Effects of Sirolimus on Lipids in Renal Allograft Recipients: An Analysis Using the Framingham Risk Model

Authors


  • This work was presented, in part, at the First Joint Transplant Meeting of the American Society of Transplant Surgeons and the American Society of Transplantation, May 13–17, 2000, Chicago, IL, USA, and at the 18th International Congress of the Transplantation Society, Rome, Italy, August 27–September 1, 2000.

* Corresponding author: Conrad B. Blum

Abstract

This report describes the effects of sirolimus on plasma lipids, and uses the Framingham risk model to assess the clinical importance of these effects.

Lipid data from two large controlled studies of 1295 renal transplant patients were analyzed retrospectively. Sirolimus 2 mg/day and 5 mg/day were compared with placebo or azathioprine, and administered concomitantly with steroids and cyclosporine over 12 months.

Hypercholesterolemia and hypertriglyceridemia occurred in all treatment groups and were maximal at 2–3 months. The sirolimus groups evidenced higher lipid levels than the controls, but the elevations diminished over time. At 1 year, the patients given sirolimus 2 mg/day had a mean cholesterol level 17 mg/dL greater and a mean triglyceride level 59 mg/dL greater than the controls. Among the patients given sirolimus 5 mg/day, mean cholesterol was 30 mg/dL greater and mean triglycerides were 103 mg/dL greater than the controls. Treatment with statins and fibrates was effective in reducing cholesterol and triglyceride levels, respectively, in the sirolimus-treated patients. The Framingham risk model predicted that the 17 mg/dL elevation in cholesterol would increase the incidence of coronary heart disease (CHD) by 1.5 new cases per 1000 persons per year and CHD death by 0.7 events per 1000 persons per year.

Lipid elevations observed in the sirolimus-treated patients were manageable, improved over time, and responded to lipid-lowering therapy. Based on the Framingham risk model, the CHD risks associated with these cholesterol elevations are small compared with the baseline risks of the transplant population.

Introduction

Coronary heart disease (CHD) poses a major threat to patients who have undergone renal transplantation. It is the largest single cause of death for these patients; among middle-aged individuals 45–64 years of age, the CHD death rate for transplant patients is approximately 0.6% per year (1), more than five times that for the general population (2).

The major risk factors for CHD in the general population appear to contribute to the risk in transplant patients (3–6). These include age, male sex, hypertension, hypercholesterolemia, low high-density lipoprotein (HDL) cholesterol, diabetes mellitus, and cigarette use. The immunosuppressant medications, cyclosporine, corticosteroids, and tacrolimus, are known to exacerbate some of these risk factors (7–15).

Sirolimus, a macrocyclic lactone isolated from Streptomyces hygroscopicus, has recently been introduced as an immunosuppressant medication. Phase III trials in renal transplantation showed that the addition of sirolimus to cyclosporine-prednisone-based immunosuppression reduced the frequency of acute rejection and did not adversely affect the frequency of malignancies and most infections (16, 17). Furthermore, at 12 months, patient survival (> 92%) and graft survival (> 87%) have been excellent. The overall safety and efficacy results of these trials have been presented elsewhere (16, 17).

The adverse effects of sirolimus on cardiovascular risk factors appear to be limited to increased plasma lipid levels. Studies have shown that diabetes was not increased when sirolimus was added to cyclosporine and corticosteroids (16–18). Previous studies in which sirolimus was compared with cyclosporine showed improved renal function, less hypertension, and a similar incidence of new post-transplant insulin-dependent diabetes (19, 20). Early data had indicated that sirolimus can increase cholesterol and triglyceride levels (19–21). The current report describes the influence of sirolimus on plasma lipid levels in two large-scale controlled clinical trials. Furthermore, the likely impact on CHD of sirolimus-associated changes in cholesterol levels was assessed using the Framingham risk model (22).

Materials and Methods

One thousand two hundred and ninety-five renal allograft patients were enrolled in two randomized multicenter phase III trials of sirolimus. The US trial involved 719 patients (16), and the global trial involved 576 patients (17). These patients were randomly assigned in a 2 : 2 : 1 ratio to treatment with sirolimus 2 mg/day, sirolimus 5 mg/day, or a comparator treatment, which was azathioprine for the US trial and placebo for the global trial. All patients received corticosteroids and cyclosporine microemulsion (Neoral®, Novartis Pharmaceuticals Corporation, East Hanover, NJ, USA). Entry criteria included a pretransplantation fasting serum triglyceride level ≤ 500 mg/dL and a fasting cholesterol level ≤ 350 mg/dL.

