Polycystic ovary syndrome (PCOS) is commonly associated with insulin resistance, dyslipidemia and increased inflammation, which all benefit from dietary intake of monounsaturated and n-3 polyunsaturated fatty acids (MUFA and n-3 PUFA). Our goal was to compare the effects of MUFA-rich almonds vs n-3/n-6 PUFA-rich walnuts on metabolic and endocrine parameters in PCOS.
Thirty-one PCOS patients randomly received either walnuts or almonds containing 31 g of total fat per day for 6 weeks. At the beginning and at the end, anthropometric parameters, fasting lipids, phospholipid-fatty acids, inflammatory markers, androgens, oral glucose tolerance tests (OGTT) and frequently sampled intravenous-GTT were obtained.
Weight remained stable. Within group, walnuts increased the n-3/n-6 essential PUFA in the diet and plasma phospholipids. Walnuts decreased low-density lipoprotein-cholesterol by 6% from 3.76±0.27 to 3.38±0.22 mmol/l (P=0.05) and apoprotein B by 11% from 0.72±0.04 to 0.64±0.05 g/l (P<0.03). Although almonds also reduced low-density lipoprotein-cholesterol by 10% and apoprotein B by 9%, these were not significant. Walnuts increased insulin response during OGTT by 26% (P<0.02). Both walnuts and almonds increased adiponectin (walnuts from 9.5±1.6 to 11.3±1.8 μg per 100 ml, P=0.0241; almonds from 10.1±1.5 to 12.2±1.4 μg/dl, P=0.0262). Walnuts decreased HgBA1 from 5.7±0.1 to 5.5±0.1% (P=0.0006) with significant intergroup difference from almonds (P=0.0470). Walnuts increased sex hormone-binding globulin from 38.3±4.1 to 43.1±4.3 nmol/l (P=0.0038) and almonds reduced free androgen index from 2.6±0.4 to 1.8±0.3 (P=0.0470).
Nut intake exerted beneficial effects on plasma lipids and androgens in PCOS.
Polycystic ovary syndrome (PCOS) affects 1 out of 16 young women (Ehrmann, 2005). These patients have irregular menstrual periods and infertility, and elevated androgen levels causing excess facial and body hair. In the USA, majority of PCOS patients are obese and ∼50% have metabolic syndrome (Glueck et al., 2003; Ehrmann et al., 2006)—characterized by insulin resistance, hyperlipidemia, hypertension and increased inflammation (Dunaif, 2000). Consequently a large number of young women seeking help from nutritionists for treatment of obesity and hyperlipidemia may have PCOS.
Metabolic disorders associated with PCOS can benefit from monounsaturated and polyunsaturated fatty acids (MUFA and PUFA). The MUFA and PUFA improve plasma lipids, and PUFA (especially n-3) increases insulin sensitivity, decreases blood pressure and inflammation (Vessby, 2000; Sacks and Campos, 2006; Berglund et al., 2007). As nuts are rich sources of MUFA and PUFA, PCOS patients are frequently advised to increase nut intake. In general, such recommendations do not consider significant differences in fatty acid compositions of nuts. For example, almonds contain 30% MUFA whereas walnuts contain 9% MUFA. Their PUFA contents and composition also differ. Although almonds contain 12% PUFA (all n-6 class), walnuts contain 47% PUFA (with 1:4 n-3/n-6 ratio). American diet can have n-3:n-6 PUFA ratio up to 1:20 (Simopoulos et al., 2000). This is important because n-3 vs n-6 PUFA exert contrasting biological effects. For example, although n-3 PUFA increases insulin sensitivity and reduces insulin levels, n-6 PUFA may stimulate insulin secretion. Although n-3 PUFA are anti-inflammatory and anti-coagulant, n-6 PUFA are pro-inflammatory and pro-coagulant (Abbate et al., 1996). In order to compare the metabolic and endocrine effects of MUFA vs PUFA rich nuts, we investigated the effects of almonds vs walnuts on anthropometric parameters, plasma fatty acids, plasma lipids, inflammatory markers, glucose homeostasis and androgens in PCOS.
Subjects and methods
This study was approved by the Institutional Review Board of University of California, Davis, CA, USA. Women between the ages 20–45 years and with a body mass index of &1QJ;25–45 kg/m2 fulfilling the NIH criteria for PCOS (Azziz, 2005) were recruited. Patients were excluded if they used oral contraceptives, anti-androgens, insulin sensitizers, d-chiro inositol, or any other medicines or supplements affecting weight or insulin sensitivity during the preceding 2 months; had diabetes mellitus, untreated thyroid disease and any other systemic illness such as renal, hepatic and gastrointestinal disease; smoke; or drink >2 alcoholic drinks per week.
