Effect of rye consumption on markers of glycemic control: evidence on the “rye factor”: a systematic review and meta-analysis of randomized controlled trials

Abstract

Rye, as a source of dietary fiber, may have beneficial effects in glycemic control. In the current meta-analysis, we collected randomized controlled trials (RCTs) that examined the effect of rye consumption on glucose and insulin markers. PubMed, Scopus, and Web of Science databases were searched to find the RCTs. Random-effects model was used to calculate mean difference and 95% confidence intervals. Thirty-one RCTs, including 922 participants, passed the screening and eligibility stages and were included in the meta-analysis. Rye consumption did not have a significant effect on glucose indices including fasting, postprandial, and area under the curve (AUC). Subgroup analysis did not make a difference in the results, except that there was trends for increased postprandial glucose in two subgroups: individuals aged > 50 y (weighted mean difference (WMD) = 0.93, 95% CI: -0.03, 1.90 mmol/l, P = 0.058) and short intervention lengths (≤ 270 min) (WMD = 0.48, 95% CI: -0.03, 0.99 mmol/l, P = 0.066), and a trend for decreased AUC for glucose in rye fiber doses ≥ 12 g (WMD = -0.22, 95% CI: -0.46, 0.01 mmol/l, P = 0.059). Rye consumption did not show an effect on fasting and postprandial insulin but indicated a reduction in AUC for insulin (WMD = -0.48, 95% CI: -0.66, -0.30 mU/l, P < 0.001). Overall, results of this meta-analysis suggest that rye consumption may reduce insulin postprandial AUC without affecting glucose markers. Prospective cohorts are needed to determine the clinical importance of the finding.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12986-025-00901-8.

Introduction

Glycemic control refers to the maintenance of blood glucose levels within an appropriate range [1]. Glycemic control is achieved when blood glucose levels, including fasting and postprandial glucose, remain within the normal range during the 24-h period [2, 3].

Glycemic control is dysregulated by a number of illnesses such as hyperthyroidism, polycystic ovary syndrome, infections, and particularly diabetes [4]. In addition, some medications, such as corticosteroids, and psychological factors, such as stress, can negatively impact glycemic control. Unhealthy diets, insufficient physical activity, and obesity are also among factors that disrupt glycemic control [5].

The prevalence of poor glycemic control is high, even among healthy individuals [6]. A systematic review and meta-analysis of observational studies in Ethiopia showed that 61.9% of patients with diabetes had poor glycemic control [6]. Also, in a cross-sectional study, 41% of rural Zambian residents had elevated glycated hemoglobin (HbA1c) (≥ 5.7%) [7]. Poor glycemic control has also been observed among adolescents. A school-based cross-sectional study of 37,804 Brazilian adolescents revealed that 20.5% had high HbA1c levels (≥ 5.7%) [8].

Poor glycemic control can lead to a number of complications, especially in the case of diabetes [9]. Consistently elevated levels of glucose increase the risk of forming advanced glycation end products [10]. These products then bind to their receptors on macrophages, endothelial cells, and smooth muscle cells, initiating cascades that lead to increased intracellular oxidative stress and inflammation [9].

Nutrition therapy plays a critical role in establishing glycemic control [11]. In fact, based on the results of a meta-analysis, poor adherence to dietary recommendations for diabetes was the strongest predictor of poor glycemic control, after poor drug adherence [6]. Deviation from this control increases the risk of microvascular and macrovascular complications, which are known consequences of poor glycemic control [9, 12].

Rye (Secale cereale) is a member of cereal family [13]. Rye has the highest amount of fiber among cereals [13, 14]. It constitutes at least 20% of rye dry matter, while other cereals such as wheat, oat, and barley have 10–15% fiber. Dietary fibers are known to reduce the glycemic load of foods and lower the risk of diabetes [15]. They delay gastric emptying, thereby prolonging gastric and intestinal transit time and slowing down the absorption rate in the small intestine [16].

Since rye has a high fiber content, we hypothesized that it may have beneficial effects on glucose and insulin markers. In the current meta-analysis, we collected randomized controlled trials (RCTs) that examined the effect of rye consumption on markers of glycemic control. Given the high prevalence of poor glycemic control among both diabetic and non-diabetic individuals, and considering the role of diet in the management of glycemic control, conducting such meta-analyses may strengthen our knowledge and provide helpful recommendations for achieving good glycemic control.

Methods

The systematic review and meta-analysis was performed based on guidelines of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) at all stages of design, implementation, and reporting [17].

Search

PubMed, Scopus, and Web of Science databases were searched from the earliest online available data until 17th of March 2024. RCTs examining the effect of rye consumption on blood glucose, insulin, glycosylated hemoglobin (HbA1c), and Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) in adults aged 18 years and older. Search terms included: glucose, insulin, insulin resistance, glycemic index, and glycemic control. Search strategy is provided as a supplementary material (Supplementary Table 1). The search was performed by one investigator, and the screening of titles and abstracts was conducted by two independent investigators. There was no restriction on language.

Study selection and eligibility criteria

RCTs investigating the effect of rye consumption on blood glucose, insulin, HbA1c, and HOMA-IR were included. Randomization was a requirement for eligibility, but blindness was not necessary because in the majority of studies, rye had been administered in the form of a food product, such as bread.

Trials were included if they had reported means and standard deviation (SD) or error (SE) for baseline and post-intervention values of at least one treatment arm and a control group, or post minus pre-intervention mean difference for each of treatment and control groups along with SD/SE. They should also have established identical conditions in the treatment and control groups (except for rye consumption), stated the amounts of test products as well as the fiber content of the products, and used comparable carbohydrate content in treatment and control products. Trials were excluded if only ingredients of rye such as rye bran were administered as treatment, rye was largely mixed with other fiber-containing grains such as oat or barley, in the absence of a control group consuming an inert product, when baseline values were not reported or participants had diabetes, and in repeated publications.

