Abstract
Background
Functional magnetic resonance imaging (fMRI) has not been used to assess the effects of statins on the brain. We assessed the effect of statins on cognition using standard neuropsychological assessments and brain neural activation with fMRI on two tasks.
Methods
Healthy statin-naïve men and women (48±15 yr) were randomized to 80 mg/day atorvastatin (n=66; 27 men) or placebo (n=84; 48 men) for 6 months. Participants completed cognitive testing while on study drug and 2 months after treatment cessation using alternative test and task versions.
Results
There were few changes in standard neuropsychological tests with drug treatment (all p>0.56). Total and delayed recall from the Hopkins Verbal Learning Test-Revised increased in both groups (p<0.05). The Stroop Color-Word score increased (p<0.01) and the 18-Point Clock Test decreased in the placebo group (p=0.02) after drug cessation. There were, however, small but significant group-time interactions for each fMRI task: participants on placebo had greater activation in the right putamen/dorsal striatum during the maintenance phase of the Sternberg task while on placebo but the effect was reversed after drug washout (p<0.001). Participants on atorvastatin had greater activation in the bilateral precuneus during the encoding phase of the Figural Memory task while on-drug but the effect was reversed after drug washout (p<0.001).
Conclusion
Six months of high dose atorvastatin therapy is not associated with measurable changes in neuropsychological test scores, but did evoke transient differences in brain activation patterns. Larger, longer-term clinical trials are necessary to confirm these findings and evaluate their clinical implications.
Clinical Trial Registration
clinicaltrials.gov; NCT00609063.
Keywords: Functional magnetic resonance imaging, brain activation, statin therapy, central nervous system
Introduction
Hydroxy-methyl-glutaryl (HMG) CoA reductase inhibitors (statins) are the most effective medications for managing elevated concentrations of low-density lipoprotein cholesterol (LDL-C). Resultantly, they are among the most prescribed drugs in the United States and world. Statins are well-tolerated, but mild central nervous system (CNS) symptoms such as memory loss and attention decrements are second to muscle complaints as the most commonly reported adverse effects of these drugs (1, 2).
Studies assessing cognitive effects of statins vary widely and have produced inconclusive findings, in part because many studies have been observational and thereby prone to prescription bias(3, 4). In general, randomized control trials have found little or no effect of statins on cognitive outcomes(5–7), and meta-analyses have also found little evidence of adverse cognitive side effects associated with statins(8, 9). Despite the paucity of data on adverse cognitive side effects with statin therapy, the FDA in 2012 added memory loss and confusion to the safety label for statins, based on case reports (2, 10–12) and data from the Adverse Events Reporting System(13).
Studies involving the effects of statins on cognition have typically assessed cognitive function using traditional cognitive tests, which may yield small effect sizes and demonstrate high intra-participant variability (14). This may explain the discrepancy between clinical trials and patient self-reports, and could be addressed by utilizing CNS tests that directly assess brain parameters. Accordingly, in the current study we used functional magnetic resonance imaging (fMRI) to assess neural activation in adults treated with 80 mg atorvastatin or placebo for six months. We hypothesized that changes in neural activation with high dose atorvastatin treatment would differ from the changes in neural activation with placebo.
Methods
Study Overview
Participants aged ≥ 20 yr were recruited from a randomized, double-blind, placebo-controlled clinical trial (Effect of Statins On Skeletal Muscle Performance (STOMP); NCT00609063) which was designed to assess the incidence of muscle side-effects in 420 healthy, statin-naïve adults (15). Baseline measurements were obtained and participants were randomized in a double-blind fashion to either 80 mg atorvastatin or placebo daily for 6 months as per the published methods of the parent trial (16). The STOMP cognitive sub-study was funded three years after the parent study and thus the study design was based on recruiting participants already in the study, performing on-drug study cognitive assessments at the end of the treatment period, and repeating these measurements after participants had completed a drug washout period. Consequently, participants who agreed to participate in the STOMP cognitive sub-study underwent cognitive testing after 6 months on study drug consisting of paper-based neuropsychological tests and functional magnetic resonance imaging (fMRI) to assess neural activation during two in-scanner tasks. These neuropsychological and fMRI tests were repeated after the participants ceased study drug for two months. Blood lipid levels were assessed at both timepoints. The study was approved by the Institutional Review Board at Hartford Hospital and monitored by an NHLBI-approved Data Safety and Monitoring Board. Funded was provided by NHLBI 1R01HL098085-1. One author (BAT) had full access to all the data in the study and takes responsibility for its integrity and the data analysis. The authors are solely responsible for the design and conduct of this study, all study analyses, the drafting and editing of the paper and its final contents.
Participant Inclusion/Exclusion Criteria
Possible participants (n=293) were screened for the STOMP cognitive sub-study with 91 randomized to placebo and 75 to atorvastatin (Figure 1). Participants in the STOMP study were not included if they had cancer within 5 years, a baseline alanine aminotransferase (ALT) value >2 times the upper limit of normal (ULN), a creatinine level >2 mg/L, an abnormal thyroid-stimulating hormone (TSH) level, a history of cardiovascular disease, or diabetes mellitus. Women who were pregnant or planned to become pregnant were not recruited, nor were individuals treated with medications known to affect skeletal muscle, to alter statin metabolism, or to impact blood cholesterol levels. Participants with hypertension were recruited if their blood pressure was controlled and ≤140/90 mm|Hg at baseline. Additional exclusion criteria specific to the cognitive sub-study included: consumption of > 2 alcoholic drinks per day, reported illicit drug use, and/or history of auditory or visual impairment, past traumatic brain injury, or any neurologic illness that could affect the brain. Participants who reported any form of mental illness (e.g., depression, anxiety, schizophrenia), developmental or learning disorder, or with cognitive impairment determined by the Mini Mental State Exam (MMSE) during the initial screening process (score < 23) were also excluded. Participants were removed from the study at any point of assessment if their creatine kinase (CK) level exceeded 10 times UNL, if their ALT level exceeded 2 times UNL for two measurements within one week or if their TSH level exceeded 5.00 U/L.
