Comparison of the Effects of Pravastatin and Lovastatin on Sleep Disturbance in Hypercholesterolemic Subjects Downloaded from https://academic.oup.com/sleep/article/22/1/117/2731686 by guest on 29 September 2021

Bruce L. Ehrenberg,1 Stefania Lamon-Fava,3 Kate E. Corbett,1 Judith R. McNamara,3 Gerard E. Dallal,4 and Ernst J. Schaefer2,3

(1) Department of Neurology and the (2) Division of Endocrinology, Diabetes, Metabolism, and Molecular Medicine†, New England Medical Center, Boston, Mass; (3) the Lipid Metabolism Laboratory and (4) Division of Biostatistics, Jean Mayer USDA Human Nutrition Research Center on Aging at Tufts University, Boston, Mass

Summary: We have studied the effects of two cholesterol-lowering medications, lovastatin and pravastatin, on different sleep parameters in hypercholesterolemic subjects. These medications are 3-hydroxy-methylglutaryl coenzyme A inhibitors. Only subjects who had complained of sleep disturbance while on previous treatment with lovastatin were enrolled. Sixteen subjects (11 men and 5 women) underwent a randomized, double-blind, three-way crossover treatment with lovastatin, pravastatin, and placebo. Each phase of the study lasted 4 weeks. A placebo wash-out period of 4 weeks separated each treatment phase. At the end of each treatment phase, subjects were admitted to the sleep laboratory for 2 consecutive nights. No statistical differences were detected during treatment with lovastatin, pravastatin, and placebo for sleep parame- ters such as total sleep time, total awake time, wake time after sleep onset, efficiency of sleep, and percent of different phas- es of sleep. Our study suggests that lovastatin and pravastatin do not have a significant effect on sleep parameters in hyper- cholesterolemic subjects that could explain their complaints of insomnia. Nevertheless, the subjects did have moderate sleep disturbances that could account for insomnia and most likely predate the use of HMG-CoA reductase inhibitors. Key words: Hydroxy-methylglutaryl coenzyme A; lovastatin; pravastatin; hypercholesterolemia

HYDROXY-METHYLGLUTARYL coenzyme A (HMG- has not been detected in cerebrospinal fluid.4 It has also CoA) reductase is a key enzyme in the early steps of the been shown that the rate of transport of lovastatin in extra- intracellular synthesis of cholesterol. Inhibitors of this hepatic cells is 100-fold greater than that of pravastatin.5 enzyme have been developed and have become the most Some studies have indicated that treatment with lovas- widely used medications in the treatment of elevated plas- tatin may be associated with increased wake time after ma cholesterol levels, or hypercholesterolemia.1 Lovastatin sleep onset6 or with impaired daytime performance7 when (Mevacor®) and pravastatin (Pravachol®) belong to this compared to pravastatin. It has been hypothesized that group of medications. The chemical composition of lovas- these effects of lovastatin may be caused by its lipophilici- tatin differs slightly from that of pravastatin due to the pres- ty and a direct action in the central nervous system. ence of a methyl instead of a hydroxyl group side chain.2,3 The purpose of this study was to test the hypothesis This difference, though small, is sufficient to confer differ- that lovastatin and pravastatin differ in their effect on sev- ent physico-chemical characteristics to these molecules: eral sleep parameters in a group of hypercholesterolemic lovastatin is lipophilic and has been shown to cross the men and women who had previously complained of sleep blood-brain barrier, while pravastatin is hydrophilic and disturbance while on lovastatin.

