ARTIIRIl'IS & RI1EIJMhl'ISM Vol. 39, Yo. 11. November 1996. pp 1808-1817 1808 0 19Y6, American Collegc of Rhcumatology

____ -. . .. ~ -

GAMMA-LINOLENIC ACID TREATMENT OF RHEUMATOID

A Randomized, Placebo-Controlled Trial

ROBERT B. ZURIER. RONALD G. ROSSETTI, ERIC W. JACOBSON, DEBORAH M. DEMAKCO. NANCY Y. LIU. JOSEPH E. TEMMING. BERNADETTE M. WHITE, and MICIIAEL LAPOSATA

Objective. To assess the clinical efticacy and ad- GLA is available as a component of several plant seed verse effects of y-linolenic acid (GLA),a plant seed oils and is usually taken in far lower doses than were oil-derived unsaturated fatty acid that suppresses in- used in this trial. It is not approved in the United States flammation and joint tissue injury in animal models, in for the treatment of any condition, and should not be the treatment of active (RA). viewed as therapy for any disease. Further controlled Methods. Fifty-six patients with active RA were studies of its use in RA are warranted. randomized to treatment groups in a 6-month, double- blind trial of GLA versus placebo. This was followed by Gamma-linolenic acid (GLA; 18:3 06) is an a 6-month, single-blind trial during which all patients essential fatty acid (Figure 1) found in certain plant seed received GLA. Patients were treated with 2.8 gm/day of oils. It is metabolizcd to dihomo-y-linolcnic acid GLA as the free fatty acid or with sunflower seed oil (DGLA; 20:3 06). the immediate precursor of prosta- (placebo) administered in identical capsules. glandin E, (PGE,), an cicosanoid with known antiin- Resulrs. Treatment with GLA for 6 months re- flammatory and immunoregulatory properties (1-4). sulted in statistically significant and clinically relevant GLA and DGLA also modulate immune rcsponses in an reductions in the signs and symptoms of disease activity eicosanoid-independent manner by acting directly on T in patients with RA. Overall meaningful responses (at lymphocytes (5-7). GLA suppresscs acute and chronic least 25% improvement in 4 measures) were also better , including arthritis, in several animal mod- in the GLA treatment group (14 of 22 patients versus 4 els (8). In a randomized, placebo-controlled trial of 6 of 19 in the placebo group; P = 0.015). During the months' duration in a small numbcr of rheumatoid second 6 months, both groups exhibited improvement in arthritis (pA)patients, 1.4 gdday of GLA in disease activity. Thus, patients taking GLA during the seed oil reduced synovitis in a statistically and clinically entire study showed progressive improvement during significant manncr (9), which suggests that GLA is a the second 6 months. In this group, 16 of 21 patients potentially effectivc treatment for RA. Results of exper- showed meaningful improvement at 12 months com- iments in animals suggest that the antiinflammatory pared with study entry. effects of GLA arc dose dependent. Therefore, we Conclusion. GLA at doses used in this study is a carried out a 1-year trial of GLA treatment at a dosage well-tolerated and effective treatment for active RA. of 2.8 gm/day in patients with RA and activc synovitis to

~ .- ... - assess the clinical efficacy and tolcrability of GLA Supported in part by FDA grant FD-R-000756 and by NIH administered as the frce fatty acid. grant RO- 1-AR-38501. Robcrt B. Zurier. MD. Konald G. Kossctti. MPH. Eric W. Jacobson, MD. Dcborah M. DcMarco. MD, Nancy Y. Liu. MD, PATIENTS AND METHODS Joseph E. Temming, MD, Bernadette M. White. RN: University of Massachusetts Medical Center, Worcester; Michael Laposata. MD. Study design. The first portion of the study was a PhD: Massachusetts GCnfXdl Hospital, Boston. 6-month, randomized, double-blind comparison of GLA and Addrcss rcprint rcqucsts to Robcrt B. Zurier, MD. University of Massachusetts Medical Center, 55 Lake Avenue North, Worcester. placebo (sunflower seed oil) in RA patients with aclive syno- MA 01655-0335. vitis. During months 6-12, all patients were given <;LA. Submitted for publication February 21. 1996; accepted in Although the patients knew they were guaranteed at least 6 rcviscd form May 20, 1YY6. months of GLA during the 13-month trial, they did not know GAMMA-LINOLENIC ACID IN RA 1809

LlNOLElC ACID (182)

DELTA-6 DESATURASE zI GAMMA LlNOLENlC ACID (183)

ELONCASE

COOH

DELTA-5 DESATURASE DIHOMOGAMMA b LlNOLENlC ACID (203) ARACHIDONIC ACID (204)

