	This month’s edition of Cardiology Review features an article in the diabetes section by Dr. Harold L. Lazar and associates on the beneficial effect of maintaining blood glucose levels lower than 200 mg/dL. This study of 141 diabetic patients undergoing coronary artery bypass graft (CABG) surgery included 72 patients who were treated with a modified protocol for a glucose-insulin-potassium infusion and 70 control patients. Drs. Justin Levisay and Peter M. Buttrick comment on this benefical therapy.

	In a combined article in the lipid disorders and heart failure sections, Drs. Andrew L. Clark and Stefan D. Anker of the United Kingdom evaluate the relationship between serum cholesterol levels and survival in patients with chronic heart failure. Their study included two groups of patients (an initial “derivation” group and a subsequent “validation” group). Their results appear to show that the lower the serum cholesterol level, the worse the prognosis. Dr. Roger S. Blumenthal and his associate Dr. Thomas Cimato offer their own thoughts on this provocative study in their commentary, which emphasizes the need for confirmatory trials.

	Our entry for the stroke section this month is by Dr. Carlos J. Rodriguez and colleagues. These investigators report on medical therapy (warfarin versus aspirin) for preventing stroke recurrence or death in 69 patients with atrial septal aneurysm and 203 patients with patent foramen ovale. Forty-four patients had both abnormalities. Dr. Elliott G. Gross provides the commentary for this study.

	In the hypertension section, Dr. Ramón C. Hermida and associates from Spain evaluate the seasonal variation in fibrinogen levels in 1,006 hypertensive patients. A relationship between these levels and risk of subsequent vascular events was found. The commentary is by Dr. Wendy S. Post.

	Finally, Dr. Stephen P. Glasser presents the first in a series of articles on the fundamentals of clinical research. His first article provides a brief overview of the history of clinical research.

Articles

Cardiology Review Online

Tight glycemic control  in diabetic CABG  surgery patients

	TORONTO—Carvedilol appears to have renoprotective properties in patients with chronic systolic heart failure (CHF), as evidenced by increases in glomerular filtration rate (GFR), said Anil Verma, MD, at the 8th Annual Scientific Meeting of the Heart Failure Society of America.

	“Patients with CHF and renal dysfunction have a worse prognosis and should be treated aggressively with medications that have a favorable risk/benefit ratio, such as carvedilol, which may inhibit a decline of renal function and improve morbidity and mortality,” said Dr. Verma, lead investigator of a study presented here, and attending physician at Candler Internal Medicine in Statesboro, Georgia.

	Renal dysfunction is a common complication for patients with CHF, and is associated with a poorer prognosis in heart failure (HF) associated with left ventricular (LV) systolic dysfunction. In hypertensive animal models and humans, carvedilol has been shown to reduce urinary albumin excretion and expression of profibrotic factors such as renal tissue growth factor (TGF)-beta, suggesting that it may inhibit the decline in renal function in patients with CHF.

	The changes in GFR over time were assessed in a study of 94 men and women with CHF defined by an LV ejection fraction of less than 45%. Forty-nine patients taking carvedilol and 45 not taking carvedilol were randomly selected from five physician practices to participate in the study. All patients were on standard HF therapies, with the initiation of carvedilol at the discretion of the treating physician. GFR was estimated at baseline and after a mean duration of 20 months in both users and nonusers of carvedilol.

	There was no difference in GFR at baseline between the carvedilol users and nonusers. After a median follow-up of 20 months, the mean GFR in carvedilol recipients was 67.10 mL/min per 1.73 m2 and the mean GFR in the nonusers was 50.00 mL/min per 1.73 m2, for a mean average difference in GFR of 17.10 mL/min per 1.73 m2 (P = .011). The mean average GFR increased by 3.31 mL/min per 1.73 m2 in the carvedilol group and declined by 11.30 mL/min per 1.73 m2 in patients not taking carvedilol.

	Of the patients who were not taking carvedilol, 69% had moderate chronic kidney disease at the end of the study, defined as GFR less than 60 mL/min per 1.73 m2, compared with 45% who were taking carvedilol, corresponding to a 35% reduction in relative risk with carvedilol use.

	The improvement in GFR with carvedilol was independent of the improvement in LV ejection fraction, said Dr. Verma, which suggests that carvedilol is independently renoprotective. “The reason could be that carvedilol inhibits TGF and prevents fibrosis,” he said. “It also reduces albumin excretion in the kidney and has antioxidant properties.”

	The alpha- and beta-blocking properties of carvedilol result in both intravascular and intrarenal blood pressure control, he added.

	In determining the prognosis of a patient with HF, the GFR should be integrated with other prognostic factors, such as LV ejection fraction, patient functional status, serum sodium, and B-type natriuretic peptide levels, said Dr. Verma. “The take-home message is that by diagnosing and aggressively treating HF early, you can prevent much of the decline in renal function that is associated with chronic HF,” he said.

Carvedilol slows renal dysfunction in systolic HF

	TORONTO—Thiazolidinediones (TZDs) have no adverse effect on outcomes in patients with concomitant heart failure (HF) and diabetes. To the contrary, TZD use is associated with a reduction in hospitalizations and hospital days in a community setting, according to study data presented at the 8th Annual Scientific Meeting of the Heart Failure Society of America.

	Fluid retention occurs in 3% to 5% of patients treated with TZDs, which increases to 10% to 15% when TZDs are used in combination with insulin. Observational studies have also suggested an increased risk of HF in diabetic patients treated with TZDs. For these reasons, the American Heart Association and American Diabetes Association have recommended against the use of TZDs in patients with New York Heart Association (NYHA) Functional Class III or IV HF. Despite these recommendations, TZD use in patients with HF is increasing, said John Golden, MD, chief of cardiology at Kaiser Permanente of the Mid-Atlantic States, Fairfax, Virginia. The use of TZDs in diabetic Medicare beneficiaries with a primary diagnosis of HF increased from 7.2% in 1998 and 1999 to 16.1% in 2000 and 2001, he noted.

	The fluid retention induced by TZDs appears to be largely peripheral and not central, he said. “A retrospective analysis of TZD use in patients with left ventricular [LV] systolic dysfunction revealed no association between TZD-related fluid retention and HF severity as assessed by baseline LV size, systolic function, or functional class,” he said.

	Furthermore, TZDs may have substantial cardiovascular benefits. In models of HF, pioglitazone has been shown to improve LV remodeling and function, reduce LV dilatation and hypertrophy, and improve systolic function.

	Because the significant nonhypoglycemic cardiovascular effect of TZDs raises the possibility of potential clinical benefit in HF, Dr. Golden and colleagues at Kaiser Permanente analyzed their experience with TZD use in a comprehensive HF disease management program.

	They studied 97 consecutive diabetic patients referred to their HF treatment program, 37% of whom were treated with TZDs (pioglitazone) and 15% of whom were treated with both TZDs and insulin. All patients were being treated with angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, and 97% were treated with beta blocking agents. Clinical outcomes were assessed and compared by diabetic treatment groups at 1 year. The 1-year survival was 100% in all groups.

