- Email: [email protected]
Elevation of fasting morning glucose relative to hemoglobin A1c in normoglycemic patients treated with niacin and with statins
Journal of Clinical Lipidology (2012) 6, 168–173
Elevation of fasting morning glucose relative to hemoglobin A1c in normoglycemic patients treated with niacin and with statins Veena Rajanna, PharmD, Kristen B. Campbell, PharmD, Jeffrey Leimberger, PhD, Bibhu D. Mohanty, MD, John R. Guyton, MD* Beaumont Hospital, Royal Oak, MI, USA (Dr. Rajanna); Duke University Medical Center, Box 3510, Durham, NC 27710, USA (Drs. Campbell and Guyton); Duke Clinical Research Institute, Durham, NC, USA (Dr. Leimberger); and Mount Sinai Hospital, New York, NY, USA (Dr. Mohanty) KEYWORDS: Fibrates; Glucose; Hemoglobin A1c; Niacin; Statins
BACKGROUND: Niacin increases fasting glucose levels, and statins modestly increase the rate of new-onset diabetes. The clinical importance and mechanisms of these effects are not fully explored. OBJECTIVE: On the basis of anecdotal observations, we hypothesized that elevated morning fasting glucose may be accompanied by relatively normal hemoglobin A1c (HbA1c) in patients treated with niacin and other lipid-modifying drugs. We conducted a retrospective cohort analysis to test this hypothesis. METHODS: The Duke Lipid Clinic database (1994–2007) was screened for simultaneous determinations of fasting morning glucose and HbA1c, yielding 1483 data pairs among 554 subjects. Subjects with diabetes, by clinical diagnosis, medication, or any HbA1c $6.5%, or nondiabetes were analyzed separately. Repeated-measures linear regression featured glucose as dependent variable and included terms for HbA1c, drug(s), and their interaction. RESULTS: Regression lines for glucose on HbA1c had altered slopes in the presence of niacin and/ or statin use in normoglycemic subjects. The corresponding interaction terms (drug and HbA1c) were significant (niacin P 5 .026, statin P 5 .013). Fibrate use had no effect (interaction P 5 .49). When modeled together, niacin and statin effects were independent. Regression curves in diabetic patients were not affected by lipid medications. CONCLUSION: Elevated fasting glucose may be accompanied by relatively normal HbA1c in niacin- and statin-treated patients. HbA1c reflects average daily glucose levels and is likely a better measure of the glycemic effect of lipid medications. Because our data were retrospective, confirmation from randomized trials is needed. Ó 2012 National Lipid Association. All rights reserved.
Statins and niacin improve the lipoprotein profile and reduce cardiovascular events. Despite the lack of efficacy * Corresponding author. E-mail address: [email protected] Submitted September 13, 2011. Accepted for publication December 20, 2011.
for niacin found in a recent large randomized clinical trial, four earlier randomized trials in which the authors used niacin in monotherapy or in combination therapy demonstrated significant cardiovascular event reductions.1–5 In the Coronary Drug Project, niacin monotherapy lowered rates of nonfatal myocardial infarction and cerebrovascular events.1
1933-2874/$ - see front matter Ó 2012 National Lipid Association. All rights reserved. doi:10.1016/j.jacl.2011.12.008
Rajanna et al
Fasting morning glucose and lipid medications
Niacin is known to increase plasma glucose levels, an effect that sometimes leads health care providers to avoid or discontinue niacin therapy.2,6–8 Initial concerns about glucose raising by niacin have been allayed by several new findings: (1) Niacin treatment revealed the same trends for event reduction in the Coronary Drug Project regardless of the glycemic status of patients.9 (2) The lower doses of niacin used currently (1500–2000 mg total daily dose) induce only modest hyperglycemia and worsening of diabetic control, compared with doses (3000–4500 mg) used in the past.6,10,11 (3) The hyperglycemic effect of niacin peaks at 8 to 12 weeks after initiation and subsides thereafter.12 Nevertheless, for the rational use of niacin, clinicians need a clearer understanding of the balance between effects on lipoproteins and glucose. Recently, statins have also been shown to affect glucoseinsulin homeostasis. Meta-analyses of large randomized trials demonstrate that rates of new-onset diabetes are modestly increased in statin-treated subjects.13,14 In an academic lipid clinic, we encountered instances of high fasting glucose in niacin-treated patients accompanied by relatively normal hemoglobin A1c (HbA1c). HbA1c reflects daily average glucose levels and thus provides a better estimate of overall glucose homeostasis than morning fasting glucose. A computer database maintained since 1992 in our lipid clinic contained 2446 pairs of glucose and HbA1c measurements as well as information on diagnoses and drug treatment, providing the opportunity for a retrospective cohort study. We developed an analysis to ask whether the relationship between fasting morning glucose and HbA1c was altered in association with treatment by 3 major classes of lipid drugs: statins, niacin, and fibrates.
169 33%, unknown timing of blood draw for glucose, or the lack of a corresponding clinic visit note. We also noted (1) diagnosis of type 1 diabetes mellitus, type 2 diabetes mellitus, impaired fasting glucose ($100 mg/dL), impaired glucose tolerance, or apparently normal glucose homeostasis, (2) lipid and cardiovascular diagnoses, (3) niacin dose, formulation (immediate-release, extended-release, or slow-release), and duration of treatment, and (4) whether the patient was on statin or fibrate therapy. In designing the analysis, we recognized that subjects with and without diabetes should be analyzed separately. Fasting glucose levels are much more variable in individuals with diabetes compared with those without diabetes. If certain medications are used disproportionately in patients with and without diabetes, artifacts can be introduced. This is precisely the case for niacin, because the clinical tendency is to avoid the use of niacin in patients with diabetes. Thus, subjects with and without diabetes were analyzed separately. Subjects were considered to have diabetes if there was a clinical diagnosis, if antidiabetic medication was used, or if HBA1c was $ 6.5%. We decided to regress fasting glucose on HbA1c, rather than vice versa, because HbA1c in a given patient is more stable than fasting glucose. Repeated-measures linear regression comparing patients on niacin, statin, or fibrate to patients not taking each medication was used to determine any statistically significant differences in the regression of glucose on HbA1c (alpha ,.05, two-tailed). Fasting glucose was logtransformed to approximate more closely a normal distribution. Predictive values from modeling were transformed to the original scale for plotting. All analyses were conducted within SAS Proc Mixed v. 8.2 (SAS Institute, Cary, NC).