In the US trial, randomization was stratified by center and by race (black or nonblack) and occurred 24–48 h after transplantation. In the global trial, randomization was stratified by center and by donor source (living vs. cadaveric) and occurred immediately before transplantation.

In the sirolimus 2 mg/day group, an initial (loading) dose of sirolimus 6 mg was given orally 24–48 h after transplantation; subsequent doses were 2 mg daily. In the sirolimus 5 mg/day group, an initial dose of 15 mg was given, and subsequent doses were 5 mg daily.

All patients received corticosteroids, initially a maximum of 250 mg (global trial) or 500 mg (US trial) of methylprednisolone on the day of transplantation, then tapering to 30 mg prednisone daily within 1 week of transplantation. The dose was further tapered to 10 mg daily by month 6 and was maintained at 5–10 mg daily from month 6 onward.

The dose of cyclosporine was adjusted during the first month to maintain a trough concentration of 200–400 ng/mL in the global study and 200–350 ng/mL in the US study. During months 2–3, the cyclosporine trough concentration was adjusted to 200–300 ng/mL and subsequently to 150–250 ng/mL.

In the US study, azathioprine (or a placebo for azathioprine) was given in a loading dose of 5 mg/kg; subsequent daily doses were 2–3 mg/kg.

The study protocol required a reduction in the dose of the blinded study drug (sirolimus, azathioprine, or placebo) in the event of severe hyperlipidemia. The dose was reduced by 50% for cholesterol levels exceeding 450 mg/dL or triglyceride levels exceeding 750 mg/dL, by 75% for cholesterol exceeding 600 mg/dL or triglycerides exceeding 1200 mg/dL, and the drug was temporarily discontinued for cholesterol levels exceeding 750 mg/dL or triglyceride levels exceeding 1500 mg/dL. If severe hyperlipidemia persisted for 21 days despite discontinuation of the study medication, the medication was not resumed. Certain other laboratory abnormalities also mandated reduction or discontinuation of the study drug.

Serum cholesterol and triglyceride concentrations in fasting blood samples were obtained monthly for the first 4 months and then at months 6, 9, and 12. Differences in cholesterol and triglyceride levels at the various post-transplant time points were analyzed after adjusting for mean baseline values by analysis of covariance. High-density lipoprotein cholesterol was not measured.

Pooling of data from the US and Global Studies was felt to be appropriate because these studies did not differ in any way that would be expected to alter lipid responses.

The Framingham risk model was used to assess the potential impact of sirolimus-associated increases in serum cholesterol concentration (22). This multivariate parametric model allows estimation of the risk associated with changes in specific values of the following major coronary risk factors: age, sex, cigarette use, total and HDL cholesterol levels, systolic blood pressure, diabetes, and left ventricular hypertrophy. Baseline 10-year risk for CHD (fatal and nonfatal) and for CHD death was first calculated, and then a second calculation was performed using a higher value for total cholesterol. The difference between these two results represents the expected impact of the increase in cholesterol concentration. All risk calculations were made for a 10-year period and results are presented on an annualized basis.

Results

Study population

The demographic characteristics of the patients randomized to treatment have been described in previous publications (16, 17). A slightly higher percentage of women and blacks were randomized to the azathioprine group. The treatment groups were well balanced with respect to age.

In each group, approximately 50% of patients persisted with the blinded therapy for 1 year. The reasons for discontinuation of therapy are outlined in Table 1. Efficacy failure (acute rejection, graft loss, or death) was less common as a cause of discontinuation among the sirolimus patients, while other adverse events were somewhat more common among the sirolimus 5 mg/day patients.

Table 1. Discontinuation of treatment (% of randomized patients)
 SRL (2 mg/day)
(n = 511)
SRL (5 mg/day)
(n = 493)
AZA
(n = 161)
Placebo
(n = 130)
  1. AZA = azathioprine

  2. SRL = sirolimus

Total44.250.950.950.0
Efficacy failure14.212.221.723.8
Protocol violation 3.1 3.4 2.5< 1
Medical/other events17.218.515.518.5
Adverse reaction 9.716.811.2 6.9

Mean cholesterol and triglyceride levels

Figure 1 shows the observed mean cholesterol levels during the first 12 months after transplantation. In all treatment groups, cholesterol levels were maximal at month 2 and then decreased during the remainder of the 12-month follow-up period.

