Diabetes is a chronic disease and one of the most prevalent public health concerns globally. The current number of people suffering from diabetes exceeds 425 million in the world, and there is still a large number of people who remain undiagnosed [1
]. Furthermore, many people do not meet the accepted criteria for diabetes [2
], yet their blood glucose test results are too high to be considered as normal and, thus, they are diagnosed as being in the prediabetes stage. These subjects are very likely to go on to develop type 2 diabetes in the coming years if they do not change their eating habits. Diabetes and prediabetes can be screened based on plasma glucose criteria, either with fasting plasma glucose, 2-h glucose concentrations after an oral glucose tolerance test (OGTT), or glycated hemoglobin (HbA1c) criteria [1
The American Diabetes Association’s (ADA) recommendations strongly emphasize that there is no single eating pattern that is best for those with diabetes; however, they do suggest that the diet should mainly be based on products with a low glycemic index and should exclude refined sugars and processed food [2
]. Many studies revealed that diets high in protein can improve diabetes-related parameters. For example, after a three-month intervention, Luger et al. [3
] observed improvements in losing weight, fasting glucose, and insulin concentrations in a group of subjects consuming 30% energy from protein in comparison with the control group that consumed 15% energy from protein. Results were comparable with a meta-analysis in which the participants with high-protein eating plans had 2-kg greater weight loss and 0.5% greater improvement in HbA1c [4
]. The consumption of carbohydrate foods is also a crucial factor in patients with diabetes or in the pre-diabetes stage. Foods containing carbohydrates have a wide range of effects on the glycemic response because they differ in their proportions of sugars, starches, and fiber [5
The Paleolithic diet, also known as a hunter–gatherer diet or stone-age diet, is an estimated nutritional pattern considered to be typical of people living during the paleolithic era, from approximately 2.5 million to 10,000 years ago. It encourages the consumption of meat, fish, eggs, vegetables, fruits, roots, and nuts, while many cultivated products, such as dairy products, oils, cereals, and legumes (as well as salt and refined sugars) are excluded [6
]. A traditional Paleolithic diet contains an estimated 35% of energy from fats, 35% of energy from carbohydrates, and 30% of energy from proteins. Therefore, the Paleolithic diet typically resembles a low-carbohydrate diet. However, the hunter–gatherer diet provides a higher amount of dietary fiber (up to 45–100 g per day) than a low-carbohydrate diet [7
]. Recently, the Paleolithic diet received attention due to its possible health benefits [8
]. Paleolithic nutrition is suggested to be associated with an improvement in lipid profile and with the reduction of blood pressure [8
]. Moreover, this type of diet is suggested to have a positive impact on weight loss [10
]. Additionally, it has anti-inflammatory benefits and might reduce oxidative stress [11
]. It was also suggested that the Paleolithic diet might have a beneficial effect on carbohydrate metabolism and insulin homeostasis [8
], thus making it relevant for patients with diabetes. However, the results of studies assessing the effect of the diet on glucose and insulin levels are ambiguous, with conflicting findings being reported [10
Therefore, the aim of this meta-analysis was to compare the effect of the Paleolithic diet with other types of diets commonly perceived as healthy on glucose and insulin homeostasis in studies conducted in adults with altered glucose metabolism. Because altered glucose metabolism is frequently observed in subjects with metabolic syndrome and metabolic syndrome is also associated with a high risk of progression to type 2 diabetes mellitus [21
], this meta-analysis also included studies in which the majority of participants had at least two characteristics of metabolic syndrome.
This systematic review and meta-analysis was carried out and reported in accordance with the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA, see Supplementary Materials
]. Prior to initiating the review process, the protocol was registered with PROSPERO, registration number: CRD42019126412 [23
2.1. Search Strategy
PubMed, Web of Knowledge, Scopus, and the Cochrane Library databases were searched from time of inception until September 2019, using the following MeSH terms: (((“paleo diet” OR “paleolithic diet” OR “stone age diet” OR “caveman diet” OR “hunter–gatherer diet”) AND (“insulin” OR “insulin resistance” OR “hyperinsulinism” OR “glucose metabolism disorders” OR “hypoglycemia” OR “hyperglycemia” OR “blood glucose” OR “diabetes mellitus” OR “glucose intolerance” OR “glucose tolerance test” OR “glycated hemoglobin A” OR “prediabetic state”)) NOT “animals”). In addition, hand searches of the reference lists of included papers identified further potential studies not captured in the electronic database searches. No language restrictions were applied.