Methods We conducted a single-center, retrospective cohort analysis of the effect of lipid medications on the relationship between fasting morning glucose and HbA1c values in adult patients treated in the Duke Lipid Clinic between 1994 and July 2007. Patients were identified through an electronic database which is continuously maintained for Lipid Clinic patients, and data were confirmed by reference to the Duke University electronic medical record as well as paper office charts. Data were included if fasting morning glucose and HbA1c were determined within 1 week of each other (usually from the same blood draw). Most glucose levels were not specifically stated to be fasting or nonfasting, but fasting status was assumed if the specimen was drawn between 7:00 a.m. and 12:00 noon in a patient who simultaneously had a lipid panel performed. Inpatient blood draws were excluded. HbA1c levels were excluded if the upper limit of normal fell outside the range of 5.7% to 6.3%. Other exclusions were total dose of niacin less than 200 mg per day, use of inositol hexanicotinate, serum creatinine greater than 3 mg/dL, or hematocrit less than
Results Screening of the Lipid Clinic database provided 2446 pairs of glucose and HbA1c measurements derived from 977 patients. After review of original laboratory reports and clinic notes, 1483 paired measurements of morning fasting glucose and HbA1c from 554 subjects were included. Frequent reasons for exclusion were nonfasting or afternoon glucose, nonsimultaneous HbA1c, inpatient laboratory tests, and lack of a corresponding clinic note. Niacin treatment preceded 589 pairs of glucose/HbA1c, statin treatment 985 pairs, and fibrate treatment 554 pairs in overlapping fashion. Baseline demographics and diagnoses according to drug class are shown in Table 1. The most common lipid diagnosis was hypertriglyceridemia, which is known to be associated with greater glucose levels that would prompt measurement of HbA1c. Results from regression of fasting glucose on HbA1c in subjects without diabetes are shown in Figure 1, according to treatment with and without niacin, statin, and fibrate. The
170 Table 1
Journal of Clinical Lipidology, Vol 6, No 2, April 2012 Patient characteristics by class of medications taken
N Mean age (SD) Male/female Ethnicity White Black Other Clinical atherosclerosis Lipid diagnosis Combined hyperlipidemia Hypertriglyceridemia Hypercholesterolemia Other
Niacin, n 5 224
Not on niacin, n 5 408
Statin, n 5 356
Not on statin, n 5 266
Fibrate, n 5 191
57.7 (10.6) 65/35
56.4 (11.5) 65/35
57.2 (10.8) 62/38
56.3 (11.6) 58/42
55.0 (11.0) 62/38
93 6 1 45
85 10 5 38
86 10 3 48
87 8 6 28
89 5 6 42
34 45 14 7
32 44 19 5
33 42 20 5
32 48 14 6
26 66 6 2
Percentages of patients are shown according to gender, ethnicity, clinically manifested atherosclerotic disease, and lipid diagnosis. Other lipid diagnoses included high lipoprotein(a), low high-density lipoprotein cholesterol, and lipoprotein-X.
regression lines were altered by treatment with niacin or with statin but not with fibrate. Morning fasting glucose levels were elevated with niacin or statin treatment, but the effect occurred only at lower levels of HbA1c. Statistically, the effect of niacin was revealed by a significant interaction term HbA1c and niacin treatment (P 5 0.026), whereas the niacin effect on fasting glucose at mean HbA1c was not significant. Likewise, for statin treatment the interaction term HbA1c and statin treatment (P 5 .013) was significant whereas the statin effect at mean HbA1c was not. For fibrates, neither the effect of treatment nor the interaction term (P 5 .49) were significant. Because niacin and statin treatment, considered singly, were each associated with significant effects, we repeated the analysis in subjects without diabetes including both variables. Regression lines for no drug treatment, statin only, niacin only, and statin 1 niacin are shown in Figure 2. Drug effects on the regression lines appear to be similar at lower levels of HbA1C. In the statistical model, niacin and statin treatment both interacted significantly with level of HbA1C in predicting fasting glucose (HbA1c and niacin treatment, P 5 .042, and HbA1c and statin treatment, P 5 0.023). This finding suggests that the effects of niacin and statins on regression slopes were independent. The predictive models for fasting glucose among subjects with diabetes are shown in Figure 3. No significant
differences were found related to treatment versus nontreatment with niacin, statin, or fibrate. Niacin preparations differ in terms of release characteristics and dosing regimens. Extended-release niacin is recommended in our clinic for use as a single daily dose at bedtime, consistent with the approved prescribing information. Dietary supplement niacin, which is almost always immediate-release in our clinic, is recommended for use twice daily with breakfast and supper. Because the dosing regimen and release characteristics might affect diurnal glucose variation, we compared the regression lines of bedtime-dosed extended-release niacin (on the basis of 96 data points) versus twice daily dietary supplement niacin (on the basis of 85 data points). Figure 4 gives no evidence for differences in the regression lines. Thus, niacin formulation and dosing did not appear to play a role. The regression lines for both niacin dosing regimens appear to differ from the regression line for subjects not taking niacin, but in this analysis the numbers of measurements were too small to show statistically significant differences.