	Nonusers and users of TZDs, excluding those patients on combination therapy, had similar improvements in ejection fraction, “but all-cause hospitalizations, HF hospitalizations, and total hospital days were significantly reduced in the patients on TZDs, interestingly,” said Dr. Golden.

	The number of all-cause hospitalizations per patient was 0.19 in the TZD recipients compared with 0.71 in the nonusers (P = .01); the number of HF hospitalizations per patient was 0.03 in the TZD group and 0.16 in the nonusers (P = .04); and the total number of hospital days per patient was 0.67 and 2.72 in the TZD users and nonusers, respectively (P = .02). Among the TZD/insulin users, there was no significant reduction in all-cause or HF hospitalizations, but total hospital days per patient were significantly less in the TZD/ insulin group compared with those not on the combination (0.07 versus 2.20; P = .005).

	Among the subgroup of patients with NYHA Class III or IV HF, the effects of TZD with or without insulin were comparable to those observed in the overall study population. There was a trend toward increased diuretic requirements in those treated with a combination of TZD and insulin (69.3 versus 45.5 mg/day of furosemide;

	P = .08), which was not noted in the patients treated with TZDs alone.

	“These findings support the concept that the nonhypoglycemic cardiovascular effects of TZDs may have favorable clinical impact in HF,” said Dr. Golden. “Clinical trials will determine the specific effects of TZDs in HF and the extent to which these findings can be extrapolated to broader populations.”

	TZD use in HF “is a real issue because TZDs are extremely popular, they do a lot of good things in diabetic patients, and they’re going to be used, so I think we have to find which patients we can use them in safely and which patients have a chance to get into trouble,” said Prakash Deedwania, MD, chief of the cardiology section, and director of cardiovascular research, University of California, San Francisco, Fresno.

TZDs: No adverse clinical effects in diabetic HF patients

	TORONTO—Therapy with HMG-CoA reductase inhibitor drugs (statins) improves survival in patients with heart failure (HF) regardless of the etiology of the HF, according to two studies presented at the 8th Annual Scientific Meeting of the Heart Failure Society of America.

	Retrospective data from the University of California, Los Angeles (UCLA) show an improvement in survival among patients with systolic HF who took statins compared with those who did not take statins, and the favorable effect of statins was observed in patients with either low or high total cholesterol levels, said Tamara Horwich, MD, cardiology fellow at UCLA, and lead investigator of the study.

	Horwich and colleagues analyzed data on 473 patients with systolic HF of multiple etiologies who were evaluated at their clinic. Total cholesterol levels were measured at the time of initial referral. Statin therapy was defined as a statin begun prior to referral or started within 3 months of referral and continued throughout the study period.

	Fifty-four percent of the patients were treated with a statin. Those on a statin were more likely to have coronary artery disease, hypertension, and diabetes.

	In the total cohort, 1-year survival was 81% among the patients taking a statin compared with 67% for those not taking a statin. When adjusted for other variables, patients with total cholesterol of 162 mg/dL or less who were on a statin had a 48% reduction in mortality risk, and those with total cholesterol greater than 162 mg/dL who were on a statin had a 76% reduced risk.

	Lower total cholesterol levels have been associated with impaired HF survival in previous studies, noted Dr. Horwich. “Even in low cholesterol, we saw continued advantage with a statin,” she said. “In our clinic, we have been using statins in HF of ischemic etiology.”

	Nonlipid-lowering effects may explain the benefit in HF. Some of the potential mechanisms are inhibition of cardiac hypertrophy and remodeling, improvement of autonomic nervous system function, increased endothelial nitric oxide, improved endothelial function, and anti-inflammatory effects. Animal studies show improvements in left ventricular ejection fraction of 5% to 10% with simvastatin, she said.

	In a second retrospective analysis presented here, higher cholesterol levels were associated with improved survival in patients with chronic HF, but only in those taking a statin, said Gad Cotter, MD.

	Cotter and co-investigators from Assaf-Harofeh Medical Center in Israel analyzed 807 patients from the placebo arm of a study designed to assess the efficacy of an endothelin antagonist in chronic HF. To be eligible, patients had to have New York Heart Association Functional  Class IIIb-IV chronic HF despite optimal medical therapy, an ejection fraction less than 35%, and either hospitalization for HF in the previous 12 months or the ability to walk less than a 375-meter distance on the 6-minute walk test. Fifty-six percent of the patients received statins.

	Having a cholesterol level above the median (193 mg/dL) was associated with decreased mortality, although this association became weaker as follow-up progressed to 130 weeks after study randomization. The survival benefit associated with higher baseline cholesterol was observed only in the patients taking statins, however. Statin treatment was associated with a 67% reduction in the risk of all-cause mortality (P < .001), and the risk reduction with statins became stronger over time, but statin treatment was paradoxically not correlated with readmission for HF, said Dr. Cotter, who is now an associate professor of medicine at Duke University in Durham, North Carolina.

	Unlike the UCLA study, the effect of statins was more pronounced in patients with higher baseline cholesterol levels. In patients not receiving statin therapy, baseline cholesterol level had no effect on mortality. The beneficial effect of statins on outcome was equal in patients with HF of ischemic or nonischemic origin, he said.

	“One of the problems is that there is no prospective, randomized study yet looking at statins in HF, and we’re awaiting the results of

	the CORONA [Controlled Rosuvastatin Multinational Trial in Heart Failure] study to give us the definitive answer to this question,” said Dr. Cotter.

	CORONA is a 3-year, randomized, double-blind study comparing the effects of rosuvastatin, 10 mg daily, with placebo in about 5,000 patients with chronic symptomatic systolic HF of ischemic etiology on the end points of overall survival and cardiovascular morbidity and mortality. Until the results of CORONA are known, Dr. Cotter said that he uses statin treatment in chronic HF patients with high cholesterol regardless of HF etiology.

Statins extend survival in heart failure

	We understand a great deal about the etiology of cardiovascular disease; however, the triggers that lead to acute events often remain elusive. Why does a plaque rupture or a ventricular arrhythmia occur today and not a few months from now, or why at this moment and not 6 hours from now? The timing of events appears to follow both a circadian (daily) and circannual (seasonal) pattern. In the early waking hours, there is an increase in catecholamines, which is associated with a variety of other neurohormonal changes. This is one of the potential explanations for an increase in the incidence of events in the early waking hours.

	Another source of variation is related to the seasons. More events occur in the winter than in the summer months.1 The explanation for this observation is not clear, but it might be related to changes in the duration of daylight hours, temperature variation, activity levels, or the emotional state related to holidays.2 An animal model of sudden death and inducible ventricular arrhythmias in dogs, however, suggests a seasonal variation that is not related to these factors.3

	Blood pressure changes throughout a 24-hour period have been well characterized. During sleep, the normal pattern includes a decrease in blood pressure, which is called “dipping.” Many studies have demonstrated that “nondippers” are at increased risk for cardiovascular disease. Thus, lack of the “normal” circadian variation in this instance leads to a greater 24-hour load on the heart.