METHODS Accepted for publication August, 1998 Subjects Address correspondence and requests for reprints to Bruce L. Ehrenberg, MD, Department of Neurology, New England Medical Center, 750 Washington Subjects were recruited from a population of patients Street, Box #4, Boston, MA 02111 attending our Lipid Clinic and being treated with lovas-

SLEEP, Vol. 22, No. 1, 1999 117 Lovastatin and pravastatin in sleep—Ehrenberg et al tatin. Subjects were asked to complete a 13-question sleep questionnaire.8 We only recruited subjects who had three or Table 1.—Design of the randomized, crossover, placebo-controlled more of the following complaints on at least 3 nights per trial of the effect of lovastatin and pravastatin on sleep disturbances. week while taking lovastatin for over 30 days: 1) sleep onset latency equal to or greater than 30 minutes; 2) five or Treatment phase more nocturnal awakenings lasting more than 1 minute, or subjective total wake time after sleep onset of 30 minutes 1 2 3 or more; 3) awakening 30 minutes or more before desired Sequence n wake-up time; 4) sleeping less than 6 hours per night; or 5) 1 Placebo Lovastatin Pravastatin 1 daytime fatigue 2-3 hours after wake-up time. A total of 20 2 Placebo Pravastatin Lovastatin 2 subjects were recruited. Three subjects were removed from 3 Lovastatin Placebo Pravastatin 4 the study prior to completion—one for developing chest 4 Lovastatin Pravastatin Placebo 3 5 Pravastatin Placebo Lovastatin 3 pain and skin rash before the randomization phase of the Downloaded from https://academic.oup.com/sleep/article/22/1/117/2731686 by guest on 29 September 2021 study, and two because of poor compliance. In addition, at 6 Pravastatin Lovastatin Placebo 3 the end of the study one subject admitted to taking pre- scribed sleep medications (benzodiazepine) throughout the study period. While inclusion of this subject’s data did not change the results, results are presented for the 16 remain- Myers Squibb Pharmaceutical Research Institute. Also, ing subjects only. All subjects had elevated blood choles- subjects were told not to take medications such as steroids, terol levels (>240 mg/dl), and six subjects had triglyceride other hypolipidemic drugs, anticoagulants, theophylline, levels >250 mg/dl. Subjects who had recently experienced barbiturates or other hypnotics, anxiolytics, and aluminum- a myocardial infarction, or had homozygous familial containing anti-acids during the study. hypercholesterolemia, non-type-II hyperlipoproteinemia, dysproteinemia, hypertension (>160/100 mm Hg), obesity Sleep Measurements (body weight >40% above ideal weight), diabetes mellitus, At the end of the placebo lead-in phase and each treat- and thyroid, kidney, or liver disease were excluded from ment phase (lovastatin, pravastatin, and placebo), subjects the study. Eleven of the 16 subjects were men, with a mean were admitted to the sleep laboratory and polysomno- age of 53.8±11.5 years, and 5 were women, all post- graphic evaluations (EEG, EOG, EKG, EMG, and four res- menopausal, with a mean age of 57.4±11.3 years. piratory monitors) on 2 consecutive nights were performed. Briefly, polysomnograms were recorded using Nihon- Study Design Kohdon EEG amplifiers connected to a Telefactor semi- Subjects were stabilized on a reduced-fat, reduced- automatic sleep scoring system. The data from each night cholesterol diet (<30% of energy as total fat, <7% as satu- were scored by direct visual analysis of each 30-second rated fat, and <200 mg/day of cholesterol) prior to entry epoch of data visualized on the reformatter’s TV screen or into the study, and were asked to follow this diet through- on tape backups. Measurements included: total sleep time out the study. The study consisted of six distinct phases, (TST), sleep efficiency (total sleep time/total time in bed, each lasting 4 weeks, for a total of 24 weeks. After an ini- or EFF), amount of wake time after sleep onset (WASO), tial single-blind period of 4 weeks during which subjects stage 1 sleep (S1), stage 2 sleep (S2), stages 3-4 sleep were given placebo tablets matching the lovastatin and (SWS), rapid eye movement (REM), irregular respiratory pravastatin tablets to establish their compliance to the reg- sleep events of apnea or hypopnea (APN), periodic leg imen, subjects were randomized in a double-blind, three- movements without arousal (PLM), periodic leg move- way crossover protocol to treatment with lovastatin, ments with arousal (PLMA), and arousals unrelated to APN pravastatin, and placebo. The dose for each study drug was or PLM (AR). In addition, we calculated the following 40 mg, given in one of three ways: 1) one active lovastatin indexes: tablet at dinner and one pravastatin-matching placebo periodic movement index (or PMI) = (PLM+PLMA)/TST tablet at bedtime; 2) one lovastatin-matching placebo tablet apnea/hypopnea index (or AHI) = APN/TST total arousal index (or TAI) = (PLMA+AR+APN)/TST. at dinner and one active pravastatin tablet at bedtime; or 3) We consider the PMI normal if lower than 10/hour, the AHI appropriate matching placebos at dinner and at bedtime. normal if lower than 5/hour, and the TAI normal if lower Each of these treatment phases lasted 4 weeks and was sep- than 5/hour. arated from the following treatment phase by an additional 4-week placebo “washout” period. The sequence of these Laboratory Measurements randomized treatments is shown in Table 1. All medication and matching placebo tablets were provided by the Bristol- Blood was drawn from each subject at the end of each SLEEP, Vol. 22, No. 1, 1999 118 Lovastatin and pravastatin in sleep—Ehrenberg et al Table 2.—Effects of lovastatin and pravastatin on sleep measurements (n=16)