Discussion This retrospective cohort study examining 554 patients confirms our casual observations that niacin-treated patients
Figure 1 Regression of morning fasting glucose on HbA1c in subjects without diabetes, according to treatment (solid lines) or nontreatment (dashed lines) with niacin, statins, or fibrates. The predicted regression lines (thick) and 95% confidence limits (thin) are shown.
Rajanna et al
Fasting morning glucose and lipid medications
Figure 2 Regression of morning fasting glucose on HbA1c in subjects without diabetes in a model including both niacin and statin as treatment variables.
tend to have high fasting morning glucose levels accompanied by relatively normal HbA1c. However, the effect did not occur in people with diabetes. Among those without diabetes, the regression lines showed separation according to niacin treatment or nontreatment only in those with more robust glucose homeostasis, that is, hemoglobin A1c in the lower half of the normal range. This effect was statistically significant (P , .026 for the interaction term). Bedtime dosing of extended-release niacin might be expected to affect fasting glucose levels differently from twice-daily immediate-release niacin taken with meals, but we found no evidence for a difference between the two dosing regimens (Fig. 4). Statin treatment showed the same association with greater fasting morning glucose levels relative to HbA1c, and this effect was also limited to patients who had HbA1c in the lower half of the normal range (P , .013 for the interaction term). Moreover, niacin and statin effects were independent (Fig. 2). Our retrospective analysis has limitations because decisions to use certain drugs and to obtain measurement of hemoglobin A1c were made in the course of clinical practice. The data do not exclude the possibility that some aspect of lipid treatment, perhaps even dietary, rather than medication may be responsible for relatively high morning glucose levels. However, not all lipid-modifying drugs were associated with the phenomenon because fibrates (gemfibrozil or fenofibrate) had no effect (Fig. 1).
171 It seems likely that the associations seen with niacin and statins represent pharmacologic effects, but confirmation is required from randomized controlled trials. Niacin is known to increase fasting glucose levels.2,6 Recent data suggest that most of the increase in fasting glucose is transient, peaking at 8 to 12 weeks of niacin administration and decreasing thereafter, although a persistent 5% increase in fasting glucose has been suggested for long-term use of niacin.8,12,15 For statins, clinical reports of increased fasting glucose have been inconsistent and generally not confirmed in randomized controlled trials.16–18 However, statins modestly increase the risk of developing diabetes (relative risk of 1.09 in large meta-analyses).13,14 Because HbA1c reflects average glucose concentration throughout 24 hours, our data suggest that niacin and statins change diurnal glucose variation, such that morning fasting glucose levels are elevated with respect to glucose levels through most of the day. The clinical implication is that a moderately elevated fasting glucose in a patient without diabetes does not necessarily imply 24-hour hyperglycemia. HbA1c should be measured to assess overall impairment glucose homeostasis. The present study reinforces the notion that lipid medications have pharmacologic effects beyond lipoproteins, but mechanisms remain speculative. Fasting morning glucose levels largely reflect hepatic glucose metabolism, which is influenced by nocturnal variation of counterregulatory hormones. Increasing levels have been described for growth hormone at 2400 to 0400 hours, for cortisol after 0330 hours, and for catecholamines after 0330 hours.19 In normal humans, hepatic glucose production increases after 0530 hours with a concomitant rise in circulating insulin, so that morning glucose levels either remain constant or increase modestly.20,21 The ‘‘dawn phenomenon’’—marked glucose elevation in patients with diabetes and fixed insulin availability—is thought to be an exaggerated manifestation of normal physiology without the braking effect of pancreatic insulin responsiveness.22 Likewise, increased morning fasting glucose levels in our patients taking niacin and statins might represent an unrestrained response to nocturnal counterregulatory hormone secretion. Niacin could trigger counterregulatory hormone responses in the following manner: At pharmacologic doses, niacin binds the G-protein coupled receptor GPR109A in adipocytes, decreasing cyclic AMP and strongly inhibiting
Figure 3 Regression of morning fasting glucose on HbA1c in subjects with diabetes, according to treatment (solid lines) or nontreatment (dashed lines) with niacin, statins, or fibrates. The predicted regression lines (thick) and 95% confidence limits (thin) are shown.
172
Journal of Clinical Lipidology, Vol 6, No 2, April 2012 speaker’s honoraria from Abbott and Merck; and consultant fees from Merck and sanofi aventis and owns stock in Eli Lilly. This research study was funded by the Merck Investigator Study Program.
References
Figure 4 Regression of morning fasting glucose on HbA1c in subjects without diabetes, according to treatment with differing niacin regimens versus no niacin treatment. Niacin given once daily was generally extended-release niacin at bedtime. Niacin given twice daily was generally immediate-release niacin at breakfast and supper.
lipolysis.23,24 Consequently, plasma nonesterified fatty acid (NEFA) levels decrease sharply, but this is followed by a rebound and overshoot of NEFA substantially above baseline levels.25 The maintenance of relatively normal glucose levels during the time that NEFA are unavailable may be due to counterregulatory hormones. The same hormonal responses may lead to the rebound and overshoot of NEFA.26 Statin effects on glucose-insulin homeostasis have been studied in a variety of settings, but a clear explanation for the increased risk of diabetes is not yet available.17 Both atorvastatin and rosuvastatin were found to increase fasting insulin levels, suggesting perhaps an interplay with nocturnal counterregulatory hormone dynamics.27
Conclusion This retrospective cohort study suggests that in patients without diabetes, niacin and statins increase fasting morning glucose levels out of proportion to HbA1c. The effect requires robust glucose homeostasis (HbA1c in the lower half of the normal range), and it is not seen at all in patients with diabetes. Because our data are retrospective, confirmation in randomized clinical trials is needed. Recognizing the potential effect of niacin and statins on fasting glucose levels should help to inform clinical decisions, particularly the need for HbA1c measurement to judge the overall impact on glycemia, and may help to direct research on mechanisms.