PGEI TXA 1 (150H)DCLA LTCJ

Figure 1. Metabolic pathway of 06 fatty acids. The pathway is one of progressive desaturation, alternating with elongation (addition of 2 carbons). Numbers in parentheses represent the number of carbon atomsthe number of double bonds. PGE, = E,; TXA, = thromboxane A,; (15OH)DGLA = 15-hydroxydihomo-y- linolenic acid: LTC, = C3. when that would occur. Also, during the second 6 months, the treated with any investigational drug within 1 month of entry, examining rheumatologist did not know which group the with commercially available fish or plant seed oil products patient had been in during the first 6 months. Patients were within 3 months of entry, or with an intraarticular injection of evaluated at 3-month intervals. GLA was stopped at 12 corticosteroids within 1 month of entry. months, and patients were evaluated again 3 months later (at Patients were randomized to receive either GLA 15 months). (GLA-70) or placebo capsules. GLA-70 (provided by Scotia The protocol was reviewed and approved by the Com- Pharmaceuticals Ltd., Guildford, Surrey, England) is produced mittee for the Protection of Human Subjects in Research at from borage seed oil, which is hydrolyzed, separated from the University of Massachusetts Medical Center and by the reactants and byproducts, and purified by washing with water. Food and Drug Administration. Written informed consent was After fractional crystallization, short-chain and saturated fatty obtained from each patient. acids are removed by filtration. The filtrate is distilled twice Eligibility. Patients were eligible to participate in the under vacuum, and the GLA is isolated in the distillate. study if they had KA according to the revised criteria of the GIA-70 contains -70% GLA as a percentage of the total fatty American College of Rheumatology (ACR; formerly, the acids; the other major fatty acids are linoleic acid (20%) and American Rheumatism Association) (lo), were between ages oleic acid (5%). The remaining 5% is made up of smaller 18 and 90, and were of functional class I, 11, or 111 (11). The quantities of the fatty acids present in borage seed oil. Each dosage of aspirin or other nonsteroidal antiinflammatory drug capsule contained 0.5 ml of oil and 13.6 IU of vitamin E, and (NSAID) and of corticosteroid (not to exceed 10 mg of provided 4.5 kcal daily. Vitamin E alone does not benefit RA prednisone or equivalent per day) had to be unchanged for 1 patients (12). The total daily dose (8 capsules) provided 2.8 gm month before study entry, and dosages of second-line drugs of GLA had to be stable and unchanged for 3 months before entry. All The placebo capsules, which were identical in appear- patients had active disease. as defined by the presence of at ance and color and were taken according to the same schedule least 3 of the following 4 criteria: 26 painful or tender joints; (4 capsules twice a day, after a meal), contained sunflower seed 23 swollen joints; morning stiffness of at least 45 minutes' oil (Scotia Pharmaceuticals Ltd.), which is 52% linoleic acid, duration; and a Westergren erythrocyte sedimentation rate 35% oleic acid, 6% palmitic acid, 6% stearic acid, and 1% (ESR) of at least 28 mm/hour. Medications required for other fatty acids. Neither the GLA-70 nor the placebo capsules coexisting medical conditions were permitted; these conditions had any taste. An excess of capsules was given, and the were mild and well controlled. Patients were instructed to capsules were counted at each visit to monitor compliance. maintain their typical diet. GLA was administered under an Investigational New Drug Patients were ineligible for the study if they had beeti Approval (#28867) issued to one of the investigators (RHZ). 1810 ZURIER ET AL

Clinical assessments. The patients, primary care phy- above): joint tenderness count or score, joint swelling count or sicians, referring rheumatologists, and study investigators were score, global assessment by physician, global assessment by unaware of the study group assignments. Assessments were patient, pain assessment by patient, duration of morning performed by 4 board-certified rheumatologists (EWJ, DMD, stiffness, and IIAQ score. NYL, and JET) with the aid of a nurse (BMW) in the Clinical Preparation of platelets. Blood was obtained by veni- Research Center at the University of Massachusetts .Medical puncture and collected in tubes containing EDTA. Platelet- Center. Two of the rheumatologists evaluated 44 of the 56 rich plasma was prepared by centrifugation of whole blood at patients who entered the study. Each patient’s condition was l0Og for 15 minutes at room temperature. Platelet counts were assessed by the same rheumatologist through the entire course performed in a hemocytometer. Aliquots (1 ml) of platelet-rich of the study to avoid inter-examiner variability. The primary plasma and of plasma were stored under nitrogen at -80°C outcome measure was the tender joint count as defined by the until fatty acid analyses were done. ACR (13). Fatty acid analysis. Lipids from each sample were The referring rheumatologist continued to manage the extracted and methylatcd as described by Lepage and Roy patient’s RA. Medication dosages could be reduced, but any (19). The mixture of fatty acid methyl esters was analyzed by increase in the medication for RA or any change in or addition electron-impact gas chromatography/mass spectroscopy using of a drug was a protocol violation. Hewlett Packard 5890 Series I1 gas chromatograph (Hewlett Disease activity was assessed clinically at baseline and Packard, McMinnville, OK) equipped with a Supelcowax 10 at 3-month intervals and included 1) physician’s global assess- capillary column (Supelco, Bellefonte, PA) coupled to a ment of disease activity (rated as 0 = none, 1 = mild, 2 = Hewlett Packard 5971 mass spectrometer. Peaks were idcnti- moderate, 3 = severe, and 4 = very severe); 2) patient’s global fied by comparing retention times and mass spectra with assessment of disease activity and a pain assessment, both by known fatty acid methyl ester reference standards (NuCheck- 100-mm visual analog scale (VAS) and by 0-4 scale as above; Prep, Elysian, MN). Quantitation of fatty acids in the cells was 3) number of joints with tenderness and/or pain on pressure or determined by comparing the area under each peak with the passive motion (total of 68 diarthrodial jolnts); 4) number of area of a known internal standard (methyl 17:O) added as joints with swelling (total of 66 diarthrodial joints); 5) joint tri-17:O at the beginning of the lipid extraction. paidtenderness score (rated as 0 = none, 1 = mild, 2 = Statistical analysis. Sample size considerufiom. With moderate, and 3 = severe); 6) joint swelling score (same scale the tender joint count used as the primary obtcome, a sample as for pain/tenderness); 7) duration of morning stiffness (min- of 35 patients was needed to conduct a test with a significance utes); 8) grip strength (mm Hg) in each hand, using a level of 0.05, a probability of detecting a significant difference sphygmomanometer inflated initially to 20 mrn Hg; and 9) of 90%, and a desire to show a 35% reduction in tender joint degree of disability, as measured with the Health Assessment count. We entered 56 patients into the study. Based on data Questionnaire (HAQ) (14,15). from several therapeutic trials (16-18), -20% of patients in A laboratory assessment was done at baseline and at the placebo group would be expected to experience meaningful 6-month intervals, and included a complete blood cell (CBC) improvement. count, platelet count, rheumatoid factor (RF) level, and ESR. Clinical responses. Statistical analysis was performed Patients were asked about adverse experiences at each using chi-square testing (with Yates’ correction for continuity, visit. Criteria for patient withdrawal from the study included 1) when appropriate), and the Wilcoxon sign rank test for within- adverse drug reaction that required the patient to miss doses group comparisons of final visit data with baseline data. for more than 7 days; 2) violation of study protocol; 3) patient’s Nonparametric tests were used because most variables were desire to withdraw from the study; and 4) onset of a serious not normally distributed. Data were analyLed for both the medical condition. entire group of patients entered into the study and for patients Categories of response. The following definitions were who completed the study. used to characterize changes for several clinical measures. Important improvement in any meisure was defined as at least a 25% improvement in continuous variables or an improve- RESULTS ment of 2 levels or an improvement from a baseline value of I Among the patients who were potentially eligible to a value of 0 in the scaled variables. Important worsening in any measure was defined as at least 25% worsening in contin- for the study, 56 patients were randomized. Ariother 10 uous variables or a worsening of 2 levels or a worsening from patients were screened but not randomized for the a baseline value of 3 to 4 in the Scaled variables. A change following reasons: did not meet ACR criteria for entry (increase or decrease) of 20% in th’e HAQ score was consid- (5 patients); not taking stable dosage of medicine (2 ered important. patients); recent intraarticular injection of cortico- Patients who completed the srudy were classified into 1 steroids (1 patient); enrolled in another clinical trial (1 of 4 overall response categories, as described by several groups of investigators (16-18): remission, meaningful improvement, patient); unwilling to risk randomization into placebo no meaningful change, or deterioration. RA was considered to group (1 patient). be in remission if global and pain assessments by the physician Of the 56 patients who enrolled in the study, 28 and the patient were 0, no swollen or tender joints were received GLA and 28 received the sunflower seed oil present, and morning stiffness did not exceed 15 minutes. Overall meaningful improvement (or deterioration) occurred placebo during the first 6 months. The demographic and if the patient had important improvement (or important clinical characteristics, treatment at study entry, and worsening) in at least 4 of the following measures (as described laboratory test values at study entry were similar in the GAMMA-LINOLENIC ACID IN RA 1811