Data extraction

Data extraction was performed by two investigators. Information extracted from each study was as follows: general publication information (authors’ and journal’s name, and publication year), study location (i.e. country of origin), trial design (parallel or crossover), intervention length, sample size (enrolled and completers), participants’ characteristics (age, sex, and body mass index), the type and dose of products administered to treatment and control groups, the dose of fiber in treatment and control groups, pre- and post-intervention values and SD (or SE) of the outcomes in treatment and control groups, and the type of data (fasting, postprandial, and area under the curve (AUC)). For those data which were presented as graphs, Plot Digitizer software (Free Software Foundation Inc., USA, version 2.6.9) was used to extract the data. Postprandial data were not utilized if they were presented in the form of unreadable graphs.

Risk of bias assessment

The quality of the trials was assessed using the revised Cochrane risk of bias tool for randomized trials [18], with supplemental updates in November 2020 for cluster randomized trials and in December 2020 for crossover trials (available at https://sites.google.com/site/riskofbiastool/welcome/rob-2-0-tool?authuser=0). The criteria used by these tools include: risk of bias arising from randomization process, risk of bias due to deviations from intended interventions, risk of bias due to missing outcome data, risk of bias in measurement of the outcome, and risk of bias in selection of the reported results. For crossover design, there is an additional item for the risk of bias arising from period and carryover effects.

Statistical analysis

STATA software version 13.0 (StataCorp, USA) was used for data analysis. Mean difference and SD of the difference between pre- and post-intervention values for each of treatment and control groups was used to estimate the pooled effects. Because there was a high heterogeneity in some of the results, the random-effects model was used to estimate the overall effects. Heterogeneity was assessed by the I² test using random inverse-variance heterogeneity, and < 40%, 40–69%, and 70–100% were considered as low, moderate, and considerable heterogeneity, respectively [19]. Subgroup analysis was performed based on participants’ age (< 40 y, ~ 40–65 y, or > 50 y), sex (males, females, or males/females), weight status ((normal (BMI < 25 kg/m²), overweight (BMI = 25–29.9 kg/m²), or obese (BMI ≥ 25 kg/m²)(, rye dosage (< 12 g or ≥ 12 g rye fiber), study duration (≤ 270 min or > 270 min), and fermentation of the product (yes or no). The classification of rye dose was performed based on the median of the doses. In the classification of study duration, less than one night was considered as “short” and longer durations were labeled as “long” study duration. To determine the extent to which the analyses relied on a specific study, a sensitivity analysis was performed using the One Study Removed method.

Results

Systematic review of the trials

In this meta-analysis, RCTs examining the effect of rye consumption on glycemic control and insulin levels were investigated. In total, 1583 RCTs from PubMed (n = 304), Scopus (n = 680), and Web of Science (n = 599) databases were found. After removing duplicates (n = 639) and screening titles and abstracts (n = 822), 122 citations remained to be assessed for eligibility. During reading the full texts. 95 citations were excluded due to reasons described in Fig. 1. Ultimately, 27 citations including 31 trials were included in the meta-analysis.

Fig. 1.

Flowchart of the screening and selection process of RCTs included in the meta-analysis of the effect of rye on glycemic control indicators

The summary of the RCTs included in the systematic review and meta-analysis is presented in Table 1. Out of 31 trials, 8 reported fasting glucose [20–27], 12 reported postprandial glucose [23, 27–36], 18 calculated glucose AUC [27, 29, 30, 34, 36–45], 7 measured fasting insulin [20–22, 24–27], 12 reported postprandial insulin [27–36, 46], and 16 determined insulin AUC [27, 29, 30, 32, 34, 36–38, 40, 41, 43, 45, 46]. Each of HbA1c [25], HOMA-IR [21], and QUICKI [20] were reported by 1 trial, and thus were not included in the meta-analysis.

Table 1.

Summary of the RCTs included in the meta-analysis of the effect of rye consumption on serum glucose and insulin levels