Financial disclosure Dr. Guyton has received research funding from Abbott, Amarin, Isis/Genzyme, Merck, Regeneron/sanofi aventis;
1. Coronary Drug Project Group. Clofibrate and niacin in coronary heart disease. J Am Med Assoc. 1975;231:360–381. 2. Carlson LA, Rosenhamer G. Reduction of mortality in the Stockholm Ischaemic Heart Disease Secondary Prevention Study by combined treatment with clofibrate and nicotinic acid. Acta Med Scand. 1988; 223:405–418. 3. Brown BG, Zhao XQ, Chait A, et al. Simvastatin and niacin, antioxidant vitamins, or the combination for the prevention of coronary disease. N Engl J Med. 2001;345:1583–1592. 4. Whitney EJ, Krasuski RA, Personius BE, et al. A randomized trial of a strategy for increasing high-density lipoprotein cholesterol levels: effects on progression of coronary heart disease and clinical events. Ann Intern Med. 2005;142:95–104. 5. Boden WE, Probstfield JL, Anderson T, et al; AIM-HIGH Investigators. Niacin in patients with low HDL cholesterol levels receiving intensive statin therapy. N Engl J Med. 2011;365:2255–2267. 6. Garg A, Grundy SM. Nicotinic acid as therapy for dyslipidemia in non-insulin-dependent diabetes mellitus. J Am Med Assoc. 1990;264: 723–726. 7. Gibbons LW, Gonzalez V, Gordon N, Grundy S. The prevalence of side effects with regular and sustained-release nicotinic acid. Am J Med. 1995;99:378–385. 8. Goldberg RB, Jacobson TA. Effects of niacin on glucose control in patients with dyslipidemia. Mayo Clin Proc. 2008;83:470–478. 9. Canner PL, Furberg CD, Terrin ML, McGovern ME. Benefits of niacin by glycemic status in patients with healed myocardial infarction (from the Coronary Drug Project). Am J Cardiol. 2005;95:254–257. 10. Grundy SM, Vega GL, McGovern ME, et al. Efficacy, safety, and tolerability of once-daily niacin for the treatment of dyslipidemia associated with type 2 diabetes. Arch Intern Med. 2002;162:1568–1576. 11. Elam MB, Hunninghake DB, Davis KB, et al, Diabetes Multicenter Research Group. Effect of niacin on lipid and lipoprotein levels and glycemic control in patients with diabetes and peripheral arterial disease: the ADMIT study: A randomized trial. Arterial Disease Multiple Intervention Trial. J Am Med Assoc. 2000;284:1263–1270. 12. Guyton JR, Brown BG, Fazio S, Polis A, Tomassini JE, Tershakovec AM. Lipid-altering efficacy and safety of ezetimibe/ simvastatin coadministered with extended-release niacin in patients with type IIa or type IIb hyperlipidemia. J Am Coll Cardiol. 2008; 51:1564–1572. 13. Mills EJ, Wu P, Chong G, et al. Efficacy and safety of statin treatment for cardiovascular disease: a network meta-analysis of 170,255 patients from 76 randomized trials. Q J Med. 2011;104:109–124. 14. Sattar N, Preiss D, Murray HM, et al. Statins and risk of incident diabetes: a collaborative meta-analysis of randomised statin trials. Lancet. 2010;375:735–742. 15. Guyton JR, Goldberg AC, Kreisberg RA, et al. Effectiveness of oncenightly dosing of extended-release niacin alone and in combination for hypercholesterolemia. Am J Cardiol. 1998;82:737–743. 16. Sasaki J, Iwashita M, Kono S. Statins: beneficial or adverse for glucose metabolism. J Atheroscler Thromb. 2006;13:123–129. 17. Koh KK, Sakuma I, Quon MJ. Differential metabolic effects of distinct statins. Atherosclerosis. 2011;215:1–8. 18. Ridker PM, Danielson E, Fonseca FA, et al, JUPITER Study Group. Rosuvastatin to prevent vascular events in men and women with elevated C-reactive protein. N Engl J Med. 2008;359:2195–2207. 19. Campbell PJ, Bolli GB, Cryer PE, Gerich JE. Sequence of events during development of the dawn phenomenon in insulin-dependent diabetes mellitus. Metabolism. 1985;34:1100–1104.
Rajanna et al
Fasting morning glucose and lipid medications
20. Schmidt MI, Lin QX, Gwynne JT, et al. Fasting early morning rise in peripheral insulin: evidence of the dawn phenomenon in nondiabetes. Diabetes Care. 1984;7:32–35. 21. Bolli GB, De Feo P, De Cosmo S, et al. Demonstration of a dawn phenomenon in normal human volunteers. Diabetes. 1984;33: 1150–1153. 22. Bolli GB, Gerich JE. The ‘‘dawn phenomenon’’—a common occurrence in both non-insulin-dependent and insulin-dependent diabetes mellitus. N Engl J Med. 1984;310:746–750. 23. Guyton JR. Niacin in cardiovascular prevention: mechanisms, efficacy, and safety. Curr Opin Lipidol. 2007;18:415–420.
173 24. Tunaru S, Kero J, Schaub A, et al. PUMA-G and HM74 are receptors for nicotinic acid and mediate its anti-lipolytic effect. Nat Med. 2003; 9:352–355. 25. Carlson LA. Nicotinic acid: the broad-spectrum lipid drug. A 50th anniversary review. J Intern Med. 2005;258:94–114. 26. Vega GL, Cater NB, Meguro S, Grundy SM. Influence of extended-release nicotinic acid on nonesterified fatty acid flux in the metabolic syndrome with atherogenic dyslipidemia. Am J Cardiol. 2005;95:1309–1313. 27. Thongtang N, Ai M, Otokozawa S, et al. Effects of maximal atorvastatin and rosuvastatin treatment on markers of glucose homeostasis and inflammation. Am J Cardiol. 2011;107:387–392.