Table 1. Characteristics of the rheumatoid arthritis (RA) patients at study entry* GLA group Placebo group Characteristic (n = 28) (n = 28)

~~ ~~ ~ ~~ Age, years 57.4 (9.98) 54.6 (13.58) Sex No. male 2 5 No. female 26 23 Duration of RA, years 12.0 (6.1) 10.3 (8.7) Current medications, n NSAIDs 20 23 Corticosteroids (mean daily dose) 15 (5.2 mg) 13 (5.1 mg) Second-line drugs 17 15 Previous treatment with second-line drug, n 16 14 Swollen joint count, 0-66 18.5 (9.7) 23.228.7 (9.1)(12.4) Swollen joint score, 0-3 23.5 (12.4) Tender joint count, 0-68 25.6 (11.2) 23.1 (11.7) Tender joint score, 0-3 31.5 (13.9) 29.4 (15.2) Morning stiffness. minutes 109.4 (99.7) 143.6 (128.1) HAQ score, 1-4 1.8 (0.42) 1.7 (0.45) Global assessment, 0-4 Physician 2.29 (0.94) 2.57 (0.63) Patient 2.43 (0.69) 2.35 (0.75) Pain assessment, 0-4 2.46 (0.86) 2.61 (0.83) Visual analog scale. 1-10 Global disease 5.35 (2.1) 5.69 (2.2) Pain 5.14 (2.5) 6.06 (2.1) Grip strength, mm Hg Right hand 142.1 (74.9) 142.7 (65.0) Left hand 154.0 (67.6) 139.3 (69.6) Erythrocyte sedimentation rate, mm/hour 45.2 (29.9) 58.1 (26.6) Rheumatoid factor, IU 301.4 (587) 344.7 (572.7) Platelet count ( x 1og/liter) 343.3 (107) 303.6 (79.1) * Values are the frequency or the mean (SD). P values for between-group comparisons were all >0.3, by Mann-Whitney U test or Yates’ corrected chi-square. GLA = y-linolenic acid; NSAIDs = nonsteroidal antiinflammatory drugs; HAQ = Health Assessment Questionnaire. two groups (Table 1).At entry into the study, 10 patients come measure) and score, the duration of morning in the GLA group were taking methotrexate, 3 gold, and stiffness, patient’s global assessment (by scale and by 4 hydroxychloroquine. In the placebo group, 5 patients VAS), patient’s assessment of pain (by scale and by were taking methotrexate, 5 gold, and 5 hydroxy- VAS), and HAQ score at 6 months compared with chloroquine. baseline (Table 2). The degree of improvement was During the first 6 months of the study, 6 patients moderate. When compared with placebo, GLA treat- in the GLA group withdrew (lack of efficacy in 2, ment resulted in significantly more improvement in the diarrhea in 1, surgery in 1, an unrelated illness in 1, and swollen joint count, the tender joint count and score, protocol violation in l), and 9 patients in the placebo pain (by VAS), and the HAQ score. Data from all group withdrew (lack of efficacy in 4, constipation in 1, patients who entered the study, regardless of whether an unrelated illness in 2, rash in 1, and protocol viola- they completed the protocol, were analyzed (intention to tion in 1). treat analysis) and confirmed that the results seen in The clinical responses are presented as absolute patients treated with GLA were not due to bias intro- change and percentage of change. The group of patients duced by patients leaving the study prematurely. who completed the 6-month double-blind portion of the Although improvements from baseline were sig- study (41 patients) included 22 patients in the GLA nificant for the swollen joint score, morning stiffness, group and 19 patients in the placebo group (Table 2). and patient’s global assessment and pain assessment, The demographic characteristics did not differ between these changes did not reach statistical significance when the two groups. Whereas the placebo group did not show compared with the placebo group at 6 months. No significant improvement in any measure, the GLA group significant changes from entry to 6 months or between showed significant improvement in the swollen joint groups were seen for grip strength, the ESR, the platelet count and score, the tender joint count (primary out- count, or the RF level. Thus, for example, the mean f 1812 ZURIER ET AL