Author, year	Location	Participants	Sample size sex	Age (y)	BMI (kg/m2)	Study duration	Treatment	Control	Rye fiber dose (g)	Control fiber dose (g)	Outcomes
[28]	Finland	Healthy	16 M/F	23 ± 3.7	22 ± 1.8	240 min	Refined endosperm rye bread	Refined wheat bread	7.6	3.7	GPP, IPP
[37]	UK	Healthy	6 M/F	19–23	20–25	180 min	Rye bread	Wholemeal wheat bread	8.6	8.7	GAUC, IAUC
[20]	Italy	MetS	123 M/F	40–65	31.5 ± 4.5	12 weeks	Whole grain + endosperm rye bread	Refined wheat bread	24.3	10.4	FBG, Fasting insulin QUICKI
[29]	Denmark	MetS	15 M/F	62.8 ± 4.2	31.1 ± 3.2	270 min	Whole rye bread	Refined wheat bread	12.2	2.9	GPP, GAUC, IPP, IAUC
[38]	Denmark	MetS	15 M/F	63.5 ± 5	31.3 ± 2.7	120 min	Semolina porridge + rye kernels	Semolina porridge	12.7	5.7	GAUC, IAUC
[30]	Finland	MetS	8 M/F	55.6 ± 1.8	33.7 ± 0.7	120 min	Rye bread meal	Refined wheat bread	10.2	3.4	GPP, GAUC, IPP, IAUC
[39]	Sweden	Healthy	10 M/F	26 ± 1	24.1 ± 0.8	90 min	Whole-meal rye bread	White wheat bread	3.75	0	GAUC
[21]	Sweden	OW/OB	207 M/F	56.7 ± 9.6	30.1 ± 0.3	12 weeks	High-fiber rye products	Refined wheat products	32.7	8.7	FBG, Fasting insulin, HOMA
[40]	Sweden	Healthy	23 M/F	60.1 ± 12.1	23.8 ± 3.4	230 min	Whole grain rye bread (unfermented)	Refined wheat bread	11.3	2.9	GAUC, IAUC
[40]	Sweden	Healthy	23 M/F	60.1 ± 12.1	23.8 ± 3.4	230 min	Whole grain rye bread (fermented)	Refined wheat bread	10.2	2.9	GAUC, IAUC
[22]	Finland	Postmenopausal	20 F	59 ± 6	27.5 ± 2.9	8 weeks	High-fiber rye bread	White wheat bread	22	3	FBG, Fasting insulin
[41]	Finland	Postmenopausal	19 F	61 ± 4.8	26.0 ± 2.5	180 min	Traditional rye bread	Refined wheat bread	15.2	2.7	GAUC, IAUC
[41]	Finland	Postmenopausal	19 F	61 ± 4.8	26.0 ± 2.5	180 min	High-fiber rye bread	Refined wheat bread	29	2.7	GAUC, IAUC
[23]	Finland	OW/OB MetS	47 M/F	55 ± 6	32.1 ± 3.8	12 weeks	Rye pasta diet	Oat-wheat-potato diet	24.1	11.8	FBG, GPP
[31]	Finland	OW/OB MetS	19 M/F	55.1 ± 6.4	31.9 ± 3.8	180 min	Rye bread	Oat and wheat bread	9.7	6.2	GPP, IPP
[24]	Finland	Prostate cancer	17 M	74.6 ± 4.7	27.5 ± 4.6	6 weeks	Whole rye products	Refined wheat products + cellulose	64	64	FBG, Fasting insulin
[32]	Finland	Healthy	15 M/F	57 ± 7.6	21–32	240 min	Wholegrain rye bread	White wheat bread	16.4	3.8	GPP, IPP, IAUC
[42]	Sweden	Healthy	21 M/F	38.6 ± 11.8	24.9 ± 3.3	8 h	Whole grain rye porridges	Refined wheat bread	7.1	3.4	GAUC
[42]	Sweden	Healthy	21 M/F	38.6 ± 11.8	24.9 ± 3.3	8 h	Whole grain rye porridges	Refined wheat bread	9.7	3.4	GAUC
[43]	Finland	Healthy	20 M/F	29.5 ± 4	22.4 ± 1.7	180 min	Whole kernel rye bread	White wheat bread	13.5	2.3	GAUC, IAUC
[25]	China	H. pylori infection	84 M/F	51.5 ± 12.6	23.8 ± 4.1	12 weeks	Whole grain rye with rye bran products	Refined wheat products	37.4	12.2	FBG, Fasting insulin, HbA1c
[33]	Finland	Ileostomy	10 M/F	34–65	28.4 ± ?	1 week	Whole rye bread diet	Wheat bread diet	51.4	15.5	GPP, IPP
[26]	Australia	OW/OB	28 M	40–65	30 ± 0.9	4 weeks	Whole grain rye foods	Low-fiber, refined cereals	18	5	FBG, Fasting insulin
[44]	Sweden	Healthy	12 M/F	28.3 ± 5.1	22.1 ± 2.0	120 min	Rye kernel breakfast	Wheat kernel breakfast	14.2	8	GAUC
[34]	Sweden	Healthy	9 M	28.2 ± 6.3	23.9 ± 1.1	180 min	Wholemeal rye bread	Refined wheat bread	15.9	4.3	GPP, GAUC, IPP, IAUC
[35]	Sweden	Healthy	12 M/F	25.3 ± 0.8	23.1 ± 0.6	180 min	Whole grain rye bread	White wheat bread	9.6	1.8	GPP, IPP
[45]	Sweden	Healthy	14 M/F	23.6 ± 0.5	22 ± 0.5	180 min	Whole grain rye bread	White wheat bread	13	3.4	GAUC, IAUC
[27]	Sweden	Healthy	21 M/F	25.3 ± 3.9	22.7 ± 2.3	1 evening	Whole grain rye bread	White wheat bread	18.1	3.6	FBG, Fasting insulin, GPP, GAUC, IPP, IAUC
[46]	Finland	Healthy	20 F	57 ± 12	24.2 ± 2	120 min	Whole rye bread	White wheat bread	19	5	IPP, IAUC
[36]	Sweden	Healthy	24 M/F	30 ± 11	23 ± 5	240 min	Whole grain rye crispbread	Refined wheat crispbread	11.7	2.9	GPP, GAUC, IPP, IAUC
[36]	Sweden	Healthy	24 M/F	30 ± 11	23 ± 5	240 min	Sourdough-fermented rye	Refined wheat crispbread	9.5	2.9	GPP, GAUC, IPP, IAUC

AUC, area under the curve; F, female; FBG, fasting blood glucose; GAUC, glucose AUC; HOMA or HOMA-IR, homeostatic model assessment for insulin resistance; IAUC, insulin AUC; IPP, postprandial insulin; M, males; MetS, metabolic syndrome; OW/OB, overweight/obese; PP, postprandial; QUICKI, quantitative insulin sensitivity check index

Most of the RCTs were conducted in European countries, especially Sweden and Finland, and only two studies were conducted in China [25] and Australia [26]. Four trials [20, 21, 23, 25] had parallel and 27 had crossover design. Twelve trials tested shorter durations (≤ 270 min), while others examined longer-term (8 h and longer) effects of rye consumption. In short-term category, study durations ranged from 90 to 270 min, and in long-term trials, study durations ranged from 8 h to 12 weeks.

