A Randomized Trial of High-Dose Compared with Low-Dose Omega-3 Fatty Acids in Severe IgA Nephropathy

Study Design

The study was a randomized, open-label, parallel-group comparison of treatment for 2 yr with high-dose and low-dose ω-3 fatty acids. As this was an open-label study, patients who completed a minimum of 2 yr of treatment (in the absence of failure) were continued on their same regimens, either low dose or high dose, until study closure (January 2000). The quantities of ω-3 fatty acids we chose for the low-dose group approximated the amounts of C20:5ω3 (EPA) and C22:6ω3 (DHA) that were given to patients in the clinical trial (1) and extended observational study (2). In these earlier studies, patients were given 1.68 to 1.87 g/d EPA and 0.97 to 1.36 g/d DHA. In the present study, patients in the low-dose group received 1.88 g/d EPA and 1.47 g/d DHA; in the high-dose group, patients received 2× this amount. We did not consider it justifiable to have a placebo-control group because the primary renal end point occurred in only 6% of ω-3 fatty acid—treated patients compared with 33% in the placebo group in our previously reported clinical trial (1). The trial was performed in 14 centers of the Mayo Nephrology Collaborative Group (see the Appendix). The study protocol was approved by the institutional review boards at each site, and all patients gave written, informed consent.

Patient Selection and Treatment

Patients aged 18 yr and older with a baseline serum creatinine of 1.5 to 4.9 mg/dl were eligible for the study if they had renal biopsy-proven IgA nephropathy. The diagnosis of IgA nephropathy was based on histologic assessment of renal biopsy tissue performed by one investigator (J.P.G.) and was confirmed by immunofluorescence studies showing predominant or co-dominant mesangial deposition of IgA (5,6). Histopathologic assessment of renal injury was done by a semiquantitative scoring system, as described previously (5). In a previous study of 148 patients with IgA nephropathy, total glomerular score, derived from extent of mesangial cell proliferation, matrix increase, capillary loop narrowing or disruption, glomerular sclerosis, cellular crescents, and fibrous adhesions, was an independent predictor of adverse outcome (5). Therefore, total glomerular score at time of renal biopsy was assessed in this study.

At study entry, patients were randomized within strata determined by (1) two ranges of elevated serum creatinine concentrations (1.5 to 2.9 mg/dl and 3.0 to 4.9 mg/dl), (2) previous treatment with ω-3 fatty acids, and (3) previous treatment with corticosteroids. One patient who had a central laboratory baseline serum creatinine of 1.4 mg/dl was randomized to low-dose Omacor (Pronova Biocare, Oslo, Norway). This patient's initial serum creatinine was 1.5 mg/dl measured 1 mo before in her local laboratory. We elected to include the patient in all analyses (in the 1.5 to 2.9 serum creatinine group). Omacor soft gelatin capsules, the study medicine used in this study, contain 1 g of a long-chain polyunsaturated ω-3 fatty acid ethyl ester concentrate including 4 mg of α-tocopherol, which is added as an antioxidant. The concentrate is produced from high-quality fish oil and contains 47% EPA, 37% DHA, and 5 to 10% of other ω-3 fatty acids. In the manufacturing process, impurities, cholesterol, vitamins A and D, and potentially toxic compounds such as heavy metals, dioxins, and pesticides are removed below detection limits (7). The patients were randomly assigned to receive either Omacor 8 g/d given as eight 1-g soft gelatin capsules containing 3.76 g of EPA and 2.94 g of DHA for a total of 6.7 g of ω-3 fatty acids—the high-dose group—or Omacor 4 g/d given as four 1-g soft gelatin capsules containing 1.88 g of EPA and 1.47 g of DHA for a total of 3.35 g of ω-3 fatty acids—the low-dose group. Dosing instruction was eight capsules or four capsules once a day ingested with a meal. Patients with hypertension were treated with the angiotensin-converting enzyme inhibitor enalapril (Vasotec; Merck Sharp & Dohme Laboratories, West Point, PA). When the target BP of 140/85 mmHg was not achieved, other antihypertensive drugs were added at the physician's discretion. Mild dietary sodium restriction limited to 90 mmol/d was advised.

Patient Monitoring

At study entry, complete medical histories were taken and physical examinations were performed for all patients. Initial clinical and laboratory results were sent to the coordinating center. Follow-up patient examinations and measurements of serum creatinine; total, high-density lipoprotein, and low-density lipoprotein cholesterol; triglycerides; and 24-h urine total protein were scheduled after 1.5, 6, 12, 18, and 24 mo of treatment. Patients who remained at risk for a renal failure event after 24 mo of treatment were continued on the same dose regimens to which they were originally assigned, and they had scheduled visits and the above named laboratory tests performed at 6-mo intervals until study closure (January 2000). Also, first morning urinalysis, hemoglobin, hematocrit, peripheral blood leukocytes, platelets, and serum potassium were collected and analyzed at the local center at each scheduled visit. All clinical and laboratory results were recorded on case report forms, forwarded to the coordinating center, and entered for data processing. To reduce variability, we analyzed each sample for serum creatinine and lipids and 24-h urine total protein at the Mayo Medical Laboratories (Rochester, MN) using standard methods. Patient compliance was ascertained by measuring the plasma phospholipid fatty acids—EPA, DHA, and C20:4ω6 (AA)—at 6 wk and 6 mo versus baseline within and between the two Omacor-supplemented groups. The EPA/AA ratio, a particularly sensitive indicator of compliance, was also calculated. The fatty acid composition including ω-6 and ω-3 fatty acids of plasma phospholipid was measured by capillary gas-liquid chromatography (8,9).