	In this issue, Hermida and colleagues (page 38) describe interesting results related to both circadian and circannual variation from an observational study of over 1,000 patients with hypertension. Patients were characterized as nondippers or dippers using ambulatory blood pressure monitoring. Nondippers were found to have higher fibrinogen levels than dippers. Furthermore, fi-brinogen levels were higher in the winter months, peaking in February, than in the summer months, in both dippers and nondippers. As elevated fibrinogen is a predictor of cardiovascular events, the authors conclude that changes in fibrinogen might be one explanation for the increased risk associated with altered circadian variation in blood pressure (nondipping) and also the increased risk associated with the winter months.

	Similar results related to seasonal variation in cholesterol levels were also recently reported, with higher levels observed in the winter months.4 That finding was ascribed primarily to seasonal hemodilutional changes, with an increase in blood volume during the summer. The authors appropriately question whether the important factor is the total concentration of cholesterol (or fibrinogen, in this case) or the total amount that is present regardless of hemoconcentration.

	There is clearly a need to develop a more thorough understanding of triggers for cardiovascular events in order to develop strategies toward risk reduction. These studies emphasize the potential importance of circadian and circannual variation in risk factors as one component of potential triggers for events.

Variation of fibrinogen in  hypertensive patients

	From the Bioengineering and Chronobiology Laboratories, University of Vigo, Campus Universitario, Vigo, Spain; and the Hypertension and Vascular Risk Unit, Hospital Clínico Universitario, Santiago de Compostela, Spain

	Studies have shown that increased plasma fibrinogen levels are closely associated with a higher rate of cardiovascular events. Therefore, an increased fibrinogen level is considered a risk factor for acute myocardial infarction (MI) and stroke.1 In small samples of patients, primarily in elderly patients, plasma fibrinogen levels have also been shown to fluctuate with the seasons, with higher levels found during the winter.2-4 Some studies have also shown a seasonal variation in the incidence of onset of acute MI, with a peak during the coldest months.5 Separate studies, therefore, appear to show that the seasonal distribution of fibrinogen is related in time with the circannual fluctuation in coronary events.5 The diminished nocturnal decline in blood pressure (nondipping6) has also been shown to be related to end-organ damage and cardiovascular events.7-9

	We studied the seasonal variation of plasma fibrinogen levels in a sample of hypertensive patients based on their circadian patterns of blood pressure fluctuations using 48-hour ambulatory blood pressure monitoring, taking into account the fact that fibrinogen is an independent risk factor for cardiovascular disease.

	Our study consisted of 1,006 sequential patients (482 men) who had been diagnosed with stage 1 or stage 2 essential hypertension and were evaluated for a span of 2.5 consecutive years. The mean age was 53.0 ± 13.4 years (range, 19 to 87 years). On enrollment into the study, 407 patients (40.5%) were not receiving therapy for their blood pressure. After overnight fasting, blood samples from an antecubital vein were taken between 8:00 am and

	9:00 am the day ambulatory blood pressure monitoring began. Systolic and diastolic blood pressure measurements were taken for 48 consecutive hours.  Between 7:00 am and 11:00 pm, measurements were taken every 20 minutes; at night, they were taken every 30 minutes. Patients were told to carry

	out their normal activities and to follow a similar schedule on both days of monitoring. Patients were assessed during their customary daytime activities, which was between 8:00 am and 11:00 pm for most participants, and during sleep at night. They were given actigraphs to wear on their dominant wrists to assess physical activity every minute. The means for diurnal and nocturnal blood pressure were ascertained from the data from the actigraphs.

	Based on the data collected over the 48 hours, participants were divided into two groups according to their nocturnal blood pressures: nondippers and dippers. Those with less than a 10% decline in sleep mean systolic blood pressure compared with the awake mean were categorized as nondippers (n = 513). The rest were considered dippers

	(n = 493). A seasonal fluctuation in plasma fibrinogen levels was found for both nondippers and dippers using multiple components analysis of longitudinal time series,10 a method that can be used with nonsinusoidal-shaped time series consisting of values distributed at equal or unequal intervals. A nonparametric test designed to compare measurements from multiple component analysis was used to compare seasonal variations between dippers and nondippers.11

	A marked circannual fluctuation in plasma fibrinogen levels was shown for the entire group of patients (P < .001), with a mean

	fibrinogen level of 318 mg/dL, an extent of predictable change throughout the year (amplitude) of 40 mg/dL, and the time of peak value (orthophase) in February.

	The seasonal variation in plasma fibrinogen levels was analyzed using a model that included comparison of elements with time periods of 6 and 12 months. We also compared the monthly means of plasma fibrinogen using analysis of variance (ANOVA) to additionally confirm the statistically sig-

	nificant circannual distribution (P < .001). Nondippers and dippers were analyzed separately using the same model and time periods. Throughout the year, nondippers had a higher plasma fibrinogen level compared with dippers (P < .001, from the test for comparison of seasonal mean between both groups; figure). However, despite this difference, the pattern in seasonal variation was similar for both groups. The time of peak value, therefore, was comparable for nondippers and dippers. The greatest plasma fibrinogen levels occurred during the winter months, regardless of the degree of the nighttime drop in blood pressure. The extent of predictable change through out the year was also comparable for both groups, as shown by an analysis of rhythm parameters. The circannual variation was about 42 mg/dL for both nondippers and dippers (figure). After adjust-ing for possible confounding factors, such as sex, drinking alcohol, smoking, exercise, and antihypertensive medication, ANOVA test results showed that nondippers had a marked increase in plasma fibrinogen (P = .003).

	Our study showed that hypertensive patients classified as both nondippers and dippers had a significant increase in plasma fibrinogen levels in February and had lower levels in September and October. This corroborates the findings of other studies that a seasonal variation in plasma fibrinogen levels exists, including two separate 1-year studies of elderly patients and a cross-sectional study of patients 55 years of age and older.4 In a study by Stout and Crawford, the predictable circannual variation in plasma fibrinogen levels was higher (71 mg/dL) than in our study (40 mg/dL).2 However, in studies by Woodhouse and colleagues3 and van der Bom and colleagues,4 the plasma fibrinogen levels were lower than in our study (13 mg/dL and 31 mg/dL, respectively). These disparities may be the result of environmental factors or differences in the study samples.

	A 67% higher risk of death from ischemic heart disease was shown when plasma fibrinogen levels increased above one standard deviation (59 mg/dL) in 5 years of follow-up in the Northwick Park Heart Study.12 Based on this finding and using the seasonal predictable variation shown in our study, our subjects would have a 45% increased risk of ischemic heart disease during the coldest months. These figures show how the circannual variation in plasma fibrinogen levels can have a

	significant influence on death rates from cardiovascular disease, although projection of results between studies must be done carefully.