SD ESR at 6 months was 48 ? 27 mm/hour for the Table 3. Changes in clinical and laboratory measures of rheumatoid GLA-treated group and 50 t 24 mmhour for the arthritis during months 6-12 of the study* placebo-treated group. Similarly, the mean ? SD RF GLA group PlacebolGLA group levels at 6 months were 448 t 581 IU for the GLA group Variable (n = 21) (n = 14) and 204 ? 306 IU for the placebo group. Adverse Swollen joint count, 0-66 -3.557 -8.00t reactions among the patients who completed the first 6 % change -23.3 (41.8) -38.2 (41.2) months of the study included belching (3 in GLA group; Swollen joint score, &3/joint -4.3t - 6.9t % change -23.8 (31.2) -28.9 (42.5) 2 in placebo group) and diarrhea (4 in GLA group; 1 in Tender joint count, 0-68 -2.751- - 1.64t placebo group). % change -17.6 (68.3) -9.2 (50.9) During the second 6 months of the study, all Tender joint score, O-3/joint -3.8t -4.61- % change -20.2 (87.9) -19.3 (49.9) patients were treated with GLA; 3 patients from the Morning stiffness, minutes -3.3 -69.9t placebo group withdrew from the study after the % change -6.1 (90.1) -47.0 (1.5) Global assessment, 0-4 6-month evaluation because of exacerbation of arthritis. Physician -0.2 -0.36 A total of 38 patients entered the second 6 months of the % change -10.2 (35.6) - 14.4 (34.0) trial, 22 from the GLA treatment group and 16 from the Patient -0.20 -0.15 placebo group. Of these 38, withdrew from the GLA % change -10.2 (35.6) -6.4 (46.4) 1 Pain assessment, 0-4 0 -0.15 group at 9 months because of protocol violation and 2 % change 0 (40.4) -6.2 (41.2) withdrew from the placebo group, 1 at 8 months for Visual analog scale, 1-10 surgery unrelated to RA and 1 at 9 months for protocol Global -0.52 -0.34t % change -12.2 (54.2) -6.2 (47.4) violation. Pain 0.15 - 0.75 Patients who were taking GLA during the first 6 % change 4.0 (64.2) - 12.9 (49.4) HAQ score, 1-4 -0.012 -0.064 % change -0.8 (33.3) -3.5 (24.5)

Table 2. Changes from baseline and comparison of clinical and * Values are the mean absolute change or the mean (SD) percentage laboratory measures of rheumatoid arthritis during months 1-6 of the of change from the month-6 data. The placebo/GLA (7-linolenic acid) study* group took GLA only during months 6-12. HAQ = Health Assess- ment Questionnaire. GLA group Placebo group 1- P < 0.05 versus month-6 data, by Wilcoxon sign rank test. Variable (n = 22) (n = 19) Pt Swollen joint count, 0-66 -4.1$ -1.82 0.034 % change -20.9 (34.3) -8.3 (22.1) 0.026 months of study continued to improve during the second Swollen joint score, 0-3/joint -6.1$ -4.6 0.17 % change -23.1 (31.2) -16.1 (28.5) 0.101 6 months, and patients who were switched from placebo Tender joint count, 0-68 -8.5$ - 2.4 0.006 to GLA at 6 months exhibited improvement during the % change -35.2 (32.5) - 11.9 (36.8) 0.015 second 6 months that was comparable to that in patients Tender joint score, O-3/joint -11.4$ -2.8 0.021 % change -37.5 (35.6) - 10.7 (45.4) 0.041 who had been given GLA during the first 6 months. In Morning stiffness, minutes - 67.5$ 10.6 0.109 this latter group, there was significant improvement in % change -55.4 (0.98) 7.9 (1.4) 0.092 the swollen joint count, swollen joint score, tender joint Global assessment, 0-4 Physician 0.045 -0.21 0.65 count, tender joint score, duration of morning stiffness, % change 2.1 (34.6) -7.8 (29.3) 0.44 and patient’s global assessment (by VAS) (Table 3). Patient -0.45$ -0.11 0.18 Patients who took GLA for the entire 12 months had % change -18.9 (35.6) -4.5 (34.9) 0.2 Pain assessment, 0-4 -0.59$ -0.37 0.44 significant improvement in the swollen joint count, % change -24.1 (42.9) - 13.7 (41.5) 0.43 swollen joint score, tender joint count, and tender joint Visual analog scale, 1-10 score during months 6-12. Progressive improvement was Global -0.91$ 0.31 0.48 % change - 17.5 (38.0) 5.9 (33.4) 0.31 also reflected in the clinical responses at 12 months Pain - 1.363 0.09 0.004 compared with the data at study entry (Table 4). % change -26.8 (58.5) 1.6 (40.6) 0.082 No patient experienced remission of RA during HAQ score, 1-4 -0.192$ 0.082 0.027 % change -11.0 (19.0) 4.75 (17.9) 0.041 the first 6 months of the study. In the GLA group, 14 of 22 patients exhibited meaningful improvement (at least * Values are the mean absolute change or the mean (SD) percentage of change from baseline. GLA = ylinolenic acid; HAQ = Health 25% improvement in at least 4 measures, as described in Assessment Questionnaire. Patients and Methods), whereas 4 of 19 patients in the 1- Determined by Mann-Whitney U test for between-group compari- placebo group experienced meaningful improvement sons. 3 P < 0.05 for intragroup change from baseline, by Wilcoxon sign rank (Table 5); the overall clinical responses for the two test. groups were significantly different (P = 0.015). The data GAMMA-LINOLENIC ACID IN RA 1813