Figure 1.
Study Flow Diagram. STOMP cognitive sub-study flow diagram grouped by number of patients participating in screening, randomization, drug allocation, follow-up and analysis.
Neuropsychological Measurements
A trained neuropsychologist (ST) performed the cognitive testing by using the Hopkins Verbal Learning Test-Revised (HVLT-R; auditory memory)(17), the Brief Visuospatial Memory Test-Revised (BVMT-R; visual memory)(18), the Symbol Digit Modalities Test (SDMT; general cognitive function and neurological impairment)(19), the Lafayette Grooved Pegboard (manual dexterity)(20), Stroop Color-Word (21) and Trail-Making (22) Test Parts A and B (attention, reasoning and executive functioning), the Wechsler Adult Intelligence Scale III: Total Digit Span (simple auditory attention and working memory) and Reliable Digit Span (cognitive effort)(23), the 18-point Clock Test (mild cognitive impairment)(24), and the Cognitive Failures Questionnaire (CFQ; self-reported failures in perception, memory, and motor function)(25). Alternative test versions of the HVLT-R and BVMT-R were used pre- and post-study to avoid a learning effect.
Functional Magnetic Resonance Imaging (fMRI) Assessments
We used two well-established paradigms to probe both working memory and long-term memory function, in verbal and nonverbal domains, using blood oxygen level dependent (BOLD) fMRI. The Figural Memory task (26, 27) is a visual encoding and recognition paradigm that consistently activates medial temporal, cingulate, inferior frontal, and posterior parietal brain regions during correct encoding across a wide age range(28). The task stimuli consists of black abstract line drawings, designed to minimize verbal encoding, presented against a white background. Participants performed separate encoding and recognition phases during fMRI scanning. During encoding, participants were presented with 20 target stimuli that are hard-to-encode-verbally (duration 3s, inter-stimulus interval [ISI] 4s) to memorize. Following a 5-minute rest (with no intervening cognitive task), participants completed a recognition phase, during which 20 target and 20 similar -appearing distractor stimuli were presented in a fixed pseudo-random order, each for 3000ms with an ISI of 4000ms. Participants used a press button on a fiber-optic response device to indicate whether or not each stimulus was/was not in the encoding phase set. The figural memory task was analyzed as an event-related design, with responses during recognition used to determine the event types during the encoding phase.
The Verbal Working Memory task is a modified version of the Sternberg item selection paradigm(29), which produces reliable activation of dorsolateral prefrontal, anterior cingulate, posterior parietal, and subcortical brain regions (30). This fMRI task consisted of three separate phases: encoding, maintenance, and retrieval. During encoding, participants were presented with sets of consonants (duration 2500ms) and were instructed to memorize them. Working memory load was manipulated by varying the number of consonants in the set among 2, 4, and 6. Each encoding phase was followed by a maintenance phase of 12 seconds. In the retrieval phase, participants were presented with a probe letter and asked to press a button to indicate whether or not the probe was included in the memorized set. The task comprised 18 trials consisting of randomly generated stimulus letter sets, with half of retrieval trials presenting target probes and half presenting distractor probes, and an equal number of trials for each load. Participants performed two runs of the task at each measurement point. The verbal working memory task was also an event-related design, with each trial divided into encoding, maintenance, and retrieval phases.
Image Acquisition
Imaging data were collected on a Siemens 3T Allegra system. Structural imaging was acquired with a sagittal T1 MPRAGE protocol using the following parameters: TR = 2500 ms, TE=2.74 ms, flip angle = 8°, FOV=176 × 256 mm, matrix = 256 × 208, voxel size = 1 mm3, 176 slices, total scan time =7:20. Functional images were collected in the axial plane using a T2*-weighted echoplanar image (EPI) gradient-echo pulse sequence covering the whole brain. The figural memory task was acquired as follows: TR = 1860 ms, TE = 27 ms, flip angle = 70°, FOV = 240 mm, matrix=64 × 64, in-plane resolution=3.44 × 3.44 mm, slice thickness = 3 mm, gap = 1 mm, 36 slices, total scan time = 12:33. Participants performed two runs of the verbal working memory task with the following parameters: TR=1500 ms, TE=27 ms, flip angle=60°, FOV=24×24 cm, acquisition matrix=64×64, in-plane resolution=3.4×3.4 mm, slice thickness=4 mm, gap = 1 mm, 29 slices, total scan time=7:50.
