Cardiovascular health and cancer risk associated with plant based diets: An umbrella review

Group 1: Cardiovascular endpoints and risk factors

I. Total cholesterol (TC). Eight studies examined the levels of total serum cholesterol (TC) in vegetarians. Two focused on the general population and included 5,561 [14] and 576 [15] respectively, and, based on data meta-analysis, found a significant reduction in TC among vegetarians and people who assumed plant-based proteins (MD: -1.56 mmol/L; 95%CI: −1.73, −1.39; and -0.11 mmol/L; 95%CI: −0.22, −0.01, respectively). Data were confirmed by Wang et al. (N = 832 total; Ovolacto/lacto-vegetarians: 291) [16], showing a greater dietary effect in subjects with a BMI ranging from 18.5 to 25 kg/m2 (mean TC reduction: −0.94 mmol/L; 95%CI: −1.33, −0.55), and from 25 to 30 kg/m2 (−0.58 mmol/L; 95%CI: −0.89, −0.27), than in those with a BMI >30 kg/m2 (−0.16 mmol/L; 95%CI: −0.30, −0.01), and by Xu et al. (N = 783) [17], reporting lower TC in overweight and obese people (WMD: −0.37 mmol/L; 95%CI: −0.52, −0.22) adopting a vegetarian diet. Another systematic review by Elliott et al., including 27 randomized controlled trials on plant based vs. normal western diets [18], found lower TC levels in vegetarians. These results were in line with other two descriptive reviews, the first including 2,890 overweight/obese adults [19], the second 8,969 vegetarian children aged 0–18 years [20]. Furthermore, a meta-analysis by Liang et al. described significantly lower TC (from -0.36 to -0.24 mmol/L) in people adopting plant based diets vs. people adopting western habitual diets [21]. Moreover, the review and meta-analysis by Dinu et al. [14], based on 19 studies for a total of 1,272 adults, reported significantly lower levels of TC among vegans than in omnivores (WMD: −1.72 mmol/L; 95%CI: −1.93, −1.51).

II. High-density lipoprotein cholesterol (HDL-C). Eight reviews focused on the effects of vegetarian diet on serum high-density lipoprotein cholesterol (HDL-C) levels. Six [15, 17, 18, 21–23] found no significant difference between vegetarians and omnivores, when considering normal weight and overweight/obese people. On the contrary, the study by Dinu et al. [14], based on 51 studies, for a total of 6,194 vegetarian adults, reported a WMD −0.15 mmol/L (95%CI: −0.19, −0.11). Liang et al. [21] analyzed 4 studies and reported a pooled estimated MD of −0.10 mmol/L (95%CI: −0.14, −0.05; p<0.001) in vegetarian diet adopters vs. western diets adopters. Finally, Zhang et al. [22] did not find any statistically significant differences in HDL-C levels when assessing vegetarian diets compared to non-vegetarians; on the same note Dinu et al. [14], analyzing data from 15 studies, for a total of 1,175 adults, found no significant differences in HDL-C levels between vegans and people following other dietary regimens.

III. Low-density lipoprotein cholesterol (LDL-C). Ten reviews summarized the effect of vegetarian diets on serum levels of low-density lipoprotein cholesterol (LDL-C). Seven [14–18, 21, 23] found significantly lower LDL-C levels associated with vegetarian diet, both in the general population and in diabetic patients. In particular, Elliot et al. [18], analyzing 43 observational and interventional studies, described lower LDL-C in people adopting plant based diets; a significant difference was reported by the study of Liang et al. [21] based on 68 studies (MD: -0.29 to -0.17), and similar to data by Lamberg et al. [15], based on 13 RCTs including for a total of 576 participants (MD: -0.14 mmol/L; 95%CI: -0.25, -0.02). The impact of vegetarian diet appeared even greater in overweight or obese people, according to the analysis by Xu et al. [17], based on 7 RCTs (N = 783; MD: -0.31 mmol/L; 95%CI: -0.46, -0.16). Two reviews [19, 20] reported similar results in overweight/obese patients and children aged 0–18 years, but no meta-analyses were conducted. Wang et al. [16] reported a MD of −0.34 mmol/L (95%CI: −0.57, −0.11; p<0.001) in the general adult population. Ferdowsian et al. [23] reported an overall reduction of LDL-C associated with vegetarian diet, but no synthesis analyses were performed. Dinu et al. [14] analyzed 46 studies encompassing 5,583 vegetarians and found a WMD of -1.18 mmol/L (95%CI: -1.34, -1.01). Finally, Viguiliouk et al. [24] reported a MD of −0.12 mmol/L (95%CI: −0.20, −0.04) in 6 trials involving 602 diabetic patients. Four reviews identified a significant reduction in LDL-C in vegans as compared to omnivores [14, 19, 23, 25]. Benatar et al. [25] analyzed 31 studies, for a total of 3,355 healthy vegan adults and 53,393 non-vegan controls and found MD of -0.49 mmol/L (95%CI: -0.62, -0.36; p<0.0001). Ferdowsian et al. [23] reported a reduction of LDL-C in healthy vegans, and Ivanova et al. [19] in overweight patients, but no meta-analysis was performed. Finally, Dinu et al. [14] analyzed 13 studies, for a total of 728 healthy vegan adults, and found a significant LDL-C reduction (WMD: −1.27 mmol/L; 95%CI: −1.66, −0.88).