2.2. Study Selection
Original studies were included if they met the following inclusion criteria:
Types of studies: randomized controlled trial (RCTs; parallel or crossover), irrespective of publication status;
Types of interventions: Paleolithic diet (regardless of the duration of the intervention) versus another type of diet (e.g., the Mediterranean diet, diabetes diets, national dietary recommendation);
Population: studies conducted in humans with glucose metabolism disorders (diabetes mellitus (criteria for the diagnosis: fasting plasma glucose concentrations ≥126 mg/dL (7.0 mmol/L) or 2-h glucose levels ≥200 mg/dL (11.1 mmol/L) during OGTT or HbA1c ≥6.5% (in the absence of unequivocal hyperglycemia; for these parameters, diagnosis requires two abnormal test results from the same sample or in two separate test samples) or a random plasma glucose ≥200 mg/dL (11.1 mmol/L)), prediabetes state (impaired fasting glucose (fasting plasma glucose concentrations from 100 mg/dL (5.6 mmol/L) to 125 mg/dL (6.9 mmol/L)), or impaired glucose tolerance (2-h plasma glucose levels during OGTT from 140 mg/dL (7.8 mmol/L) to 199 mg/dL (11.0 mmol/L)) or HbA1c from 5.7% to 6.4%) [2
], or studies which included participants where the majority had at least two characteristics of metabolic syndrome (waist circumference ≥102 cm for men and ≥88 cm for women, triglyceride levels ≥150 mg/dL (1.7 mmol/L), high-density lipoprotein (HDL) cholesterol <40 mg/dL (1.0 mmol/L) for men and <50 mg/dL (1.3 mmol/L) for women, hypertension or blood pressure ≥130/85 mmHg, or fasting plasma glucose ≥100 mg/dL (5.6 mmol/L)) [24
], with no restrictions on age, gender, and race/ethnicity of study participants, location of study, or sample size.
The exclusion criteria were as follows:
Types of studies: non-RCTs, uncontrolled trials, observational studies (e.g., ecologic study, cohort study, case–control study, case reports, case series, editorials, commentaries, letters to the editor, qualitative research), conference papers, or publications available only in abstract form (no possible contact with authors);
Population: studies conducted in animal models or studies performed in healthy subjects or a specific group of patients (e.g., pregnant or breastfeeding women).
Any RCTs that assessed the effect of the Paleolithic diet on glucose and insulin homeostasis qualified for consideration.
2.3. Quality Assessment
Three investigators (M.J., A.J., and B.K.) evaluated each article independently in the three main stages of the extraction process (Figure 1
). Firstly, article titles were screened, followed by abstracts and finally full texts for eligibility for inclusion in the systematic review and meta-analysis. Disagreements were resolved by discussion between the investigators and included a discussion with two other authors (C.S. and J.W.) until a consensus was reached. All investigators agreed on the final decision of the studies to be included. Primary authors of relevant articles were contacted directly if the data sought were unavailable or if the study was only published in abstract form.
2.4. Data Extraction
Eligible studies were reviewed, and the following data were independently extracted by three authors (M.J., A.J., and B.K.):
General information: first author’s name, publication year, country;
Study characteristics: study design and method of blinding;
Characteristics of study participants: sample size (total sample size and number of subjects in each group), age, sex, body mass index (BMI), body weight, ethnicity, and health status (diabetes mellitus type 1, diabetes mellitus type 2, impaired fasting glucose, and impaired glucose tolerance or other);
Type of intervention: type of diet, the macronutrient composition of diet (the energy value of the diet, percentage energy from carbohydrate, protein and fat, dietary fiber intake (g/day)), recommended and excluded food products, time of intervention, duration of intervention;
Pre- and post-intervention fasting glucose and insulin levels, HbA1c values, the area under the curve (AUC; 0–120 min) for glucose and insulin during OGTT and homeostasis model assessment of insulin resistance (HOMA-IR).