Figure 1.

Mean cholesterol levels by treatment group over 12 months. Data at each time point are from all patients taking study drugs for whom fasting cholesterol concentrations were available.

A dose-dependent increase in mean cholesterol levels was consistently seen in the sirolimus groups. The values for each of the sirolimus groups were significantly greater than those for the comparator groups at months 3, 6, and 12 (p < 0.05).

In the placebo and azathioprine groups, mean cholesterol levels stabilized by month 6. In the two sirolimus groups, mean cholesterol levels continued to decrease through month 12. Thus, the difference between the sirolimus groups and the comparator groups diminished with time.

A similar pattern was seen for mean triglyceride concentrations (Figure 2). In each group, values were maximal at month 2 or 3. Mean triglyceride concentrations then diminished in the sirolimus groups. In the placebo and the azathioprine groups, however, they changed very little following the third month after transplantation. Thus, the difference in mean triglyceride levels between the sirolimus groups and the placebo and azathioprine groups diminished over time.

Figure 2.

Mean triglyceride levels by treatment group over 12 months. Data at each time point are from all patients taking study drugs for whom fasting triglyceride concentrations were available.

At 12 months, the sirolimus 2 mg/day group had a mean cholesterol concentration 17 mg/dL greater than the comparators, and the 5 mg/day group had a mean cholesterol concentration 30 mg/dL greater than the comparators. The sirolimus 2 mg/day group had a mean triglyceride concentration 59 mg/dL greater than that of the comparators, and the sirolimus 5 mg/day group had a mean triglyceride concentration 103 mg/dL greater than that of the comparators.

To provide the precision that a larger sample size affords, the analysis at each time point included all patients for whom fasting lipid values were available at that time point. Thus, this was not a cohort analysis. Similar plots were made for the adjusted mean cholesterol and triglyceride levels of the much smaller cohort for whom fasting lipid data were available at every time point (data not shown). Sample size varied between 10 (azathioprine group, triglyceride analysis) and 43 (sirolimus 5 mg/day group, cholesterol analysis). Data from this group gave results similar to those seen in the much larger noncohort analysis. Thus, the observations made with the larger group of subjects do not appear to be artifacts attributable to patients who discontinued treatment.

Prevalence of hyperlipidemia

Table 2 shows the prevalence of cholesterol concentrations of at least 240 mg/dL and triglyceride concentrations of at least 400 mg/dL. These values have been established by the National Cholesterol Education Program as defining high cholesterol and high triglyceride levels (23). Hyperlipidemia, thus defined, was more prevalent in the sirolimus groups than in the azathioprine or placebo groups. In each treatment group, however, the prevalence of high cholesterol or high triglyceride levels diminished between month 3 and month 12 after transplantation.

Table 2. Prevalence (%) of cholesterol 240 mg/dL and triglycerides 400 mg/dL
 Month 3Month 12
 CHOL ≥ 240TG ≥ 400CHOL ≥ 240TG ≥ 400
  1. CHOL = cholesterol

  2. SRL = sirolimus

  3. AZA = azathioprine

  4. TG = triglycerides

SRL (2 mg/day)74.822.350.314.2
SRL (5 mg/day)82.731.254.825.9
AZA60.0 7.436.2 9.3
Placebo61.9 8.143.1 8.0

The box-and-whisker plots shown in Figures 3 and 4 provide additional detail. By month 12, only a small percentage of the sirolimus patients had values exceeding the range of the control patients.

Figure 3.

Fasting cholesterol concentrations in the US (A) and global (B) studies. The boxes define the middle two quartiles of the cholesterol distribution. The line across the box defines the mean, and the cross within the box defines the median. The vertical lines above and below each box end at the 90th and 10th percentiles. Values exceeding the 90th percentile or below the 10th percentile are plotted individually. Data represent crude (unadjusted) cholesterol concentrations.

Figure 4.

Fasting triglyceride concentrations in the US (A) and global (B) studies. The boxes define the middle two quartiles of the triglyceride distribution. The line across the box defines the mean, and the cross within the box defines the median. The vertical lines above and below each box end at the 90th and 10th percentiles. Values exceeding the 90th percentile or below the 10th percentile are plotted individually. Data represent crude (unadjusted) triglyceride concentrations.