Rye treatment was mostly rye bread, but rye porridge, rye products (rye puffs, rye flakes, and rye crisp bread), and rye pasta were also used. In the control arms, white wheat bread or refined cereals were generally used. The form of reporting rye dose largely varied between studies. It was reported as grams of rye products (ranging from 52 to 235 g per day), calories of rye products (ranging from 280 to 1400 kcal per day), grams of rye products’ carbohydrate (40–60 g/day), or the percentage of energy consumed from rye (≥ 20%). Consequently, rye fiber in the treatment arm varied from 3.8 g to 64 g per day. Similarly, participants in the control arm consumed various amounts of control products, containing fiber (generally from wheat sources) ranging from 0 to 15.5 g. However, only in one study [24] refined wheat products were used with added cellulose in an amount of 64 g of fiber in the control group, which was comparable to the amount of rye fiber in the treatment group.

A total of 922 participants were recruited in the trials. Participants were chiefly healthy, but postmenopausal women and individuals with overweight/obesity and metabolic syndrome, and in few cases apparently healthy individuals with Helicobacter pylori infection, ileostomy, and survivors of prostate cancer were also included.

Effect of rye consumption on fasting glucose

Eight studies [20–27] investigated the effect of rye consumption on fasting glucose (Fig. 2). There was no significant effect in the overall analysis (weighted mean difference (WMD) = −0.03 mmol/l, 95% CI: −0.22, 0.16 mmol/l, P = 0.740; I² = 22.4%, P = 0.251) or in any of the subgroup analyses, although some subgroups did not have sufficient number of trials (Table 2). No trials examined rye fiber dosages < 12 g and intervention duration ≤ 270 min on fasting glucose. Heterogeneity between trials was generally low. Sensitivity analysis revealed that excluding one study at a time did not change the results (Supplementary Fig. 1).

Fig. 2.

Forest plots of RCTs examining the effect of rye consumption on fasting glucose, postprandial glucose, and glucose area under the curve (AUC)

Table 2.

Subgroup analysis for the effect of rye consumption on fasting glucose (mmol/l)

Subgroups	Studies (n)	Weighted mean difference (95% CI)	P value	Heterogeneity	P for heterogeneity
Age
Youth (< 40 y)	1	−0.22 (−0.82, 0.39)	0.481	−	−
Middle ages	2	−0.14 (−0.93, 0.64)	0.722	83.5%	0.014
Aged (> 50 y)	5	−0.02 (−0.21, 0.18)	0.864	0	0.633
Sex
Males	2	−0.37 (−0.85, 0.11)	0.135	22.5%	0.256
Females	1	0.28 (−0.35, 0.90)	0.384	−	−
Males/Females	5	0.01 (−0.17, 0.18)	0.926	0	0.413
BMI1
Normal-weight	2	−0.18 (−0.54, 0.17)	0.314	0	0.891
Overweight	2	0.11 (−0.34, 0.57)	0.622	0	0.453
Obese	4	−0.02 (−0.34, 0.30)	0.622	58.9%	0.063
Dose2
< 12 g	−	−	−	−	−
≥ 12 g	8	−0.03 (−0.22, 0.16)	0.740	22.4%	0.251
Intervention duration
Short (≤ 270 min)	−	−	−	−	−
Long (> 270 min)	7	−0.06 (−0.26, 0.13)	0.630	22.8%	0.255
Fermentation
Yes	1	−0.17 (−0.61, 0.28)	0.465	−	−
No	7	−0.01 (−0.23, 0.20)	0.915	30.3%	0.197
Total effect	8	−0.03 (−0.22, 0.16)	0.740	22.4%	0.251

¹Normal-weight: BMI < 25 kg/m², overweight: BMI = 25–29.9 kg/m², and obese: BMI ≥ 25 kg/m²

²The classification of rye dose was performed based on the median of the doses. CI: confidence interval

Effect of rye consumption on postprandial glucose

Eleven studies, including 12 trials [23, 27–36] examined the effect of rye products on postprandial glucose (Fig. 2). No significant effect was detected in the pooled analysis (WMD = 0.26 mmol/l, 95% CI: −0.15, 0.67 mmol/l, P = 0.211; I² = 73.0%, P < 0.001). In the subgroup analysis, participants older than 50 y (WMD = 0.93 mmol/l, 95% CI: −0.03, 1.90 mmol/l, P = 0.058; I² = 86.5%, P < 0.001; n = 5) and intervention durations ≤ 270 min (WMD = 0.48 mmol/l, 95% CI: −0.03, 0.99 mmol/l, P = 0.066; I² = 75.6%, P < 0.001; n = 9) had a trend for indicating an increasing effect (Table 3). However, there was high heterogeneity between trials in the overall effect and in most subgroups for this outcome. Sensitivity analysis did not detect a major influence from any single study on the pooled estimate (Supplementary Fig. 2).

Table 3.