A total of 142 PCOS patients were assessed for eligibility; 90 subjects failed to meet the inclusion criteria: 52 were consented and screened; 16 did not qualify; remaining 36 were randomized to treatment groups (n=18 per group). Four in the almond and one in the walnut group dropped out (one due to pregnancy, one transportation, one diarrhea, one personal reasons and one for having difficulty with blood draw). Thirty-one women (22 White, four Hispanic, two African American, two Asian and one mixed) completed the study.
This was a randomized, prospective study with two parallel arms. The goal was to exchange 31 g of the habitual dietary fats per 1800 kcal with nuts (equivalent of 46 g of almonds or 36 g of walnuts) for 6 weeks, which is adequate for changes in plasma lipids (Schaefer et al., 1995). Although 46 g almonds or 36 g walnuts each contain 31 g oil, their compositions differ: 31 g of almond-oil delivers 2.4 g saturated fat, 19.5 g MUFA, 7.5 g linoleic acid (LA; 18:2 n-6) and no α-linolenic acid (ALA; 18:3 n-3; Abbate et al., 1996) whereas 31 g walnut-oil delivers 2.9 g saturated fat, 4.5 g MUFA, 19.2 g LA and 4.3 g ALA. The fat exchange was accomplished as described previously (Kasim-Karakas et al., 2004). Amount of nut intake was based on average daily energy intake, which was assessed using 7-day food records (Food Processor SQL software; ESHA Research, Salem, OR, USA). The subjects were counseled by the Clinical Research Center dietitian to maintain fat and energy intakes constant. Daily allotments of the nuts were individually packaged for each subject. To assess compliance, the participants were asked to return unused portions of the nuts.
Data were obtained at the beginning and at the end of the study. The oral glucose tolerance test (OGTT) and frequently sampled intravenous glucose tolerance test were performed 1 week apart.
Weight was measured in light clothing using the Tanita BWB800-P Digital Medical Scale (Tokyo, Japan). Height was measured using an Ayrton Model S100 stadiometer (Ayrton Corp, Prior Lake, MN, USA).
After an overnight fast, baseline samples were obtained; participants drank 75 g of glucose (Glucola, Fisher Healthcare, Houston, TX, USA) and additional blood samples were obtained every 30 min for 2 h. Average carbohydrate intake was 237±16 g/day before OGTT.
Frequently sampled intravenous glucose tolerance test
After an overnight fast, an intravenous catheter was placed in their forearm and kept open with normal saline. Heating pads were used in order to maximize blood flow. Three blood samples were obtained at time −15, −10 and −5 min. Glucose (0.3 U/kg as 25% dextrose) was given intravenously at time 0 min. Intravenous insulin 0.03 U/kg (Humulin Regular, Eli Lilly, Indianapolis, IN, USA) was given at time 20 min after the glucose administration. Blood samples were obtained at time 0, 2, 3, 4, 5, 6, 8,10, 12, 14, 16, 19, 22, 23, 24, 25, 27, 30, 40, 50, 60, 70, 90, 100, 120, 140, 160 and 180 min. Acute insulin response to glucose (AIRGlucose: an index of insulin secretion), β-cell function, sensitivity index and disposition index were calculated using MiniMod Millennium software (Dr Bergman, Los Angeles, CA, USA).
Glucose was measured using YSI 2300 STAT Plus Glucose and Lactate Analyzer (YSI Life Sciences, Yellow Springs, OH, USA), with coefficient of variation (CV) of 1%. Insulin was measured by radioimmunoassay (Millipore, St Charles, MO, USA) with a CV of 8.2%. The homeostatic model assessment and Matsuda Index, surrogate measures of hepatic and peripheral insulin sensitivity, were calculated as previously published (Karakas et al., 2010). Triglyceride, cholesterol and high-density lipoprotein-cholesterol were measured using Poly-Chem System Analyzer (Polymedco, Inc., Cortlandt Manor, NY, USA) with CVs of 3.5, 4 and 3.6%, respectively. Leptin and adiponectin were measured using radioimmunoassay (Millipore) with CVs of 4.3 and 6.5%. The high-sensitivity C-reactive protein was measured using a highly sensitive latex-enhanced immunonephelometric assay with CV <5%. Interleukin (IL)-6, IL-1β, tumor necrosis factor-α were measured using the High Sensitivity Human Cytokine Panel-3 Plex (Milliplex) kit (Millipore) with CV of 11%. Total testosterone, sex hormone binding globulin (SHBG) and dehydroepiandrosterone sulfate were measured by radioimmunoassay (Diagnostic Systems Laboratories, Webster, TX, USA) with CVs of 8.3, 4.4 and 9.6%, respectively. Fatty acid composition was measured using gas chromatography by the Lipid Technologies, LLC (Austin, MN, USA).