Placebo Lovastatin Pravastatin F P Minimum Minimum significant significant ratio difference mean+SD mean+SD mean+SD TST (h) 6.2+0.8 6.4+0.6 6.3+0.8 0.98 0.38 0.40 AW (h) 0.68+0.60 0.64+0.59 0.79+0.67 1.72 0.19 1.35 WASO (h) 0.53+0.49 0.53+0.49 0.63+0.49 0.21 0.81 0.23 EFF (%) 0.90+0.09 0.91+0.08 0.88+0.09 1.02 0.37 0.04 APN (n) 16.7+23.8 21.8+30.6 22.2+31.9 2.03 0.14 1.51 PLM (n) 37.8+42.9 34.6+26.0 53.7+101.1 0.19 0.83 1.70 Downloaded from https://academic.oup.com/sleep/article/22/1/117/2731686 by guest on 29 September 2021 PLMA (n) 46.9+67.8 33.4+35.3 35.3+32.7 0.64 0.53 1.53 AR (n) 4.0+3.6 5.9+4.9 4.8+4.7 0.67 0.52 1.60 REM (%) 0.18+0.07 0.19+0.09 0.19+0.05 1.03 0.36 0.03 S1 (%) 0.28+0.07 0.28+0.12 0.26+0.10 0.92 0.40 0.05 S2 (%) 0.44+0.07 0.42+0.08 0.45+0.07 0.27 0.77 0.05 SWS (%) 0.10+0.04 0.09+0.05 0.13+0.07 -- --* 1.26** PMI 13.9+18.4 10.6+7.9 13.9+18.6 0.19 0.83 1.66 AHI 2.8+3.9 3.7+5.3 3.7+5.4 -- --* 1.54** TAI 11.3+13.9 9.8+6.8 10.1+5.9 0.85 0.44 1.31 TST, total sleep time; AW, total awake time; WASO, amount of wake time after sleep onset; EFF, sleep efficiency; APN, irregular respiratory rate events (apnea); PLM, periodic leg movement without arousal; PLMA, periodic leg movements with arousal; AR, arousal without leg movements: REM, rapid eye movement sleep; S1, stage 1 sleep; S2, stage 2 sleep; SWS, slow wave sleep (sleep stage 3 and 4, or delta sleep); PMI, periodic movement index; AHI, apnea/hypoxia index; TAI, total arousal index Model includes subject, treatment, phase, day of phase, and treatment-by-phase, day-by-phase, day-by-treatment, day-by-treatment-by-phase interactions. * with a significant treatment effect and a day-by-treatment interaction (P<0.05), ** minimum significant ratio on day 2. treatment phase. After a 12-hour fast, blood was drawn into Treatment differences were evaluated using analysis of tubes containing EDTA (final concentration of 0.15%), and variance with subject, treatment, study phase (phase of this then was centrifuged at 2,500 rpm for 30 minutes at 4°C to crossover study), day of phase (either day 1 or day 2) as separate plasma. Plasma high-density lipoprotein (HDL) study factors, and treatment-by-phase, day-by-phase, day- cholesterol levels were measured after dextran sulfate- by-treatment, and day-by-treatment-by-phase interactions 9 MgCl2 precipitation. Triglyceride levels in total plasma in the model. Results were considered statistically signifi- and cholesterol levels in total plasma and in the HDL cant if the two-sided observed significance level (p value) supernatant were measured by standardized enzymatic was less than 0.05. techniques.10 Low-density lipoprotein (LDL) cholesterol In order to obtain approximate treatment differences levels were calculated by using the Friedewald formula as that would have been declared statistically significant follows: LDL cholesterol = total cholesterol - (HDL choles- despite the complication of missing data, missing values terol + triglyceride/5).11 were imputed from the analysis-of-variance models fitted to the incomplete data. The reported minimum detectable Statistical Analyses differences are differences that would be significantly dif- ferent by Tukey’s honestly significant differences in the Data were analyzed using the SAS statistical pack- completed data set. In the case of logarithmically trans- age, version 6.04 (SAS Institute Inc., Cary, NC). The dis- formed variables, the antilogarithm of the difference is tributions of AW, SWS, APN, PLM, PLMA, AR, PMI, reported as the smallest ratio of geometric means that AHI, and TAI were markedly skewed, so a logarithmic would have achieved statistical significance. transformation was used prior to formal analysis in order to make the distributions more symmetric. The value 0.1 was RESULTS added to AW, SWS, PMI, and AHI, while 1 was added to APN, PLM, PLMA, and AR prior to the logarithmic trans- Polysomnographic recordings at the end of the placebo formation so that occasional values of 0 would not be lost. lead-in phase of the study indicated that 14 of the 16 sub-

