;92:e2286-e2294
AC
23.
RI
Med. 2015;372:2387-2397
ACCEPTED MANUSCRIPT FIGURE LEGEND
Fig. 1: Meta-Regression of Relationship Between Percent Change in Total Cholesterol Achieved in Cholesterol-Lowering Intervention Trials and the Relative Risk of Stroke: Meta-regression of the relationship between the percent change in total
PT
cholesterol (Delta TC) achieved in various cholesterol-lowering intervention trials (on the abscissa) and the relative risk (RR) of stroke (on the ordinate), on a logarithmic (ln) scale.
RI
The single circles represent individual trials as the primary source of data, where light green dots represent trials with diet, light blue circles represent trials with statins, dark
SC
green circles represent trials with fibrates, deep blue circles are trials with CETP inhibitors, black circles are trials with other drugs, gray circles represent trials with PCSK9 inhibitors,
NU
and other colors represent trials with surgery and other interventions. The size each dot is proportional to the number of subjects involved in the trial. Labels specifically indicate the
MA
FOURIER, SPIRE 1/2 and ODISSEY OUTCOMES trials. Labels of the other dots can be inferred by corresponding values of the x and y axis as reported in Online Table 1. The graph shows the regression line and its upper and lower 95% confidence limits. The
D
shaded area in the graph denotes the area corresponding to a RR below the point where
CE PT E
this becomes statistically significant in the meta-regression, with 95% boundaries detaching from the unity line. The significance of the regression is indicated in the upper inset. CI denotes confidence interval. Updated from Ref.
AC
appeared in recent years.
11
, with the inclusion of new trials
ACCEPTED MANUSCRIPT Table 1: Total cholesterol and stroke in the randomization groups in FOURIER, SPIRE-1/2 and ODISSEY OUTCOMES
Total cholesterol at study end mg/dL
Total stroke
Δ%
n
Δ%
35.7
262 207 55
21,0
Observed Total stroke RR in the original study publication
Predicted Total stroke RR in 11
Predicted Total stroke RR with the new equation*
FOURIER placebo evolocumab Δ = placebo - evolocumab RR
168 108 60
SPIRE-1 and -2 5 placebo bococizumab Δ = placebo - bococizumab RR
164 105 59
0.76
0.79
0.60
0.76
0.79
0.79
0.85
0.84
MA
21.6
168 120 48
28,6
D
185 145 40
40
NU
36
75 45 30
SC
RI
0.79
ODISSEY OUTCOMES 6 placebo alirocumab Δ = placebo - alirocumab RR
PT
4
CE PT E
Abbreviations; HR = hazard ratio; RR = relative risk ln (TOTAL STROKE RR) = 0,00518 - 0,00793 x (% total cholesterol reduction)
AC
New equation: LnRR = - 0.061 – 0.005*(% total cholesterol reduction)
ACCEPTED MANUSCRIPT
CLINICAL SIGNIFICANCE
Among lipid-lowering treatments, stroke reduction has been observed with statins, and can be attributed to interference with the mevalonate pathway.
In a meta-regression of total cholesterol changes with various treatments (including PCSK9 inhibitors) vs changes in stroke compared with control, a reduction of total cholesterol conforms to a Cholesterol lowering proportionally reduces stroke risk, without the need of invoking “pleiotropic”
CE PT E
D
MA
NU
SC
RI
effects of any such treatment.
AC
PT
log-linear relationship with stroke reduction.