	In addition to showing a relationship between circannual fluctuation and many other parameters related to risk of cardiovascular disease, other studies have found that seasonal fluctuation of plasma fibrinogen levels is associated with yearly variation in the incidence of mortality from coronary artery disease and of onset of acute MI.5 Further research is needed to reveal the mechanisms involved. Activation of the acute phase response due to respiratory infections, which are more common in winter, may be one cause, as suggested by Woodhouse and colleagues.3 There is no evidence to implicate infections as a cause of the circannual fluctuation in fibrinogen, however, as shown by studies on interleukin-6, soluble P-selectin, factor VII, C-reactive protein, plasminogen activator inhibitor, and white blood cell count.13

	Our study is the first to show that nondipper hypertensive patients have a marked increase in plasma fibrinogen levels throughout the year compared with dipper hypertensive patients. This study also confirms the results of other studies showing the seasonal variation of high amplitude that characterizes plasma fibrinogen. Nondipper hypertensive patients are at much greater risk of having a stroke compared with dippers, as shown by O’Brien and colleagues, although the reasons for this are not understood.6 In another study, the number of adverse events was increased threefold for nondippers compared with dippers over a follow-up of 3.2 years.7 In the Systolic Hypertension in Elderly in Europe (Syst-Eur) trial, nondippers were shown to have a higher rate of MI and stroke than dippers in a subsample of 808 patients receiving ambulatory blood pressure monitoring.8 In another study, hypertensive patients with a decrease in the nocturnal blood pressure drop of 5% had a 31% higher risk of mortality from cardio-

	vascular disease after a follow-up of 9.2 years.9 Furthermore, dipper hypertensive and nondipper normotensive subjects had a comparable relative hazard of cardiovascular mortality (2.37 for dippers and 2.16 for nondippers).9 This shows that risk of cardiovascular events is affected by abnormal circadian blood pressure variability, not just increased blood pressure, and the treatment of patients who lack a nocturnal decline in blood pressure is extremely important.9

	In our study, we used ambulatory blood pressure monitoring for 48 hours, compared with the more common 24-hour assessment. Although less practical, 48-hour monitoring provides better data regarding blood pressure fluctuations, diagnosis of disease, and assessment of response to medication.14 Studies of 24-hour blood pressure monitoring have shown that this period of time may not be long enough to correctly evaluate circadian patterns. A marked reduction in the diurnal mean of blood pressure was shown on the second day of monitoring in patients who were undergoing ambulatory blood pressure monitoring for the first time. During the first 4 hours of monitoring, systolic blood pressure increased an average of 7 mm Hg and diastolic blood pressure increased an average of 5 mm Hg. This effect persisted for at least the first 9 hours of measurement.15 Thirty-five percent of patients who were classified as dippers on the first day of measurement became nondippers on the second day because of the decrease in blood pressure during diurnal activity but not during nocturnal resting hours.15 This blood pressure monitoring effect, which occurs in studies based on

	24-hour ambulatory monitoring, explains why categorizing patients based on their dipping status has not been reproducible and why the percentage of nondippers in many studies has been underestimated.

	Our study showed that there was a marked increase in plasma fibrinogen levels during the colder months of the year in nondipper and dipper hypertensive patients. In addition, nondipper hypertensive patients had increased plasma fibrinogen levels throughout the year compared with dipper hypertensive patients. These findings may support other studies that have correlated an increase in the occurrence of cardiovascular events with a lack of nocturnal decline in blood pressure and may also support the finding that the seasonal variation in plasma fi-brinogen levels is associated with the yearly variation in the incidence of mortality from coronary artery disease and onset of acute MI. This study highlights the importance of the need to identify patients with altered circadian blood pressure patterns, especially nondipper hypertensive patients, and to provide appropriate treatment.  n

Seasonal variation  of fibrinogen in  hypertensive patients

	The notion that the infusion of glucose, insulin, and potassium (GIK) could reduce infarct size was first proposed by Sodi-Pallares and others in the 1960s.1,2 The proposed (and validated) mechanisms of action of GIK involved shifting myocardial oxidative metabolism from free fatty acids, which are increased in ischemia and toxic to myocardium, to glucose, which is more energy-efficient. Approximately 10% more adenosine triphosphate is synthesized per mole of oxygen when glucose is oxidized as opposed to free fatty acids.3

	Despite a number of positive observations in a variety of ischemic circumstances, including several large multicenter trials such as the Diabetes and Insulin Glucose Infusion in Acute Myocardial Infarction (DIGAMI) study,4 which showed a 30% and 25% relative mortality reduction at 1 and 3 years, respectively, and studies by the extracorporeal lung assist (ECLA) group5 and by van der Horst and colleagues,6 which showed that GIK infusion improves survival in the setting of acute myocardial infarction, the strategy has not become “the standard of practice.” The cynical view of this oversight would be that the therapy is not industry-driven; the more benign view would be that alternative reperfusion strategies have a more dramatic impact on survival and preservation of myocardial function.

	Nonetheless, GIK refuses to go away, and the article by Lazar and colleagues (page 17) again reminds us of the important role that cardiac metabolism and preferential glucose utilization can play in the preservation of cardiac function, especially in the surgical management of diabetic patients. The authors investigated whether the same strategy of GIK infusion that affected cardiac function in the setting of acute ischemia and infarction might improve postoperative outcomes in diabetic patients undergoing coronary artery bypass graft (CABG) surgery. This is obviously a very important patient group, both because diabetic patients, by definition, have impaired glucose metabolism and are vulnerable to free fatty acid toxicity and because diabetic patients remain a subgroup of patients who appear to benefit from CABG surgery as opposed to percutaneous revascularization.

	The authors randomly assigned diabetic patients undergoing CABG surgery to a GIK group (n = 72) or a No-GIK group (n = 69). GIK infusion was initiated before induction of anesthesia and terminated 12 hours after arrival in the intensive care unit. Results of the study showed that GIK patients had significantly better glycemic control not only during the infusion (134.3 ± 3.7 mg/dL in the GIK group versus 266.8 ± 6.3 mg/dL in the No-GIK group; P < .001), but for at least 18 hours afterward (170.6 ± 5.3 mg/dL in the GIK group versus 257.1 ± 6.0 mg/dL in the No-GIK group; P < .001). This resulted in lower serum lactate and free fatty acid levels. In the perioperative period, GIK-treated patients had higher cardiac indices (2.91 ± 0.06 L/min/m2 in the GIK group versus 2.43 ± 0.06 L/min/m2 in the No-GIK group) and required less inotropic support. GIK patients had a lower incidence of atrial fibrillation, need for pacing, ventilatory support, pneumonia, and wound infection. These benefits translated into a shorter postoperative hospital stay (6.5 ± 0.1 days for the GIK group versus 9.2 ± 0.3 days for the No-GIK group; P = .03). Surprisingly, follow-up over a 5-year period showed a significant (P = .04) survival advantage and decreased number of ischemic events in the GIK group (5% for the GIK group versus 19% for the No-GIK group).