Only 0.4% of the sirolimus 2 mg/day patients and 2.5% of the sirolimus 5 mg/day patients required discontinuation of treatment because of hyperlipidemia.

Use of lipid-lowering medication

In these studies, lipid-lowering drugs were prescribed at the discretion of the investigators. Table 3 shows the frequency of use of statins and fibrates at 12 months after transplantation. While the specific drugs were recorded, the dosages were not. Lipid-lowering drugs were used more often in the sirolimus patients than in the controls and more often in the US study than in the global study. In both studies, statins were used in 20% to 25% more sirolimus patients than controls, and fibrates were used in 5% to 10% more sirolimus patients than controls. The use of these agents was not substantially greater in the sirolimus 5 mg/day group compared with the sirolimus 2 mg/day group.

Table 3. Frequency (%) of lipid-lowering drug use 12 months after transplantation
 US studyGlobal study
 StatinFibrateStatinFibrate
  1. SRL = sirolimus

  2. AZA = azathioprine

SRL (2 mg/day)48.211.942.7 8.8
SRL (5 mg/day)53.213.142.910.5
AZA31.0 4.9
Placebo16.9 3.0

Figure 5 shows the effects of statins on total serum cholesterol levels. In each of the treatment groups, a statistically significant reduction in total serum cholesterol concentration occurred in response to treatment with statins. The magnitude of the cholesterol reduction was similar in all treatment groups: 50 mg/dL, 35 mg/dL, 47 mg/dL, and 50 mg/dL for patients treated with placebo, azathioprine, sirolimus 2 mg/day, and sirolimus 5 mg/day, respectively.

Figure 5.

Response to statin therapy in sirolimus-treated patients.

No sirolimus-treated patient developed rhabdomyolysis, a complication noted with increased frequency when statins have been used with cyclosporine (24).

Response to fibrates was assessed in a small number of patients. A preponderance of the patients responded with a reduction in serum triglyceride concentration. The responses of individual patients in the sirolimus 5 mg/day group to fibrate therapy are shown in Figure 6, indicating only a small number of nonresponders.

Figure 6.

Response to fibrates in individual patients treated with sirolimus 5 mg/day.

Because of the absence of information on the agents and doses used, we cannot compare the efficacy of statins and fibrates in these patients to that seen in nontransplant populations.

Clinical consequences of lipid elevation

The frequency of potential clinical complications of hyperlipidemia in each treatment group is shown in Table 4. Over the 1-year follow-up period, there were no important differences in the rates of pancreatitis, myocardial infarction, or cerebrovascular accidents across the treatment groups. Moreover, there were no cases of pancreatitis resulting from hypertriglyceridemia in any of the sirolimus-treated patients.

Table 4. Frequency [% (n)] of potential hyperlipidemia complications
 PancreatitisMyocardial infarctionStroke
  1. SRL = sirolimus

  2. AZA = azathioprine

SRL (2 mg/day)1.4 (7)1.6 (8)1.4 (7)
SRL (5 mg/day)0.4 (2)1.7 (8)1.7 (8)
AZA1.3 (2)1.3 (2)2.5 (4)
Placebo0.8 (1)0 (0)0.8 (1)

Coronary risk assessment

The Framingham risk model was used to estimate the long-term consequences of the sirolimus-associated increase in serum cholesterol (22). This multivariate parametric model allows assessment of the simultaneous impact of several risk factors. The Framingham model was developed for a general population. In applying it to a transplant population, it seemed essential to set baseline levels of risk factors in a manner that would mimic the very high coronary risk of transplant patients.

Data from the US Renal Data System (USRDS, Ann Arbor, MI, USA) for the period 1995–97 indicated that the CHD death rate (sum of deaths from acute myocardial infarction and cardiac arrest) for 45- to 64-year-old renal transplant patients is 6.7 deaths per 1000 patients per year (1). This statistic, however, may understate the actual coronary death rate because the cause of death was missing or unknown in half of all cases.