Subgroup analysis for the effect of rye consumption on postprandial glucose (mmol/l)

Subgroups	Studies (n)	Weighted mean difference (95% CI)	P value	Heterogeneity	P for heterogeneity
Age
Youth (< 40 y)	6	0.05 (−0.22, 0.32)	0.707	0	0.877
Middle ages	1	−0.97 (−1.90, −0.04)	0.041	−	−
Aged (> 50 y)	5	0.93 (−0.03, 1.90)	0.058	86.5%	< 0.001
Sex
Males	1	0.38 (−0.55, 1.32)	0.421	−	−
Females	−	−	−	−	−
Males/Females	11	0.26 (−0.18, 0.69)	0.250	75.3%	< 0.001
BMI1
Normal-weight	7	0.24 (−0.14, 0.63)	0.211	0.53.0%	0.047
Overweight	1	−0.97 (−1.90, −0.04)	0.040	−	−
Obese	4	0.78 (−0.30, 1.86)	0.158	86.4%	< 0.001
Dose2
< 12 g	6	0.46 (−0.15, 1.08)	0.147	76.7%	0.001
≥ 12 g	6	0.09 (−0.49, 0.68)	0.759	73.2%	0.002
Intervention duration
Short (≤ 270 min)	9	0.48 (−0.03, 0.99)	0.066	75.6%	< 0.001
Long (> 270 min)	3	−0.24 (−0.77, 0.29)	0.373	44.9%	0.163
Fermentation
Yes	2	−0.06 (−0.50, 0.38)	0.800	0	0.756
No	10	0.36 (−0.14, 0.86)	0.161	77.2%	< 0.001
Total effect	12	0.26 (−0.15, 0.67)	0.211	73.0%	< 0.001

¹Normal-weight: BMI < 25 kg/m², overweight: BMI = 25–29.9 kg/m², and obese: BMI ≥ 25 kg/m²

²The classification of rye dose was performed based on the median of the doses. CI: confidence interval

Effect of rye consumption on glucose AUC

Fourteen studies, including 18 trials [27, 29, 30, 34, 36–45] examined the effect of rye consumption on glucose AUC (Fig. 2). Similar to the effect of rye on fasting and postprandial glucose, the pooled effect was not statistically significant (WMD = −0.05 mmol/l, 95% CI: −0.24, 0.14 mmol/l, P = 0.600; I² = 27.7%, P = 0.133). In the subgroup analysis, still some subgroups were with less than 3 trials, but among those with sufficient number of trials, a trend for a decreasing effect was observed in rye fiber doses ≥ 12 g (WMD = −0.22 mmol/l, 95% CI: −0.46, 0.01 mmol/l, P = 0.059; I² = 0%, P = 0.988; n = 9) (Table 4). Heterogeneity between trials was generally low. Sensitivity analysis indicated that the estimates did not depend on any single study (Supplementary Fig. 3).

Table 4.

Subgroup analysis for the effect of rye consumption on glucose area under the curve (mmol/l)

Subgroups	Studies (n)	Weighted mean difference (95% CI)	P value	Heterogeneity	P for heterogeneity
Age
Youth (< 40 y)	9	−0.08 (−0.31, 0.16)	0.521	0	0.984
Middle ages	2	0.19 (−0.47. 0.85)	0.572	57.4%	0.125
Aged (> 50 y)	7	−0.02 (−0.47, 0.44)	0.947	66.4%	0.007
Sex
Males	1	−0.29 (−1.22, 0.64)	0.546	−	−
Females	2	−0.28 (−0.73, 0.17)	0.224	0	0.695
Males/Females	15	−0.00 (−0.23, 0.22)	0.967	35.2%	0.080
BMI1
Normal-weight	11	−0.07 (−0.27, 0.14)	0.528	0	0.996
Overweight	4	−0.04 (−0.43, 0.35)	0.846	35.5%	0.199
Obese	3	0.51 (−0.98, 2.00)	0.506	88.1%	0
Dose2
< 12 g	9	0.15 (−0.20, 0.50)	0.394	55.2%	0.022
≥ 12 g	9	−0.22 (−0.46, 0.01)	0.059	0	0.988
Intervention duration
Short (≤ 270 min)	15	−0.09 (−0.31, 0.13)	0.420	29.2%	0.137
Long (> 270 min)	3	0.11 (−0.31, 0.52)	0.617	27.8%	0.250
Fermentation
Yes	2	−0.00 (−0.41, 0.40)	0.986	0	0.792
No	16	−0.06 (−0.28, 0.17)	0.617	35.9%	0.076
Total effect	18	−0.05 (−0.24, 0.14)	0.600	27.7%	0.133

¹Normal-weight: BMI < 25 kg/m², overweight: BMI = 25–29.9 kg/m², and obese: BMI ≥ 25 kg/m²

²The classification of rye dose was performed based on the median of the doses. CI: confidence interval

Effect of rye consumption on fasting insulin

Seven studies [20–22, 24–27] investigated the effect of rye consumption on fasting insulin (Fig. 3). No significant effect was observed in the overall effect (WMD = 0.24 mU/l, 95% CI: −0.46, 0.94 mU/l, P = 0.501; I² = 93.4%, P < 0.001) and the subgroups, except for normal-weight subjects, where pooled analysis of two studies showed a significant increase in fasting insulin (WMD = 0.37 mU/l, 95% CI: 0.01, 0.73 mU/l, P = 0.044; I² = 0%, P = 0.908; n = 2). However, similar to the situation for fasting glucose, some subgroups of fasting insulin did not have sufficient number of trials, and as a result a successful subgrouping could not be performed (Table 5). Heterogeneity was generally high between trials. Sensitivity analysis suggested that none of the studies significantly affected the results (Supplementary Fig. 4).

Fig. 3.

Forest plots of RCTs examining the effect of rye consumption on fasting insulin, postprandial insulin, and insulin area under the curve (AUC)

Table 5.

