The effect of vitamin C supplementation on lipid profile of type 2 diabetic patients: a systematic review and meta-analysis of clinical trials

Background and aims We conducted a systematic review and meta-analysis of clinical trials evaluating the role of vitamin C supplementation on lipid profiles among diabetic patients to summarize the available findings. Methods A comprehensive search of PubMed, ScienceDirect, Google Scholar, and Cochrane Library databases was performed. Clinical trials conducted on adult type 2 diabetic patients evaluating the effect of vitamin C supplementation and reported lipid profiles (cholesterol (TC), triglycerides (TG), low density lipoprotein (LDL), high density lipoprotein (HDL)) were included. Weighted mean difference (WMD) was calculated. Results Vitamin C supplementation had no significant effect on TC (WMD = − 4.36 mg/dl (95% CI − 10.24, 1.52) p-value = 0.146), LDL level (WMD = 2.73 mg/dl (95% CI − 1.72, 7.17) p-value = 0.229), and HDL level (WMD = 0.91 mg/dl (CI − 0.45, 2.27) p-value = 0.191). However, it reduced TG and secondary outcomes (FBS and HgA1C): TG (WMD = − 11.15 mg/dl (95% CI − 21.58, − 0.71) p-value = 0.036), FBS (WMD = − 16.94 mg/dl CI − 21.84, − 12.04, p-value = 0.000), and HgA1C (WMD = − 1.01% CI − 1.18, − 0.83, p-value = 0.001. Subgroup analysis also depicted younger patients, longer duration of treatment and higher dose were important factors. In addition, meta-regression analysis indicated the significant role of patient age, duration of treatment, supplementation dose, BMI and other baseline variables. Conclusion There is no adequate evidence to support vitamin C supplementation for dyslipidemias in diabetic patients. Specific group of patients might have benefited including younger diabetic patients. Future researches should give emphasis on the duration of treatment, the dose of vitamin C and baseline values. Supplementary Information The online version contains supplementary material available at 10.1186/s13098-021-00640-9.


Background
Association of insulin resistance with elevated total triglycerides (TG) and low-level high-density lipoprotein cholesterol (HDL-C) is well established in many former studies [1][2][3]. Lipoprotein abnormalities such as small dense low-density lipoprotein (LDL) and excess TG rich remnants were attributed to insulin resistance [4]. In light of the insulin resistance in Type 2 Diabetes Mellitus (T2DM), dyslipidemia is a common problem in these patients. Worsening dyslipidemia and inflammation overtime among T2DM patients rise concern regarding the premature development of atherosclerosis [5].
Insulin resistance with attendant increase free fatty acid flux into the liver pays a central role in triad of dyslipidemias in the form of high TG and high LDL-C and

Open Access
Diabetology & Metabolic Syndrome Tareke and Hadgu Diabetol Metab Syndr (2021) 13:24 low HDL-C [6]. Other pathogenic mechanism of diabetic dyslipidemia relates to the increased production of inflammatory cytokines such as tumor necrosis factora. These cytokines increase insulin resistance and down regulate HDL production while increasing the activity of key enzymes promoting hypertriglyceridemia [7,8].
The prevalence of dyslipidemia in diabetic patients ranges up to 90%, especially in patients with poor glycemic control [9]. Lipids are classically associated with cardio vascular disease (CVD) and they are involved in diabetic microvascular complications. Elevated total cholesterol (TC) and TG levels have deleterious effects on kidney resulted in nephropathy [10,11], whereas, increase in HDL level is suggested to be protective of renal injury [12]. Development of retinal hard exudate with diabetic macular edema has been noticed associated with elevated serum lipids [13]. Diabetic neuropathy with possible vascular etiology was pointed out as result of dyslipidemia [14]. The proportion of diabetic patients receiving lipid lowering agents is increasing [15] indicating the urge to decrease lipids in these patients.
Ascorbic acid is a water soluble vitamin with antioxidant activity, which, can also regenerate other antioxidants like vitamin E [16]. However, some in vivo studies implicated potential pro-oxidative effects of this vitamin [17,18]. Patients with diabetes have lower circulating levels of vitamin C and higher level of lipid peroxides, reports also suggest improvement in insulin action, glycemic control and endothelial function [19]. Contrary to these findings, high dose vitamin C may increase the rate of cardiovascular complications, as observed in postmenopausal women [20]. Such findings are inconclusive to synthesize scientific evidence.
Besides clinical trials conducted from 1980′s till now, we failed to retrieve a systematic review and meta-analysis study summarizing the effects of vitamin C supplementation on lipid profile of T2DM patients. Thus, we aimed to conduct a comprehensive systematic review and meta-analysis of clinical trials to summarize the available findings.