The Framingham model predicts a CHD death rate of 6.7 per 1000 persons per year for individuals with the following pattern of risk factors: male, 51 years of age, cigarette smoker, systolic blood pressure 145 mmHg, total cholesterol 235 mg/dL, HDL cholesterol 35 mg/dL, no diabetes, and absence of left ventricular hypertrophy on ECG. This pattern of risk factors was used as the baseline in the modeling and was associated with a total CHD incidence (fatal and nonfatal) of 23.5 new cases per 1000 persons per year.

The impact of sirolimus on major coronary risk factors was limited to an increase in the total serum cholesterol level. The blinded studies reported here demonstrated that sirolimus had no impact on the frequency of diabetes or (with the exception of sirolimus 2 mg/day vs. azathioprine) on hypertension. Data on the impact of sirolimus on HDL cholesterol are limited to two small studies (unpublished observations, data on file, Wyeth Pharmaceuticals, Collegeville, PA, USA, 2001). In a short-term metabolic ward study of six renal transplant patients, treatment with sirolimus had no effect on HDL cholesterol, except in one patient whose baseline level increased by 19.6%. In a group of patients with psoriasis, sirolimus 3 mg/m2/day (n = 5) as monotherapy was compared with placebo (n = 4); HDL cholesterol decreased by 10.3% with placebo and by 8.1% with sirolimus. For the modeling therefore, it was assumed that sirolimus has no effect on HDL cholesterol.

Figure 7 shows the predicted impact on coronary risk of the sirolimus-associated increases in mean serum cholesterol (17 mg/dL for the 2 mg/day dose and 29 mg/dL for the 5 mg/day dose). The 17 mg/dL increase in mean cholesterol associated with sirolimus 2 mg/day, if not further treated, is predicted to result in 0.7 CHD deaths per 1000 patients per year and 1.5 CHD (fatal or nonfatal) cases per 1000 patients per year. The 30 mg/dL increase in mean cholesterol associated with sirolimus 5 mg/day is predicted to result in 1.2 CHD deaths per 1000 patients per year and 2.6 CHD (fatal or nonfatal) cases per 1000 patients per year.

Figure 7.

Framingham risk model: predicted effects of sirolimus-associated increases in total cholesterol. The lower portion of each bar reflects baseline risk. The upper portion reflects the incremental risk expected as a result of the sirolimus-associated increases in cholesterol: 17 mg/dL for sirolimus 2 mg/day (two left bars), and 29 mg/dL for sirolimus 5 mg/day (two right bars). CHD = coronary heart disease.

Discussion

Cardiovascular disease is the greatest life-threatening risk for renal allograft recipients. While this is also true for the general population, death rates from cardiovascular diseases for middle-aged renal allograft recipients are substantially greater than those seen in the general population. As with the general population (2), data from the US Renal Data System indicate that over two-thirds of cardiac deaths are the result of CHD (1).

The same factors that increase the risk for CHD in the general population also cause increased risk in the transplant population. In particular, several reports indicate that hypercholesterolemia contributes to cardiovascular risk after transplantation, as in nontransplant patients (3, 4, 25).

Renal transplant patients carry an unusually high burden of cardiovascular risk factors. In a large series, Kasiske et al. found that diabetes was the cause of end-stage renal disease in 28% of transplant patients (5). A comprehensive review by a task force of the National Kidney Foundation concluded that the prevalence of hypertension in renal transplant recipients is 70% to 85% (6). Certainly, the pre-existence of these conditions frequently sets the stage for end-stage renal disease and transplantation. Additionally, the immunosuppressant medications on which transplant recipients depend often exacerbate coronary risk status. Steroids and tacrolimus are diabetogenic (7–9), and steroids as well as cyclosporine cause elevations of blood pressure (10–15).

Notably, the HDL cholesterol levels of transplant recipients have been somewhat higher (approximately 55 mg/dL) than the population mean (3, 5, 26). This is a consequence of the use of corticosteroids, which increase HDL cholesterol levels (27, 28).

However, total and low density lipoprotein (LDL) cholesterol levels tend to be elevated in transplant populations. Dimeny et al. reported a mean serum cholesterol level of 262 mg/dL and an LDL cholesterol level of 175 mg/dL at 6 months after transplantation in 105 renal transplant recipients (29). Aakhus et al. reported a mean total cholesterol level of 279 mg/dL in 403 transplant patients (26). Aker et al. reported similar findings in 427 renal transplant recipients (3), with a mean cholesterol concentration of 273 mg/dL and a mean LDL cholesterol concentration of 189 mg/dL at 24 months after transplantation. Kasiske et al. in a 10-year follow up, reported somewhat lower mean cholesterol levels, with a total cholesterol of 247 mg/dL and LDL cholesterol of 153 mg/dL (5). These investigators commented on the frequent use of lipid-lowering medications in their population. Nonetheless, these means reported by Kasiske et al. are clearly above the desirable range for a population at such high risk for CHD.