	This study unequivocally supports the use of GIK infusion in this surgical cohort. The mechanism by which this benefit is achieved is unclear, however, and the broad spectrum of salutary effects suggests that it might encompass more biologic moments than were initially recognized. Certainly, the acute infusion of GIK should shift substrate utilization and could contribute to the hemodynamic effect. Moreover, the reduced utilization of sympathomimetic amines might have an impact on perioperative electrical instability. Whether these outcomes might have been affected solely by tight glycemic control is unclear, as it has been shown that control of glucose levels per se, rather than levels of exogenous insulin, appear to account for the mortality benefit associated with insulin therapy in critically ill patients.7 There is also evidence to suggest that postoperative CABG surgery patients who achieved tight glycemic control by continuous insulin infusion have better outcomes than those with poor control treated with intermittent subcutaneous insulin.8

	Beyond this, the authors postulate that additional mechanisms to explain the beneficial effect of GIK might include improved endothelial function, reduced vascular inflammation, and thrombogenicity leading to enhanced graft patency. It is well known that diabetic patients have increased inflammatory biomarkers and that exposure of endothelial cells to proinflammatory cytokines induces procoagulant activity,9 leads to the expression of cell surface adhesion molecules,10 and impairs endothelium-dependent vascular relaxation.11 Unfortunately, no inflammatory biomarkers or measures of endothelial function were obtained, so this potentially important interface cannot be evaluated. Additionally, little information regarding the adequacy of medical treatment prior to and at the time of discharge is provided, nor are there follow-up data on graft patency. These omissions are important, as a growing number of interventions known to decrease cardiovascular risk have been shown to improve endothelial function and reduce inflammatory biomarkers, including HMG-CoA reductase inhibitors  (statins)12 and angiotensin-converting enzyme inhibitors.13

	These limitations notwithstanding, the study certainly provides a simple and rational adjunctive strategy for surgical management of

	diabetic patients. Whether the benefit is broad-based and applies to all patients undergoing CABG surgery or whether it is efficacious only in subgroups of diabetic patients (eg, those with heart failure) remains to be established. In addition, a more nuanced understanding of the mechanisms of benefit will need to emerge from further studies evaluating the relative role of inflammation and metabolic balance. But as it stands, the benefits are already clear. One hopes that the cardiothoracic community will do better than the cardiology community and embrace this simple, inexpensive therapy, which appears to have a very substantial upside and no apparent downside.

Glucose-insulin-potassium: A therapy whose time has come again

	From the Departments of Cardiothoracic Surgery, Biostatistics, and Cardiology, Boston University School of Medicine and Boston Medical Center, Boston, Massachusetts; and the Division of Endocrinology, Baystate Medical Center, Springfield, Massachusetts Diabetes mellitus is an independent risk factor for increased morbidity and mortality, reduced long-term survival, and decreased freedom from recurrent angina following coronary artery bypass graft (CABG) surgery.1 There is now evidence to suggest that techniques that result in tighter glycemic control may improve survival in diabetic patients during acute coronary syndromes. The Diabetes and Insulin-Glucose Infusion in Acute Myocardial Infarction (DIGAMI) study showed that mortality could be reduced by 30% in diabetic patients sustaining an acute myocardial infarction (MI) with the infusion of a glucose-insulin solution to achieve serum glucose levels below 200 mg/dL.2 We designed a modified glucose-insulin-potassium (GIK) solution to determine whether tighter perioperative glycemic control in diabetic CABG surgery patients would improve perioperative outcomes.

	Patients and methods Following approval from the Boston University Medical Center Institutional Review Board, 141 diabetic patients undergoing primary or

	reoperative CABG surgery performed on cardiopulmonary bypass were randomly assigned to a GIK (n = 72) or No-GIK (n = 69) group. The GIK consisted of 500 mL of 5% dextrose in water (D5W) with 80 units of regular insulin and 40 mEq of potassium chloride infused at 30 mL/hr, adjusted using a defined protocol to keep serum glucose at or below 200 mg/dL (table 1). The GIK was started at the time of anesthetic induction and continued for 12 hours after arrival in the intensive care unit (ICU). The No-GIK group received D5W at 30 mL/hr and subcutaneous insulin designed to keep the blood glucose level below 250 mg/dL (table 2). Data are represented as the mean ± standard error of the mean and were considered significant at P values of .05 or lower.

	There was no difference in age, sex, ejection fraction, urgency of surgery, or the type of diabetes (insulin versus noninsulin) between the two groups. There was also no difference in the number of vessels bypassed (3.26 ± 0.10 for the GIK group compared with 3.29 ± 0.09 for the No-GIK group) or in the cross clamp time (47.7 ± 1.5 min for the GIK group compared with 44.5 ± 1.4 min for the No-GIK group) and cardiopulmonary bypass time (90.4 ± 2.5 min for the GIK group compared with 87.5 ± 2.4 min for the No-GIK group). All patients received at least one internal mammary artery bypass graft.

	Before induction of anesthesia, there was no difference in glucose levels between the two groups (180.4 ± 6.9 mg/dL for the GIK group and 179.0 ± 3.8 mg/dL for the No-GIK group). Just before cardiopulmonary bypass surgery, however, patients treated with GIK had glucose levels that were considerably more favorable compared with the No-GIK group (169.2 ± 4.9 mg/dL and 209.2 ± 5.3 mg/dL, respectively; P < .001). After 18 hours in the ICU, the effect was still evident (170.6 ± 5.3 mg/dL for the GIK group and 257.1 ± 6.0 mg/dL for the No-GIK group; P < .001). Cardiac indices were considerably higher for patients treated with GIK, which was still evident 18 hours after the GIK was stopped (2.91 ± 0.06 L/min/m2 for the GIK group compared with 2.43 ± 0.06 L/min/m2 for the No-GIK group;

	P < .001). In addition, serum lactate levels were lower as a result of greater glycemic control. At 6 hours, the serum lactate level for the GIK group was 1.75 ± 0.1 mmol/L compared with 2.33 ± 0.31 mmol/L for the No-GIK group (P = .04). The effect remained and was still evident at 12 hours, with 1.36 ± 0.11 mmol/L for the GIK group compared with 2.24 ± 0.19 mmol/L for the No-GIK group (P < .001). After 6 hours, free fatty acid levels were also lower, with 0.33 ± 0.03 mEq/L for the GIK group and 0.57 ± 0.05 mEq/L for the No-GIK group (P < .001).

	The postoperative results are shown in table 3. GIK-treated patients had less need for pacing and inotropic support, gained less weight, spent less time on the ventilator, and had a lower incidence of atrial fibrillation, pneumonia, and wound infection. This contributed to a shorter postoperative hospital length of stay for GIK patients (6.5 ± 0.1 days for the GIK group compared with 9.2 ± 0.3 days for the No-GIK group; P = .003).