Literature searches
A comprehensive search of PubMed, ScienceDirect, Google Scholar, and Cochrane Library databases was performed. Reports were compiled according to the prefered reporting items for systematic reviews and metaanalysis (PRISMA) guideline [21]. Unpublished articles were searched from clinical trials registration platforms. Preprint articles were also retrieved from websites. Manual search was conducted by screening the reference lists of inclusive studies. The search strategy involves patients with diabetes mellitus and intervention relevant terms including vitamin C or ascorbic acid with both Medical Subject Headings (MeSH) and free text. The following search terms were used with Boolean search operator. Diabetes, "diabetes mellitus", "type 2 DM, DM, "noninsulin dependent diabetes mellitus", "late onset diabetes mellitus" vitamins, "vitamin C", "ascorbic acid", "antioxidant vitamins", "lipid profile", "lipid indices", cholesterol, "total cholesterol", triglycerides, "high density lipoproteins" HDL, "low density lipoproteins" LDL. All relevant articles published or unpublished available until the commencement of the study were included.

Study selection and outcomes Inclusion criteria
The studies were considered for inclusion if they are clinical trials, performed among adult patients with T2DM; evaluating the effect of vitamin C supplementation; reporting mean and standard deviation, median with maximum and minimum or median with 1 st and 3 rd quartiles of lipid profiles (TC, TG, LDL, HDL) of intervention and control groups. The time of intervention should be at least two weeks. Due to feasibility, articles published in English language/have English version was included. All above mentioned researches conducted on adults and before May 1, 2020 was included.

Exclusion criteria
The control group should be placebo or no treatment at all, studies which have non-placebo control group (active substance) were excluded. Cross over trials were excluded from this meta-analysis. Studies lacking control group or with non-diabetic control were excluded. This meta-analysis excluded in vitro and animal studies, review articles, studies with cohort, case control or crosssectional design, trials with combination therapy (combine vitamin C with other nutrients). Incomplete articles, conference procedings, and duplicates was excluded. The two authors independently screened the titles, abstracts and full-text of retrieved articles to identify their eligibility and disagreements were resolved by discussion and consensus among the two authors.

Outcome variables
The primary outcome of the current study was to assess the effect of vitamin C supplementation on lipid profile [measured in terms of TC, TG, HDL and LDL] among type 2 DM patients. As secondary outcomes we also evaluated the effect of vitamin C on FBS and HgA1C among those patients.

Data extraction and quality assessment
Data extraction were conducted by the two authors independently using a predefined data extraction format. Relevant variables extracted were study characteristics, population characteristics, intervention characteristics, and outcomes (TC, TG, LDL, HDL, FBS and HgA1C). Disagreement on data extraction between the two investigators was solved by discussion and consensus. The risk of bias of inclusive studies were assessed in accordance with the Cochrane Collaboration Risk of Bias Tool [22]. The risk of bias of individual study was rated as low, moderate, or high.

Statistical analysis
Since the outcome variables were continuous variables, weighted mean difference (WMD) with 95% confidence interval (CI) was calculated using random effects, inverse variance model. I squared statistic (I 2 ) was used to measure heterogeneity, with a I 2 > 50% representing notable heterogeneity [23]. Subgroup analysis was performed based on the dose of vitamin C, duration of supplementation and participants mean age. For statistical feasibility (considering the number of studies in each subgroup) the dose of vitamin C was categorized in to two; (1) less than 1000 mg/day and (2) greater than or equal to 1000 mg/day. The duration of supplementation was dichotomized in to less than 12 weeks and greater than or equal to 12 weeks. Whereas, the mean age of participants was categorized after calculating grand mean, which was 52.87 years. Studies reported median with maximum and minimum or median with 1st and 3rd quartiles of lipid profiles (TC, TG, LDL, HDL) and FBS or HgA1C of intervention and . . . .  control groups converted into means and standard deviation according to Wan et al. [24]. For studies which have more than one experimental arms, the sample size of the control group is adjusted by dividing the control group sample by the number of experimental arms.
To identify possible covariates, meta-regression analysis was performed for each primary and secondary outcome of the study. Possible covariates were baseline measurements (Tc, TG, HDL, LDL, FBS, HgA1C, body mass index (BMI)), dose, duration and age. For detection of robustness of the results sensitivity analysis was conducted by sequential elimination of each study from the pool. The sensitivity analysis was conducted for each primary and secondary outcome. Potential publication bias was assessed by visual inspection of funnel plot, and considering the number of studies for each outcome we claimed egger's regression test for bias detection. All statistical analysis was performed using STATA software Version 16. (Statacorp, Texas, USA) with p-value P < 0.05 indicating a statistically significant difference of twosided test.