SLEEP, Vol. 22, No. 1, 1999 119 Lovastatin and pravastatin in sleep—Ehrenberg et al DISCUSSION

Table 3.—Effects of lovastatin and pravastatin on plasma lipid levels There have been few reports of an effect of lovastatin (n=16). on sleep. Schaefer12 reported a higher prevalence of sleep complaints in hypercholesterolemic patients on lovastatin Placebo Lovastatin Pravastatin F P (17.6%) than in patients on pravastatin (0%). In controlled mean+SD mean+SD mean+SD studies comparing the effects of lovastatin and pravastatin Total cholesterol 298+68 218+49 230+42 27.1 0.0001 on sleep, Vgontzas et al6 found that continued administra- (mg/dl) Triglycerides 222+132 159+72 221+138 4.2 0.05 tion of lovastatin, but not pravastatin, increased wake time (mg/dl) after sleep onset, and Roth et al7 found that lovastatin, but LDL cholesterol 207+59 141+46 149+43 27.8 0.0001 not pravastatin, significantly impaired daytime perfor- (mg/dl)* HDL cholesterol 43+14 47+14 45+12 3.7 0.06 mance while not affecting sleep during the night. It has (mg/dl) been hypothesized that these effects of lovastatin may be LDL cholesterol was calculated only on 14 subjects, due to triglyceride attributed to its lipophilicity and its capability of crossing Downloaded from https://academic.oup.com/sleep/article/22/1/117/2731686 by guest on 29 September 2021 levels >400 mg/dl in 2 subjects the blood-brain barrier. Similarly, a decrease in sleep time in hypercholesterolemic subjects who had previously reported a normal sleep pattern was observed after admin- istration of simvastatin, another HMG CoA reductase inhibitor with lipophilic characteristics.13 These effects jects participating in our study had an abnormal sleep pat- were not observed after administration of other cholesterol- tern, as measured by the total arousal index (TAI greater lowering medications.13 On the other hand, in a study in than 5/hour). In one of the two cases without any abnor- 409 hyperlipidemic subjects treated with different classes malities on the sleep studies during the placebo lead-in of cholesterol-lowering medications, no significant differ- phase, the diagnosis of sleep-misperception syndrome was ences were observed when sleep questionnaire responses of made. subjects treated with HMG CoA reductase inhibitors (most- Mean values for each of the sleep parameters at the end ly lovastatin and simvastatin) were compared to those of of each of the randomized treatment phases of the study are subjects on a diet or subjects on other medications.14 shown in Table 2. Values represent the mean for each In the present study, sleep disturbances were present in parameter in the 2 consecutive nights during each phase. In most of the participating subjects during the placebo lead- a few cases, only one measurement was available for one in phase. A questionnaire was used to identify hypercholes- phase. Most sleep parameters were not affected by treat- terolemic subjects who had experienced sleep disturbances ment with either lovastatin or pravastatin when compared during prior lovastatin treatment, and the finding of a high to placebo. Almost no statistically significant differences prevalence of polysomnographic sleep disturbances at the between treatments were noted, perhaps in part because of end of the placebo lead-in phase suggests that our selected the small sample size. Table 2 contains the smallest population had pre-existing sleep problems that were inde- observed differences and ratios that would have been pendent of the lovastatin treatment or only subjectively judged statistically significant at the level of 0.05. The related to the lovastatin treatment. We cannot exclude the effect of treatment on SWS, which includes sleep stages 3 possibility that the placebo lead-in phase, which lasted only and 4 (or delta sleep), was statistically significant (p=0.04). 