Figure 1
Tanya Salvatore, Riccardo Morganti, Roberto Marchioli, Raffaele De Caterina PII: DOI: Reference:
S0002-9343(19)30564-9 https://doi.org/10.1016/j.amjmed.2019.06.029 AJM 15267
To appear in:
The American Journal of Medicine
Please cite this article as: T. Salvatore, R. Morganti, R. Marchioli, et al., Cholesterol lowering and stroke: no longer room for pleiotropic effects of statins – confirmation from PCSK9 inhibitor studies, The American Journal of Medicine, https://doi.org/10.1016/ j.amjmed.2019.06.029
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Cholesterol lowering and stroke: no longer room for pleiotropic effects of statins – confirmation from PCSK9 inhibitor studies Short title: Cholesterol lowering and stroke Tanya Salvatorea, MD, Riccardo Morganti, PhDb, Roberto Marchioli, MDc,
PT
and Raffaele De Caterina, MD, PhDb, d Institute of Cardiology, “G. d’Annunzio” University, Chieti, Italy
b
Section of Statistics, Azienda Ospedaliero-Universitaria Pisana, Pisa University Hospital,
RI
a
Pisa, Italy
CV, Metabolic & Renal Disease, Medical & Scientific Services, IQVIA, Milan, Italy
d
Chair of Cardiology, University of Pisa, Pisa, Italy
NU
SC
c
MA
Word count: 3261 – 1 Table – 1 Figure – 1 Online Supplement
Funding sources related to this manuscript: none Conflict of interest statement for all authors: none
Correspondence:
CE PT E
D
All authors had access to the data and a role in writing the manuscript.
Prof. Raffaele De Caterina, MD, PhD
Chair of Cardiology, University of Pisa, and Cardiovascular Division, Pisa University Hospital
AC
Via Paradisa 2 56124 Pisa, Italy
Tel.: +39-050-996-751
E-mail: [email protected]
ACCEPTED MANUSCRIPT ABSTRACT
Background: The relationship between cholesterol levels and stroke has been much less clear than between cholesterol levels and coronary heart disease. This is likely mostly due to the inadequate power of older studies and the low intensity of cholesterol-lowering interventions at that time available. Because a reduction in stroke has been, conversely, clearly observed in trials with statins, for long “pleiotropic” effects of such drugs, unrelated
PT
to cholesterol lowering, have been invoked. In a previous analysis of all randomized trials of cholesterol-lowering treatments reporting on stroke we had, however, reached the
RI
conclusion that any cholesterol lowering is related to a significant reduction of stroke, in a
SC
relationship that appeared to exist for both statin and non-statin cholesterol-lowering interventions. Outcome results of the FOURIER trial with evolocumab, SPIRE-1 and -2
NU
with bococizumab, and ODISSEY OUTCOMES trial with alirocumab now offer the opportunity of clearly confirming or confuting this concept.
MA
Methods: We here report on an updated meta-regression of the relationship of total cholesterol changes occurred with various drugs or treatments and changes in the risk of
D
stroke compared with control.
CE PT E
Results: Figures of the RR here found in FOURIER, SPIRE-1/2 and ODISSEY OUTCOMES (0.79, 0.60 and 0.79) are extremely close to the RR of 0.79, 0.79 and 0.84 predicted by our new meta-regression, respectively. Conclusions: These findings offer a definitive proof that the pure total (and low-density lipoprotein) cholesterol lowering, with any available lipid-lowering intervention, reduce
AC
stroke risk proportional to the extent of cholesterol reduction, without the need of invoking “pleiotropic” effects of any such treatment.
Graphical abstract
Key words: cholesterol; stroke; atherosclerosis; lipid-lowering; statins; PCSK9 inhibitors; ezetimibe.
ACCEPTED MANUSCRIPT INTRODUCTION Hypercholesterolemia is a powerful risk factor for coronary heart disease 1. Cholesterollowering drugs and non-pharmacological treatments significantly reduce morbidity from coronary heart disease, thus proving a causal role for cholesterol in coronary events 1. Conversely, the relationship between cholesterol levels and stroke has been much less clear. Trials with statins have shown decreased stroke incidence in treated populations 1, 2,
PT
but this had not been observed with non-statin drugs or treatments 2, supporting the concept that such effects are attributable to statin-related “pleiotropism” – cholesterol-
RI
independent (neuro)protective properties related to the interference by statins with the
SC
mevalonate pathway 3.