Subgroup analysis for the effect of rye consumption on fasting insulin (mU/l)

Subgroups	Studies (n)	Weighted mean difference (95% CI)	P value	Heterogeneity	P for heterogeneity
Age
Youth (< 40 y)	1	0.34 (−0.27, 0.95)	0.271	−	−
Middle ages	2	0.43 (−2.65, 3.51)	0.784	98.7%	< 0.001
Aged (> 50 y)	4	0.07 (−0.14, 0.28)	0.502	0	0.445
Sex
Males	2	−0.61 (−1.69, 0.48)	0.273	83.6%	0.013
Females	1	0.13 (−0.49, 0.75)	0.691	−	−
Males/Females	4	0.67 (−0.27, 1.61)	0.164	93.1%	< 0.001
BMI1
Normal-weight	2	0.37 (0.01, 0.73)	0.044	0	0.908
Overweight	2	0.05 (−0.41, 0.51)	0.827	0	0.727
Obese	3	0.28 (−1.31, 1.87)	0.733	97.8%	< 0.001
Dose2
< 12 g	0	−	−	−	−
≥ 12 g	7	0.24 (−0.46, 0.94)	0.501	93.4%	< 0.001
Intervention duration
Short (≤ 270 min)	0	−	−	−	−
Long (> 270 min)	7	0.24 (−0.46, 0.94)	0.501	93.4%	< 0.001
Fermentation
Yes	1	0.39 (−0.06, 0.83)	0.090	−	−
No	6	0.21 (−0.63, 1.06)	0.620	94.5%	< 0.001
Total effect	7	0.24 (−0.46, 0.94)	0.501	93.4%	< 0.001

¹Normal-weight: BMI < 25 kg/m², overweight: BMI = 25–29.9 kg/m², and obese: BMI ≥ 25 kg/m²

²The classification of rye dose was performed based on the median of the doses. CI: confidence interval

Effect of rye consumption on postprandial insulin

Eleven studies, including 12 trials [27–36, 46] examined the effect of rye on postprandial insulin (Fig. 3). No effect was observed in the pooled effect (WMD = 0.04 mU/l, 95% CI: −0.23, 0.30 mU/l, P = 0.759; I² = 39.1%, P = 0.080). In subgroup analysis, both dose subgroups showed a trend, although in opposite directions, for being significant: while doses < 12 g showed a trend for increased postprandial insulin (WMD = 0.33 mU/l, 95% CI: −0.04, 0.71 mU/l, P = 0.084; I² = 42.2%, P = 0.124; n = 6), doses ≥ 12 g indicated a trend for decreasing it (WMD = −0.26 mU/l, 95% CI: −0.56, 0.03 mU/l, P = 0.083; I² = 0%, P = 0.731; n = 6) (Table 6). There was low heterogeneity between studies. Sensitivity analysis did not identify any significant influence from an individual study on the pooled estimate (Supplementary Fig. 5).

Table 6.

Subgroup analysis for the effect of rye consumption on postprandial insulin (mU/l)

Subgroups	Studies (n)	Weighted mean difference (95% CI)	P value	Heterogeneity	P for heterogeneity
Age
Youth (< 40 y)	6	0.14 (−0.13, 0.42)	0.296	0	0.600
Middle ages	1	−0.39 (−1.28, 0.49)	0.383	−	−
Aged (> 50 y)	5	0.00 (−0.58, 0.59)	0.992	67.6%	0.015
Sex
Males	1	0.26 (−0.67, 1.19)	0.582	−	−
Females	1	−0.08 (−0.70, 0.54)	0.791	−	−
Males/Females	10	0.05 (−0.27, 0.35)	0.582	49.4%	0.038
BMI1
Normal-weight	8	0.05 (−0.18, 0.29)	0.659	0	0.555
Overweight	1	−0.39 (−1.28, 0.49)	0.383	−	−
Obese	3	0.27 (−0.85, 1.40)	0.634	82.4%	0.003
Dose2
< 12 g	6	0.33 (−0.04, 0.71)	0.084	42.2%	0.124
≥ 12 g	6	−0.26 (−0.56, 0.03)	0.083	0	0.731
Intervention duration
Short (≤ 270 min)	10	0.11 (−0.19, 0.42)	0.471	45.0%	0.059
Long (> 270 min)	2	−0.28 (−0.78, 0.22)	0.272	0	0.760
Fermentation
Yes	2	0.33 (−0.11, 0.77)	0.143	0	0.505
No	10	−0.03 (−0.33, 0.27)	0.853	41.4%	0.082
Total effect	12	0.04 (−0.23, 0.30)	0.759	39.1%	0.078

¹Normal-weight: BMI < 25 kg/m², overweight: BMI = 25–29.9 kg/m², and obese: BMI ≥ 25 kg/m²

²The classification of rye dose was performed based on the median of the doses. CI: confidence interval

Effect of rye consumption on insulin AUC

Thirteen studies in 16 trials [27, 29, 30, 32, 34, 36–38, 40, 41, 43, 45, 46] examined the effect of rye consumption on insulin AUC (Fig. 3). There was a significant decrease in insulin AUC associated with rye consumption in the pooled analysis (WMD = −0.48 mU/l, 95% CI: −0.66, −0.30 mU/l, P < 0.001; I² = 9.4%, P = 0.346) and most of the subgroups. However, the group of trials with < 12 g rye fiber doses (WMD = −0.27 mU/l, 95% CI: −0.68, 0.14 mU/l, P = 0.195; I² = 51.6%, P = 0.067; n = 6) and trials with obese individuals (WMD = −0.18 mU/l, 95% CI: −1.28, 0.92 mU/l, P = 0.752; I² = 80.5%, P = 0.006; n = 3) did not show a significant effect (Table 7). Heterogeneity between trials was mostly low and medium. Sensitivity analysis indicated that no individual study had a significant impact on the overall results (Supplementary Fig. 6).