Demographic, Blood Lipid and Neuropsychological Analyses
All statistical analyses were conducted with IBM SPSS Statistics for Windows, Version 22.0 (Armonk, NY: IBM Corp). Standard diagnostics including histograms, normal probability plots, and residual analysis were used to determine whether the parametric assumptions (e.g., variance homogeneity, normality) of the models were met. Appropriate transformation of the data and/or analogous non-parametric methods were used as necessary. Baseline characteristics were compared using analysis of variance, the Kruskal-Wallis test, or the chi-squared test, as appropriate. A linear mixed effects model for repeated measurements which incorporated time as the within-participants factor and statin treatment as the between-participants factor was used. The main variable of interest in the model was the estimate for the interaction between time and statin treatment. Participants were defined as the random factor; all other variables were fixed within the model. Sex was included as a between-participant factor and age as a covariate. The proportion of participants demonstrating changes in the Cognitive Failures Questionnaire was compared between the statin and placebo groups using chi-square statistics. For neuropsychological test scores, scores were analyzed in three ways: as raw scores (Table 2), t-scores, and z-scores (data not shown).
Table 2.
Neuropsychological measurements on/off treatment presented in mean (SD).
| On Treatment | Off Treatment | |||||
|---|---|---|---|---|---|---|
|
| ||||||
| Measure | Atorvastatin | Placebo | Total | Atorvastatin | Placebo | Total |
| BVMT-R | ||||||
| Total Recall | 24.37 (6.87) | 25.72 (6.37) | 25.12 (6.61) | 25.14 (70.42) | 27.03 (6.69) | 26.18 (7.06) |
| Delayed Recall | 9.56 (2.53) | 10.10 (2.19) | 9.85 (2.36) | 9.83 (2.49) | 10.00 (2.23) | 9.92 (2.34) |
| HVLT-R | ||||||
| Total Recall | 25.82 (5.04) | 25.49 (4.68) | 25.64 (4.84) | 27.23 (9.27) | 27.24 (4.86) | 27.24 (7.23)* |
| Delayed Recall | 9.25 (2.05) | 9.39 (2.17) | 9.32 (2.11) | 9.21 (2.03) | 9.51 (2.15) | 9.37 (2.09)* |
| TMT | ||||||
| Trial A | 25.56 (9.15) | 25.92 (11.78) | 25.30 (10.66) | 22.06 (5.96) | 24.32 (7.28) | 23.30 (6.79) |
| Trial B | 60.90 (28.35) | 62.01 (42.82) | 61.53 (37.10) | 54.14 (19.77) | 53.57 (18.96) | 53.82 (19.24) |
| CLOCK | 16.02 (2.67) | 16.64 (1.60) | 16.37 (2.16) | 16.44 (2.94) | 15.73 (3.22)* | 16.04 (3.11) |
| WAIS-III | ||||||
| Total Digit Span | 17.15 (3.66) | 17.56 (4.10) | 17.38 (3.91) | 17.73 (4.23) | 18.94 (4.25) | 18.39 (4.27) |
| Reliable Digit Span | 9.82 (2.15) | 10.19 (2.28) | 10.03 (2.22) | 10.30 (2.06) | 10.69 (2.24) | 10.52 (2.16) |
| Stroop | ||||||
| Color | 71.68 (12.05) | 74.34 (11.97) | 73.12 (12.03) | 71.07 (12.22) | 76.96 (10.58) | 74.25 (11.70) |
| Word | 101.82 (16.9) | 102.74 (16.61) | 102.32 (16.68) | 102.68 (14.04) | 106.37 (12.68) | 104.69 (13.40) |
| Color-Word | 45.11 (13.45) | 42.57 (10.42) | 43.73 (11.93) | 42.89 (12.11) | 46.18 (13.08) † | 44.67 (12.7) |
| LGPT | ||||||
| Right | 65.35 (13.69) | 65.15 (10.74) | 65.23 (12.05) | 63.71 (14.48) | 66.34 (17.56) | 65.20 (16.28) |
| Left | 68.98 (18.76) | 70.13 (12.92) | 69.63 (15.70) | 70.10 (16.29) | 68.88 (10.12) | 69.42 (13.16) |
| Symbol Digit | 54.00 (12.30) | 55.35 (12.69) | 54.76 (12.49) | 55.63 (12.27) | 57.82 (12.67) | 56.86 (12.50) |
| CFQ | 32.48 (8.75) | 31.43 (10.78) | 31.92 (9.80) | 33.44 (8.32) | 34.07 (8.35) | 33.77 (8.26) |
p<0.05;
p<.01.
Asterisk(s) in the total column represents a significant change in score on treatment vs off treatment for the total sample. Asterisk(s) in the group column represents a significant difference in change of score comparing treatment groups on vs. off treatment. BVMT-R = Brief Visuospacial Memory Test-Revised; HVLT-R = Hopkins Verbal Learning Test-Revised; TMT = Trail Making Test Parts A and B; CLOCK = 18-Point Clock Test; WAIS-III = Wechsler Adult Intelligence Scale-III; LGPT = Layfayette Grooved Pegboard Test; Symbol Digit = Symbol Digit Modalities Test; CFQ = Cognitive Failures Questionnaire.