2.5. Risk of Bias
Risk of bias was assessed independently by two authors (M.J. and A.J.) using the Cochrane Collaboration tool, where the following domains are included: selection bias, performance bias, detection bias, attrition bias, reporting bias, and other bias. Criteria for low risk, unclear risk, and high risk of bias per the Cochrane Handbook for Systematic Reviews of Interventions were used [25
2.6. Data Analysis
Subjects were categorized according to the BMI cut-off (BMI = body weight (kg)/body height (m2
)) values proposed by the World Health Organization for adolescents and adults [26
]. According to these criteria, underweight was defined as BMI <18.5 kg/m2
, normal body weight as 18.5–24.9 kg/m2
, overweight as 25.0–29.9 kg/m2
, and obesity as ≥30.0 kg/m2
The ADA recommendations [2
] were used to assess fasting glucose and insulin levels, as well as AUC 0–120 for glucose and insulin. Impaired glucose tolerance is defined as plasma concentrations of glucose at 120 min in the OGTT ranging from 7.8 to 11.0 mmol/L, while impaired fasting glucose is defined as fasting glucose levels from 5.6 to 6.9 mmol/L, normal glucose tolerance is defined as glucose levels at 120 min in the OGTT <7.8 mmol/L, and normal fasting glucose is defined as fasting glucose levels ranging from 3.9 to 5.5 mmol/L. Diabetes mellitus was diagnosed when fasting glucose levels were ≥7.0 mmol/L or glucose levels at 120 min in the OGTT were ≥11.1 mmol/L or HbA1c ≥6.5%. A reference range for fasting insulin of <174 pmol/L was assumed [27
]. The changes in the HOMA-IR index during the intervention period were used to assess the alterations in insulin resistance within the studied populations. According to ATP III-Met, we defined cut-off values of HOMA-IR for the diagnosis of insulin resistance as ≥1.8 [28
Our primary outcomes were post-intervention fasting glucose, insulin levels, HbA1c values, and HOMA-IR index. Our prespecified secondary outcomes were AUC 0–120 for glucose and insulin.
2.8. Statistical Analysis
Data are presented as means ± standard deviations (SD). We undertook data synthesis, including a calculation of effect sizes with 95% confidence intervals (CIs), using fixed-effects models (if no heterogeneity was present) and random-effects models (to analyze outcomes moderate and high with heterogeneity) with inverse variance weighting [29
]. A meta-analysis of the studies was carried out when at least two studies were included that analyzed data for the specific outcome. Standardized mean differences (SMDs) were used as a summary statistic to allow comparison of effect sizes across studies. The SMD is estimated from the difference between the mean outcome values of the intervention and control groups divided by the pooled SD of the outcome values. Forest plots were generated to illustrate the study-specific effect sizes along with 95% CI [25
]. Sensitivity analyses were also performed by removing each study one by one and recalculating the pooled estimates. Due to a small number of studies included in this meta-analysis, subgroup analysis was not performed.
Heterogeneity between studies was evaluated using Cochran Q statistics; p
< 0.1 indicates significant heterogeneity. The I2
test was also used to evaluate consistency between studies in which a value <25% indicates a low risk of heterogeneity, 25%–75% indicates a moderate risk of heterogeneity, and >75% indicates a high risk of heterogeneity [30
All analyses were performed using the Comprehensive Meta-Analysis software, version 3.0 (Biostat, Inc., Englewood, NJ, USA). A p-value <0.05 was considered to be statistically significant.
The present meta-analysis demonstrates that the Paleolithic diet did not differ from other types of diets commonly perceived as healthy regarding its effect on fasting glucose and insulin concentrations, AUC 0–120 glucose and AUC 0–120 insulin levels, HbA1c values, and the HOMA-IR index.
Despite the fact that, at the end of the intervention period, a decrease in fasting glucose concentrations in the Paleolithic group was observed in most of the studies included in this systematic review [9
], our meta-analysis did not show significant differences between the effect of the Paleolithic diet and control diets on fasting glucose levels. Similar results were observed in the previous meta-analysis by Manheimer et al. [8
] who also found that the Paleolithic diet did not significantly improve fasting glucose levels. Nevertheless, previous studies showed that a low-carbohydrate diet can contribute to an improvement in fasting glucose levels. In addition, this effect was pronounced in subject with type 2 diabetes [32
]. Recently, Otten et al. [34
] also observed that the Paleolithic diet decreased fasting glucose concentrations in overweight and obese subjects with type 2 diabetes mellitus; however, the authors reported no differences between patients following the Paleolithic diet with standard-care exercise recommendations and the Paleolithic diet together with a 3-h weekly supervised exercise training (−17% vs. −26%). The lack of differences between the effect of the Paleolithic diet and control diets on glucose concentrations observed in this meta-analysis might be partly explained by the negative effect of the Paleolithic diet on microbiota composition. Indeed, it is now well established that an imbalanced gut microbiota is linked to host glycemic control impairment and type 2 diabetes development [35
]. In addition, Genoni et al. [36
] suggested that long-term adherence to the Paleolithic diet may not be beneficial for gut health, due to the association with lower relative abundances of known beneficial bacterial genera, and the increased relative abundance of trimethylamine-N
-oxide producing genus Hungatella
. This is, however, only a speculation since we did not have access to data relating to the gut microbiome.