The hyperlipidemia of transplant patients is at least partly a result of the use of immunosuppressant medications. Steroids (13, 27, 30), cyclosporine (28, 31), and sirolimus can increase plasma cholesterol concentrations. Cyclosporine (Sandimune but perhaps not Neoral formulation) (32, 33) and tacrolimus (33) may also foster atherogenesis by increasing the oxidizability of LDL. Additionally, cyclosporine may foster atherogenesis by increasing plasma levels of homocysteine (34, 35).

Knowing this, the purpose of the current work was to describe the effects of sirolimus on plasma lipid levels and to assess the clinical importance of these effects.

A dose-related increase in cholesterol and triglyceride levels occurred in each of the treatment groups; however, this diminished with the passing of time. The increases were greatest in the sirolimus groups, but the excess of lipid levels in sirolimus vs. control patients also diminished over time. The mechanisms by which sirolimus fosters hyperlipidemia have not been fully delineated. However, kinetic studies show that treatment with sirolimus reduces the clearance of very low density lipoprotein and of LDL from plasma (unpublished observations, data on file, Wyeth Pharmaceuticals, Collegeville, PA, USA, 2001).

Several factors are responsible for the reduction of lipid levels over time, including protocol-mandated reductions in the doses of steroids and cyclosporine. However, the changes in doses of cyclosporine and steroids were applied to the control groups as well as to the sirolimus groups. Cyclosporine levels were virtually identical in each of the treatment groups, as was the steroid dosing. Thus, these cannot be responsible for the reduction in the difference between the control and sirolimus patients. The use of lipid-lowering drugs accounts for some of the reduction in lipid levels in each of the four treatment groups; this was more frequent among the sirolimus-treated patients. The current studies cannot determine whether other factors in addition to lipid-lowering drugs may have contributed to the time-related reduction in the lipid elevations of the sirolimus-treated patients compared with the controls.

The study protocols mandated decreases in the dose of sirolimus for patients who developed severe hyperlipidemia or certain other laboratory abnormalities. However, reduced dosing with sirolimus was not the cause of the diminution of lipid elevations (sirolimus vs. azathioprine and placebo groups) that occurred over time. Between months 3 and 12, the mean sirolimus dose that patients actually received decreased by only 4% in the group assigned to sirolimus 2 mg/day (from 1.89 mg/day to 1.81 mg/day) and by 8% in the group assigned to sirolimus 5 mg/day (from 4.50 mg/day to 4.15 mg/day). During this same time period, however, the sirolimus 2 mg/day group sustained a 50% reduction in cholesterol elevation (compared with azathioprine and placebo groups) and a 36% reduction in triglyceride elevation. The sirolimus 5 mg/day group sustained a 51% reduction in cholesterol elevation and a 34% reduction in triglyceride elevation. It is inconceivable that the small (4% and 8%) reductions in sirolimus doses that occurred between months 3 and 12 could have been responsible for these very major reductions in lipid elevation.

It should be emphasized that these results reflect the overall effect of all treatments given to these patients. Thus, the lipid-elevating effects of sirolimus are somewhat mitigated by efforts to treat hyperlipidemia with drugs and possibly also with diet. Thus, these data allow a judgment of the impact that sirolimus has on coronary risk in a ‘real world’ setting. One might presume that with more intensive efforts at therapy, cholesterol and triglyceride levels could be further improved. These data indicate that a more intensive treatment of hyperlipidemia may have been appropriate for all treatment groups: both sirolimus groups, as well as the azathioprine and placebo groups.

The important clinical sequelae of hyperlipidemias are atherosclerotic vascular disease and, for severe hypertriglyceridemia, pancreatitis. It is noteworthy that neither clinical vascular events nor pancreatitis were observed with greater frequency in association with sirolimus treatment. However, the 1-year follow up was brief for an assessment of vascular risk. Additionally, the sample size of these studies would not provide adequate power for an assessment of vascular events.