The SAS software, version 9.1 (SAS Institute Inc., Cary, NC, USA) was used. Descriptive statistics were calculated for each outcome, for each group, before and after intervention. The data were log-transformed in order to improve the normality of residuals and homoscedasticity of errors as appropriate before analysis. Paired t-test was performed to determine the significance of within-group change. Two-group comparison was performed by analysis of covariance, adjusted for the baseline values. The longitudinal trajectories for changes in glucose and insulin during OGTT were estimated by repeated measures analysis of variance. Individual trajectories for changes in glucose and insulin were estimated from linear random effects models. Each observed level was entered as the dependent variable. Treatment, time and treatment × time interaction term were entered as independent variables. The coefficients for the interaction term were to estimate the additional changes in glucose and insulin level over time associated with treatment. To account for between-subject heterogeneity in the change of glucose or insulin level, intercept and time were modeled as random effects. Multiple comparisons were controlled by the Bonferroni method. An as-treated analysis was performed. We chose not to perform an intent-to-treat analysis because those who dropped out had only the baseline, but no interim, data. At the planning stage, the sample size was calculated to detect differences between groups in insulin at 6 weeks. A sample size of 18 in each group would yield 80% power to detect differences of 0.85 s.d. of the mean for each of these outcomes with a 0.05 two-sided α-level. At the end of the study, the number of completers was not adequate to detect the significant differences in all parameters.
Changes in weight and diet
Weight did not change in either group (Tables 1 and 3). The participants consumed all the nuts. The 7-day food records demonstrated that percent fat intake did not change; before and after values for percentage of fat were 36.9±1.9 vs 39.5±1.9 (P=0.2839) in the almond group and 36.6±1.4 vs 38.6±1.4 (P=0.3595) in the walnut group. The two groups had significantly different LA and ALA intakes (P=0.0163 and P=0.0002, respectively). Within group changes indicated that walnuts increased ALA intake by 11-folds (from 0.3±0.1 to 3.5±0.03 g, P<0.0001) and LA intake by fivefolds (from 3.2±0.5 to 16.0±1.5 g, P<0.0001), but did not alter saturated fat or MUFA intakes. Almonds decreased saturated fat intake (from 29.2±4.7 to 22.2±2.7 g, P=0.0419) and tended to increase MUFA (from 17.7±3.4 to 23.2±2.7 g, P=0.1991).
Changes in fatty acid composition of plasma phospholipids
There were significant differences between the groups for changes in LA (P=0.0477), ALA (P=0.0142) and arachidonic acid (P=0.03555) (Table 2). Within group, walnuts increased LA (18:2 n-6) by 7% from 16.44±0.91 to 17.60±0.55% and ALA (18:3 n-3) by 16% from 0.19±0.01 to 0.22±0.01%. Walnuts did not increase the long chain n-3 PUFA eicosapentanoic acid (EPA) or docosahexanoic acid (DHA). Almonds increased (P=0.0240) and walnuts decreased (P=0.0473) arachidonic acid (AA; 20:4 n-6). Although almonds increased the oleic acid (18:1 n-9) by 9% and walnuts decreased it by 8%, these changes were not significant.
Changes in glucose homeostasis, insulin secretion and action
The only significant difference between the treatment effects was seen in HgBA1c (P=0.047) while the change in AUCInsulin approached significance (P=0.0628) (Table 3; Figures 1 and 2). Walnuts decreased HgBA1 (from 5.7±0.1 to 5.5±0.1%, P=0.0006). Although almonds also caused a slight decrease in HgBA1c from 5.8±0.1 to 5.7±0.1%, this change was not significant (P=0.1309). Neither treatment changed fasting glucose, insulin or homeostatic model assessment.
As seen in Figure 1, neither almonds nor walnuts changed glucose response during OGTT. Walnuts increased insulin response (Figure 2), and AUCInsulin by 26% (from 2556±351 to 3227±519 pmol/l per 2 h, P=0.0182), whereas almonds did not affect insulin. Walnuts or almonds did not change any of the intravenous glucose tolerance test parameters. Both treatments increased adiponectin (almonds by 21%; 2.1 μg per 100 ml, P=0.0192 and walnuts by 19%; 18 μg per 100 ml, P=0.0179) and walnuts increased leptin by 18% (from 25.7±3.0 to 30.3±3.7 ng/ml, P=0.0416).