Elevation of fasting morning glucose relative to hemoglobin A1c in normoglycemic patients treated with niacin and with statins Veena Rajanna, PharmD, Kristen B. Campbell, PharmD, Jeffrey Leimberger, PhD, Bibhu D. Mohanty, MD, John R. Guyton, MD* Beaumont Hospital, Royal Oak, MI, USA (Dr. Rajanna); Duke University Medical Center, Box 3510, Durham, NC 27710, USA (Drs. Campbell and Guyton); Duke Clinical Research Institute, Durham, NC, USA (Dr. Leimberger); and Mount Sinai Hospital, New York, NY, USA (Dr. Mohanty) KEYWORDS: Fibrates; Glucose; Hemoglobin A1c; Niacin; Statins
BACKGROUND: Niacin increases fasting glucose levels, and statins modestly increase the rate of new-onset diabetes. The clinical importance and mechanisms of these effects are not fully explored. OBJECTIVE: On the basis of anecdotal observations, we hypothesized that elevated morning fasting glucose may be accompanied by relatively normal hemoglobin A1c (HbA1c) in patients treated with niacin and other lipid-modifying drugs. We conducted a retrospective cohort analysis to test this hypothesis. METHODS: The Duke Lipid Clinic database (1994–2007) was screened for simultaneous determinations of fasting morning glucose and HbA1c, yielding 1483 data pairs among 554 subjects. Subjects with diabetes, by clinical diagnosis, medication, or any HbA1c $6.5%, or nondiabetes were analyzed separately. Repeated-measures linear regression featured glucose as dependent variable and included terms for HbA1c, drug(s), and their interaction. RESULTS: Regression lines for glucose on HbA1c had altered slopes in the presence of niacin and/ or statin use in normoglycemic subjects. The corresponding interaction terms (drug and HbA1c) were significant (niacin P 5 .026, statin P 5 .013). Fibrate use had no effect (interaction P 5 .49). When modeled together, niacin and statin effects were independent. Regression curves in diabetic patients were not affected by lipid medications. CONCLUSION: Elevated fasting glucose may be accompanied by relatively normal HbA1c in niacin- and statin-treated patients. HbA1c reflects average daily glucose levels and is likely a better measure of the glycemic effect of lipid medications. Because our data were retrospective, confirmation from randomized trials is needed. Ó 2012 National Lipid Association. All rights reserved.
Statins and niacin improve the lipoprotein profile and reduce cardiovascular events. Despite the lack of efficacy * Corresponding author. E-mail address: [email protected] Submitted September 13, 2011. Accepted for publication December 20, 2011.
for niacin found in a recent large randomized clinical trial, four earlier randomized trials in which the authors used niacin in monotherapy or in combination therapy demonstrated significant cardiovascular event reductions.1–5 In the Coronary Drug Project, niacin monotherapy lowered rates of nonfatal myocardial infarction and cerebrovascular events.1
1933-2874/$ - see front matter Ó 2012 National Lipid Association. All rights reserved. doi:10.1016/j.jacl.2011.12.008
Rajanna et al
Fasting morning glucose and lipid medications
Niacin is known to increase plasma glucose levels, an effect that sometimes leads health care providers to avoid or discontinue niacin therapy.2,6–8 Initial concerns about glucose raising by niacin have been allayed by several new findings: (1) Niacin treatment revealed the same trends for event reduction in the Coronary Drug Project regardless of the glycemic status of patients.9 (2) The lower doses of niacin used currently (1500–2000 mg total daily dose) induce only modest hyperglycemia and worsening of diabetic control, compared with doses (3000–4500 mg) used in the past.6,10,11 (3) The hyperglycemic effect of niacin peaks at 8 to 12 weeks after initiation and subsides thereafter.12 Nevertheless, for the rational use of niacin, clinicians need a clearer understanding of the balance between effects on lipoproteins and glucose. Recently, statins have also been shown to affect glucoseinsulin homeostasis. Meta-analyses of large randomized trials demonstrate that rates of new-onset diabetes are modestly increased in statin-treated subjects.13,14 In an academic lipid clinic, we encountered instances of high fasting glucose in niacin-treated patients accompanied by relatively normal hemoglobin A1c (HbA1c). HbA1c reflects daily average glucose levels and thus provides a better estimate of overall glucose homeostasis than morning fasting glucose. A computer database maintained since 1992 in our lipid clinic contained 2446 pairs of glucose and HbA1c measurements as well as information on diagnoses and drug treatment, providing the opportunity for a retrospective cohort study. We developed an analysis to ask whether the relationship between fasting morning glucose and HbA1c was altered in association with treatment by 3 major classes of lipid drugs: statins, niacin, and fibrates.
169 33%, unknown timing of blood draw for glucose, or the lack of a corresponding clinic visit note. We also noted (1) diagnosis of type 1 diabetes mellitus, type 2 diabetes mellitus, impaired fasting glucose ($100 mg/dL), impaired glucose tolerance, or apparently normal glucose homeostasis, (2) lipid and cardiovascular diagnoses, (3) niacin dose, formulation (immediate-release, extended-release, or slow-release), and duration of treatment, and (4) whether the patient was on statin or fibrate therapy. In designing the analysis, we recognized that subjects with and without diabetes should be analyzed separately. Fasting glucose levels are much more variable in individuals with diabetes compared with those without diabetes. If certain medications are used disproportionately in patients with and without diabetes, artifacts can be introduced. This is precisely the case for niacin, because the clinical tendency is to avoid the use of niacin in patients with diabetes. Thus, subjects with and without diabetes were analyzed separately. Subjects were considered to have diabetes if there was a clinical diagnosis, if antidiabetic medication was used, or if HBA1c was $ 6.5%. We decided to regress fasting glucose on HbA1c, rather than vice versa, because HbA1c in a given patient is more stable than fasting glucose. Repeated-measures linear regression comparing patients on niacin, statin, or fibrate to patients not taking each medication was used to determine any statistically significant differences in the regression of glucose on HbA1c (alpha ,.05, two-tailed). Fasting glucose was logtransformed to approximate more closely a normal distribution. Predictive values from modeling were transformed to the original scale for plotting. All analyses were conducted within SAS Proc Mixed v. 8.2 (SAS Institute, Cary, NC).