Imaging Analyses
Functional images were preprocessed and modeled using an in-house pipeline incorporating FSL (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FSL) and SPM (http://www.fil.ion.ucl.ac.uk/spm/software/), utilizing similar methods as our previous work (31, 32). The first six volumes were discarded to allow for T1 saturation effects. Images were realigned and spatially normalized to Montreal Neurological Institute (MNI) standardized space. For the figural memory task, images were resampled to 3×3×3 mm voxels, and smoothed with a 6 mm full-width, half-maximum Gaussian filter. For the verbal working memory task, images were resampled to 2×2×2 mm voxels, and smoothed with a 6 mm full-width, half-maximum Gaussian filter. Datasets were inspected for motion, and those with >3mm displacement were excluded. In addition, scan and behavioral data were examined for completeness and valid behavioral responses (e.g., not responding “yes” to all probe items). For the figural memory task, BOLD response was modeled as in our previous work (28), based on behavioral performance, while covarying for motion and linear trends. Trials from the encoding phase were modeled as correctly encoded if they were subsequently identified as targets during the recognition phase. Targets that were identified as distractors (i.e., misses) in the recognition phase were categorized as incorrectly encoded. Events in the recognition phase were coded as hits, misses, false alarms, and correct rejections. BOLD response for each event type was modeled using a canonical hemodynamic response function fitted to the onset and duration of each event, with duration of 3 seconds for encoding phase events, and reaction time determining the duration of recognition phase events. For the verbal working memory task, BOLD response was modeled for encoding, maintenance, and retrieval phases based on a synthetic hemodynamic response function and temporal derivative, while covarying for motion and linear trends. In addition, BOLD response contrast was modeled for the parametric effect of increasing working memory load.
Whole-brain group level analyses of BOLD response contrasts were conducted using Analysis for Functional NeuroImages (AFNI) version 2016_01_16_16.00 (33). BOLD response contrast was examined at the group level for correct encoding and for recognition hits on the figural memory task. For the verbal working memory task, group level analyses examined BOLD response during encoding, maintenance, and retrieval phases, as well as the parametric interaction term representing change in BOLD response with increasing working memory load during each phase. For each contrast of interest, we ascertained the group × time interaction while controlling for age and sex. In order to avoid inflated false positive rates (34), we used rigorous multiple-comparisons corrections with cluster volume and voxel thresholding (voxel-wise p<.001), determined through Monte-Carlo simulation (35), yielding a whole-brain α=.05 (34). To interpret significant results, we extracted average BOLD response in each cluster for each participant and examined the pattern of simple effects in SPSS. In addition to our main outcome measure, we also conducted single sample t-tests (voxel-wise p<.001, uncorrected) for the off-medication scan for each BOLD response contrast of interest in order to confirm that typical activation patterns were elicited in our sample.
Results
Participants (84 placebo, 66 atorvastatin) completed the study with neuropsychological and blood lipids data (Figure 1). Participants were excluded from fMRI analysis if they did not have complete scans at both timepoints due to significant motion artifact, invalid behavioral responses, inability to complete in-scanner testing, or incomplete/corrupted fMRI data. This yielded a sample size of 77 participants (42 placebo, 35 atorvastatin; 73 excluded) for the Figural Memory task and 120 participants (68 on placebo, 52 on atorvastatin; 30 excluded) for the Verbal Working Memory task.
Total cholesterol, LDL cholesterol and triglycerides at 6 months were lower with atorvastatin (all p<0.01; Table 1), but not different between groups 2 months after treatment cessation (all p>0.42).
Table 1.
Participant Characteristics
| Demographics | Atorvastatin (n=66) | Placebo (n=84) |
|---|---|---|
| Women (n) | 39 | 36 |
| Age (years) | 47.8 (14.7) | 48.8 (14.7) |
| Education (years) | 15.2 (3.5) | 16.2 (2.0) |
| Systolic Blood Pressure (mmHg) | 120.4 (12.9) | 120.0 (13.7) |
| Diastolic Blood Pressure (mmHg) | 77.2 (9.3) | 75.6 (8.9) |
| Body Mass Index (kg/m2) | 27.1 (5.2) | 27.2 (4.9) |
| On-Treatment Lipids | ||
| Total-C (mg/dL) | 128.3 (29.0) | 195.6 (37.8)* |
| LDL-C (mg/dL) | 54.9 (21.3) | 115.2 (35.7)* |
| HDL-C (mg/dL) | 55.6 (18.6) | 58.1 (16.1) |
| TRIG (mg/dL) | 82.7 (35.5) | 101.0 (42.3)* |
| Off-Treatment Lipids | ||
| Total-C (mg/dL) | 193.3 (34.8) | 191.0 (35.2) |
| LDL-C (mg/dL) | 114.6 (31.7) | 111.9 (29.8) |
| HDL-C (mg/dL) | 58.2 (16.3) | 59.4 (15.0) |
| TRIG (mg/dL) | 102.4 (52.1) | 95.9 (43.9) |
Data expressed as mean (SD).
indicates significant difference between atorvastatin and placebo groups at p < 0.01.
C = cholesterol; LDL = low-density lipoprotein; HDL = high-density lipoprotein; TRIG = triglycerides.
Neuropsychological test scores were largely unchanged with drug therapy (raw scores only shown in Table 2); HVLT-R raw, z- and t- scores for total and delayed recall in the total sample improved when participants were retested off study drug (p=0.01 and 0.04, respectively). The only significant drug*time interactions were observed for raw scores on Stroop Color-Word testing (p<0.01) where the placebo group improved with treatment cessation relative to on treatment, and the 18-Point Clock Test (p=0.02), where the placebo group demonstrated a reduction in score off relative to on treatment.
There was no difference in change in the CFQ scores on/off treatment between the atorvastatin and placebo groups (Table 1). There were also no differences between treatment groups in individual questions except for Q17 (Do you forget where you put something like a newspaper or book?), in which the placebo group displayed a slight increase in mean score off treatment (change score±SD: 0.35±0.63) compared to the statin group (−0.04±0.75, p=0.04). A chi-square analysis compared the number of participants who answered 3 “quite often” or 4 “very often” on all questions between treatment groups on vs off treatment and did not find any atorvastatin effect (all p>0.10). Additional repeated measures and chi square analyses for total scores on questions grouped into 4 different domains (memory, distractibility, blunders, and names) also showed no significant differences between groups on vs off treatment (all p>0.12).