The Framingham risk model was utilized to assess the potential long-term impact on CHD of sirolimus-associated increases in cholesterol. This model has some limitations. It does not account for regression dilution bias, by which the strength of a risk factor is somewhat underestimated when that risk factor has been determined by a single measurement rather than by multiple measurements (36). The Framingham population was of European origin, and in certain non-European groups data from the Framingham population either overestimate or underestimate absolute risk: absolute risk tends to be underestimated in south Asian populations (37–39) and overestimated in East Asian populations (40, 41). Nevertheless, even in these populations the Framingham model can provide useful estimates of relative risk and the effects of changes in risk factors. Finally, the Framingham model does not include certain risk factors whose importance is less well established. These include plasma homocysteine levels, physical inactivity, obesity, lipoprotein (a) levels, hypertriglyceridemia, and a predominance of small, dense LDL particles.

The Framingham model was developed using prospective data collected from 1968 to 1975 from 5573 participants in the Framingham Heart Study and the Framingham Offspring Study and thus reflects a general population, not a transplant population. Unfortunately, no model has been developed using data from a transplant population. To make the Framingham model most applicable to a transplant population, baseline risk factors were stipulated in a manner that would mimic the reported baseline risk of transplant patients.

For the 17-mg/dL cholesterol elevation seen with sirolimus 2 mg/day, the model predicts an annual increase of 1.5 new cases of CHD per 1000 persons per year; additionally, the model predicts an increase of 0.7 CHD deaths per 1000 patients/year. These increases are small in an absolute sense and in comparison with the baseline risk of transplant patients. In summary, lipid elevations observed in sirolimus-treated patients are manageable, improve over time, and respond to lipid-lowering therapy. It is reasonable to presume that with more intensive antihyperlipidemic treatment, the expected risks could be further diminished.

Acknowledgments

This work was supported by clinical funds from Wyeth-Ayerst Research, Philadelphia, PA, USA.

The author thanks Bernadette Maida, Susan Nastasee, and Robert Goldberg-Alberts of Wyeth-Ayerst Research for assistance in preparation of the manuscript. The author further acknowledges inspiration and support for this and considerable other work given by the late Dr Robert I. Levy.

The following investigators participated in the RAPAMUNE® (Wyeth-Ayerst Pharmaceuticals, Philadelphia, PA, USA) US study group

Patricia Adams, MD (Bowman Gray School of Medicine, Winston-Salem, NC); Francisco Badosa, MD (Albert Einstein Medical Center, Philadelphia, PA); Stephen Bartlett, MD (University of Maryland Medical System, Baltimore, MD); J. Andrew Bertolatus, MD (University of Iowa Hospitals, Iowa City, IA); Kenneth Brayman, MD, PhD (Hospital of the University of Pennsylvania, Philadelphia, PA); Khalid Butt, MD (Westchester Medical Center, Valhalla, NY); David Conti, MD (Albany Medical Center, Albany, NY); Gabriel Danovitch, MD (UCLA School of Medicine, Los Angeles, CA); John Dunn, MD (UCSD Medical Center, San Diego, CA); Robert Ettenger, MD (UCLA School of Medicine, Los Angeles, CA); Ralph Fairchild, MD (Tufts University School of Medicine, Boston, MA); Robert Fisher, MD (Medical College of Virginia, Richmond, VA); A. Osama Gaber, MD (University of Tennessee Medical Center, Memphis, TN); Paul Gores, MD (Carolinas Medical Center, Charlotte, NC); Sharon Inokuchi, MD, PharmD (California Pacific Medical Center, San Francisco, CA); Barry D. Kahan, PhD, MD (University of Texas School of Medicine, Houston, TX); David Laskow, MD (Allegheny University Hospitals, Philadelphia, PA); Alan Leichtman, MD (University of Michigan Medical Center, Ann Arbor, MI); Marc Lorber, MD (Yale University School of Medicine, New Haven, CT); Arthur Matas, MD (University of Minnesota, Minneapolis, MN); Robert Mendez, MD (National Institute of Transplantation, Los Angeles, CA); Stuart Meyers, MD (Temple University School of Medicine, Philadelphia, PA); Joshua Miller, MD (University of Miami, Miami, FL); Anthony Monaco, MD (Deaconess Hospital, Boston, MA); John Neylan, MD (Emory University Hospital, Atlanta, GA); Mark Pescovitz, MD (Indiana University Medical Center, Indianapolis, IN); Raymond Pollack, MB, FRCS, FACS (University of Illinois College of Medicine, Chicago, IL); John Scandling, MD (Stanford University Medical Center, Stanford, CA); Steven Steinberg, MD (Sharp Memorial Hospital, San Diego, CA); Terry Strom, MD (Beth Israel Hospital, Boston, MA); Rodney Taylor, MD (University of Nebraska Medical Center, Omaha, NE); Stephen Tomlanovich, MD (UCSF Medical Center, San Francisco, CA); David VanBuren, MD (Vanderbilt University Medical Center, Nashville, TN); Jorge Velosa, MD (Mayo Clinic, Rochester, MN); Samuel Weinstein, MD (Fletcher Medical Center, Tampa, FL); Duane Wombolt, MD (Clinical Research Associates of Tidewater, Norfolk, VA); E. Steve Woodle, MD (University of Chicago Medical Center, Chicago, IL); and Francis Wright, MD (San Antonio Community Hospital, San Antonio, TX).