Methods We conducted a single-center, retrospective cohort analysis of the effect of lipid medications on the relationship between fasting morning glucose and HbA1c values in adult patients treated in the Duke Lipid Clinic between 1994 and July 2007. Patients were identified through an electronic database which is continuously maintained for Lipid Clinic patients, and data were confirmed by reference to the Duke University electronic medical record as well as paper office charts. Data were included if fasting morning glucose and HbA1c were determined within 1 week of each other (usually from the same blood draw). Most glucose levels were not specifically stated to be fasting or nonfasting, but fasting status was assumed if the specimen was drawn between 7:00 a.m. and 12:00 noon in a patient who simultaneously had a lipid panel performed. Inpatient blood draws were excluded. HbA1c levels were excluded if the upper limit of normal fell outside the range of 5.7% to 6.3%. Other exclusions were total dose of niacin less than 200 mg per day, use of inositol hexanicotinate, serum creatinine greater than 3 mg/dL, or hematocrit less than
Results Screening of the Lipid Clinic database provided 2446 pairs of glucose and HbA1c measurements derived from 977 patients. After review of original laboratory reports and clinic notes, 1483 paired measurements of morning fasting glucose and HbA1c from 554 subjects were included. Frequent reasons for exclusion were nonfasting or afternoon glucose, nonsimultaneous HbA1c, inpatient laboratory tests, and lack of a corresponding clinic note. Niacin treatment preceded 589 pairs of glucose/HbA1c, statin treatment 985 pairs, and fibrate treatment 554 pairs in overlapping fashion. Baseline demographics and diagnoses according to drug class are shown in Table 1. The most common lipid diagnosis was hypertriglyceridemia, which is known to be associated with greater glucose levels that would prompt measurement of HbA1c. Results from regression of fasting glucose on HbA1c in subjects without diabetes are shown in Figure 1, according to treatment with and without niacin, statin, and fibrate. The
170 Table 1
Journal of Clinical Lipidology, Vol 6, No 2, April 2012 Patient characteristics by class of medications taken
N Mean age (SD) Male/female Ethnicity White Black Other Clinical atherosclerosis Lipid diagnosis Combined hyperlipidemia Hypertriglyceridemia Hypercholesterolemia Other
Niacin, n 5 224
Not on niacin, n 5 408
Statin, n 5 356
Not on statin, n 5 266
Fibrate, n 5 191
57.7 (10.6) 65/35
56.4 (11.5) 65/35
57.2 (10.8) 62/38
56.3 (11.6) 58/42
55.0 (11.0) 62/38
93 6 1 45
85 10 5 38
86 10 3 48
87 8 6 28
89 5 6 42
34 45 14 7
32 44 19 5
33 42 20 5
32 48 14 6
26 66 6 2
Percentages of patients are shown according to gender, ethnicity, clinically manifested atherosclerotic disease, and lipid diagnosis. Other lipid diagnoses included high lipoprotein(a), low high-density lipoprotein cholesterol, and lipoprotein-X.
regression lines were altered by treatment with niacin or with statin but not with fibrate. Morning fasting glucose levels were elevated with niacin or statin treatment, but the effect occurred only at lower levels of HbA1c. Statistically, the effect of niacin was revealed by a significant interaction term HbA1c and niacin treatment (P 5 0.026), whereas the niacin effect on fasting glucose at mean HbA1c was not significant. Likewise, for statin treatment the interaction term HbA1c and statin treatment (P 5 .013) was significant whereas the statin effect at mean HbA1c was not. For fibrates, neither the effect of treatment nor the interaction term (P 5 .49) were significant. Because niacin and statin treatment, considered singly, were each associated with significant effects, we repeated the analysis in subjects without diabetes including both variables. Regression lines for no drug treatment, statin only, niacin only, and statin 1 niacin are shown in Figure 2. Drug effects on the regression lines appear to be similar at lower levels of HbA1C. In the statistical model, niacin and statin treatment both interacted significantly with level of HbA1C in predicting fasting glucose (HbA1c and niacin treatment, P 5 .042, and HbA1c and statin treatment, P 5 0.023). This finding suggests that the effects of niacin and statins on regression slopes were independent. The predictive models for fasting glucose among subjects with diabetes are shown in Figure 3. No significant
differences were found related to treatment versus nontreatment with niacin, statin, or fibrate. Niacin preparations differ in terms of release characteristics and dosing regimens. Extended-release niacin is recommended in our clinic for use as a single daily dose at bedtime, consistent with the approved prescribing information. Dietary supplement niacin, which is almost always immediate-release in our clinic, is recommended for use twice daily with breakfast and supper. Because the dosing regimen and release characteristics might affect diurnal glucose variation, we compared the regression lines of bedtime-dosed extended-release niacin (on the basis of 96 data points) versus twice daily dietary supplement niacin (on the basis of 85 data points). Figure 4 gives no evidence for differences in the regression lines. Thus, niacin formulation and dosing did not appear to play a role. The regression lines for both niacin dosing regimens appear to differ from the regression line for subjects not taking niacin, but in this analysis the numbers of measurements were too small to show statistically significant differences.
Discussion This retrospective cohort study examining 554 patients confirms our casual observations that niacin-treated patients
Figure 1 Regression of morning fasting glucose on HbA1c in subjects without diabetes, according to treatment (solid lines) or nontreatment (dashed lines) with niacin, statins, or fibrates. The predicted regression lines (thick) and 95% confidence limits (thin) are shown.