Characteristics of included studies
Eleven studies [25][26][27][28][29][30][31][32][33][34][35], conducted between 2003 and 2020 included a total of 606 T2DM patients and their mean age of participants in the studies ranged 37.8-60 years. The dose of vitamin C supplementation ranged 200 mg/day to 2000 mg/day. The minimum duration of supplementation was 4 weeks and the maximum were 52 weeks. All 11 studies reported TC, LDL, HDL and FBS, and 10 studies reported TG and HgA1C. Characteristics of included studies summarized in Table 1.

Vitamin C and TG
Vitamin C supplementation to T2DM patients significantly reduce serum TG with notable heterogeneity, WMD = − 11.15 mg/dl (95% CI − 21.58, − 0.71) p-value = 0.036 and I 2 = 85.7 (Fig. 4). Subgroup analysis also pointed out that the reduction was higher with longer duration, higher daily dose and younger patients.  (Fig. 5). Sensitivity analysis for TG indicated no single study had significant influence on the outcome (Additional file 1: Fig. S2 and Table S2).

Vitamin C and LDL
Generally, vitamin C supplementation to diabetic patients didn't reduce LDL level (Fig. 6). But, findings from subgroup analysis showed opposite effects; the lower the dose of vitamin C, the longer the duration of supplementation, or the younger the patients there were a significant reduction in LDL level. Higher dose, shorter duration and older age vitamin C supplementation significantly increased LDL level ( Table 2). Meta-regression analysis of LDL and covariates (Table 3), showed mean age, dose of vitamin C, duration of supplementation and HgA1C were significant covariates. Vitamin C supplementation better works for younger patients, with smaller dose, longer period of treatment and higher HgA1C level (Fig. 7). In sensitivity analysis dropping a single study can change the effect size of the study, WMD = − 10.00 CI − 15.18, − 4.83 and there was a significant reduction of LDL level due to supplementation of vitamin C (Additional file 1: Fig. S3 and Table S3).  Table 2). Moreover meta-regression analysis results showed baseline variables and dose have no influence on the response of patients regarding HDL. Younger age and longer duration of supplementation had better responses; (β = − 1.03, CI − 1.63, − 0.42 p-value = 0.004) and (β = 0.59 CI 0.29, 0.89 p-value = 0.001) respectively (Fig. 9). Sensitivity analysis detected one study which have much influence and elimination of this study showed the beneficial effect of vitamin C to boost HDL level [WMD = 1.63 CI 0.21, 3.05] as shown in Additional file 1: Fig. S4 and Table S4.

Secondary outcomes
There were a significant reduction in FBS (Fig. 10) and HgA1C level (Fig. 11), (WMD = − 16.94 mg/dl CI − 21.84, − 12.04, p-value = 0.000 and I 2 = 80.3%), and (WMD = − 1.01% CI − 1.18, − 0.83, p-value = 0.001 and I 2 = 89.5%) respectively. Subgroup analysis for secondary outcomes revealed both FBS and HgA1C decreased significantly in both lower and higher doses. While we observe a significant reduction in HgA1C regardless of age and duration categorization, the FBS reduction was seen in longer duration of supplementation and younger patients ( Table 2). In meta-regression  analysis, age, duration and BMI was significant covariates for FBS (Fig. 12). Again, age, duration, baseline FBS and BMI were factors influencing HgA1C (Fig. 13). Diabetic patients having lower baseline FBS responds well for vitamin C supplementation. Reduction in FBS was associated with younger (β = 2.17 CI 0.

Publication bias
Visual inspection of funnel plot for all outcomes showed no asymmetry, given in (Additional file 1: Figs. S7-S12). Egger's regression test also indicated no significant publication bias for all outcomes (Additional file 1: Table S7).