There was a group × time interaction in bilateral paracentral lobule/precuneus during the encoding phase of the Figural Memory Task, (Table 3 and Figure 2). In this region, participants on atorvastatin had more response than participants on placebo while on study drug (F(1,73) = 8.06, p=.006), but less response than patients in the placebo group after the drug washout period (F(1,73) = 11.82, p=.001). Neither group showed a significant change in BOLD response between scans (Atorvastatin: F(1,32) = 0.11, p=.747), Placebo: F(1,39) = 0.05, p=.826)). There were no group × time effects on neural response during recognition hits. There were also no group × time effects on any behavioral data from the task, including hits, misses, correct rejections, false alarms, and d-prime (all p >0.68). Single sample t-tests of the off-treatment scan revealed activation patterns similar to those observed in our previous work utilizing this task, (28, 31) confirming typical activation patterns in the current sample.
Table 3.
Cluster details for regions showing significant group × time interactions fMRI responses during Figural Memory and Verbal Working Memory, controlling for age and sex (voxel-wise p < .001, α<.05 whole brain corrected).
| Anatomic Region (Brodmann’s Area) | Cluster Volume (μl) | MNI Coordinates | F | Effect size Partial η2 | Observed power | Group Activation Marginal Mean (SE) | |||
|---|---|---|---|---|---|---|---|---|---|
| x, y, z | On Treatment | Off Treatment | |||||||
| Atorvastatin | Placebo | Atorvastatin | Placebo | ||||||
| Figural Memory Task – Encoding Phase | df=1,73 | ||||||||
| Bilateral paracentral lobule/precuneus (5) | 891 | 0, −46, 54 | 21.04 | 0.22 | 0.995 | 0.534 (0.236) | −0.699 (0.265) | −0.397 (0.215) | 0.563 (0.240) |
| Verbal Working Memory Task – Rehearsal Phase | df=1,116 | ||||||||
| Right putamen | 632 | 24, −2, 2 | 29.95 | 0.21 | 1.00 | −0.035 (0.041) | 0.139 (0.048) | 0.125 (0.036) | −0.095 (0.042) |
MNI coordinates refer to maximum intensity voxel within each cluster; F statistic, effect size, observed power, and group activation marginal means refer to averages across each cluster, controlling for age and sex.
Figure 2.
Neural Activation During Figural Memory Task. Group × time interaction on fMRI response during Figural Memory Task correct encoding.
During the maintenance phase of the Verbal Working Memory Task, there was a group × time interaction in the right putamen extending into the right globus pallidus (Table 3 and Figure 3). In this region, participants on atorvastatin had less BOLD response compared to participants on placebo while on treatment (F(1,116) = 8.30, p=.005), but after the treatment washout period, participants in the atorvastatin group had more BOLD response than participants in the placebo group (F(1,116) = 12.84, p<.001). The atorvastatin group also showed a trend toward an increase in BOLD response between scans (F(1,49) = 3.78, p=.055), whereas the placebo group showed no change in BOLD response over time (F(1,65) = 0.95, p=.33). There were no group × time effects during the encoding phase or the retrieval phase, and there were no parametric effects of increasing memory load during any phase. There were also no group × time effects on any behavioral data from the task, including identification of incorrect or correct probes for 2, 4 and 6 consonants and accompanying reaction time (all p >0.16). Single-sample t-tests of the off-medication scans demonstrated activation patterns in the current sample similar to our previous work (32).
Figure 3.
Neural Activation During Verbal Working Memory Task. Group × time interaction on fMRI response during Verbal Working Memory Task maintenance phase.
Discussion
To the best of our knowledge and literature review, this study is the first to investigate the effects of statins on the central nervous system by utilizing fMRI to assess brain neural activation in healthy adults treated with 80 mg atorvastatin or placebo. We detected few changes attributable to statin therapy with standardized neuropsychological tests, a finding similar to that from previous clinical trials (5, 36). However, participants on atorvastatin demonstrated altered patterns of neural activation on vs. off statin compared to participants treated with placebo. Unexpectedly, the treatment groups differed at both timepoints. The clinical implications of these findings are unclear and warrant additional clinical trials.
A meta-analysis of 14 studies with 27,643 participants failed to show significant adverse effects of statins on standard neuropsychological cognitive tests. The authors questioned whether the FDA warning on statins and cognition was appropriate (8). However, case reports of memory decrements with statin therapy continue to be published (37), and patients report cognitive side effects as a leading cause of statin intolerance (38). The discrepancy between self-report and clinical trials data could be due to a nocebo effect, such as observed with some statin-associated muscle symptoms (39, 40), or it could be due to the small effect sizes and learning/practice effects associated with repeated neuropsychological tests (6, 7). Direct measures of brain structure and neural activation could supplement data from standard cognitive tests. The PROSPER trial did not observe changes in brain structure or cerebral blood flow after 3 years of pravastatin therapy (41), but prior to this no study has used fMRI to evaluate changes in brain neural activation with statin therapy.