As of today, results from the recently published FOURIER 4, SPIRE 1/2
5
and ODYSSEY
NU
OUTCOMES 6 trials allow a testing of this hypothesis because the three drugs here used – evolocumab, bococizumab and alirocumab – reduce total and low-density lipoprotein (LDL) cholesterol by a mechanism totally independent of the mevalonate pathway and to a
MA
much more substantial extent compared with previous non-statin treatments. We here therefore now present a new comprehensive meta-regression of all reported
D
cholesterol-lowering trials, (a) providing an updated meta-regression of such relationship; and (b) reporting an updated equation allowing to predict the extent of total stroke
METHODS
AC
Literature search
CE PT E
reduction per each definite change in total cholesterol.
To retrieve published randomized controlled trials (RCT), we used the database of studies developed for our reviews on cholesterol-lowering interventions, stroke and secondary prevention of coronary heart disease
2, 7-12
. The database was updated using the
commonly adopted approach of computer-aided literature search of the Pubmed database. The new search was limited to randomized controlled trials performed in humans published between January 2000 to April 2019, using the same key terms adopted for our previous publications plus those for new lipid-lowering drugs (i.e., (Random allocation) OR (clinical trial)) AND ((3-hydroxy-3-methylglutaryl-coenzyme-A reductase inhibitor) OR (HMG CoA reductase inhibitor) OR (statin) OR (simvastatin) OR
ACCEPTED MANUSCRIPT (fluvastatin) OR (atorvastatin) OR (pravastatin) OR (rosuvastatin) OR (lovastatin) OR (colestipol) OR (niacin) OR (gemfibrozil) OR (bezafibrate) OR (probucol) OR (fibrates) OR (diet) OR (ciprofibrate) OR (etofibrate) OR (nicotinic acid OR (acipimox) OR (clofibrate) OR (fenofibrate) OR (colestyramine) OR (ezetimibe) OR (CETP inhibitor) OR (cholesterol ester transfer protein inhibitor) OR (PCSK9 inhibitor). Then titles and abstracts were screened to check if the publications met the following inclusion criteria: a) outcome including the incidence of death; b) randomized study design comparing cholesterol-
PT
lowering treatment with placebo or no treatment; c) inclusion of at least 100 patients; d) follow-up duration of at least 6 months. If more than one publication was found referring to
RI
a single study, all publications were used to compile the most complete dataset. We also
SC
performed a manual search of references cited by the published original studies and recent, relevant review articles.
NU
Potentially relevant studies were reviewed independently by 2 investigators (T.S. and R.D.C.) in duplicate to determine whether they met eligibility criteria for inclusion. Where discrepancies between investigators occurred for inclusion or exclusion, a third
MA
investigator (R.M) was involved to perform an additional evaluation, and discrepancies
D
were resolved by discussion.
CE PT E
Criteria for trial selection and data abstraction
Studies were eligible for inclusion if: (1) the study design was a RCT; (2) the intervention consisted of a single treatment compared with placebo or no treatment; (3) the number of patients included in the study was greater than 100; (4) the intervention duration was at least 6 months; and (5) the number of events occurring during the study were reported by
AC
intervention and control groups. We excluded studies with a crossover design. Although PRISMA guidelines
13
were not formally followed and declared at the time of our former
meta-analyses, criteria used for checking the search accuracy and data reporting were substantially similar. Placebo-controlled trials are less prone to bias than open-conducted studies, but dietary interventions, typical and frequent side effects of cholesterol-lowering drugs, and blinding of laboratory records often make the adoption of a placebo-controlled design unfeasible or at least problematic. Both open and blinded trials were therefore accepted. In multipleintervention trials and in the case of prolongation of follow-up after the end of the study, it is not possible to clearly attribute differences in outcome between the intervention and
ACCEPTED MANUSCRIPT control groups to the actions of any specific treatment. Therefore multi-factorial trials and studies that reported data that were gathered beyond the planned trial duration were excluded. Whenever possible, intention-to-treat analysis data were used. Some RCT had randomized 2 active treatments and a single control group 14-20. For these cases, each arm was considered as a separate one, and the data of control groups were used twice. All data were independently abstracted in duplicate by 2 investigators (T.S. and R.D.C.) using a standardized data collection form. Discrepancies were resolved by discussion with
PT
a third investigator (R.M.) and by referencing the original report. We did not contact
RI
authors to request additional information.