4 weeks, was too short to allow the residual effects of However, this effect differed according to day of the study lovastatin on sleep to wear off. However, we did not phase (day-by-treatment interaction, p=0.029). On day 1 observe significant differences in any sleep parameters SWS was 20% lower on lovastatin than on pravastatin or between the lovastatin and placebo phases when subjects placebo (p<0.05), but there were no differences between had been on placebo for a total of 8 weeks (4 weeks treatments on day 2 (p=0.09). Similarly, the effect of treat- “washout” plus 4 weeks placebo). ment on AHI was statistically significant (p=0.03), but dif- Our results do not confirm previous observations of fered according to day of the study (day-by-treatment inter- disruption of sleep with lovastatin treatment. It should be action, p=0.003), with no differences in this parameter on noted that in Vgontzas’ study6 sleep disturbances, in partic- day 1 (p=0.34), and AHI values during pravastatin on day ular an increase in wake time after sleep onset and in num- 2 that were approximately twice those of placebo and ber of awakenings, were detected after 2 weeks of treat- lovastatin (p=0.05). ment, while in our study subjects were studied after 4 Lovastatin and pravastatin were equally effective in weeks on medication. It is possible that the effects of lovas- reducing total cholesterol and LDL cholesterol levels in tatin on sleep, if present, are transient and detectable only these hypercholesterolemic subjects, as indicated in Table at the beginning of treatment and may disappear by com- 3. pensatory mechanisms after few weeks. Also, most of our

SLEEP, Vol. 22, No. 1, 1999 120 Lovastatin and pravastatin in sleep—Ehrenberg et al subjects had pre-existing sleep disturbances and it is possi- REFERENCES ble that the effect of lovastatin, if any, may be dependent upon the subjects’ previous sleep status or susceptibility to 1. Schaefer EJ, Genest JJ, Ordovas JM, Salem DN, Wilson PWF. sleep disturbances. Also, pravastatin did not significantly Familial lipoprotein disorders and premature coronary artery disease. Atherosclerosis 1994;108:S41-S54. alter the sleep pattern in our subjects. Even though pravas- 2. Tobert JA. The HMG-CoA reductase inhibitors. Similarities and dif- tatin appeared to increase the proportion of slow-wave ferences. Current Patents sleep (SWS, stages 3 and 4), this effect was significant only 3. Serajuddin ATM, Ranadive SA, Mahoney EM. Relative lipophilici- on day 1, but not on day 2. Since acclimatization may play ties, solubilities, and structure-pharmacological considerations of 3- an important role in sleep measurements, the second day’s hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase inhibitors pravastatin, lovastatin, mevastatin, and simvastatin. J measurement should be a better index of SWS distur- Pharmaceut Sciences 1991;80:830-834. bances, and therefore it is likely that SWS was not affected 4. Botti RE, Triscari J, Pan HY, Zayat J. Concentrations of pravastatin by pravastatin. and lovastatin in cerebrospinal fluid in healthy subjects. Clin Neuropharm 1991;14:256-261.