On the Figural Memory Task, we observed a group × time interaction during correct encoding in a small cluster spanning the bilateral paracentral lobule and anterior precuneus. Specifically, participants on 6 months atorvastatin showed less response than participants on placebo, and this effect was reversed following drug washout. This region is associated with sensorimotor function (42) and is not typically activated by this task (28, 31) or other memory paradigms (43). Moreover, regions most consistently associated with this paradigm, such as medial temporal, posterior parietal, and inferior frontal cortices, were activated by both groups during the post-treatment scan, and showed no group × time interactions. Further, we observed no group × time effects during recognition. In contrast, during the Verbal Working Memory Task maintenance phase, we observed a group × time interaction in the left putamen and globus pallidus. Specifically, participants on 6 months of atorvastatin therapy demonstrated less response than participants on placebo, but more response after drug discontinuation. These structures contribute to a basal ganglia network involved in working memory function (44). However, both groups differed in opposite directions at each timepoint, and neither group showed a significant change between on- and off-treatment scans. In addition, there were no group × time effects during encoding or retrieval, and no group × time effects on activation related to working memory load. Moreover, while groups engaged typical working memory regions (44), including prefrontal, anterior cingulate, and posterior parietal cortices, no group × time interactions were seen in these regions for any task phase. Finally, for both tasks, there were no differences with drug treatment in the behavioral measures that represent task accuracy. In sum, these findings do not provide convincing evidence of measurable verbal or nonverbal memory dysfunction due to statin medications, and support the large body of literature suggesting that in most healthy adults, statins do not evoke measurable adverse cognitive side effects. However, it is possible that these altered patterns of neural activation could underlie the occasional reports of memory loss and confusion reported in surveillance data and as case reports and observational studies. For example, we reported a patient who complained of adverse cognitive side effects during atorvastatin therapy and demonstrated altered fMRI activation patterns during the Sternberg Task during statin treatment (12). In addition, both the precuneus and putamen contribute to memory and learning processes, with the latter impacted by pathologies such as Alzheimer’s and Parkinson’s disease (46,47). Consequently, future studies are needed to confirm and expand our findings and assess their relationship to clinical sequelae.
Limitations
The current study has several limitations. Results are likely dose and statin specific, and some data suggest that atorvastatin has the greatest frequency of cognitive complaints (2). We did not investigate the long-term effects of atorvastatin on memory function since STOMP was only a 6 month study, but the majority of cognitive complaints occur within 2 months of initiating statins (11). The median time to onset of statin-associated cognitive complaints is 5 months (2). The short duration of the study also indicates that observed changes are attributable to drug treatment rather than cognitive decline. The two-month washout period may also have been insufficient and could be responsible for our observed group-time interactions, although reported recovery from statin-associated cognitive complaints begins at a median time of 1.5 weeks and reported maximal recovery occurs at 8 weeks (2). The parent STOMP trial, which provided participants for this ancillary trial, necessitated our using an on- vs. off-drug measurements order, which could produce a learning effect associated with repeated cognitive, behavioral and fMRI measurements. The use of a placebo group, alternative versions of cognitive tests where applicable and fMRI tasks, and administration of a practice session prior to fMRI testing should have reduced this potential bias.
Conclusions and Future Directions
Published studies and reports of statin-associated cognitive complaints indicate that the potential adverse effects of statins on cognition are rare. Meta-analyses also find little effect of statins on cognition in healthy adults, although two randomized clinical trials (6, 7) found slightly diminished cognitive test scores, (exhibited largely as a lack of a learning effect, in adults not complaining of cognitive changes during low-dose simvastatin and lovastatin administration). Our study also found few observable cognitive effects of high dose, short duration atorvastatin treatment, even when CNS function was measured directly. The small but significant alterations in regional neural activation with statin therapy require confirmation in larger trials, since they may indicate a mechanism for cognitive complaints. If confirmed, future studies are needed to examine the clinical consequences of altered neural function with statin therapy, and should study those rare patients who present with statin-related cognitive complaints.
Acknowledgments
We wish to acknowledge the following individuals for their participation on the Data Safety Monitoring Board: JoAnne Foody, MD; Pamela Hartigan, PhD; and Ira Ockene, MD (chair). The STOMP parent study was funded by National Heart, Lung, and Blood Institute/National Institutes of Health grant RO1 HL081893 (Dr Thompson). The STOMP ancillary cognitive study was funded by National Heart, Lung, and Blood Institute/National Institutes of Health grant NHLBI 1R01HL098085 (Drs. Taylor and Polk).
Footnotes
Declaration of Interest
Dr. Thompson has received research support from Genomas, Roche, Sanofi, Regeneron, Esperion, Amarin and Pfizer; has served as a consultant for Amgen, Regeneron, Merck, Genomas, Runners World, Sanofi, Esperion, and Amarin; has received speaker honoraria from Merck, Astra Zeneca, Kowa, and Amarin; owns stock in Abbvie, Abbott Labs, General Electric, J&J; and has provided expert legal testimony on exercise-related cardiac events and statin myopathy. Dr. Taylor served on the Pharmacovigilance Monitoring Board for Amgen, Inc. and received grant support from Regeneron, Inc. All other authors report no disclosures. All authors approved the final manuscript.
Publisher's Disclaimer: 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 citable 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.