SC
The key data elements sought from each trial were categorized as study characteristics, patients characteristics, treatment characteristics, and clinical outcomes. Specifically, the
NU
characteristics considered for each study were: the name of the first author, journal and year of publication, the duration of follow-up, the number of patients enrolled, the type of intervention, and drug dosages. For patients, the following variables were collected: sex;
MA
mean age; total, fatal, and non-fatal strokes; the means of total cholesterol, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, and triglycerides at
D
baseline, at the end of study and during the follow-up. When the final value of serum lipids was not reported, an estimate was calculated as the mean of the observed values during
CE PT E
the follow-up. As to co-morbidities, information on gender, age, and main cardiovascular risk factors (e.g., history of smoking, diabetes mellitus, and hypertension) was collected. Previous cardiovascular events and pharmacological treatment were also recorded. The clinical outcomes considered were the incidence of total, fatal, and non-fatal stroke, as
AC
defined in the original reports of the trials. Statistical methods
We calculated estimates of the average effect and 95% confidence intervals (95% CI) of statins, fibrates and other cholesterol-lowering interventions on serum lipids estimating the relative risk (RR) to assess the effect of the interventions on the risk of stroke of the population recruited in each trial. To explore the relationship of cholesterol reduction and total stroke, we performed a metaregression by using inverse variance-weighted linear regression
21
. We used the logarithm
RR for total stroke as the dependent variable in the model against the aforementioned
ACCEPTED MANUSCRIPT variables, and weights in each study were the reciprocals of the variances for the logarithm RR for stroke
11
. In each trial, we calculated the percent reduction in total serum
cholesterol levels (%ΔTC) by subtracting end-study (or mean in-study) TC from baseline TC in treated and control groups:
%TC
TCf TCb x100 TCb
where TCf was end-study (or mean in-study) total serum cholesterol, and TCb was
PT
baseline total serum cholesterol.
SC
RI
We conducted statistical testing for efficacy at a 2-tailed α-level of 0.05.
NU
RESULTS
MA
Data on total stroke from the FOURIER, SPIRE 1/2 and ODISSEY OUTCOMES are presented in Table 1. A new meta-regression of all cholesterol-lowering trials vs the observed risk reduction of stroke is presented in Figure 1, The new equation to predict the
CE PT E
D
saving of strokes as a function of total cholesterol lowering is now: LnRR = - 0.061 – 0.005*(% cholesterol reduction)
According to this equation, one should expect a RR of 0.851, 0.810 and 0.770 for a 20%,
AC
30% and 40% reduction in total cholesterol, respectively.
DISCUSSION
We here show that the reduction in stroke by cholesterol lowering fits a log-linear relationship with the reduction in cholesterol levels. These data quite explicitly indicate that stroke reduction is a most likely consequence of cholesterol lowering. They would thus now offer a quite definitive proof that no special property of any cholesterol-lowering intervention has to be invoked to explain stroke reduction previously observed only in statin-intervention trials. These data also allow a precise estimate of the expected results on stroke in future trials affecting total cholesterol.
