Assessment of Vascular Event Prevention and Cognitive Function Among Older Adults With Preexisting Vascular Disease or Diabetes

Key Points Question Do null results on cognitive function in cardiovascular trials exclude worthwhile benefit? Findings In this secondary analysis of 3 randomized clinical trials including 45 029 participants undergoing cognitive assessment, the prevention of nonfatal cardiovascular events in 4.5% of survivors in the Heart Protection Study, by randomization to statin, yielded an estimated cognitive function difference equivalent to avoiding 0.15 years of aging. By contrast, the trial was powered to detect a difference in cognitive aging of at least 1 year. Meaning Nonsignificant findings, even from large trials, should not be taken as good evidence of a lack of worthwhile benefit on cognitive function of prolonged use of cardioprotective therapies.


Supplementary Methods eAppendix 1: Evidence Before This Study
Evidence from randomized trials of the effects of cardiovascular event prevention with bloodpressure-lowering, lipid-lowering, or antiplatelet therapy on cognitive function or dementia was systematically reviewed. Review was restricted to cardiovascular outcome trials that successfully prevented cardiovascular events, defined as having a statistically significant effect on their primary cardiovascular outcome or on non-fatal stroke or non-fatal myocardial infarction (MI) presented in the main report of the trial, and published before 30 th August 2017. Trials specifically in demented populations were excluded. For lipid-lowering and blood-pressure lowering, trials were identified from recent systematic reviews [1][2][3][4][5] of the effects of such therapies on cognitive function, together with expert review by the authors to add any more recent such trials (which would be well known). For antiplatelet therapy, we undertook a search restricting to randomized trials, using the terms 'aspirin' or 'antiplatelet*' in titles or abstracts combined with cogniti*, dement* or memory in titles or abstracts or the MeSH term cognition. Relevant reports were searched for mention of additional studies before restricting to cardiovascular outcome studies as for the other therapies. For blood pressure-lowering therapy, from 11 trials in three reviews, seven met the outcome criteria; [6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] for lipid lowering therapy from 15 trials in two reviews plus five trials from expert review, five met the outcome criteria; [21][22][23][24][25][26] for antiplatelet therapy, from 43 initial hits, two trials met the outcome criteria. [27][28][29][30] Overall, most trials failed to show a benefit of therapy on cognition (eTable 1). The two largest trials of cognitive function (about 25 000 survivors in the REVEAL trial of anacetrapib and 15 926 survivors in the Heart Protection Study of statin therapy) failed to show such benefit, but reductions in non-fatal events were only 0·7% for MI and <0·1% for stroke in REVEAL, and 2·2 % for MI and 1·1 % for stroke in HPS. The greatest reductions in non-fatal events (4-5%) were observed in the blood pressure-lowering trials, some of which did observe a statistically significant effect on a cognitive measure. However, the evidence was inconclusive overall, as concluded in a recent review on the impact of hypertension on cognitive function. 31