References
- 1.Golomb BA, McGraw JJ, Evans MA, et al. Physician response to patient reports of adverse drug effects: implications for patient-targeted adverse effect surveillance. Drug Saf. 2007;30:669–75. doi: 10.2165/00002018-200730080-00003. [DOI] [PubMed] [Google Scholar]
- 2.Evans MA, Golomb BA. Statin-associated adverse cognitive effects: survey results from 171 patients. Pharmacotherapy. 2009;29:800–11. doi: 10.1592/phco.29.7.800. [DOI] [PubMed] [Google Scholar]
- 3.Bettermann K, Arnold AM, Williamson J, et al. Statins, risk of dementia, and cognitive function: secondary analysis of the ginkgo evaluation of memory study. J Stroke Cerebrovasc Dis. 2012;21:436–44. doi: 10.1016/j.jstrokecerebrovasdis.2010.11.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Etminan M, Gill S, Samii A. The role of lipid-lowering drugs in cognitive function: a meta-analysis of observational studies. Pharmacotherapy. 2003;23:726–30. doi: 10.1592/phco.23.6.726.32184. [DOI] [PubMed] [Google Scholar]
- 5.Trompet S, van Vliet P, de Craen AJ, et al. Pravastatin and cognitive function in the elderly. Results of the PROSPER study. J Neurol. 2010;257:85–90. doi: 10.1007/s00415-009-5271-7. [DOI] [PubMed] [Google Scholar]
- 6.Muldoon MF, Ryan CM, Sereika SM, et al. Randomized trial of the effects of simvastatin on cognitive functioning in hypercholesterolemic adults. Am J Med. 2004;117:823–9. doi: 10.1016/j.amjmed.2004.07.041. [DOI] [PubMed] [Google Scholar]
- 7.Muldoon MF, Barger SD, Ryan CM, et al. Effects of lovastatin on cognitive function and psychological well-being. Am J Med. 2000;108:538–46. doi: 10.1016/s0002-9343(00)00353-3. [DOI] [PubMed] [Google Scholar]
- 8.Ott BR, Daiello LA, Dahabreh IJ, et al. Do statins impair cognition? A systematic review and meta-analysis of randomized controlled trials. J Gen Intern Med. 2015;30:348–58. doi: 10.1007/s11606-014-3115-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Swiger KJ, Manalac RJ, Blumenthal RS, et al. Statins and cognition: a systematic review and meta-analysis of short- and long-term cognitive effects. Mayo Clin Proc. 2013;88:1213–21. doi: 10.1016/j.mayocp.2013.07.013. [DOI] [PubMed] [Google Scholar]
- 10.Orsi A, Sherman O, Woldeselassie Z. Simvastatin-associated memory loss. Pharmacotherapy. 2001;21:767–9. doi: 10.1592/phco.21.7.767.34577. [DOI] [PubMed] [Google Scholar]
- 11.Wagstaff LR, Mitton MW, Arvik BM, et al. Statin-associated memory loss: analysis of 60 case reports and review of the literature. Pharmacotherapy. 2003;23:871–80. doi: 10.1592/phco.23.7.871.32720. [DOI] [PubMed] [Google Scholar]
- 12.Parker B, Polk D, Rabdiya V, et al. Changes in memory function and neuronal activation associated with atorvastatin therapy. Pharmacotherapy. 2010;30:236e. [Google Scholar]
- 13.FDA Drug Safety Communication. [Last accessed 9/20/2017];Important safety label changes to cholesterol-lowering statin drugs. Available from: http://www.fda.gov/Drugs/DrugSafety/ucm293101.htm.
- 14.Schretlen DJ, Munro CA, Anthony JC, et al. Examining the range of normal intraindividual variability in neuropsychological test performance. J Int Neuropsychol Soc. 2003;9:864–70. doi: 10.1017/S1355617703960061. [DOI] [PubMed] [Google Scholar]
- 15.Parker BA, Capizzi JA, Grimaldi AS, et al. Effect of statins on skeletal muscle function. Circulation. 2013;127:96–103. doi: 10.1161/CIRCULATIONAHA.112.136101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Thompson PD, Parker BA, Clarkson PM, et al. A randomized clinical trial to assess the effect of statins on skeletal muscle function and performance: rationale and study design. Prev Cardiol. 2010;13:104–11. doi: 10.1111/j.1751-7141.2009.00063.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Shapiro AM, Benedict RH, Schretlen D, et al. Construct and concurrent validity of the Hopkins Verbal Learning Test-revised. Clin Neuropsychol. 1999;13:348–58. doi: 10.1076/clin.13.3.348.1749. [DOI] [PubMed] [Google Scholar]
- 18.Gaines JJ, Gavett RA, Lynch JJ, et al. New error type and recall consistency indices for the Brief Visuospatial Memory Test - Revised: performance in healthy adults and multiple sclerosis patients. Clin Neuropsychol. 2008;22:851–63. doi: 10.1080/13854040701565216. [DOI] [PubMed] [Google Scholar]
- 19.Smith A. Symbol Digit Modalities Test. Las Vegas: Western Psychological Services; 1982. [Google Scholar]
- 20.Lafayette Instrument. Grooved Pegboard Test User Instructions. North Lafayette, IN: Lafayette Instrument; 2012. [Google Scholar]
- 21.Jensen AR, Rohwer WD., Jr The Stroop color-word test: a review. Acta Psychol (Amst) 1966;25:36–93. doi: 10.1016/0001-6918(66)90004-7. [DOI] [PubMed] [Google Scholar]
- 22.Lezak MD, Howieson DB, Loring DW. Neuropsychological Assessment. 4. New York: Oxford University Press; 2004. [Google Scholar]
- 23.Babikian T, Boone KB, Lu P, Arnold G. Sensitivity and specificity of various digit span scores in the detection of suspect effort. Clin Neuropsychol. 2006;20:145–59. doi: 10.1080/13854040590947362. [DOI] [PubMed] [Google Scholar]