ACCEPTED MANUSCRIPT
We had already reported such a significant relationship, with no significant association with changes of high-density lipoprotein cholesterol levels, and inconsistent associations with reduction of triglycerides
11
. We had subsequently shown
12
that IMPROVE IT, a trial
assessing the incremental cardiovascular benefit of adding the non-statin agent ezetimibe to statin therapy 22, fitted very well the regression line from our previous analysis, offering a first proof of our thesis with a non-statin treatment. Several new lipid-lowering trials have
PT
appeared since that report (see Online Supplement). Most recent trials with PCSK9 inhibitors have made use of drugs with pure effects on LDL (and total) cholesterol, and
RI
with an unprecedented extent of LDL reduction. When examined in their totality, these
SC
results (a) confirm that no special property of any cholesterol lowering intervention has to be invoked to explain the reduction in stroke, fitting a log-linear relationship; (b) allow a
NU
precise estimate of the expected results on stroke in future intervention trials affecting total cholesterol.
MA
We recognize two limits of this report. First, we have run this analysis using total cholesterol and not LDL cholesterol, while the latter that should more directly relate to stroke. We used total cholesterol because this was the only cholesterol-related parameter
D
reported in older trials, and their exclusion would have grossly flawed the entire construct
CE PT E
by excluding earlier interventions with a weaker effect on cholesterol, but also with more varied mechanisms of action. A second limitation is the use of total stroke as an endpoint: again due to the necessity of not excluding older trials, which did not report on adjudicated distinction of ischemic and hemorrhagic strokes, with only the former positively related to total or LDL cholesterol. It has been recently reported that cholesterol lowering might be 23
. If anything, such occurrence
AC
related to an increase in hemorrhagic stroke in women
would have attenuated the direct relationship of total cholesterol with total stroke here reported. An inference from our data is that the favorable effects of cholesterol lowering on ischemic stroke by far outweighs any possible detrimental effect on hemorrhagic stroke overall for most achieved cholesterol levels. Disclosures None.
ACCEPTED MANUSCRIPT REFERENCES
1.
Mihaylova B, Emberson J, Blackwell L, Keech A, Simes J, Barnes EH, Voysey M, Gray A, Collins R, Baigent C. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet. 2012;380:581-590 Di Napoli P, Taccardi AA, Oliver M, De Caterina R. Statins and stroke: evidence for
PT
2.
cholesterol-independent effects. Eur Heart J. 2002;23:1908-1921 Amarenco P, Lavallee P, Touboul PJ. Stroke prevention, blood cholesterol, and
RI
3.
statins. Lancet Neurol. 2004;3:271-278
Sabatine MS, Giugliano RP, Keech AC, Honarpour N, Wiviott SD, Murphy SA,
SC
4.
Kuder JF, Wang H, Liu T, Wasserman SM, Sever PS, Pedersen TR, for the
NU
FOURIER Steering Committee Investigators. Evolocumab and Clinical Outcomes in Patients with Cardiovascular Disease. N Engl J Med. 2017;376:1713-1722 Ridker PM, Revkin J, Amarenco P, Brunell R, Curto M, Civeira F, Flather M, Glynn
MA
5.
RJ, Gregoire J, Jukema JW, Karpov Y, Kastelein JJP, Koenig W, Lorenzatti A, Manga P, Masiukiewicz U, Miller M, Mosterd A, Murin J, Nicolau JC, Nissen S,
D
Ponikowski P, Santos RD, Schwartz PF, Soran H, White H, Wright RS, Vrablik M,
CE PT E
Yunis C, Shear CL, Tardif J-C. Cardiovascular Efficacy and Safety of Bococizumab in High-Risk Patients. N Engl J Med. 2017;376:1527-1539 6.
Schwartz GG, Steg PG, Szarek M, Bhatt DL, Bittner VA, Diaz R, Edelberg JM, Goodman SG, Hanotin C, Harrington RA, Jukema JW, Lecorps G, Mahaffey KW, Moryusef A, Pordy R, Quintero K, Roe MT, Sasiela WJ, Tamby J-F, Tricoci P, White
AC
HD, Zeiher AM. Alirocumab and Cardiovascular Outcomes after Acute Coronary Syndrome. N Engl J Med. 2018;379:2097-2107 7.