Changes in cardiovascular risk factors
There were no differences between the treatments for plasma lipids and inflammatory risk factors (Table 3). Within groups, fasting triglyceride was not affected by either treatment. Walnuts decreased low-density lipoprotein-cholesterol by 6% or 0.24±0.1 mmol/l (P=0.05) and apoprotein B by 11% or 0.08±0.03 g/l (P=0.0236). Although walnuts also decreased high-sensitivity C-reactive protein by 11.5% and IL-1β by 23%, these changes were not significant. Decreases caused by almonds in low-density lipoprotein-cholesterol (10%), apoprotein B (9%), IL-6 (19%) and tumor necrosis factor-α (20%) were not significant either (Table 3).
Changes in androgens
There were no differences between the treatments for androgens (Table 3). Within groups, walnuts increased SHBG by 12.5% (from 38.3±4.1 to 43.1±4.3 nmol/l, P=0.0104). Almonds increased SHBG by 16% (P=0.0596). As SHBG binds testosterone avidly, increased SHBG can decrease the free fraction of androgens. Therefore, free androgen index was calculated. Almonds reduced free androgen index from 2.6±0.4 to 1.8±0.3, P=0.0491. Testosterone and dehydroepiandrosterone sulfate levels did not change.
Nut intake altered fatty acid composition of the diet. Almonds increased MUFA intake by 33% and decreased saturated fat by 25% without altering n-3 or n-6 PUFA. In contrast, walnuts increased n-3 PUFA and n-6 PUFA intakes by 11-fold and 5-fold, without affecting saturated fat or MUFA intakes.
Composition of the plasma phosholipids also changed. Almonds tended to increase and walnuts tended to decrease the MUFA oleic acid. Walnuts increased LA and ALA—the essential PUFA that cannot be produced in vivo. However, LA can be converted to the longer and more unsaturated product AA (20:4) and ALA can be converted to EPA (20:5) and DHA (22:6). Despite walnuts providing large amounts of LA and ALA, the long chain product of LA (AA) decreased, whereas the long chain products of ALA (EPA and DHA) did not change. In humans, conversion of the essential PUFA (LA and ALA) to the long chain PUFA (AA, EPA and DHA ) is inefficient (Arterburn et al., 2006) because LA and ALA compete with each other on the rate limiting step (Marshall and Johnston, 1982). Ratio ALA to LA (1:4) is much higher in walnuts than in the Western diet (1:20). Thus, relative abundance of ALA may have interfered with conversion of LA into AA. Similarly, the high LA content may have prevented conversion of ALA to EPA and DHA.
A major focus was the changes in glucose homeostasis. Blood glucose levels are regulated by the balance between glucose utilization in the tissues and insulin secretion from the pancreas. When the tissues cannot use glucose efficiently (insulin resistance), insulin secretion increases (hyperinsulinemia). Insulin resistance is the major underlying pathology in PCOS, metabolic syndrome and type 2 diabetes. Although experimental data have shown that n-3 PUFA increase insulin sensitivity and glucose utilization (Kopecky et al., 2009; Martin de Santa Olalla et al., 2009), and studies in human muscle have demonstrated that MUFA and PUFA in the cell membrane correlate directly with glucose uptake (Baur et al., 1998), insulin resistance did not decrease in our study. This may be due to the specific mechanism by which n-3 PUFA overcomes insulin resistance. Glucose uptake requires binding of insulin to its receptor in the cell membrane, followed by serial phosphorylation steps that occur in the insulin receptor and its substrate. Phosphorylation of tyrosine residues in the cascade advances, and serine phosphorylation reduces insulin signaling. The n-3 PUFA overcomes insulin resistance by preventing serine phosphorylation (Ma et al., 2009). In PCOS, there is intrinsic increase in serine phosphorylation (Dunaif et al., 1995). It is conceivable that n-3 PUFA cannot reverse this intrinsic abnormality, and therefore cannot correct insulin resistance. To test this, future research will need to compare the effects of n-3 PUFA in PCOS vs healthy control women.
We demonstrated that walnuts increased insulin response during OGTT. Walnuts increased fasting insulin in type 2 diabetic patients as well (Ma et al., 2010). This may be explained by two different mechanisms. First, walnuts could have worsened insulin resistance, thus caused compensatory increase in insulin. However, there was no increase in insulin resistance in our study. Alternatively, walnuts could have stimulated insulin secretion from pancreas directly. Animal and cell culture studies have shown that the major PUFA in walnuts (LA) can stimulate insulin secretion directly (Pareja et al., 1997).