	Table 4 summarizes the follow-up data obtained over 5 years. Kaplan-Meier estimates of survival showed a significant (P = .04) survival advantage for GIK-treated patients during the initial 2 years following surgery. One GIK patient died of intractable congestive heart failure; six No-GIK patients died, five from cardiac-related causes and one from a cerebrovascular accident. GIK-treated patients developed fewer recurrent wound infections involving the sternum and leg, had decreased episodes of recurrent ischemia, and maintained a lower angina class.

	There is growing evidence that hyperglycemia is associated with poorer outcomes in patients with acute coronary syndromes. Several studies have shown that patients with serum glucose levels of 200 mg/dL or above have higher in-hospital and 1-year mortality rates, are at greater risk for reinfarction and congestive heart failure, and have less recovery of contractible function after an acute MI independent of the presence or absence of diabetes mellitus.3-5 Furnary and colleagues found that perioperative mortality was reduced by 50% when CABG surgery patients received a continuous insulin drip compared with intermittent subcutaneous injections of insulin to manage serum glucose levels (2.5% versus 5.3%; P < .001).6

	There are several reasons hyperglycemia may be detrimental to the ischemic myocardium. The inflammatory response is augmented, resulting in an increased production

	of superoxide radicals, which contributes to endothelial dysfunction. Nicotinamide adenine dinucleotide phosphate levels are decreased, ultimately resulting in less synthesis of endothelial nitric oxidase synthase, a potent vasodilator. Platelet function is impaired due to increased levels of plasminogen activator inhibitor-1 and adhesion molecules.7 This predisposes to coronary thrombosis, which ultimately affects long-term graft patency. Insulin resistance, which occurs during cardiopulmonary bypass, accentuates the detrimental effects of hyperglycemia in the diabetic CABG surgery patient by increasing the serum levels of free fatty acids, which predisposes to arrhythmias and cotributes to depressed myocardial contractility.8

	Insulin is beneficial to the ischemic diabetic myocardium by enhancing nitric oxide synthase through the PI3-kinase-Akt pathways.9 It has been shown to decrease free fatty acids and increase myocardial uptake of glucose after CABG surgery.10 It decreases the inflammatory response by lowering levels of oxygen free radicals, adhesion molecules, and C-reactive protein.11 It may also prevent coronary thrombosis by decreasing levels of plasminogen activator inhibitor-1.7

	Our study did not provide data to determine the mechanisms for the beneficial effects of GIK in prolonging survival of the diabetic CABG surgery patient. We hypothesize that GIK’s favorable effects on endothelial function and its potential for decreasing vascular inflammation and reducing thrombogenicity contributed to improved early graft patency and enhanced viability of ischemic myocardial tissue. Our future studies will attempt to define these protective mechanisms as we endeavor to develop the optimal concentration of glucose and insulin for maximum cardioprotection in these high-risk patients.

	This study has shown that perioperative morbidity can be reduced, and long-term survival enhanced, by tighter glycemic control in the diabetic CABG surgery patient. As the number of diabetic patients requiring CABG surgery increases, these interventions can result in a significant savings of health care dollars in a group of patients known to increase health care costs because of disappointing clinical outcomes. GIK is a safe, effective, and low-cost method of achieving more optimal short- and long-term results in these high-risk patients.

Effect of tight glycemic control in diabetic CABG surgery patients

	A 71-year-old woman was admitted to the coronary care unit with unstable angina requiring intravenous heparin and nitroglycerin. Her risk factors included type 2 diabetes mellitus controlled with oral agents, hypertension, hyperlipidemia, a 40—pack-year smoking history, and obesity. Cardiac catheterization showed a 70% mid-left main lesion, an 80% proximal left anterior descending lesion, a 75% proximal right coronary artery lesion, and a total occlusion of the mid-circumflex artery with right-to-left collaterals filling a large obtuse marginal branch. Her ejection fraction was 35%, with posterolateral hypokinesis and 2  mitral regurgitation. Her serum glucose level averaged 240 mg/dL before catheterization.

	At the time of surgery, after anesthetic induction, a glucose-insulin-potassium (GIK) solution consisting of 5% dextrose in water, 80 U of regular insulin, and 40 mEq of potassium chloride was started at 30 mL/hr and titrated to keep the blood glucose level below 200 mg/dL. The patient’s serum glucose level averaged 150 mg/dL during the surgery. Myocardial revascularization was achieved on cardiopulmonary bypass with blood cardioplegic arrest using the left internal mammary artery to bypass the mid-left anterior descending artery and single saphenous vein grafts to the distal right coronary artery and obtuse marginal branches. After bypass, the intraoperative transesophageal echocardiogram showed an ejection fraction of 45%, markedly improved posterolateral wall motion, and only trace mitral regurgitation. The GIK solution was continued for 12 hours after the patient’s arrival in the intensive care unit. She was started on a fixed dose of isophane insulin with a regular insulin sliding scale designed to keep serum glucose levels below 200 mg/dL.

	The patient had a maximal weight gain of 5 lb. She was extubated within 6 hours of arrival in the intensive care unit and required no inotropic support to maintain a cardiac index over 2 L/min/m2. She was discharged on the sixth postoperative day, with prescriptions for isophane insulin, an HMG-CoA reductase inhibitor (statin), an angiotensin-converting enzyme inhibitor, aspirin, a beta blocking agent, and a diuretic. 

Case report: Tight glycemic control for a  diabetic CABG patient

	Atrial septal aneurysm and patent foramen ovale (PFO) have been

	separately associated with stroke, and the combination confers an especially high risk for stroke. Additionally, the eustachian valve, a remnant of the right valve of the sinus venosus, which directs blood flow from the inferior vena cava to the fossa ovalis area in the fetal period, causes a significant right-to-left shunt in the presence of interatrial communication when prominent in adult life. Filamentous strands in the right atrium, which can span the eustachian valve to the atrial wall, are also a remnant of the right valve of the sinus venosus. An extensive network of strands known as Chiari’s network has been associated with paradoxical embolization, also presumably by directing inferior vena cava flow toward the fossa ovalis area.

	Rodriguez and colleagues (page 31) hypothesized that septal aneurysm is associated with a large PFO and a prominent eustachian valve or right atrial filamentous strands. Thus, the combination of a large PFO and a structure in the right atrium that diverts the inferior vena cava flow toward the PFO would likely lead to an increased chance of paradoxical embolization.

	The authors found that the presence of septal aneurysm was associated with a large PFO and a prominent eustachian valve or right atrial filamentous strands. Patients with septal aneurysm and PFO taking either warfarin or aspirin after stroke did not have an increase in the incidence of recurrent stroke, transient ischemic attacks (TIAs), or death. Patients with septal aneurysm, PFO, and right atrial anatomy predisposing to paradoxical embolization also had no difference in outcome compared with those without these factors. There was no significant difference in treatment results between those taking warfarin and those taking aspirin.