A similar plasma total cholesterol and LDL choles- Downloaded from https://academic.oup.com/sleep/article/22/1/117/2731686 by guest on 29 September 2021 5. Mahoney EM, Child MJ, Smith-Monroy CA. Differential transport of terol-lowering effect for lovastatin and pravastatin was pravastatin and lovastatin by hepatocytes and fibroblasts. Circulation observed in these subjects. The LDL cholesterol-lowering 1990;82:III-6 (abstract). achieved during treatment with lovastatin and pravastatin 6. Vgontzas AN, Kales A, Bixler EO, Manfredi RL, Tyson KL. Effects (32% and 28%, respectively) was comparable to that of lovastatin and pravastatin on sleep efficiency and sleep stages. Clin reported in other studies with these medications,15,16 indi- Pharmacol Ther 1991;50:730-737. 7. Roth T, Richardson GR, Sullivan JP, Lee RM, Merlotti L, Roehrs T. cating that patients were compliant. During treatment with Comparative effects of pravastatin and lovastatin on nighttime sleep and pravastatin, subjects did not experience significant changes daytime performance. Clin Cardiol 1992;15:426-432. in plasma triglyceride levels, as compared to placebo; this 8. Johnson LC, Church MW, Seales DM, Rossiter VS. Auditory arousal finding is somewhat in disagreement with previous reports threshold of good sleepers and poor sleepers with and without flu- razepam. Sleep 1979;1:250270. of reductions of plasma triglyceride by pravastatin of 5% to 9. Warnick GR, Benderson J, Albers JJ. Dextran sulfate-Mg2+ precipi- 12%.16 tation procedure for quantitation of high density lipoprotein cholesterol. It is important to point out that large long-term clinical Clin Chem 1982;28:1379-1388. safety trials of lovastatin15 and large prospective studies of 10. McNamara JR, Schaefer EJ. Automated enzymatic standardized the prevention of cardiovascular disease mortality and mor- lipid analyses for plasma and lipoprotein fractions. Clin Chim Acta 1987;166:1-8. 16 bidity with pravastatin have been performed, and no over- 11. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the con- all identifiable effects of either medication on sleep have centration of low-density lipoprotein cholesterol in plasma, without use been observed when compared to placebo. of the preparative ultracentrifuge. Clin Chem 1972;18:499-502. In conclusion, our results do not indicate a significant 12. Schaefer EJ. HMG-CoA reductase inhibitors for hypercholes- terolemia. N Enlg J Med (letter) 1988;319:1222. effect of lovastatin or pravastatin on objective sleep mea- 13. Barth JD, Kruisbrink OAE, van Dijk AL. Inhibitors of hydrox- sures. However, the response to lovastatin treatment in our ymethylglutaryl coenzyme A reductase for treating hypercholestero- subjects might have been affected by the presence of pre- laemia. BMJ (letter) 1990;301:669. existing sleep problems. Nevertheless, in view of our find- 14. Black DM, Lamkin G, Olivera EH, Laskarzewski PM, Stein EA. ings and those of other studies, there do not appear to be Sleep disturbance and HMG CoA reductase inhibitors. JAMA (letter) 1990;264:1105. any indication of strong and consistent effects of lovastatin 15. Mantell G, Burke MT, Staggers J. Extended clinical safety profile on objective sleep measures. of lovastatin. Am J Cardiol 1990;66:11B-15B. 16. Sheperd J, Cobbe SM, Ford I et al. Prevention of coronary heart ACKNOWLEDGMENTS disease with pravastatin in men with hypercholesteroemia. N Engl J Med 1995;333:1301-1307. This work was supported by a research grant from Bristol-Myers Squibb.

SLEEP, Vol. 22, No. 1, 1999 121 Lovastatin and pravastatin in sleep—Ehrenberg et al