Table 7.

Subgroup analysis for the effect of rye consumption on area under the curve for insulin (mU/l)

Subgroups	Studies (n)	Weighted mean difference (95% CI)	P value	Heterogeneity	P for heterogeneity
Age
Youth (< 40 y)	7	−0.40 (−0.66, −0.14)	0.002	0	0.742
Middle ages	0	−	−	−	−
Aged (> 50 y)	9	−0.53 (−0.82, −0.24)	< 0.001	35.5%	0.134
Sex
Males	1	−0.63 (−1.58, 0.32)	0.192	−	−
Females	3	−0.62 (−1.00, −0.25)	0.001	0	0.783
Males/Females	12	−0.44 (−0.67, −0.20)	< 0.001	27.8%	0.172
BMI1
Normal-weight	11	−0.49 (−0.69, −0.29)	< 0.001	0	0.845
Overweight	2	−0.57 (−1.03, −0.11)	0.016	0	0.574
Obese	3	−0.18 (−1.28, 0.92)	0.752	80.5%	0.006
Dose2
< 12 g	6	−0.27 (−0.68, 0.14)	0.195	51.6%	0.067
≥ 12 g	10	−0.60 (−0.82, −0.38)	< 0.001	0	0.933
Intervention duration
Short (≤ 270 min)	15	−0.50 (−0.69, −0.31)	< 0.001	13.1%	0.307
Long (> 270 min)	1	−0.28 (−0.89, 0.33)	0.365	−	−
Fermentation
Yes	2	−0.22 (−0.62, 0.19)	0.294	0	0.522
No	14	−0.54 (−0.74, −0.34)	< 0.001	8.1%	0.363
Total effect	16	−0.48 (−0.66, −0.30)	< 0.001	9.4%	0.346

¹Normal-weight: BMI < 25 kg/m², overweight: BMI = 25–29.9 kg/m², and obese: BMI ≥ 25 kg/m²

²The classification of rye dose was performed based on the median of the doses. CI: confidence interval

Quality assessment and risk of bias

Quality assessment of the trials revealed that 65.2% of crossover studies had good quality, but 34.8% of them had weaknesses mostly due to unclear information about blindness of participants and research staff (Supplementary Table 2). Also, 30.4% of them used < 5 days washout period or did not report such information. Out of five trials with a parallel-group design, two had a high risk of bias due to degrees of dropout which may have compromised the validity of their findings (Supplementary Table 3).

Discussion

Main findings

Results of this meta-analysis showed that rye consumption did not have a significant effect on glucose indices including fasting, postprandial, and AUC measures. Subgroup analysis did not make a difference in the results, except that there were trends for increased postprandial glucose in two subgroups: individuals aged > 50 y and short intervention lengths (≤ 270 min), as well as a trend for decreased AUC for glucose in the subgroup of rye fiber doses ≥ 12 g. Rye consumption did not show an effect on fasting and postprandial insulin but indicated a reduction in insulin AUC.

Effect on glucose measures

Despite advocacy in the literature for the beneficial effects of dietary fibers on glycemic control [15, 47], the pooled results of the RCTs included in this meta-analysis did not show significant benefits of rye, a grain with the highest fiber content among cereals [13], on glucose markers. These results are consistent with findings of another meta-analysis that found no significant reduction in blood glucose AUC following consumption of whole rye compared to endosperm rye [48].

Due to its high fiber content, rye is expected to have benefits for glycemic control. Dietary fibers in rye include arabinoxylan, cellulose, β-glucan, fructan, and lignan [49, 50]. Therefore, rye contains both types of soluble and insoluble dietary fibers. However, soluble fibers constitute approximately one third of the total fiber content in rye [14]. The beneficial effect of dietary fibers in improving glycemic control is likely due to their ability to delay gastric emptying and reduce glucose absorption rate in the small intestine. However, these effects are attributed to soluble fibers, which are relatively low in rye [14]. Thus, the low content of soluble fibers in rye may have prevented it from exhibiting beneficial effects on glucose measures. This explanation is consistent with findings of a previous meta-analysis that found no effect on glucose AUC from whole wheat and whole rye, but did find a significant effect from whole rice [48], which possesses a higher percentage of soluble fibers compared to whole wheat and whole rye [14].

Despite the lack of significant effect of rye on glucose measures, trends for increased postprandial glucose concentrations were observed in subgroups of participants aged over 50 y and with intervention lengths of up to 270 min (less than one night). Individuals older than 50 are at increased risk of hyperglycemia. Due to lower adiposity and better insulin sensitivity, younger individuals are in better glycemic control conditions, but higher rate of adiposity in older adults increases the risk of diabetes [51, 52]. Hence, the possible increased risk of hyperglycemia in older individuals could be due to their higher susceptibility to increased glucose levels following consumption of carbohydrates rather than an adverse effect of rye.

The reason for the trend of increased postprandial glucose in shorter interventions is not clear, but there were only 3 studies in the subgroup of longer interventions. Therefore, it is not clear if shorter interventions have an incremental effect on postprandial glucose when compared to longer interventions. Nevertheless, continuous consumption of dietary fibers, which occurs over extended consumptions, may alter colonic microbiota in favor of species that produce short-chain fatty acids [53, 54]. The short-chain fatty acids that are produced by colonic bacteria can enter blood circulation and stimulate secretion of the gut hormones, glucagon-like peptide-1 (GLP-1) and peptide YY (PYY) [55, 56]. These hormones delay gastric emptying and increase gut transit time, thereby slowing glucose absorption. The second meal effect could be another reason for the lower levels of postprandial glucose observed in longer interventions [57].