Table 4. Changes in clinical and laboratory measures of rheumatoid Table 6. Findings in rheumatoid arthritis patients 3 months after arthritis for 21 patients given GLA for 12 months* y-linolenic acid (GLA) treatment stopped (month 15)* Variable Change GLA Placebo/GLA Measure (n = 16) Swollen joint count, 0-66 -8.05t (n = 9) % change -41.8 (39.0) Swollen joint count 56.1 70.6 Swollen joint score, 0-3/joint -11.2t Swollen joint score 51.4 79.7 % change -41.9 (33.9) Tender joint count 39.8 30.1 Tender joint count, 0-68 -12.0t Tender joint score 41.9 46.3 % change -49.7 (46.2) Tender joint score, 0-3/joint -16.3t * The placebo/GLA group took GLA only during months 6-12. Values % change -53.6 (45.1) are the percentage of increase in the measure. Morning stiffness, minutes -81.St % change -66.9 (0.98) Global assessment, 0-4 the second 6 months had meaningful improvement Physician -0.25 % change -11.2 (32.2) (Table 5). During the second 6 months of the study, 2 Patient -0.65t patients who had been taking GLA for 12 months and 1 % change -26.9 (33.7) who had been taking it for 6 months experienced Pain assessment, 0-4 -0.6St % change -26.5 (46.3) deterioration in their condition. In the group of patients Visual analog scale, 1-10 given GLA for 12 months, 16 of 21 showed meaningful Global -0.651 improvement. The dosage of NSAID and/or prednisone % change -31.6 (40.9) Pain - 1.35t was reduced in 7 patients during the course of the study. % change -29.8 (54.2) All 7 patients were in the group treated with GLA for 12 HAQ score, 1-4 -0.21 months. % change - 12.2 (26.9) At month 12 of the study, GLA was stopped in all * Values are the mean absolute change or the mean (SD) percentage patients. An evaluation was performed at 15 months of change from baseline. GLA = y-linolenic acid; HAQ = Health Assessment Questionnaire. after study entry. Most patients experienced exacerba- t P < 0.05 versus baseline (entry), by Wilcoxon sign rank test. tion of disease activity, as measured by the swollen joint count and score and the tender joint count and score (Table 6). In addition, an increase in the duration of indicate that the GLA group was 6.5 times (cross morning stiffness was noted in both groups. At 15 product ratio) more likely than the placebo group to months, the mean swollen joint counts (18.6 for the experience meaningful improvement. No patients in the GLA group; 20.8 for the placebo group) and swollen GLA group experienced deterioration in their condition joint scores (21.5 for GLA; 24.4 for placebo) were (see criteria described in Patients and Methods), similar to the values at study entry (see Table 1). whereas 6 of 19 patients in the placebo group showed However, the degree of tenderness did not return to deterioration. Results of the analysis indicated that the baseline values. placebo group was 3.4 times more likely than the GLA We also applied (in retrospect) to the data pre- group to exhibit deterioration in disease. sented here the ACR preliminary core set of disease The overall responses during the second 6 activity measures for FL4 clinical trials (13). These months of the study indicated that 9 of the 21 patients criteria require that patients show improvement by 20% initially randomized to receive GLA continued to im- in both the number of swollen and the number of tender prove, and 5 of the 14 patients given GLA only during joints and in 3 of the 5 remaining end points (pain,

Table 5. Overall clinical response to treatment of rheumatoid arthritis with y-linolenic acid (GLA)* First 6 months Second 6 months GLA for Overall GLA Placebo GLA Placebo/GLA 12 months response (n = 22) (n = 19) (n = 21) (n = 14) (n = 21) Remission 0 0 0 0 0 Meaningful improvement 141- 4 9 5 16 No meaningful change 8 9 10 8 3 Deterioration 0 6 7 1 2 * The placebo/GLA group took GLA only during months 6-12. Values are the number of patients. t P = 0.015 versus placebo, by chi-square with Yates’ correction. 1814 ZURIER ET AL

Table 7. Plasma concentrations of GLA and DGLA in a rheumatoid months of the study showed statistically significant im- arthritis patient treated with y-linolenic acid* provement in the number of tender and swollen joints, Fatty acid, Plasma level % of total associated tender and swollen joint scores, patient’s assessment time (PmV fatty acids assessment of pain, and duration of morning stiffness. GLA (18:3 w6) Not all of these measures were significantly better than Study entry 4.0 0.5 placebo. It is possible that the concentration of oleic acid 4 months 14.7 1.3 DGLA (20:3 w6) in the sunflower seed oil placebo (35%) was sufficient to Study entry 15.8 1.8 provide some symptomatic relief. It has been reported 4 months 37.3 3.1 that olive oil (68% oleic acid) does appear to benefit RA * Data are from 1 patient who took y-linolenic acid (GLA) for 4 patients (20-22). In the GLA-treated group, the tender months. DGLA = dihomo-y-linolenic acid. joint count, which was the primary outcome measure, was significantly better than at baseline and than in the patient’s global assessment, physician’s global assess- placebo group at 6 months. Improvement continued ment, degree of disability [by HAQ], and ESR). None of during the second 6 months, yielding a nearly 50% the patients in the placebo group met the ACR criteria improvement (12 joints) after 12 months of GLA treat- at 6 months, whereas 5 of 22 patients in the GLA group ment. Indeed, 11 of 21 patients had >50% improvement fulfilled the ACR criteria at 6 months, and 7 of the 21 in the tender joint count after 12 months of GLA patients treated with GLA for 12 months met the ACR treatment. criteria. After the first 6 months of the trial, and before Analyses of the fatty acid content of plasma and the second 6 months of the study began, 3 patients platelet lipids were not done on every patient. However, withdrew because of exacerbation of arthritis; all 3 representative examples of the findings of such analyses patients had been in the placebo group. Thus, at that are shown in Tables 7 and 8 to document substantial point in the study, 2 of the 28 patients randomized to enrichment of plasma and platelet lipids with GLA and receive GLA therapy and 7 of the 28 patients random- DGLA during GLA therapy. GLA and DGLA levels did ized to receive placebo had withdrawn from the study not change appreciably during administration of the because of exacerbation of disease activity. During the sunflower seed oil placebo. second 6 months of the study, all patients were given GLA; thus, patients remained blinded. Physicians who evaluated the patients knew that all patients received DISCUSSION GLA during months 6-12, but did not know which group Treatment with 2.8 gm of GLA per day (as the (GLA or placebo) the patients were in during the first 6 free fatty acid) for 6 months resulted in clinically rele- months. A cross-over design was not used because of the vant and statistically significant reduction in the signs possibility of residual effects of GLA therapy, as has and symptoms of disease activity in patients with RA. In been shown for fish oil treatment (23). The results contrast, patients given a placebo (sunflower seed oil) indicated progressive improvement in patients given did not show significant improvement in disease activity. GLA for 12 months. Patients in the GLA group who completed the first 6 Improvements during months 6-12 were, for the

Table 8. Platelet levels of GLA and DGLA over 15 months of study in 3 rheumatoid arthritis patients” Month 15 Patient, treatment Study entry Month 3 Month 6 Month 9 Month 12 (3 months without GLA)