Marchioli R, Marfisi RM, Carinci F, Tognoni G. Meta-analysis, clinical trials, and transferability of research results into practice. The case of cholesterol-lowering interventions in the secondary prevention of coronary heart disease. Arch Intern Med. 1996;156:1158-1172
8.
Marchioli R, Marfisi RM. [Primary and secondary prevention of ischemic cardiopathy. A meta-analysis with meta-regression as a biologic mechanism: reflections on clinical practice based on scientific evidence]. G Ital Cardiol. 1998;28:434-459
ACCEPTED MANUSCRIPT 9.
Di Mascio R, Marchioli R, Tognoni G. Cholesterol reduction and stroke occurrence: an overview of randomized clinical trials. Cerebrovasc Dis. 2000;10:85-92
10.
Lewington S, Whitlock G, Clarke R, Sherliker P, Emberson J, Halsey J, Qizilbash N, Peto R, Collins R. Blood cholesterol and vascular mortality by age, sex, and blood pressure: a meta-analysis of individual data from 61 prospective studies with 55,000 vascular deaths. Lancet. 2007;370:1829-1839
11.
De Caterina R, Scarano M, Marfisi R, Lucisano G, Palma F, Tatasciore A, Marchioli
PT
R. Cholesterol-lowering interventions and stroke: insights from a meta-analysis of randomized controlled trials. J Am Coll Cardiol. 2010;55:198-211 De Caterina R, Salvatore T, Marchioli R. Cholesterol-lowering interventions and
RI
12.
13.
SC
stroke: Insights from IMPROVE-IT. Atherosclerosis. 2016;248:216-218 Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic
14.
NU
reviews and meta-analyses: the PRISMA statement. PLoS Med. 2009;6:e1000097 The Coronary Drug Project Research Group. The Coronary Drug Project. Initial findings leading to modifications of its research protocol. JAMA. 1970;214:1303-
15.
MA
1313
The Coronary Drug Project Research Group. Clofibrate and niacin in coronary heart
16.
D
disease. JAMA. 1975;231:360-381
The Coronary Drug Project Research Group. A co-operative trial in the primary
CE PT E
prevention of ischaemic heart disease using clofibrate. Report from the Committee of Principal Investigators. Br Heart J. 1978;40:1069-1118 17.
Canner PL, Berge KG, Wenger NK, Stamler J, Friedman L, Prineas RJ, Friedewald W. Fifteen year mortality in Coronary Drug Project patients: long-term benefit with niacin. J Am Coll Cardiol. 1986;8:1245-1255 Brown G, Albers JJ, Fisher LD, Schaefer SM, Lin JT, Kaplan C, Zhao XQ, Bisson
AC
18.
BD, Fitzpatrick VF, Dodge HT. Regression of coronary artery disease as a result of intensive lipid-lowering therapy in men with high levels of apolipoprotein B. N Engl J Med. 1990;323:1289-1298 19.
Sawayama Y, Shimizu C, Maeda N, Tatsukawa M, Kinukawa N, Koyanagi S, Kashiwagi S, Hayashi J. Effects of probucol and pravastatin on common carotid atherosclerosis in patients with asymptomatic hypercholesterolemia. Fukuoka Atherosclerosis Trial (FAST). J Am Coll Cardiol. 2002;39:610-616
ACCEPTED MANUSCRIPT 20.
Mohler ER, 3rd, Hiatt WR, Creager MA. Cholesterol reduction with atorvastatin improves walking distance in patients with peripheral arterial disease. Circulation. 2003;108:1481-1486
21.
Thompson SG, Higgins JP. How should meta-regression analyses be undertaken and interpreted? Stat Med. 2002;21:1559-1573
22.