Rajanna et al
Fasting morning glucose and lipid medications
Figure 2 Regression of morning fasting glucose on HbA1c in subjects without diabetes in a model including both niacin and statin as treatment variables.
tend to have high fasting morning glucose levels accompanied by relatively normal HbA1c. However, the effect did not occur in people with diabetes. Among those without diabetes, the regression lines showed separation according to niacin treatment or nontreatment only in those with more robust glucose homeostasis, that is, hemoglobin A1c in the lower half of the normal range. This effect was statistically significant (P , .026 for the interaction term). Bedtime dosing of extended-release niacin might be expected to affect fasting glucose levels differently from twice-daily immediate-release niacin taken with meals, but we found no evidence for a difference between the two dosing regimens (Fig. 4). Statin treatment showed the same association with greater fasting morning glucose levels relative to HbA1c, and this effect was also limited to patients who had HbA1c in the lower half of the normal range (P , .013 for the interaction term). Moreover, niacin and statin effects were independent (Fig. 2). Our retrospective analysis has limitations because decisions to use certain drugs and to obtain measurement of hemoglobin A1c were made in the course of clinical practice. The data do not exclude the possibility that some aspect of lipid treatment, perhaps even dietary, rather than medication may be responsible for relatively high morning glucose levels. However, not all lipid-modifying drugs were associated with the phenomenon because fibrates (gemfibrozil or fenofibrate) had no effect (Fig. 1).
171 It seems likely that the associations seen with niacin and statins represent pharmacologic effects, but confirmation is required from randomized controlled trials. Niacin is known to increase fasting glucose levels.2,6 Recent data suggest that most of the increase in fasting glucose is transient, peaking at 8 to 12 weeks of niacin administration and decreasing thereafter, although a persistent 5% increase in fasting glucose has been suggested for long-term use of niacin.8,12,15 For statins, clinical reports of increased fasting glucose have been inconsistent and generally not confirmed in randomized controlled trials.16–18 However, statins modestly increase the risk of developing diabetes (relative risk of 1.09 in large meta-analyses).13,14 Because HbA1c reflects average glucose concentration throughout 24 hours, our data suggest that niacin and statins change diurnal glucose variation, such that morning fasting glucose levels are elevated with respect to glucose levels through most of the day. The clinical implication is that a moderately elevated fasting glucose in a patient without diabetes does not necessarily imply 24-hour hyperglycemia. HbA1c should be measured to assess overall impairment glucose homeostasis. The present study reinforces the notion that lipid medications have pharmacologic effects beyond lipoproteins, but mechanisms remain speculative. Fasting morning glucose levels largely reflect hepatic glucose metabolism, which is influenced by nocturnal variation of counterregulatory hormones. Increasing levels have been described for growth hormone at 2400 to 0400 hours, for cortisol after 0330 hours, and for catecholamines after 0330 hours.19 In normal humans, hepatic glucose production increases after 0530 hours with a concomitant rise in circulating insulin, so that morning glucose levels either remain constant or increase modestly.20,21 The ‘‘dawn phenomenon’’—marked glucose elevation in patients with diabetes and fixed insulin availability—is thought to be an exaggerated manifestation of normal physiology without the braking effect of pancreatic insulin responsiveness.22 Likewise, increased morning fasting glucose levels in our patients taking niacin and statins might represent an unrestrained response to nocturnal counterregulatory hormone secretion. Niacin could trigger counterregulatory hormone responses in the following manner: At pharmacologic doses, niacin binds the G-protein coupled receptor GPR109A in adipocytes, decreasing cyclic AMP and strongly inhibiting
Figure 3 Regression of morning fasting glucose on HbA1c in subjects with diabetes, according to treatment (solid lines) or nontreatment (dashed lines) with niacin, statins, or fibrates. The predicted regression lines (thick) and 95% confidence limits (thin) are shown.
172
Journal of Clinical Lipidology, Vol 6, No 2, April 2012 speaker’s honoraria from Abbott and Merck; and consultant fees from Merck and sanofi aventis and owns stock in Eli Lilly. This research study was funded by the Merck Investigator Study Program.
References
Figure 4 Regression of morning fasting glucose on HbA1c in subjects without diabetes, according to treatment with differing niacin regimens versus no niacin treatment. Niacin given once daily was generally extended-release niacin at bedtime. Niacin given twice daily was generally immediate-release niacin at breakfast and supper.
lipolysis.23,24 Consequently, plasma nonesterified fatty acid (NEFA) levels decrease sharply, but this is followed by a rebound and overshoot of NEFA substantially above baseline levels.25 The maintenance of relatively normal glucose levels during the time that NEFA are unavailable may be due to counterregulatory hormones. The same hormonal responses may lead to the rebound and overshoot of NEFA.26 Statin effects on glucose-insulin homeostasis have been studied in a variety of settings, but a clear explanation for the increased risk of diabetes is not yet available.17 Both atorvastatin and rosuvastatin were found to increase fasting insulin levels, suggesting perhaps an interplay with nocturnal counterregulatory hormone dynamics.27
Conclusion This retrospective cohort study suggests that in patients without diabetes, niacin and statins increase fasting morning glucose levels out of proportion to HbA1c. The effect requires robust glucose homeostasis (HbA1c in the lower half of the normal range), and it is not seen at all in patients with diabetes. Because our data are retrospective, confirmation in randomized clinical trials is needed. Recognizing the potential effect of niacin and statins on fasting glucose levels should help to inform clinical decisions, particularly the need for HbA1c measurement to judge the overall impact on glycemia, and may help to direct research on mechanisms.