Patient 108, placebo/GLA group GLA 11 (3.1) - 22 (6.1) 75 (20.9) 19 (24.2) 24 (6.7) DGLA 81 (24) - 81 (24) 135 (41.3) 177 (54) 103 (31) Patient 110, placebo/GLA group GLA 22 (6.1) 16 (4.4) - 101 (28.1) 83 (25.6) 15 (4.2) DGLA 86 (26) 72 (22) - 227 (69) 165 (50.5) 89 (27.2) Patient 112, GLA group GLA 13 (3.6) 69 (19.2) 92 (25.6) 81 (22.5) 80 (22.3) DGLA 90 (27.5) 151 (46.2) 192 (58.8) 218 (66.7) 191 (58.5) * Gamma-linolenic acid (GLA) was given at a dosage of 2.8 gm/day; identical preparations of sunflower seed oil were given as placebo. Patients 108 and 110 took placebo for 6 months and then GLA for 6 months; patient 112 took GLA for 12 months. Values are nmoles (pg)/1.5 X 10’ platelets. DGLA = dihomo-y-linolenic acid. GAMMA-LINOLENIC ACID IN RA 1815

most part, similar to or less than the responses during that a reduction of 1 or 2 involved joints does not result months 0-6, which strongly suggests that there was no in a 33% reported improvement. Authors of composite significant examiner bias when the evaluating physician indices for estimating clinical improvement in individual was not completely blinded. Data derived from carefully RA patients during trials of slow-acting disease- done clinical observational studies also indicate that RA modifying antirheumatic drugs (DMARDs) suggest that patients can be evaluated in an open manner without patients who exhibit >20% improvement in at least 4 significant bias (15). selected measures may be considered to have received Most patients who finished the entire 15 months meaningful benefit from the drug (16-18). We consid- of the study experienced exacerbation of joint pain and ered a 25% change (20% in the HAQ score) to be swelling between months 12 and 15, during which time important in our assessment of overall responses. Ana- they did not receive GLA. Several patients chose not to lysis of overall responses indicated that the GLA group enter this phase of the study, and some patients took was 6.5 times more likely than the placebo group to additional medication at the first hint of joint inflamma- experience meaningful improvement at 6 months, and tion during month 13; thus, only 25 patients completed that the placebo group was 3.4 times more likely than the this phase of the study. The severity of the relapse may GLA group to experience deterioration. Such worsening have been modulated had a placebo been used, and the of disease activity is not unexpected during a 6-month study design should have included administration of interval in patients with active RA who are managed placebo capsules to all patients between months 12 and 15. without changes in drug therapy. Physicians not involved in the study, but who We also determined overall responses in retro- continued to manage the patients outside the trial, were spect, using the ACR preliminary core set of disease reluctant to reduce medications because they did not activity measures for RA clinical trials (13). The conser- want to “interfere with the trial.” Despite that sentiment, vative physician evaluations, the modest improvement in NSAJD and/or prednisone dosages were reduced in 7 HAQ scores, the lack of an effect of GLA on the ESR, patients. All 7 patients were in the group treated with and the fact that the composite index we used included GLA for 12 months. morning stiffness, which was improved substantially by A “disconnect” appears between the physician’s GLA treatment, may account for the fact that only 7 of global assessment of patients and the patients’ clinical 21 patients treated with GLA for 12 months met the responses. That is likely due to the fact that the exam- ACR criteria for improvement (13). However, none of iners viewed most patients’ RA as neither very severe (a the patients in the placebo group met the ACR criteria score of 4) nor mild (a score of 1); rather, they were seen at 6 months, whereas 5 of the 22 patients who took GLA as having moderate or severe (a 2 or 3) disease activity, for 6 months fulfilled the ACR criteria at 6 months. and changes of 1 point on the 0-4 scale were most Cyclosporin A, a potent immunosuppressive agent, also common. Also, despite improvement in the swollen and does not reduce the ESR in RA patients. whether used tender joint counts and scores, patients may have de- alone (25) or in combination with methotrexate (26). All scribed cxtraarticular problems (iritis, myofascial pain) these factors serve to reinforce the importance of the which tempered the physicians’ enthusiasm for global need for trials of at least 3 years’ duration, using large improvement. numbers of patients, and including radiographic evalua- Although the HAQ is considered the most effec- tion of joint damage. tive measure for documenting and monitoring the long- Fatty acid analysis of plasma or circulating cells term clinical status of RA patients (24), our experience was not used as a measure of compliance because the in this limited study suggests a certain arbitrariness to amount of GLA or DGLA present after administration the process over the short term. For example, we of a known amount of GLA varies from person to person observed patients score a particular response as a “3,” (Rossetti RG, Zurier RB: unpublished observations) erase the “3,” and then score the response as a “1,” or and would not necessarily reflect the number of capsules vice versa. Although evaluating and reporting disease ingested. However, it is important to know that plasma activity for a 3-month period, the patients’ scores appear and cell concentrations of GLA and/or DGLA during to be very much influenced by current (previous day or GLA therapy can reach levels equivalent to concentra- previous week) events (trouble at home, problems at tions of the fatty acids which, in vitro, suppress lympho- work, poor night’s sleep, etc.). It also seems clear that cyte activation and synovial cell proliferation. Plasma eligibility for entry should include 12 or more swollen levels of GLA and DGLA were increased 2-4-fold in the joints and 12 or more tender joints (rather than 3 and 6, GLA-treated patients, to concentrations which exceeded respectively, as outlined by the ACR criteria [lo]), so those needed to exert effects in vitro (4-7). Similarly, 1816 ZURIER ET AL concentrations of GLA and DGLA in platelets in- In addition to their role as eicosanoid precursors, creased as much as 8-fold during GLA treatment. Lower fatty acids are of major importance in maintaining cell doses of GLA used in previous studies (480 to 540 membrane structure and are key determinants of the mg/day) were either not effective (27) or reduced only behavior of membrane-bound enzymes and receptors pain (28), without effects on objective physical findings. (35). The fatty acid precursors can exert these functions Results of a study in which we treated RA patients with directly, and therefore may themselves be important 1.4 gm/day of GLA in borage seed oil suggested that the regulators of immune responses. For example, DGLA higher dosage is necessary to influence active synovitis suppresses interleukin-2 production by human peri- (9). The results presented here indicate that 2.8 gm/day pheral blood mononuclear cells in vitro, suppresses the of GLA may be more effective than 1.4 gm/day of GLA. proliferation of interleukin-2-dependent peripheral However, it is probably not accurate to compare admin- blood and synovial T lymphocytes, and reduces the istration of GLA as free fatty acid to administration of expression of activation markers on T lymphocytes di- GLA in triglycerides of seed oil. rectly, in a manner independent of its conversion to In controlled clinical trials in which GLA was (5-7). These observations indicate that shown to benefit patients with atopic eczema (29), some fatty acids can modulate immune responses by acting patients responded brilliantly, others modestly. Clinical directly on T cells, and suggest that alteration of cellular responses seemed to correlate with plasma levels of fatty acids may be a worthwhile approach to control DGLA. Indeed, in a study in which benefit from GLA inflammation. Other relevant direct effects of unsatur- was not observed (30), plasma levels of DGLA did not ated fatty acids include regulation of ion channels (36) rise. Thus, responses to GLA do appear to be dose and suppression of smooth muscle cell proliferation (37). dependent. We have considered, and others have shown in a Results of the evaluations of the primary out- controlled clinical trial (28), that GLA can be an NSAID come measure at 3 months indicated that the mean substitute. However, our observations that DGLA sup- reduction in the number of tender joints was 6 in the presses synovial cell proliferation (4), and the results of GLA-treated group and 4 in the placebo-treated group. this study which indicate that RA patients are better Thus, differentiation from a placebo response did not after 12 months of GLA treatment than after 6 months, occur until after 3 months of therapy. It is, of course, not suggest that GLA might function as a so-called clear whether patients received the full benefit of GLA DMARD. Thus, controlled studies of longer duration on therapy by 12 months or whether they would continue to larger numbers of RA patients, in which GLA is com- improve with further GLA treatment. It does appear pared with more established “second-line therapy,” and from the 15-month evaluation results (Table 6) that which use radiographs to monitor disease progression GLA administration must be maintained for synovitis to are warranted. be suppressed. The potential regulation of cell activation, im- We did not address the mechanisms responsible mune responses, and inflammation by certain fatty acids for the antiinflammatory effects of GLA in this clinical is exciting to consider at the clinical, cellular, and trial, but several may be possible. Gamma-linolenic acid molecular levels. A better understanding of how fatty is converted rapidly to DGLA, the immediate precursor acids modulate the function of cells involved in host of the monoenoic PGE,, which has potent antiinflam- defense might lead to development of new, benign matory activity (1). In addition, DGLA cannot be con- treatments for such diseases as RA, which are charac- verted to ihflammatory by 5-lipoxygenase terized by acute and chronic inflammation. (31); instead, it is converted to 15-hydroxydihomo-y- linolenic acid (see Figure l), which in turn suppresses 5-lipoxygenase and 12-lipoxygenase activity (32). Ad- ACKNOWLEDGMENTS ministration of GLA to human volunteers and to pa- Mrs. Carol Mader prepared the manuscript. Ms tients with RA does result in increased DGLA levels in Joanne Cluette-Brown performed the fatty acid analyses. The phospholipids of plasma, mononuclear cells, and plate- authors thank the following physicians for assisting in patient lets (33), increased PGE, production by stimulated enrollment: Micha Abeles, MD, Norman Bress, MD, David monocytes (34), and marked reduction of PGE, and Giansiracusa, MD, Paul Gibilisco, MD, Evelyn Love, MD Ann Parke, MB, John Reed, MD, Lawrence Schiffrnan, MD, Anne leukotriene B4 synthesis (33). Moreover, arachi- Sigsbee, MD, Patricia Soscia, MD, Steve11 Strongwater, MD, acid is not increased appreciably after GLA Katherine Upchurch, MD, Eileen Winston, MD. Robert administration. Yood, MD, John Zieminski, MD. GAMMA-LINOLENIC ACID IN RA 1817