The previous meta-analysis conducted by Ghaedi et al. [37
] reported that the Paleolithic diet could significantly decrease anthropometric parameters, including body weight, waist circumference, BMI, and fat mass. These results were also confirmed by Manheimer et al. [8
], who pointed out that Paleolithic nutrition was more effective in reducing body weight in comparison to the control diet. Previous studies also reported that the Paleolithic diet is more satiating than other types of diets. The composition of the diet is likely to be an important factor in satiety and body weight management. It was suggested that the high protein content of a diet might increase satiety and weight loss [13
]. Body weight is also an important factor that might affect fasting insulin levels. In fact, body weight reduction is reported to significantly decrease fasting insulin concentrations [39
]. Here, we observed that the Paleolithic diet did not affect fasting insulin levels when compared to the control group. It is possible that these results could be explained by similar body weight changes after the intervention period observed in both groups. Indeed, in three studies included in this meta-analysis, body weight significantly decreased in both groups after the intervention period [9
]. However, two studies observed significant differences between post-intervention body weight, with lower body weight noted in the Paleolithic group [9
It is well known that fasting insulin is associated with insulin resistance, which is an essential factor in developing type 2 diabetes. Furthermore, insulin resistance is also implicated in obesity, hypertension, cancer, or autoimmune diseases [40
]. Insulin resistance is also associated with excess fat, obesity, or altered lipid profiles [41
]. High fasting insulin levels are related to the greater resistance of tissues to insulin, which is reflected through the HOMA-IR index [28
]. Several studies revealed that a diet pattern concentrated mainly on meat, fish, eggs, vegetables, fruits, berries, and nuts might be effective for improving predictors of insulin resistance such as the HOMA index [14
]. In contrast, our meta-analysis did not show a significant effect of the Paleolithic diet on the HOMA-IR index. It should be noted that insulin resistance is dependent on fasting glucose and insulin levels [44
]. Therefore, the lack of significant differences between the effect of the Paleolithic diet and the control diet on HOMA-IR can easily be explained by the lack of significant changes in fasting glucose and insulin levels [10
HbA1c is a glucose homeostasis parameter which is widely used to assess the metabolic control of diabetic subjects [2
]. It is strongly associated with severe diabetic complications [45
] and can also be used as a screening tool for subjects with prediabetes [2
]. The potential of the Paleolithic diet for a decrease in HbA1c values was observed in three studies included in this systematic review [9
]. In addition, a recent meta-analysis showed that a low-carbohydrate diet, followed by a Mediterranean diet and the Paleolithic diet, was ranked as the most optimal dietary approach for the reduction of HbA1c values [46
]. Our meta-analysis, however, did not confirm these results. It is possible that the sample size in studies included in this meta-analysis was too small or the intervention period was too short to detect significant differences between groups.
Our meta-analysis did not show any differences between the effect of the Paleolithic diet and the control diet on AUC 0–120 for glucose and insulin. Similar results were recently obtained by Otten et al. [47
] in a study conducted in healthy obese women. Otten et al. [47
] observed no difference between the effect of the Paleolithic diet and the control diet on AUC 0–120 for glucose and insulin. However, AUC 0–120 for insulin showed a tendency to decline between baseline and 24 months in both intervention groups.
Several limitations should be listed acknowledged this meta-analysis. Firstly, the number of studies included in this systematic review was relatively small, with a limited number of study participants. Secondly, there were many variations between the studies, including the ethnicity of study participants, age, various metabolic stages, methodology, and duration of the intervention period, as well as a different type of diet used by the control group in selected studies. Moreover, we observed differences in macronutrient composition amongst the included studies. Furthermore, in most included studies, there was no information about the training of the persons who performed the dietary education of the study participants [9
]. In addition, some of the referred studies at baseline showed a significant difference between groups [9
]. These factors could have influenced the findings and could partly explain why the meta-analysis data show no significant differences between the Paleolithic diet and the control diet. It is important to also note that most of the studies included in this meta-analysis were performed in Caucasian subjects with elevated BMI values. Therefore, our findings cannot be generalized to other ethnicities such as Asians, in addition to populations with lower BMI values. Moreover, we were unable to assess the long-term effect of the Paleolithic diet on glucose and insulin homeostasis.
On the other hand, the strength of this research is that it includes details on the characteristics of the study and study population, as well as the measures taken to reduce the influence of bias in the included studies. Moreover, this is the first meta-analysis to comprehensively compare the effect of the Paleolithic diet with other types of healthy diets on glucose and insulin homeostasis in subjects with altered glucose metabolism.