IV. Triglycerides (TG). Seven systematic reviews [14, 16–18, 20, 23, 26] analyzed serum triglycerides (TG) in vegetarians vs. omnivores. Specifically, Wang et al. [16] described no differences between the two, with a pooled estimated effect of 0.04 mmol/L (95%CI: −0.05, 0.13; p = 0.4). Zhang et al. [26] analyzing 12 studies for a total of 1,300 subjects, found a MD of −1.28 mmol/L (95%CI; −2.14, −0.42). Schürmann et al. and Ferdowsian et al. [20, 23] reported lower TG in vegetarians in both children and adults but did not perform data meta-analysis. Dinu et al. [14] analyzed 55 studies including 4,008 vegetarians and found a WMD of −0.63 mmol/L (95%CI: −0.97, −0.30; p = 0.02). Conversely, in the review by Elliott et al. [18] no differences were reported in triglycerides. Xu et al. [17] reported a significant increase of TG (WMD: 0.29 mmol/L; 95%CI: 0.11, 0.47) in vegetarians as compared to meat eaters. The effect of vegan diet on TG remains debated as one review [23] reported significative changes in TGs (-0.14 mmol/L, CI -0.24 to -0.05), while another [14] did not find any differences between vegans and omnivores since, after having analyzed 13 studies for 483 vegans, they reported a WMD of -0.52 mmol/L (95%CI: -1.13; 0.09).

V. C-reactive protein (CRP). Three studies reported lower C-reactive protein (CRP) levels in normal weight, overweight and obese vegetarians as compared to non-vegetarians. Craddock et al. and Menzel et al. reported a WMD of -0.61 mg/L (95%CI: -0.91, -0.32; p = 0.0001) [27]; -0.25 mg/L (95%CI: -0.49, 0; p = 0.05) [28], respectively. Data derived from the analysis by Menzel et al. [28] in vegan subjects were in line with previously mentioned studies performed in vegetarians (WMD: -0.54 mg/L; 95%CI: -0.79, -0.28; p<0.0001). Two reviews [29, 30] focused on the effects of mixed vegetarian diets on CRP levels. The first [29] included 2,689 obese patients and found a WMD of -0.55 mg/L (95%CI: -0.78, -0.32; I2 = 94.4%), while the other [30], based on 2,398 normal weight subjects found no significant differences between vegetarians and omnivores in the primary analysis; alas, when considering a minimum duration of two years vegetarianism they described lower CRP levels vs. omnivores (Hedges’ g = -0.29; 95%CI: -0.59, 0.01).

VI. Plant-based diets and lipids. Three studies [23, 26, 31] assessed the lipid profile in people following plant-based diets (without differentiating among diet subtypes) in comparison with omnivores. All of them found significantly lower levels of TC, HDL-C and LDL-C in subjects following plant-based diets. Specifically, Yokoyama et al. [31] reported a WMD of −1.62 mmol/L (95%CI: −1.92, −1.32; p< 0.001; I2 = 81.4) for TC, −1.27 mmol/L (95%CI: −1.55, −0.99; p< 0.001; I2 = 83.3) for LDL-C, −0.2 mmol/L (95%CI: −0.26, −0.14; p< 0.001; I2 = 49.7) for HDL-C, and −0.36 mmol/L; 95%CI: −0.78, 0.06; p = 0.092; I2 = 83.0) for TG when considering observational studies, and of −0.69 mmol/L (95%CI: −0.99, −0.4; p<0.001; I2 = 54.8) for TC, −0.69 mmol/L (95%CI: −0.98, −0.37; p<0.001; I2 = 79.2) for LDL-C, −0.19 mmol/L (95%CI: −0.24, −0.14; p<0.001; I2 = 8.5) for HDL-C, and a non-statistically significant increase of TG based on prospective cohort studies. Additionally, Zhang et al. [26] in their meta-analysis, including 1,300 subjects, found a SMD of -1.28 mmol/L in TG (95% CI -2.14 to -0.42). Finally, Picasso et al. [32] did not find any differences in triglycerides for mixed vegetarian diets (MD: 0.04 mmol/L; 95%CI: -0.09, 0.28), but did find statistically significant differences in HDL-C (MD: -0.05 mmol/L; 95%CI: -0.07, -0.03).