	CLOSURE I is a prospective, multicenter, randomized, controlled trial to evaluate the safety and efficacy of the STARFlex® septal closure system versus best medical therapy in patients with a stroke and/or TIA due to presumed paradoxical embolism through a PFO. It is important to note that the CLOSURE I trial has not yet shown any definitive benefit of aspirin or warfarin to prevent recurrent stroke or death in patients with cryptogenic stroke and atrial septal abnormalities. Additionally, there was insufficient evidence regarding the effectiveness of either surgical or percutaneous closure of PFO in patients with TIA or a single stroke. Devices for PFO closure have been approved for patients who have had a second stroke while taking warfarin. The details of this can be found in PDF format via hyperlink at www.nmtmedical.com/closure1.

	Patients at high risk for recurrent stroke with septal aneurysm, a PFO, and a prominent associated eusta-chian valve or with right atrial filamentous strands did not have a recurrent stroke when taking either warfarin or aspirin.

Risk of stroke in patients with right  atrial abnormalities 

	Patients with atrial septal aneurysm have been shown to have a higher risk of stroke. Those with patent foramen ovale (PFO) also have a higher stroke risk. The combination of the two conditions together increases the risk of stroke significantly.1-5 The eustachian valve is a remnant of the right valve of the sinus venosus, which, during fetal development, directs blood flow from the inferior vena cava to the fossa ovalis region.6 In adults, a marked right-to-left shunt may result when the eustachian valve is prominent in the presence of interatrial communication.7,8 Filamentous strands in the right atrium, which can extend across the eustachian valve to the atrial wall, are also remnants from the right valve of sinus venosus.9 A widespread webbing of strands, or Chiari network, has been shown to be correlated with paradoxical embolization, presumably by also diverting flow from the inferior vena cava to the fossa ovalis region.10

	We theorized that a large PFO and a prominent eustachian valve or right atrial filamentous strands were related to septal aneurysm and that the combination of a large PFO and a structure in the right atrium directing flow from the inferior vena cava to the PFO would lead to increased risk of paradoxical embolization. We assessed the characteristics of PFO in patients with septal aneurysm and evaluated whether the presence of septal aneurysm and PFO, with or without the right atrial anatomical abnormalities described above, affected the rate of stroke.

	The Warfarin-Aspirin Recurrent Stroke Study (WARSS) was designed to evaluate the effect of medical treatment on stroke risk factors as shown by transesophageal echocardiography (TEE). The Patent Foramen Ovale in Cryptogenic Stroke Study (PICSS) was a substudy of WARSS that included patients undergoing TEE for any clinical purpose, as well as WARSS patients with cryptogenic stroke subtype who were invited to undergo TEE specifically for PICSS entry.11 Participants were 30 to 85 years of age and had had an ischemic stroke within 30 days. Participants with a normal baseline international normalized ratio (INR; < l.4) were included in the study. Patients with plans to have surgery for high-grade carotid stenosis or patients with procedure-related stroke or stroke with a cardioembolic source were excluded. All strokes were categorized according to predefined criteria.12 When no specific cause was found even with an extensive diagnostic examination, patients were considered to have suffered a cryptogenic stroke.13

	Patients were randomly assigned to receive aspirin, 325 mg daily, or warfarin, given at a dosage to maintain an INR of 1.4 to 2.8, and an identical placebo. None of the participants received two active treatments or two placebos. A predefined protocol, which used either a biplane or multiplane TEE probe and included extensive characterization of the right atrial anatomy and interatrial septum, was employed. A single investigator, who was blind to treatment assignment, type of stroke, and patient outcome, reviewed the TEE tapes. To identify PFO, saline contrast was injected with and without the use of the Valsalva maneuver or cough. If one or more microbubbles was detected in the left atrium within three cardiac cycles after opacification of the right atrium, a PFO was considered to be present.14 From an image showing the maximum number of microbubbles, the total number of microbubbles was counted on a video frame within three cardiac cycles from the opacification of the right atrium. The size of all PFOs was characterized as small or large based on the size of septal separation, with 2 mm used as the cutoff.11

	A patient was considered to have a prominent eustachian valve if there was a protrusion into the right atrium of 10 mm or more of linear membrane-like structure from the junction of the right atrium and the inferior vena cava in the image plane showing the entrance of the inferior vena cava into the right atrium in a vertical plane using a biplane probe or in an equivalent plane using a multiplane probe. A septal aneurysm was defined as a protrusion of 10 mm or more of the atrial septum into the left or right atrium from the midline position.15 A patient was considered to have right atrial filamentous strands if there were freely mobile linear filamentous structures in the right atrium that were attached to the eustachian valve or to any aspect of the right atrial wall. Patients were followed up monthly in person or on the phone for a period of 24 months. At each monthly follow-up, compliance was evaluated and INRs were adjusted.

	Death from any cause and recurrent ischemic stroke were the primary end points. A secondary end point was transient ischemic attack (TIA). A patient was determined to have recurrent ischemic stroke if a new lesion was visible on a computed tomography or magnetic resonance imaging scan or, if no new lesions were present, the patient had clinical signs and symptoms of stroke for at least 24 hours. The relative hazard ratio was determined by Cox proportional hazards model. The 95% confidence intervals (CIs) for participants with septal aneurysm and PFO were compared with subjects without septal aneurysm and without PFO. Participants with the combination of PFO, septal aneurysm, and prominent eustachian valve or right atrial filamentous strands were also compared with subjects without these conditions. Analyses were performed for the entire group and for patients with cryptogenic stroke. The effects of warfarin were compared with aspirin for patients with PFO and septal aneurysm. A probability value of .05 or less was accepted as statistical significance.

	Of the 2,206 patients enrolled in WARSS, 630 took part in the PICSS trial. Among the PICSS participants, there were 265 (42.1%) cryptogenic strokes. There were 601 patients with TEE images suitable for analysis of PFO, 563 for analysis of eustachian valve, 600 for analysis of septal aneurysm, and 563 for analysis of right atrial filamentous strands. Two hundred three patients (33.8%) had PFO; 58.6% of the PFOs were small and 41.4% were large. A total of 69 of 600 patients (11.5%) had an atrial septal aneurysm, 195 of 563 (34.6%) had a prominent eustachian valve, and 106 of 563 (18.8%) had right atrial filamentous strands. Forty-four of 600 patients (7.3%) had a combination of septal aneurysm and PFO, and 252 of 561 patients (44.9%) had right atrial anatomy resulting in inferior vena cava flow to the fossa ovalis region,

	a prominent eustachian valve, or right filamentous strands. Septal aneurysms were significantly associated with the presence of a PFO, and the PFOs seen in patients with septal aneurysm were larger than those seen in patients with no septal aneurysm (table 1). A prominent eustachian valve or right atrial filamentous strand was associated with atrial septal aneurysm. A large PFO in the presence of a prominent eustachian valve or right atrial filamentous strands significantly correlated with septal aneurysm. There were 92 end points, including 71 strokes and 21 deaths, among the patients with TEE images suitable for analysis of septal aneurysm and PFO. There were also 34 TIAs, 7 of which took place before the primary event. The minor hemorrhage rate was considerably lower for patients taking aspirin compared with those taking warfarin, although the rate of major hemorrhage was comparable for both groups.11