Financial disclosure Dr. Guyton has received research funding from Abbott, Amarin, Isis/Genzyme, Merck, Regeneron/sanofi aventis;
1. Coronary Drug Project Group. Clofibrate and niacin in coronary heart disease. J Am Med Assoc. 1975;231:360–381. 2. Carlson LA, Rosenhamer G. Reduction of mortality in the Stockholm Ischaemic Heart Disease Secondary Prevention Study by combined treatment with clofibrate and nicotinic acid. Acta Med Scand. 1988; 223:405–418. 3. Brown BG, Zhao XQ, Chait A, et al. Simvastatin and niacin, antioxidant vitamins, or the combination for the prevention of coronary disease. N Engl J Med. 2001;345:1583–1592. 4. Whitney EJ, Krasuski RA, Personius BE, et al. A randomized trial of a strategy for increasing high-density lipoprotein cholesterol levels: effects on progression of coronary heart disease and clinical events. Ann Intern Med. 2005;142:95–104. 5. Boden WE, Probstfield JL, Anderson T, et al; AIM-HIGH Investigators. Niacin in patients with low HDL cholesterol levels receiving intensive statin therapy. N Engl J Med. 2011;365:2255–2267. 6. Garg A, Grundy SM. Nicotinic acid as therapy for dyslipidemia in non-insulin-dependent diabetes mellitus. J Am Med Assoc. 1990;264: 723–726. 7. Gibbons LW, Gonzalez V, Gordon N, Grundy S. The prevalence of side effects with regular and sustained-release nicotinic acid. Am J Med. 1995;99:378–385. 8. Goldberg RB, Jacobson TA. Effects of niacin on glucose control in patients with dyslipidemia. Mayo Clin Proc. 2008;83:470–478. 9. Canner PL, Furberg CD, Terrin ML, McGovern ME. Benefits of niacin by glycemic status in patients with healed myocardial infarction (from the Coronary Drug Project). Am J Cardiol. 2005;95:254–257. 10. Grundy SM, Vega GL, McGovern ME, et al. Efficacy, safety, and tolerability of once-daily niacin for the treatment of dyslipidemia associated with type 2 diabetes. Arch Intern Med. 2002;162:1568–1576. 11. Elam MB, Hunninghake DB, Davis KB, et al, Diabetes Multicenter Research Group. Effect of niacin on lipid and lipoprotein levels and glycemic control in patients with diabetes and peripheral arterial disease: the ADMIT study: A randomized trial. Arterial Disease Multiple Intervention Trial. J Am Med Assoc. 2000;284:1263–1270. 12. Guyton JR, Brown BG, Fazio S, Polis A, Tomassini JE, Tershakovec AM. Lipid-altering efficacy and safety of ezetimibe/ simvastatin coadministered with extended-release niacin in patients with type IIa or type IIb hyperlipidemia. J Am Coll Cardiol. 2008; 51:1564–1572. 13. Mills EJ, Wu P, Chong G, et al. Efficacy and safety of statin treatment for cardiovascular disease: a network meta-analysis of 170,255 patients from 76 randomized trials. Q J Med. 2011;104:109–124. 14. Sattar N, Preiss D, Murray HM, et al. Statins and risk of incident diabetes: a collaborative meta-analysis of randomised statin trials. Lancet. 2010;375:735–742. 15. Guyton JR, Goldberg AC, Kreisberg RA, et al. Effectiveness of oncenightly dosing of extended-release niacin alone and in combination for hypercholesterolemia. Am J Cardiol. 1998;82:737–743. 16. Sasaki J, Iwashita M, Kono S. Statins: beneficial or adverse for glucose metabolism. J Atheroscler Thromb. 2006;13:123–129. 17. Koh KK, Sakuma I, Quon MJ. Differential metabolic effects of distinct statins. Atherosclerosis. 2011;215:1–8. 18. Ridker PM, Danielson E, Fonseca FA, et al, JUPITER Study Group. Rosuvastatin to prevent vascular events in men and women with elevated C-reactive protein. N Engl J Med. 2008;359:2195–2207. 19. Campbell PJ, Bolli GB, Cryer PE, Gerich JE. Sequence of events during development of the dawn phenomenon in insulin-dependent diabetes mellitus. Metabolism. 1985;34:1100–1104.
Rajanna et al
Fasting morning glucose and lipid medications
20. Schmidt MI, Lin QX, Gwynne JT, et al. Fasting early morning rise in peripheral insulin: evidence of the dawn phenomenon in nondiabetes. Diabetes Care. 1984;7:32–35. 21. Bolli GB, De Feo P, De Cosmo S, et al. Demonstration of a dawn phenomenon in normal human volunteers. Diabetes. 1984;33: 1150–1153. 22. Bolli GB, Gerich JE. The ‘‘dawn phenomenon’’—a common occurrence in both non-insulin-dependent and insulin-dependent diabetes mellitus. N Engl J Med. 1984;310:746–750. 23. Guyton JR. Niacin in cardiovascular prevention: mechanisms, efficacy, and safety. Curr Opin Lipidol. 2007;18:415–420.
173 24. Tunaru S, Kero J, Schaub A, et al. PUMA-G and HM74 are receptors for nicotinic acid and mediate its anti-lipolytic effect. Nat Med. 2003; 9:352–355. 25. Carlson LA. Nicotinic acid: the broad-spectrum lipid drug. A 50th anniversary review. J Intern Med. 2005;258:94–114. 26. Vega GL, Cater NB, Meguro S, Grundy SM. Influence of extended-release nicotinic acid on nonesterified fatty acid flux in the metabolic syndrome with atherogenic dyslipidemia. Am J Cardiol. 2005;95:1309–1313. 27. Thongtang N, Ai M, Otokozawa S, et al. Effects of maximal atorvastatin and rosuvastatin treatment on markers of glucose homeostasis and inflammation. Am J Cardiol. 2011;107:387–392.