- 24.Babins L, Slater ME, Whitehead V, et al. Can an 18-point clock-drawing scoring system predict dementia in elderly individuals with mild cognitive impairment? J Clin Exp Neuropsychol. 2008;30:173–86. doi: 10.1080/13803390701336411. [DOI] [PubMed] [Google Scholar]
- 25.Broadbent DE, Cooper PF, FitzGerald P, et al. The Cognitive Failures Questionnaire (CFQ) and its correlates. Br J Clin Psychol. 1982;21(Pt 1):1–16. doi: 10.1111/j.2044-8260.1982.tb01421.x. [DOI] [PubMed] [Google Scholar]
- 26.Beason-Held LL, Golski S, Kraut MA, et al. Brain activation during encoding and recognition of verbal and figural information in older adults. Neurobiol Aging. 2005;26:237–50. doi: 10.1016/j.neurobiolaging.2004.03.014. [DOI] [PubMed] [Google Scholar]
- 27.Golski S, Zonderman AB, Malamut BL, et al. Verbal and figural recognition memory: task development and age associations. Exp Aging Res. 1998;24:359–85. doi: 10.1080/036107398244193. [DOI] [PubMed] [Google Scholar]
- 28.Jamadar S, Assaf M, Jagannathan K, et al. Figural memory performance and functional magnetic resonance imaging activity across the adult lifespan. Neurobiol Aging. 2013;34:110–27. doi: 10.1016/j.neurobiolaging.2012.07.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Sternberg S. High-speed scanning in human memory. Science. 1966 Aug 5;153(3736):652–4. doi: 10.1126/science.153.3736.652. [DOI] [PubMed] [Google Scholar]
- 30.Forsyth JK, McEwen SC, Gee DG, et al. Reliability of functional magnetic resonance imaging activation during working memory in a multi-site study: analysis from the North American Prodrome Longitudinal Study. Neuroimage. 2014;97:41–52. doi: 10.1016/j.neuroimage.2014.04.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Dager AD, Jamadar S, Stevens MC, et al. fMRI response during figural memory task performance in college drinkers. Psychopharmacology (Berl) 2014;231:167–79. doi: 10.1007/s00213-013-3219-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Meda SA, Bhattarai M, Morris NA, et al. An fMRI study of working memory in first-degree unaffected relatives of schizophrenia patients. Schizophr Res. 2008;104:85–95. doi: 10.1016/j.schres.2008.06.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Cox RW. AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Comput Biomed Res. 1996;29:162–73. doi: 10.1006/cbmr.1996.0014. [DOI] [PubMed] [Google Scholar]
- 34.Cox RW, Chen G, Glen DR, et al. FMRI Clustering in AFNI: False-Positive Rates Redux. Brain Connect. 2017;7:152–71. doi: 10.1089/brain.2016.0475. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Ward BD. Simulataneous Inference for FMRI Data. Milwaukee, WI: Biophysics Research Institute; 2000. [Google Scholar]
- 36.Bernick C, Katz R, Smith NL, et al. Statins and cognitive function in the elderly: the Cardiovascular Health Study. Neurology. 2005;65:1388–94. doi: 10.1212/01.wnl.0000182897.18229.ec. [DOI] [PubMed] [Google Scholar]
- 37.Suraweera C, de Silva V, Hanwella R. Simvastatin-induced cognitive dysfunction: two case reports. J Med Case Rep. 2016;10:83,016-0877-8. doi: 10.1186/s13256-016-0877-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Lakey WC, Greyshock NG, Kelley CE, et al. Statin intolerance in a referral lipid clinic. J Clin Lipidol. 2016;10:870, 879.e3. doi: 10.1016/j.jacl.2016.03.004. [DOI] [PubMed] [Google Scholar]
- 39.Nissen SE, Stroes E, Dent-Acosta RE, et al. Efficacy and Tolerability of Evolocumab vs Ezetimibe in Patients With Muscle-Related Statin Intolerance: The GAUSS-3 Randomized Clinical Trial. JAMA. 2016;315:1580–90. doi: 10.1001/jama.2016.3608. [DOI] [PubMed] [Google Scholar]
- 40.Taylor BA, Lorson L, White CM, et al. A randomized trial of coenzyme Q10 in patients with confirmed statin myopathy. Atherosclerosis. 2015;238:329–35. doi: 10.1016/j.atherosclerosis.2014.12.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Smit RA, Trompet S, Sabayan B, et al. Higher Visit-to-Visit Low-Density Lipoprotein Cholesterol Variability Is Associated With Lower Cognitive Performance, Lower Cerebral Blood Flow, and Greater White Matter Hyperintensity Load in Older Subjects. Circulation. 2016;134:212–21. doi: 10.1161/CIRCULATIONAHA.115.020627. [DOI] [PubMed] [Google Scholar]
- 42.Lim SH, Dinner DS, Pillay PK, et al. Functional anatomy of the human supplementary sensorimotor area: results of extraoperative electrical stimulation. Electroencephalogr Clin Neurophysiol. 1994;91:179–93. doi: 10.1016/0013-4694(94)90068-x. [DOI] [PubMed] [Google Scholar]
- 43.Kim H. Neural activity that predicts subsequent memory and forgetting: a meta-analysis of 74 fMRI studies. Neuroimage. 2011;54:2446–61. doi: 10.1016/j.neuroimage.2010.09.045. [DOI] [PubMed] [Google Scholar]
- 44.Wager TD, Smith EE. Neuroimaging studies of working memory: a meta-analysis. Cogn Affect Behav Neurosci. 2003;3:255–74. doi: 10.3758/cabn.3.4.255. [DOI] [PubMed] [Google Scholar]