The following investigators participated in the RAPAMUNE® Global study group

Paolo Altieri, MD (Ospedale San Michele, Cagliari, Italy); Manuel Arias, MD (Hospital Marques de Valdecilla, Santander, Spain); Oystein Bentdal, MD (University Hospital Rikshospitalet, Oslo, Norway); Marco Castagneto, MD (Universita Cattolica del Sacro Cuore, Rome, Italy); Jeremy Chapman, MD (Westmead Hospital, Sydney, Australia); Kerstin Claesson, MD (University Hospital, Uppsala, Sweden); Edward Cole, MD (The Toronto Hospital, Toronto, Ontario, Canada); Raffello Cortesini, MD (University of Rome La Sapienza, Rome, Italy); Pierre Daloze, MD (Hôpital Notre-Dame, Montreal, Quebec, Canada); Geoffrey Duggin, MD (Royal Prince Alfred Hospital, Sydney, Australia); Mitchell Henry, MD (Ohio State University Hospital, Columbus, OH); John Jeffery, MD (Health Sciences Centre, Winnipeg, Manitoba, Canada); Barry D. Kahan, PhD, MD (University of Texas, Houston, TX); Paul Keown, MD (Vancouver General Hospital, Vancouver, British Columbia, Canada); Michele Kessler, MD (CHRU de Nancy Brabois, Les Nancy, France); Richard Knight, MD (Mount Sinai Medical Center, New York, NY); David Landsberg, MD (St. Paul's Hospital, Vancouver, British Columbia, Canada); Nicole LeFrancois, MD (Hôpital Edouard Herriot, Lyon, France); Rolf Loertscher, MD (Royal Victoria Hospital, Montreal, Quebec, Canada); David Ludwin, MD (St. Joseph's Hospital, Hamilton, Ontario, Canada); Allan S. MacDonald, MD (Dalhousie University, Halifax, Nova Scotia, Canada); Tim Mathew, MD (Queen Elizabeth Hospital, Adelaide, Australia); Norman Muirhead, MD (University Hospital, London, Ontario, Canada); Laura Mulloy, DO (Medical College of Georgia, Augusta, GA); James Petrie, MD (Princess Alexandra Hospital, Brisbane, Australia); Claudio Ponticelli, MD (Ospedale Maggiore di Milano, Milan, Italy); Bruce Pussell, MD (Prince Henry Hospital, Sydney, Australia); Russell Rigby, MD (Princess Alexandra Hospital, Brisbane, Australia); Francesco Paolo Schena, MD (Azienda Ospedaliera Policlinico-Consorziale, Bari, Italy); Giuseppe Segoloni, MD (Azienda Ospedaliera S. Giovanni Battista, Turin, Italy); Ahmed Shoker, MD (Royal University Hospital, Saskatoon, Saskatchewan, Canada); Rakesh Sindhi, MD (Medical University of South Carolina, Charleston, SC); Charles Van Buren, MD (St. Luke's Episcopal Hospital, Houston, TX); Rowan Walker, MD (Royal Melbourne Hospital, Melbourne, Australia); and Jeffrey Zaltzman, MD (St. Michael's Hospital, Toronto, Ontario, Canada).

Ancillary