VII. Blood pressure. A. Systolic blood pressure (SBP). Various studies found significantly lower mean levels of systolic blood pressure (SBP) levels in vegetarians compared to the general population [33–36]. Specifically, Gibbs et al. [33] reported a SMD of -5.47 mmHg (95%CI: -7.60, -3.34; p<0.00001) in ovo-lacto-vegetarians, as did Lee et al. [34] reporting a SMD of -1.75 mmHg (95%CI: -5.38, 1.88; p = 0.05); furthermore, they reported a SBP decreased by -2.66 mmHg (95%CI: -3.76, -1.55), in people adopting generic vegetarian diets. Moreover, Garbett et al. [35] reported a 33% lower prevalence of hypertension in vegetarians vs. nonvegetarians. On the contrary, Schwingshackl et al. [36], analyzing data from 67 clinical trials overall including 17,230 pre-hypertensive and hypertensive adult patients with a BMI between 23.6 and 45.4 kg/m2, followed for 3 to 48 months, did not find any significant reductions in SBP associated with vegetarian diet. Four reviews investigated the differences in SBP between vegans and non-vegans. Benatar et al. and Lee et al. [25, 34] reported significantly lower mean SBP levels in vegans vs. omnivores (MD: -2.56 mmHg; 95%CI: -4.66, -0.45; and WMD: -3.12 mmHg; 95%CI: -4.54, -1.70; p<0.001, respectively). On the other hand, Gibbs et al. [-1.30 mmHg (95%CI: -3.90,1.29)] and Lopez et al. (-1.33 mmHg; 95%CI: −3.50, 0.84; P = 0.230) [33, 37] did not find any significant difference in mean SBP levels between vegans and omnivores. Both reviews [32, 38] focusing on SBP in mixed-plant-based dietary patterns found significantly lower levels in vegetarians than in omnivores. The meta-analysis by Picasso et al. [32], based on 4 RCTs did not find any differences, alas, analyzing 42 cross sectional studies, they described a MD of -4.18 mmHg (95%CI -5.57, -2.80; p<0.00001), in agreement with Yokoyama et al. [38], who reported a MD of -4.8 mmHg (95%CI: -6.6, -3.1; p<0.001; I2 = 0) according to the 7 controlled trials, 6 of which being randomized (311 participants), included in the analysis, and of -6.9 mmHg (95%CI: -9.1, -4.7; p<0.001; I2 = 91.4) based on the other 32 observational studies (21,604 participants). B. Diastolic blood pressure (DBP). Garbett et al. [35] reported reduced mean diastolic blood pressure (DBP) values in vegetarians vs. omnivores, confirmed by the analysis of Gibbs et al. [33] (WMD: –2.49 mmHg; 95%CI: –4.17, –0.80; p = 0.004; I2 = 0%) in ovo-lacto-vegetarians, by Lee et al. [34] [WMD: -1.69 mmHg (95%CI: -2.97, -0.41; p<0.001)] who included 15 randomized controlled trials (N = 856) performed in vegetarians; and by Yokoyama et al. [38], who highlighted a MD -2.2 mmHg (95%CI: -3.5, -1.0; p<0.001; I2 = 0%) and -4.7 mmHg (95%CI: -6.3, -3.1; p<0.001; I2 = 92.6%) according to data from 7 controlled trials (N = 311) and 32 observational studies (N = 21,604), respectively. Conversely, Schwingshackl et al. [36] did not find significant differences between vegetarians and non-vegetarians. Three reviews [25, 34, 37] examined the impact of vegan vs. non-vegan diet on DBP and described statistically significant reductions. Benatar et al. described reduction of DBP, corresponding to a MD of -1.33 mmHg (95%CI: -2.67, -0.02) [25]. Lee et al. described a reduction in DBP of a WMD of -1.92 mmHg (95%CI: -3.18, -0.66; p<0.001) [34]. Finally, Lopez et al. [37] described the same reduction amounting to WMD: -4.10 mmHg (95%CI: -8.14, -0.06). Four studies agreed upon the lower mean DBP levels in subjects following mixed vegetarian diets as compared to omnivores [32–34, 38], quantified as MD -3.03 mmHg (95%CI: -4.93, 1.13; p = 0.002) by Picasso et al. [32], and −2.2 mmHg (95%CI: −3.5, −1.0; p<0.001) and −4.7 mmHg (95%CI: −6.3, −3.1; p <0.001) by the analysis performed on clinical trials and observational studies, respectively, by Yokoyama et al. [38].