Increased insulin secretion without any increase in insulin resistance should lower blood glucose. Consistent with this, walnuts decreased HgBA1c. In fact, this was the only significant difference between the almond and walnut groups. The change in HgBA1c occurred within 6 weeks. Although 50% of the change in HgBA1c can be attributed to changes in blood glucose during the preceding month, full change occurs in 3 months. Therefore, a longer study might have shown a larger decrease.
Previously we have shown that walnuts increased plasma glucose without changing insulin in PCOS (Kasim-Karakas et al., 2004). However, weight has decreased in this study. Similar findings were observed in type 2 diabetic patients when body weight decreased (Tapsell et al., 2009). Taken altogether, these results support that changes in weight modify insulin response to PUFA.
Both almonds and walnuts increased adiponectin. Experimental data have demonstrated that serum adiponectin correlated directly with oleic acid (MUFA in almonds) and LA (major PUFA in walnuts) in the adipocyte membrane (Perez de Heredia et al., 2009), as well as ALA (Sekine et al., 2008). Although adiponectin correlates with insulin sensitivity (Stefan and Stumvoll, 2002; Spranger et al., 2004), insulin sensitivity did not increase in our study. It is possible that the intrinsic defect in insulin signaling in PCOS may be unresponsive to the favorable effects of adiponectin.
Walnuts also increased SHBG, which is secreted by the liver and correlates with insulin sensitivity (Stefan and Stumvoll, 2002; Bonnet et al., 2009). To our knowledge, effects of PUFA on SHBG have not been previously reported.
Walnuts increased leptin by 18%. The literature indicates that n-6 PUFA do not alter leptin whereas n-3 PUFA exert variable results (Takahashi and Ide, 2000; Reseland et al., 2001; Korotkova et al., 2002; Peyron-Caso et al., 2002). Insulin stimulates leptin production and secretion (Havel, 2002). In our study, walnuts may have increased leptin by increasing insulin.
Another area of interest was the cardiovascular risk factors. Plasma triglyceride did not change with either treatment. This is consistent with the literature reporting that MUFA or plant based essential n-3 PUFA did not decrease plasma triglyceride—unlike the long chain n-3 PUFA from fish (Finnegan et al., 2003; Ma et al., 2010). Both almonds and walnuts decreased cholesterol, as previously reported (Almario et al., 2001; Feldman, 2002; Griel and Kris-Etherton, 2006; Banel and Hu, 2009). These changes were statistically significant only with walnuts, possibly due to a smaller population size in the almond group.
Both MUFA and n-3 PUFA have anti-inflammatory effects, whereas n-6 PUFA are considered pro-inflammatory (Calder, 2001). Therefore, we anticipated almonds to decrease inflammatory markers, whereas the effects of walnuts could not be predicted due to the contrasting effects of n-3 vs n-6 PUFA. We found that although both almonds and walnuts tended to reduce these markers (high-sensitivity C-reactive protein, tumor necrosis factor-α, IL-1β and IL-6), the changes were not significant, possibly due to the smaller population size in the almond group.
We also investigated changes in androgens. Testosterone is the major ovarian androgen and its bioavailability depends upon the abundance of SHBG. Insulin resistance lowers SHBG, decreases testosterone binding and consequently increases free testosterone—the fraction which causes the undesirable clinical effects such as excess hair growth and acne (Hamilton-Fairley et al., 1995). Walnuts increased SHBG by 12.5%, without affecting insulin sensitivity, suggesting that the rise in SHBG may have been a direct effect. Almonds decreased free androgen index through a dual mechanism, both by increasing SHBG by 16% and decreasing testosterone. Thus both nuts exerted favorable effects on circulating androgens.
In summary, we compared metabolic and endocrine effects of nuts with different fatty acid compositions in high-risk women. The limitations of the study were the lack of a non-treatment PCOS arm or a non-PCOS control group, and relatively small population size. Despite these limitations, the results support inclusion of nuts in the PCOS diet because of their beneficial effects on lipids, androgens and possibly inflammatory markers. Our findings also stress the importance of monitoring changes in glucose homeostasis during nut intake. Future research can focus on the metabolic/endocrine effects of individual fatty acids in other insulin resistant states such as metabolic syndrome.
This study was supported by grants from the National Center for Complementary and Alternative Medicine (R21 AT002280) and the ALSAM Foundation to Dr SE Karakas. The clinical studies were partially supported by the UC Davis Clinical and Translational Science Center grant (RR 024146).