There was also a trend for reduced AUC for glucose in rye doses of ≥ 12 g, but not doses < 12 g. In agreement with this finding, a systematic review and meta-analysis of RCTs showed that doses ≥ 12 g fiber per day may be more effective than lower doses in improving blood glucose and lipids in pregnant women with gestational diabetes [58]. Another meta-analysis found that viscous dietary fibers at a median dose of at least 13.1 g/day were more effective for the management of type 2 diabetes [59]. However, some of meta-analyses on interventions with soluble or viscous fiber treatments have suggested lower fiber doses. For instance, in a dose–response meta-analysis, viscous fiber dosages > 8.3 g/day had a significant beneficial effect on fasting blood glucose levels in patients with type 2 diabetes [60]. Similarly, another dose–response meta-analysis recommended daily doses of 7.6–8.3 g soluble fibers for glycemic control in adults with type 2 diabetes [61]. Since the viscous part of dietary fibers is responsible for slowing intestinal glucose absorption, the net content of viscous fibers in foods is important in this context. Thus, a mixture of soluble and insoluble fibers should be consumed in larger quantities than viscous fibers alone in order to achieve the same beneficial effects.

Effect on insulin measures

Unlike glucose indices that did not show a significant change with rye consumption, insulin AUC indicated a significant decrease following rye consumption, although fasting and postprandial insulin still did not change. In line with the trend for reduced glucose AUC in rye fiber doses ≥ 12 g, reduced insulin AUC was only seen in doses ≥ 12 g rye fiber doses, further emphasizing that low doses of rye may not be effective for glycemic control.

The paradoxical finding regarding the lack of effect on glucose indices despite the reduction in insulin AUC has been repeatedly reported in the literature, where glucose concentrations have not responded to rye ingestion, but insulin levels have decreased, what is commonly known as “rye factor” [13, 50]. Östman et al., using the method of labeled glucose infusion, found that after the ingestion of wholemeal rye bread, the rate of glucose appearance in blood decreased compared to refined wheat bread, although refined wheat is not an appropriate control for comparison with whole rye [34]. However, since in the pooled analysis, we did not find an effect from rye on glucose markers, the reduction in insulin AUC may have other explanations, as described below. Reducing insulin may have benefits in prevention of aging and control of a number of diseases including obesity, type 2 diabetes, cardiovascular disease, and cancer [62].

In vitro and animal studies have shown that bioactive compounds in rye may contribute to its insulin lowering effect [13]. In this regard, lignans have shown to improve insulin resistance in ovariectomized rats [63]. Interestingly, this improvement was associated with alteration in gut microbiota composition, suggesting that gut microbiota may be involved in displaying the effects of lignans. In consolidating the role of gut microbiota in reducing circulating insulin following rye consumption, a crossover trial indicated that insulin response was lower after the intake of unfermented rye crispbread compared with sourdough-fermented rye crispbread [36]. In the colon, microbiota convert plant lignans into mammalian lignans, such as enterolactone, which can be absorbed by passive diffusion, in a similar way as short-chain fatty acids [64]. The enterolactone is the major biologically active metabolite of lignans [65] and exhibits metabolic functions [66]. A cross-sectional study on postmenopausal women showed that women with higher plasma concentrations of enterolactone had a better metabolic profile, including higher insulin sensitivity [67]. An in vitro study showed that some of lignans can activate insulin receptor substrate [68]. Alkylresorcinols, another bioactive compound in rye, have also shown to suppress insulin resistance in animals [69]. This effect may be exerted through decreasing hepatic lipid accumulation and diet-induced obesity [70].

Strengths, limitations, and future directions

This was the first meta-analysis that studied the effect of rye consumption on markers of glycemic control. Inclusion of various glycemic indicators in the search strategy and performing an extensive subgroup analysis were strengths of this work. However, we encountered some limitations. There was an insufficient number of RCTs in some subgroups (e.g., each of males and females, and individuals with overweight and obesity). Due to the nature of the treatments, blinding could not be performed in most trials. Almost all of the trials were conducted in Europe, while the results may differ for individuals from other regions. Only four trials administered fermented rye products, and thus more studies are needed to examine the effect of fermented rye on glycemic control. RCTs examining the effect of rye on HbA1c and HOMA-IR were largely scarce; these indices provide more information about glycemic control and insulin performance.

Conclusion

Overall, the results of this meta-analysis suggest that rye consumption may reduce insulin levels without affecting glucose markers. Although it cannot be stated with certainty, subgroup analysis suggests that doses of rye fiber higher than 12 g and intervention durations longer than 8 h may be more beneficial for glycemic control. The reduction of insulin levels after consumption of meals containing rye may have favorable consequences for the prevention of aging and control of metabolic diseases. However, the clinical significance of this reduction is not clear, and prospective cohorts may help clarify this issue.

Author contributions

M. G. conceived the idea for the meta-analysis and conducted the primary search and screening. S. F. and M. A. completed the search and screening processes and performed data extraction. F. G. and M. A. wrote the manuscript and prepared the figures and tables. All the authors approved the final version of the manuscript.

Funding

None.

Data availability

The datasets generated and/or analyzed during this meta-analysis are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Competing interests

The authors declare no competing interests.