VIII. Body weight and body mass index (BMI). Berkow et al. [39] identified 40 observational studies comparing weight status of vegetarians vs. non-vegetarians: 29 reported that weight/BMI of vegetarians of both genders, different ethnicities (i.e., African Americans, Nigerians, Caucasians and Asians), and from widely separated geographic areas, was significantly lower than that of non-vegetarians, while the other 11 did not find significant differences between the two groups. In female vegetarians, weight was 2.9 to 10.6 kg (6% to 17%) and BMI 2.7% to 15.0% lower than female non-vegetarians, while the weight of male vegetarians was 4.6 to 12.6 kg (8% to 17%) lower and the BMI 4.6% to 16.3% lower than that of male non-vegetarians. The review by Schürmann et al. [20], focusing on 8,969 children aged 0–18 years old found similar body weight in both vegetarian and vegan children as compared to omnivore ones. Dinu et al. [14] analyzed data from 71 studies (including 57,724 vegetarians and 199,230 omnivores) and identified a WMD BMI of -1.49 kg/m2 (95%CI: -1,72, -1,25; p<0.0001) in vegetarians when compared to omnivores. Barnard et al. [40] found a significant reduction in weight in pure ovolactovegetarians (−2.9 kg; 95% CI −4.1 to −1.6; P<0.0001), compared to non-vegetarians from control groups; furthermore, they found in vegans the mean effect was of -3.2 kg (95% CI: -4.0;-2.4, P: <0.0001); overall they included 490 subjects in their analysis, excluding subjects who did not complete the trials. Benatar et al. [25]–including 12,619 vegans and 179,630 omnivores from 40 observation studies–and Dinu et al. [14]–based on 19 cross sectional studies, for a total of 8,376 vegans and 123,292 omnivores–reported the same exact result, with a mean lower BMI in vegans vs omnivores, equal to -1.72 kg/m2 (95%CI: -2.30, -1.16) and -1.72 kg/m2 (95%CI: -2.21,-1.22; p<0.0001), respectively. The meta-analysis by Long et al. [41], performed on 27 studies, reported a MD of -0.70 kg/m2 (95%CI: -1.38, -0.01) for BMI in vegans vs. omnivores. A systematic review and meta-analysis by Agnoli et al. [42] found mean BMI to be lower in subjects adhering to mixed vegetarian diets as compared to omnivores. Additionally, Tran et al. [43] described weight reductions in clinically healthy patients, as well as in people who underwent vegetarian diets as a prescription, but no meta-analysis was performed. Finally, Huang et al. [44] found significant differences in both vegans and vegetarians, who were found to have lost weight after having adopted the diet as a consequence of being assigned to the intervention group in their randomized studies. For vegetarians the WMD was -2.02 kg (95%CI: -2.80 to -1.23), when compared to mixed diets, and for vegans the WMD was -2.52 kg (95%CI: -3.02 to -1.98), when compared to vegetarians.

IX. Glucose metabolism. Viguiliouk et al. [24] found a significant reduction in HbA1c (MD: −0.29%; 95%CI: −0.45, −0.12) and fasting glucose (MD: −0.56 mmol/L; 95%CI: −0.99, −0.13) in vegetarians vs. non-vegetarians. The meta-analysis by Dinu et al. [14], reported for vegetarians (2256) vs omnivores (2192) WMD: -0.28 mmol/L (95%CI: -0.33, -0.23) in fasting blood glucose. These findings were confirmed by Picasso et al. [32] who found a MD of -0.26 mmol/L (95% CI: -0.35, -0.17) in fasting glucose in mixed-vegetarian diets as compared to omnivores. A meta-analysis by Long et al. [41], based of 27 cross sectional studies, showed a MD for homeostasis model assessment of insulin resistance -measured as HOMA-IR, a unitless measure ideally less than one- of -0.75 (95%CI: -1.08, -0.42), fasting plasma glucose in vegetarians who adhered also to an exercise intervention as compared to omnivores. Lee & Park [45] reported a significantly lower diabetes risk (OR 0.73; 95%CI: 0.61, 0.87; p<0.001) in vegetarians vs. non-vegetarians, being the association stronger in studies conducted in the Western Pacific region and Europe/North America than in those from Southeast Asia. Regarding vegans, the review by Benatar et al. [25] determined a mean reduction of 0.23 mmol/L (95%CI: -0.35, -0.10) of fasting blood glucose in vegans (N = 12,619) as compared to omnivores (N = 179,630). The finding was in line with Dinu et al. [14], who reported a WMD of -0.35 mmol/L (95%CI: -0.69, -0.02; p = 0.04) of fasting blood glucose in vegans (n = 83) than omnivores (n = 125). A systematic review, finally, including 61 studies [42] found mean values of fasting plasma glucose, and T2D risk to be lower in subjects following mixed vegetarian diets as compared to omnivores.