Discussion
Vitamin C supplementation had no significant effect on TC, LDL and HDL. However, it reduced TG and secondary outcomes (FBS and HgA1C). Subgroup analysis also depicted younger patients, longer duration of treatment and higher dose were important factors in response of vitamin C supplementation. Despite insufficient evidence for lipid lowering capacity of vitamin C supplementation in T2DM patients, this meta-analysis gave a clue about specific group of population that might be benefited from this vitamin. Specifically, younger diabetic patients were responsive for the supplementation and likely to be benefited. In addition, meta-regression analysis indicated the A systematic review and meta-analysis previously showed insignificant role of vitamin C on lipid profile of patients including diabetic and healthy individuals generally [36]. With the same scenario, the former study also demonstrated specific groups of population benefited from vitamin C, especially lower baseline vitamin C status and individuals with higher risk of cardiovascular diseases. McRae in 2008 noted a significant reduction in serum LDL and TG without statistically significant increment in HDL [37]. Although, there are differences in the study populations, there is no concrete previous evidence both to support or rule out the role of vitamin C in management of lipid abnormalities.
In T2DM, exaggerated free radical activity and lipid peroxidation have been demonstrated, such enhanced oxidative stress was associated with poor glycemic control and appearance other metabolic complications including dyslipidemias. Vitamin C, a known anti-oxidant agent, exert its protective effect by scavenging free radicals [38,39]. It also presents in lower plasma and intracellular concentrations in diabetic patients than healthy controls [40]. Previous researches indicated daily vitamin C supplementation protects against oxidative stress in diabetic patients, macro and microangiopathy, and improve metabolic control [41]. These metabolic benefits could interfere with the lipid metabolic pathway and could reduce diabetic associated dyslipidemias.
In all four primary outcomes, the positive effects of vitamin C supplementation was witnessed in younger patients. Along with the age related pharmacokinetic and pharmacodynamic changes of this water soluble vitamin [42], one explanation could be age related difference in baseline vitamin C concentration. This metaanalysis doesn't measure the role of baseline vitamin C concentration or plasma vitamin C concentration due to insufficient data from original studies. Regarding entry to trial vitamin C concentration, older individuals have lower plasma vitamin C compared with younger patients [43]. Also younger individuals may have ascorbic acid recycling capacity to balance their redox system and to respond to the supplementation [44]. The other possible reason might be less efficient absorption in older individuals [45]. Contrary to the notion that lower baseline vitamin C in older patients could contribute to the no effect is the finding of A.W. Ashor et al. [36], indicated individuals with lower baseline vitamin C responds well. It should also be noted that the younger and older division of the population in this study is arbitrary, merely based on participants mean age. The duration of treatment might provide a clue to reach optimum plasma concentration of vitamin C for physiologic effects. The intestinal absorption of vitamin C is a saturable, dose dependent mechanism mainly by sodium dependent active transport systems [46]. It is also suggested that higher dose of vitamin C is necessary to maintain normal plasma concentration of this substance in critically ill individuals [47]. The response of diabetic patients to higher dose of vitamin C seems to be associated with the intestinal availability and the normal plasma concentration. Despite the deficiency of this meta-analysis in failing to consider plasma concentration of vitamin C, previous researches suggested the crucial impact of plasma vitamin C concentration in its physiologic effect [48].
Although it is well established that vitamin C supplementation reduces FBS and HgA1C, the reduction effect on TG is not settled. In systematic review and meta-analysis [48], vitamin C had crucial role for glycemic control especially in patients with diabetes.
Apart from quality (risk of bias given in the Additional file 1) and limitations from the primary studies, this meta-analysis had certain drawbacks. First, the baseline plasma vitamin C or concentration vitamin C at the end was not evaluated. Reports indicated the influence of plasma concentration [besides the dose of supplementation] on physiologic effects, readers should bear in mind that the reported results were not related with plasma concentration of vitamin C at baseline or final. While the patients are diabetic, they are on different anti-diabetic drugs. It is worth mentioning that different drugs may have influence in vitamin C effects, this research didn't consider specific diabetic drugs and the duration of treatment of these drugs. Other conditions that might alter the results of this research were the duration of diabetes and plasma insulin level. All these factors were not evaluated due inconsistent report or absence in the original studies.

Conclusion
There is no sufficient evidence to support vitamin C supplementation for dyslipidemias in T2DM patients. Specific group of patients might have benefited including younger diabetic patients. Future researches should give emphasis on the duration of treatment, the dose of vitamin C and baseline values.
Additional file 1. Table S1 Sensitivity analysis of the effect of vitamin C on cholesterol in diabetic patients, indicates dropping Tousoulis et al. [35] gives significant reduction in cholesterol level. Figure S1 Cholesterol. Table S2 Sensitivity analysis for triglycerides. Figure S2 TG. Table S3 Sensitivity analysis for LDL, shows the effect of vitamin C in LDL reduction by omitting Tousoulis et al. (2007). Figure S3 LDL. Table S4 Sensitivity of analysis for HDL, no single study was influencing the outcome. Figure S4 HDL. Table S5 Sensitivity analysis for FBS for detection of robustness of study. Figure S5 FBS. Table S6 Sensitivity of analysis for HgA1C, no single study was influencing the outcome. Figure S7 Funnel plot for cholesterol, showing no asymmetry. Figure S8 Funnel plot for triglycerides, showing no asymmetry. Figure S9 Funnel plot for LDL, showing no asymmetry. Figure S10 Funnel plot for HDL, showing no asymmetry. Figure S11 Funnel plot for FBS, showing no visual asymmetry. Figure S12 Funnel plot for HgA1C, showing no asymmetry.