eAppendix 2: Further Details of the Study Design
Full details of the designs of the Heart Protection Study (HPS), SEARCH, and HPS2-THRIVE studies, including their protocols [32][33][34] , have been reported previously. 21,22,[34][35][36][37] Patients were eligible for the HPS randomized trial if they had a non-fasting blood total cholesterol concentrations of at least 135mg/dL (3.5 mmol/L) and a history of coronary disease, ischemic stroke, other occlusive disease of the non-coronary arteries, diabetes mellitus, or (if men aged ≥65 years) treated hypertension, and were aged between 40 and 80 at randomization. HPS randomized 20 536 participants in the UK to 40 mg simvastatin daily or placebo, and separately in a 2x2 factorial design to antioxidant vitamins or placebo.
Patients were eligible for SEARCH if they had a history of myocardial infarction and were aged between 18 and 80 at randomization. SEARCH, randomized 12 064 participants in the UK to 80 versus 20 mg simvastatin daily, and separately in a 2x2 factorial design to folic acid plus vitamin B12 or placebo. Patients were eligible for HPS2-THRIVE if they had a history of MI, cerebrovascular disease, peripheral arterial disease, or diabetes mellitus with evidence of symptomatic coronary disease and were aged between 50 and 80 at randomization. In HPS2-THRIVE, 14 741 European and 10 932 Chinese participants were to receive 2 g niacin with 40 mg laropiprant daily versus placebo (on top of 40 mg simvastatin daily plus, when required, 10 mg ezetimibe) (eFigure 1). Baseline data recorded prior to randomization in each study included age, sex, smoking and alcohol use, history of prior disease, current medication use, height, weight, systolic and diastolic blood pressure and blood measurements of lipids, lipoproteins and creatinine.
Participants in all three trials were followed up for the incidence of events with follow-up visits every six months after the first year, and more frequently during the first year.
Participants in the HPS, SEARCH and HPS2-THRIVE were followed up for a mean of 5·0, 6·7 and 3·9 years, respectively. Information was recorded at each visit about any serious adverse event. Where patients did not attend a follow-up visit they were followed up by telephone or through their general practitioners (GPs). Further information was sought from GPs or hospital records and events were coded by co-ordinating center clinicians blinded to the treatment allocation in the trial. Severity of stroke was classified as mild where there seemed to be no help needed with everyday activities. 21,22,[34][35][36][37] For the analyses presented here, in HPS and SEARCH, onset of diabetes was defined as any hospitalisation for diabetes or taking oral hypoglycaemic mediation or insulin at any time during the trial (in SEARCH) or at the final visit (in HPS) in patients not diabetic at randomization. In HPS2-THRIVE, diabetes at baseline was defined as: self-report or baseline plasma glucose ≥200 mg/dL if fasted <8 hours or ≥126 mg/dL if fasted ≥8 hours, or baseline HbA1c ≥48 mmol/mol, or use of hypoglycaemic medication at randomization. Self-reported diabetes and use of hypoglycaemic medications were recorded directly at each visit. Onset of diabetes was defined as self-report or new use of hypoglycaemic medication.
Approval was obtained from the ethics committees of the participating institutions for each of the studies, and all participants gave written informed consent.

Expression of z-score differences as years of cognitive aging
Cognitive measures typically decline steadily with age at least from about age 60 onwards. 38,39 At younger ages relationships are less consistent and may not reflect aging. In the present study, mean age at cognitive assessment was 64 years and 68% of participants were aged over 60 at cognitive assessment. Therefore the slope of the relationship between the cognitive measures in different participants and their age in those above age 60 (which was linear) was used to express z-scores differences as years of cognitive aging.

Comparisons between simvastatin versus placebo allocated survivors in HPS:
The excess probability of surviving in the simvastatin arm compared to the placebo arm was greater in higher risk participants, resulting in an imbalance in baseline risk factors between the allocated treatment arms amongst survivors to the end of the study. Comparisons of incident events during the trial in survivors were therefore adjusted for age and baseline major vascular event risk (using a previously reported grouping 40 ). Non-fatal events avoided per person with statin were estimated as (1/(odds ratio for the event in the statin versus placebo arm) -1) x proportion with event in the statin arm. In the main analyses in this report , the observed reduction in cognitive aging between the simvastatin and placebo allocated participants in HPS ( Table 2) was adjusted for duration in trial, baseline predictors of cognitive function (for consistency with the analyses of the cognitive aging) and major vascular event risk (to allow for differences between the survivors in the two study arms). In supplementary analyses both unadjusted analyses (as in the original trial report) and adjusted analyses are presented. All analyses were conducted using SAS version 9.4.

Deductions from the Evolocumab trial
In the Further Cardiovascular Outcomes Research with PCSK9 in Subjects with Elevated Risk (FOURIER) report, 42 the treatment difference in the change in z-score for the primary end point had a standard error of about 0.036 (taken from the confidence interval in Figure 2 of the report) among the primary analysis population of 618 allocated placebo and 586 allocated Evolocmab. This is 38% smaller than the expected standard error from the treatment difference in an end z-score (which would be √[1/618+1/586]=0.058, assuming the z-score had standard deviation 1). Step a Factors forced in: Age at entry, per 10  SS=sum of squares, DBP=diastolic blood pressure, CHD=coronary heart disease, N-BNP=N-terminal Pro-B-type natriuretic peptide, HbA1c=glycated haemoglobin, ACE=angiotensin converting enzyme, ARB=angiotensin receptor blocker. SI conversion factors: To convert LDL cholesterol to mmol/L, multiply values by 0.0259, to convert homocysteine to mol/L multiply by 7.40 a Step=0 indicates the factor was forced into the model. b The stepwise model included age in single years as a categorical variable. The effect and SE for age in each study were obtained by running a further regression on the retained variables but with age as a continuous variable. Decline per year over 6

Global cognitive score
Age at test (years) Global cognitive score z-score