Cannon CP, Blazing MA, Giugliano RP, McCagg A, White JA, Theroux P, Darius H, Lewis BS, Ophuis TO, Jukema JW, De Ferrari GM, Ruzyllo W, De Lucca P, Im K,
PT
Bohula EA, Reist C, Wiviott SD, Tershakovec AM, Musliner TA, Braunwald E, Califf RM. Ezetimibe Added to Statin Therapy after Acute Coronary Syndromes. N Engl J
SC
Rist PM, Buring JE, Ridker PM, Kase CS, Kurth T, Rexrode KM. Lipid levels and
CE PT E
D
MA
NU
the risk of hemorrhagic stroke among women. Neurology. 2019;92:e2286-e2294
AC
23.
RI
Med. 2015;372:2387-2397
ACCEPTED MANUSCRIPT FIGURE LEGEND
Fig. 1: Meta-Regression of Relationship Between Percent Change in Total Cholesterol Achieved in Cholesterol-Lowering Intervention Trials and the Relative Risk of Stroke: Meta-regression of the relationship between the percent change in total
PT
cholesterol (Delta TC) achieved in various cholesterol-lowering intervention trials (on the abscissa) and the relative risk (RR) of stroke (on the ordinate), on a logarithmic (ln) scale.
RI
The single circles represent individual trials as the primary source of data, where light green dots represent trials with diet, light blue circles represent trials with statins, dark
SC
green circles represent trials with fibrates, deep blue circles are trials with CETP inhibitors, black circles are trials with other drugs, gray circles represent trials with PCSK9 inhibitors,
NU
and other colors represent trials with surgery and other interventions. The size each dot is proportional to the number of subjects involved in the trial. Labels specifically indicate the
MA
FOURIER, SPIRE 1/2 and ODISSEY OUTCOMES trials. Labels of the other dots can be inferred by corresponding values of the x and y axis as reported in Online Table 1. The graph shows the regression line and its upper and lower 95% confidence limits. The
D
shaded area in the graph denotes the area corresponding to a RR below the point where
CE PT E
this becomes statistically significant in the meta-regression, with 95% boundaries detaching from the unity line. The significance of the regression is indicated in the upper inset. CI denotes confidence interval. Updated from Ref.
AC
appeared in recent years.
11
, with the inclusion of new trials
ACCEPTED MANUSCRIPT Table 1: Total cholesterol and stroke in the randomization groups in FOURIER, SPIRE-1/2 and ODISSEY OUTCOMES
Total cholesterol at study end mg/dL
Total stroke
Δ%
n
Δ%
35.7
262 207 55
21,0
Observed Total stroke RR in the original study publication
Predicted Total stroke RR in 11
Predicted Total stroke RR with the new equation*
FOURIER placebo evolocumab Δ = placebo - evolocumab RR
168 108 60
SPIRE-1 and -2 5 placebo bococizumab Δ = placebo - bococizumab RR
164 105 59
0.76
0.79
0.60
0.76
0.79
0.79
0.85
0.84
MA
21.6
168 120 48
28,6
D
185 145 40
40
NU
36
75 45 30
SC
RI
0.79
ODISSEY OUTCOMES 6 placebo alirocumab Δ = placebo - alirocumab RR
PT
4
CE PT E
Abbreviations; HR = hazard ratio; RR = relative risk ln (TOTAL STROKE RR) = 0,00518 - 0,00793 x (% total cholesterol reduction)
AC
New equation: LnRR = - 0.061 – 0.005*(% total cholesterol reduction)
ACCEPTED MANUSCRIPT
CLINICAL SIGNIFICANCE
Among lipid-lowering treatments, stroke reduction has been observed with statins, and can be attributed to interference with the mevalonate pathway.
In a meta-regression of total cholesterol changes with various treatments (including PCSK9 inhibitors) vs changes in stroke compared with control, a reduction of total cholesterol conforms to a Cholesterol lowering proportionally reduces stroke risk, without the need of invoking “pleiotropic”
CE PT E
D
MA
NU
SC
RI
effects of any such treatment.
AC
PT
log-linear relationship with stroke reduction.
Figure 1