REFERENCES 18. Paulus HE, Egger MJ, Ward JR, Williams HJ, the Cooperative Systemic Studies of Rheumatic Diseases Group: Analysis of 1. Fantone JC, Kunkel SL, Zurier RB: Effects of prostaglandins on in improvement in individual rheumatoid arthritis patients treated vivo immune and inflammatory reactions. In. Prostaglandins and with disease-modifying antirheumatic drugs, based on the findings Immunity. Edited by JS Goodwin. Boston, Martinus Nijhoff in patients treated with placebo. Arthritis Rheum 33:477-484, Publishing, 1985 1990 2. Zurier RB: Prostaglandins, leukotrienes and related compounds. 19. Lepage G, Roy CC: Specific methylation of plasma nonesterified In, Textbook of Rheumatology. Fourth edition. Edited by WN fatty acids in a one-step reaction. J Lipid Res 29:227-235, 1988 Kelley, ED Harris, S Ruddy, CB Sledge. Philadelphia, WB Saun- 20. Darlington LG, Ramsey NW: Olive oil for rheumatoid arthritis? ders, 1993 Br J Rheumatol 26 (suppl):15, 1987 DeLuca P, Rothman D, Zurier RB: M'arine and botanical lipids as 3. 21. Kremer JM, Lawrence DA, Jubiz W, DiGiacomo R, Rynes R, immunomodulatory and therapeutic agents in the treatment of Bartholomew LE, Sherman M: Dietary fish oil and olive oil rheumatoid arthritis. Rheum Dis Clin North Am 21:759-778, 1995 4. Baker DG, Krakauer KA, Tate G, Laposata M, Zurier RB: supplementation in patients with rheumatoid arthritis: clinical and Suppression of human synovial cell proliferation by dihomo-g- immunologic effects. Arthritis Rheum 33:810-820, 1990 linolenic acid. Arthritis Rheum 32:1273-1281. 1989 22. Endres S, Meydani SN, Ghorbani S, Schindler R, Dinarello CA: 5. Santoli D, Zurier RB: Prostaglandin E (PGE) precursor fatty acids Dietary supplementation with n-3 fatty acids suppresses inhibit human IL-2 production by a PGE-independent mechanism. interleukin-2 production and mononuclear cell proliferation. J J Immunol 143:1303-1309, 1989 Leuk Biol 54599-603, 1993 6. Santoli D, Phillips PD, Colt TL. Zurier RB: Suppression of 23. Kremer JM, Jubiz W, Michalek A, Rynes RI, Bartholomew LE, interleukin 2-dependent human T cell growth by E-series prosta- Bigaouette J, Timchalk M, Beeler D, Lininger L Fish oil fatty acid glandins (PGE) and their precursor fatty acids: evidence for a supplementation in active rheumatoid arthritis: a double-blinded, PGE-independent mechanism of inhibition by the fatty acids. controlled crossover study. Ann Intern Med 106:497-503, 1987 J Clin Invest 85:424-432, 1990 24. Pincus T, Stein CM: What is the best source of useful data on the 7. DeMarco DM, Santoli D. Zurier RB: Effects of fatty acids on treatment of rheumatoid arthritis: clinical trials, clinical observa- proliferation and activation of human synovial compartment lym- tions, or clinical protocols? J Rheumatol 22:1611-1617, 1995 phocytes. J Leukoc Biol56:612-615, 1994 25. Wells G, Tugwell P: Cyclosporin A in rheumatoid arthritis: 8. Tate GA, Mandell BF, Karmali RA, Laposata M, Baker DG, overview of efficacy. Br J Rheumatol 32 (suppl):1:51-56, 1993 Schumacher HR Jr, Zurier RB: Suppression of monosodium urate 26. Tugwell P, Pincus T, Yocum D, Stein M, Gluck 0, Kraag G, crystal-induced acute inflammation by diets enriched with gamma- McKendry R, Tesser J, Baker P, Wells G: Combination therapy linolenic acid and eicosapentaenoic acid. Arthritis Rheum 31: with cyclosporine and methotrexate in severe rheumatoid arthritis. 1543-1551, 1988 N Engl J Med 333:137-141, 1995 9. Leventhal LJ, Boyce EG, Zurier RB: Treatment of rheumatoid 27. Jantti J, Nikkari T, Solakivi T, Vapaatalo H, Isomaki H: Evening arthritis with gammalinolenic acid. Ann Intern Med 119:867-873, primrose oil in rheumatoid arthritis: changes in serum lipids and 1993 fatty acids. Ann Rheum Dis 48:124-127, 1989 10. Arnett FC, Edworthy SM. Bloch DA, McShane DJ, Fries JF, 28. Belch JJF, Ansell D, Madhok R, O'Dowd A, Sturrock RD: Effects Cooper NS, Healey LA, Kaplan SR, Liang MH, Luthra HS, of altering dietary essential fatty acids on requirements for non- Medsger TA Jr, Mitchell DM, Neustadt DH, Pinals RS, Schaller steroidal anti-inflammatory drugs in patients with rheumatoid JG, Sharp JT, Wilder RL, Hunder GG: The American Rheuma- arthritis: a double-blind placebo controlled study. Ann Rheum Dis tism Association 1987 revised criteria for the classification of 47~96-104, 1988 rheumatoid arthritis. Arthritis Rheum 31:315-324, 1988 29. Wright S, Burton JL: Oral evening primrose oil improves atopic 11. Steinbrocker 0, Traeger CH, Batterman RC: Therapeutic criteria eczema. Lancet 2:1120-1122, 1982 in rheumatoid arthritis: JAMA 140:659-662, 1949 30. Bamford JT, Gibson RW, Renier CM: Atopic eczema unrespon- 12. Tulleken JE, Limburg PC, Muskiet FA, van Rijswijk MH: Vitamin sive to evening primrose oil (linoleic and gamma-linolenic acids). E status during dietary fish oil supplementation in rheumatoid J Acad Dermatol 13:959-965, 1985 arthritis. Arthritis Rheum 33:1416-1419, 1990 31. Hammarstrom S: Conversion of dihomogamma linolenic acid to 13. Felson DT. Anderson JJ, Boers M, Bombardier C. Chernoff M. an isomer of leukotriene C, oxygenated at C,. J Biol Chem Fried B, Furst Goldsmith C, Kieszak Lightfoot R, Paulus H, D, S. 256:7712-7714, 1981 Tugwell P, Weinblatt M, Widmark R, Williams HJ, Wolfe F The 32. Ziboh VA, Chapkin RS: Biologic significance of polyunsaturated American College of Rheumatology preliminary core set of dis- ease activity measures for rheumatoid arthritis clinical trials. fatty acids in the skin. Arch Dermatol 123:1686-1690, 1987 Arthritis Rheum 36:729-740, 1993 33. Pullman-Mooar S, Laposata M, Lem D, Holman RT, Leventhal 14. Fries JF, Spitz PW, Young DY: The dimensions of health out- LJ, DeMarco D, Zurier RB: Alteration of the cellular fatty acid comes: the Health Assessment Questionnaire, disability and pain profile and the production of eicosanoids in human monocytes by scales. J Rheumatol 9:789-793, 1983 gamma-linolenic acid. Arthritis Rheum 33:1526-1533, 1990 15. Pincus T: Assessment of long term outcomes of rheumatoid 34. Callegari P, Zurier RB: Botanical lipids: potential role in modu- arthritis. Rheum Dis Clin North Am 21:619-654, 1995 lation of immunologic responses and inflammatory reactions. 16. Williams HJ, Willkens RF, Samuelson CO Jr. Alarc6n GS, Gut- Rheum Dis Clin North Am 17:415-425, 1991 tadauria M, Yarboro C, Polisson RP, Weiner SR, Luggen ME, 35. McMurchie EJ: Dietary lipids and the regulation of membrane Billingsley LM, Dahl SL, Egger MJ, Reading JC, Ward JR: fluidity and function. In, Methods for Studying Membrane Fluid- Comparison of low-dose oral pulse methotrexate and placebo in ity. New York, Alan R. Liss, 1988 the treatment of rheumatoid arthritis: a controlled clinical trial. 36. Ordway RW, Walsh JV, Singer JJ: Arachidonic acid and other Arthritis Rheum 28:721-730, 1985 fatty acids directly activate potassium channels in smooth muscle 17. Weinblatt ME, Trentham DE, Fraser PA, Holdsworth DE, Fal- cells. Science 244:1176-1179, 1989 chuk KR, Weissman BN, Coblyn JS: Long-term prospective trial 37. Cornwell DG, Huttner JJ, Milo GE, Panganamal RV, Sharma HM, of low-dose methotrexate in rheumatoid arthritis. Arthritis Rheum Ceer JC Polyunsaturated fatty acids, vitamin E, and the proliferation 31:167-175, 1988 of aortic smooth muscle cells. Lipids 14:194-207, 1979