Study End Points

The primary end point of the study was the estimated annual within-patient slope in serum creatinine, a determinant of change in renal function, over the entire time in study for each patient (2-yr minimum). Secondary end points were time to ESRD (defined as chronic, repetitive dialysis or receiving a renal transplant) and time to treatment failure, defined as the first occurrence of any of the following: (1) the development of ESRD, (2) a serious side effect leading to the inability to continue study medicines and death from any cause, and (3) refusal to continue participation (noncompliance). For noncompliant patients, we attempted to obtain follow-up information on subsequent renal failure and vital status following the intent-to-treat principle. Other variables monitored included changes in BP and 24-h urine total protein and changes in serum lipid profiles, peripheral blood counts, and serum potassium for safety.

Statistical Analyses

Univariate baseline comparisons between the two dose groups were done using the rank-sum (continuous data) or χ² (nominal data). Continuous factors were summarized using means and SD. Distributions that were particularly skewed or that had severe outliers (urine protein, triglycerides, serum creatinine slopes) were summarized using medians and the estimated interquartile range (25th, 75th percentiles). For patients who developed ESRD, only laboratory values taken before starting dialysis or receiving a renal transplant are included in the analyses.

The primary end point for the study was rate of change in serum creatinine. Changes in serum creatinine over time were estimated in two ways. First, simple linear regression analysis (Y = serum creatinine, X = years since start of Omacor) was used to estimate annualized rates of change (slopes) in serum creatinine and reciprocal serum creatinine for each patient. Although reciprocal serum creatinine values were more linear over time than direct values, both measurements are presented as the interpretation of the direct changes is more familiar to readers. The rank-sum test was then used to compare dose groups with respect to serum creatinine slopes. Analyses followed intent-to-treat principles, with every attempt made to include all randomized patients in the analysis. However, one patient, randomized to high-dose Omacor, left the study before beginning treatment and was excluded from this analysis as only a single baseline serum creatinine reading was available.

Second, generalized estimating equations were used to assess the average change over time in serum creatinine readings taken from low-dose versus high-dose Omacor-treated patients (10). In contrast to the first approach, this method eliminates the need first to estimate within patient slopes, while taking into account not only the repeated nature of the data but also the variable number of readings per patient. Furthermore, all patients contributed to the analysis. The dependent variable was serum creatinine (and also reciprocal serum creatinine). Predictor variables were Omacor dose group, time since start of therapy, and dose by time interaction. As readings within patients tend to be correlated, patient was used as a clustering factor. Both an autoregressive reading (correlations between readings within a patient assumed to decrease with time) and an exchangeable reading (correlations between readings within a patient assumed constant across time) were used.

Secondary end points included time from start of therapy to development of ESRD and time to either development of ESRD or discontinuation of treatment as a result of an adverse event or noncompliance (refusal to continue). The cumulative percentage of patients who were free of these events was estimated using the Kaplan-Meier method. Dose groups were compared using the log-rank test. All randomized patients were included in these comparisons. The one patient who dropped out before beginning therapy was censored with 0 follow-up for the ESRD end point and counted as an event with 0 follow-up for the ESRD or discontinuation of treatment end point.

Within-dose group 1-yr and 2-yr changes in BP, urine protein, serum lipids, peripheral blood counts, and serum potassium were tested using the sign-rank test. All tests were two-sided with α-level 0.05.

Outcome: Annual within-Patient Slopes in Serum Creatinine

Based on within-patient serum creatinine profiles (Figure 1), we estimated within-patient annualized rates of change (slopes) for serum creatinine. These slopes are summarized in Table 2 and Figure 2. Although both groups demonstrated slopes that were significantly different from 0 (4 g, P = 0.008; 8 g, P = 0.001 [sign-rank test]), no significant differences were noted between the two Omacor dose groups (P = 0.51 [rank-sum test]). Additional analyses based on slopes derived from reciprocal serum creatinine profiles also showed no significant differences between Omacor dose groups (P = 0.58). Further analyses based on generalized estimating equations (see the Materials and Methods section) also found no significant evidence of an Omacor dose effect on follow-up serum creatinine levels (all P values >0.20). For example, the mean slope in reciprocal serum creatinine was estimated to be -0.03 dl/mg per yr for the low-dose group compared with -0.02 dl/mg per yr for the high-dose group (P = 0.58). Annualized median change in serum creatinine was not different in previously ω-3 fatty acid—treated (0.107 mg/dl per yr) versus non-previously ω-3 fatty acid—treated patients (0.064 mg/dl per yr; P = 0.36 [rank-sum test]) or in reciprocal serum creatinine slopes comparing previously ω-3 fatty acid—treated (-0.026 dl/mg per yr) with non-previously ω-fatty acid—treated (-0.020 dl/mg per yr) patients (P = 0.71 [rank-sum test]).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Serum creatinine profiles in patients who had IgA nephropathy and who were treated with low-dose (A) and high-dose (B) Omacor. The serum creatinines are plotted and connected for each patient and include all readings before renal failure (if present).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Annualized rate of change in renal function in patients who had IgA nephropathy and who were treated with low-dose and high-dose Omacor. The annual rates of change (slopes) for serum creatinine were computed for each patient by linear regression analysis using all results before renal failure (if present). The annualized median change in serum creatinine was 0.08 mg/dl per yr in the low-dose group and 0.10 mg/dl per yr in the high-dose group (P = 0.51). +, median.