	Risk factors, demographic features, and stroke characteristics of patients with PFO and septal aneurysm were similar to those patients without PFO and septal aneurysm. The time to recurrent stroke or death was the same for both groups (table 2). When TIA was considered as an end point, the groups were also similar. In addition, the subgroup of patients with cryptogenic stroke was also similar with regard to the time to end point, with and without inclusion of TIA (table 2). Comparisons were made between patients with septal aneurysm and PFO as well as with prominent eustachian valves or right atrial filamentous strands and patients without these conditions. The time to end points between the groups with and without TIA as an additional end point was similar for the overall group as well as for the group of patients with cryptogenic stroke. When the effects of aspirin and warfarin were compared in patients with septal aneurysm and PFO, the time to recurrent stroke or death was the same; it was also the same when TIA was included as an end point (table 3).

	This study showed that large PFOs are often associated with septal aneurysm, and septal aneurysm is associated with the presence of prominent eustachian valves or right atrial filamentous strands. These anatomical features may partly account for the fact that the combination of septal aneurysm and PFO carries an increased risk of stroke compared with either condition alone.

	Based on the results of this study, we were able to define for the first time the relationship between septal aneurysm with large PFO and eustachian valves or right atrial filamentous strands. In addition, we showed that aspirin and warfarin were equally effective in preventing stroke in patients with septal aneurysm and PFO, and also in patients with septal aneurysm, PFO, and right atrial anatomy favoring paradoxi-

	PICSS showed that when patients are receiving medical treatment, the adverse event rate for stroke patients is not higher for patients with a PFO and a septal aneurysm than for patients with only a PFO.11 We also showed that the event rates for patients with septal aneurysm and PFO are no different from those without septal aneurysm and without PFO. In addition, adverse event rates did not increase in the presence of a eustachian valve or right atrial filamentous strands in patients with PFO and septal aneurysm compared with patients without these conditions. Therefore, the event rate does not increase for patients with septal aneurysm and PFO who are receiving medical treatment compared with those without septal aneurysm and PFO. These results suggest that either the presence of a PFO and septal aneurysm impart a measure of risk not identified by our study or that medical therapy is efficacious in stroke prevention among patients with these conditions.

	In addition, event rates were similar between the warfarin and aspirin groups for patients with PFO and septal aneurysm, suggesting that either therapy may be considered appropriate for patients with these conditions. Patients who were given warfarin had a mean INR of 2.04.11 This INR was similar in patients with and without events, even though the range was broad.

	In our study, septal aneurysm was correlated with large PFO and right atrial anatomy favoring paradoxical embolization. This may be a possible mechanism for the significant correlation of stroke observed with the combination of PFO and septal aneurysm. These patients had outcomes comparable with patients without septal aneurysm and without PFO who were taking aspirin or warfarin. For the entire study sample and for patients with cryptogenic stroke, even when right atrial structures favoring paradoxical embolization existed, there was no difference in event rates. In addition, aspirin and warfarin were equally effective for patients with septal aneurysm and PFO. 

Effect of medical  therapy on stroke due to right atrial abnormalities

	The article by Emberson and Whincup1 and the accompanying commentary by Goldberg2 correctly point out that the three major risk factors for atherothrombotic disease are cholesterol level, systolic blood pressure, and cigarette smoking. These factors account for at least 75% of all atherothrombotic disease events in the British Regional Heart Study. The purpose of this letter is to show that if one uses dyslipidemia instead of total serum cholesterol level, prediction of the at-risk population is even better, that is, 94%.

	The total serum cholesterol level is mainly derived from its major three components: low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, and very-low-density lipoprotein cholesterol. Total serum cholesterol level may be high because of elevations in any or all of the components. At levels of total serum cholesterol associated with atherothrombotic disease (221 mg/dL in men and 239 mg/dL in women in northwest Ohio), however, most men have a low HDL cholesterol level (≤ 40 mg/dL) and most women have a high LDL cholesterol level

	(≥ 170 mg/dL) as the primary component of their dyslipidemia. Clearly, there are men with high LDL cholesterol and women with low HDL cholesterol levels. Use of a ratio between LDL and HDL cholesterol brings these two ends of the dyslipidemic spectrum together.

	The Bowling Green Study (BGS) of the Primary and Secondary Prevention of Atherothrombotic Disease produced a global risk factor graph using the cholesterol retention fraction (LDL—HDL/LDL) on the y axis and systolic blood pressure on the x axis (figure). The diagonal line is the angiographic stabilization/regression line, above which lie 85% of the cholesterol retention fraction-systolic blood pressure plots of all patients of all ages who developed some atherothrom-botic disease event in the BGS database. Of those patients who developed atherothrombotic disease despite a cholesterol retention fraction-systolic blood pressure plot below this threshold line, the vast majority were past or current cigarette smokers. That leaves only 6% of the total atherothrombotic disease population that cannot be predicted by cholesterol retention fraction-systolic blood pressure plot above the threshold line and/or cigarette smoking status.3

	In a meta-analysis of eight published angiographic regression trials (St. Thomas’ Atherosclerosis Regression study, Heidelberg study, Lipoprotein and Coronary Atherosclerosis study, Pravastatin Limitation of Atherosclerosis in the Coronary Arteries-I, Familial Atherosclerosis Treatment study, Program on Surgical Control of Hyperlipidemia,

	National Heart, Lung and Blood Institute Secondary Prevention Trial, and Lopid Coronary Angiography Trial), any therapy that brought the patient’s cholesterol retention fraction-systolic blood pressure plot below the threshold line resulted in angiographic stabilization or reversal of coronary plaque at a minimum average rate of 75%. (Had the Program on Surgical Control of Hyperlipidemia study been structured to control systolic blood pressure, the average rate would have approached 95%.)3 The BGS graph provides a superior means of identifying the population at risk for atherothrombotic disease and guides treatment to maximally stabilize or reverse atherothrombotic disease.

October 2004

Tight glycemic control in diabetic CABG surgery patients

Carvedilol slows renal dysfunction in systolic HF

TZDs: No adverse clinical effects in diabetic HF patients

Statins extend survival in heart failure

Variation of fibrinogen in hypertensive patients

Seasonal variation of fibrinogen in hypertensive patients

Glucose-insulin-potassium: A therapy whose time has come again

Effect of tight glycemic control in diabetic CABG surgery patients

Case report: Tight glycemic control for a diabetic CABG patient

Risk of stroke in patients with right atrial abnormalities

Effect of medical therapy on stroke due to right atrial abnormalities

Identifying risk of atherothrombotic disease