It is apparent that the majority of patients who had baseline serum creatinines of 1.5 to 2.9 mg/dl at study entry (n = 63) had stable renal function in both the low-dose and high-dose Omacor groups (Figure 1). Patients with initial serum creatinine levels of 1.5 to 2.9 mg/dl had significantly lower rates of deterioration in renal function (median creatinine slope = 0.08 versus 1.3 mg/dl per yr [P = 0.019]; median reciprocal creatinine slope = -0.02 versus -0.07 [P = 0.062]) and less ESRD (13 versus 85% at 3 yr; P < 0.001 [log-rank test]) when compared with those with initial serum creatinines of 3.0 to 4.9 mg/dl. Large serum creatinine slopes (>0.5 mg/dl per yr) were observed in 14 of 62 patients (23%) with baseline serum creatinines of 1.5 to 2.9 mg/dl compared with 7 of 10 (70%) patients with baseline serum creatinines of 3.0 to 4.9 mg/dl and in 17 of 18 patients who subsequently developed ESRD. Although we recognize that serum creatinine slopes >0.5 mg/dl per yr are nonlinear for expressing changes in GFR, such large slopes have clinical interpretation and correlate with the development of ESRD.

Renal Failure Events

Survival free of ESRD was similar in the two treatment groups (Table 2, Figure 3), and survival without sustaining any event—ESRD, an adverse event leading to discontinuation of treatment, or refusal to continue therapy—was also similar in the low-dose and high-dose Omacor treatment groups (Table 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Cumulative percentage of patients who had IgA nephropathy and who were treated with low-dose and high-dose Omacor and did not have renal failure during the treatment period. In the low-dose group, 73% of the patients (SEM = 8.4%) were free of renal failure at 3 yr compared with 76% (7.6) in the high-dose group (P = 0.56). Numbers of patients who were at risk at 3 yr were 12 in the low-dose group and 15 in the high-dose group.

Changes in BP, Proteinuria, Serum Lipids, Peripheral Blood Counts, and Potassium

There were no significant within-dose group time trends in BP, proteinuria, serum lipids, peripheral blood counts, and potassium, with the exception that BP and serum triglycerides were lower in the low-dose Omacor group at 12 mo compared with baseline, and high-density lipoprotein and low-density lipoprotein cholesterol were lower in the high-dose Omacor group at 12 mo compared with baseline (Table 3). Although there was a decline in urine protein excretion over time in the low-dose group compared with no change in the high-dose group, the median annual slopes in proteinuria were not significantly different between treatment groups (P = 0.17 [rank-sum test]).

Compliance with Treatment

As a measure of compliance, in 56 patients who were not taking ω-3 fatty acids at study entry, plasma phospholipid fatty acid changes from pretreatment were significant for all studied fatty acids (Table 4). There were significant increases in EPA and DHA and significant reductions in AA levels at 6 wk and 6 mo after supplementation (P < 0.01 for all values [paired t test]). With the exception of AA, fatty acid changes from baseline were, in general, higher for the high-dose compared with the low-dose group (Table 4). EPA/AA ratios were also significantly increased in both treatment groups and were higher at both 6 wk (P = 0.007, two-tailed P value from rank-sum test) and 6 mo (P = 0.002) in the high-dose group (Table 4).

Adverse Events

A total of nine patients (five at 4 g/d, four at 8 g/d) had adverse events, including the two aforementioned patients who withdrew early as a result of adverse events. Of the remaining seven patients, one patient (4 g/d) had episodes of cryptogenic gastrointestinal bleeding before, during, and after study participation. Omacor was discontinued in this patient 2 yr after study enrollment during an episode of gastrointestinal bleeding 1 mo before the patient was started on maintenance hemodialysis for ESRD. Another patient (4 g/d) discontinued therapy after development of rectal bleeding from hemorrhoids 48 mo after study entry. This patient had advanced renal failure at the time. Other single adverse events noted during the course of study participation included one episode of diverticulitis (4 g/d), an episode of pneumonitis and possible pancreatitis (8 g/d), development of asymptomatic atrial fibrillation (8 g/d), hyperkalemia (4 g/d) that resolved after discontinuing angiotensin-converting enzyme inhibitor therapy, and development of hemiparesis as a result of ischemic cerebrovascular disease (8 g/d). None of these events was thought to be related directly to study medication.