Efficacy of Intermittent or Continuous Very Low-Energy Diets in Overweight and Obese Individuals with Type 2 Diabetes Mellitus: A Systematic Review and Meta-Analyses

Objective This study is aimed at investigating the efficacy of a very low-energy diet (VLED) in overweight and obese individuals with type 2 diabetes mellitus (T2DM). Methods We thoroughly searched eight electronic resource databases of controlled studies concerning the efficacy and acceptability of intermittent or continuous VLEDs in patients with T2DM compared with other energy restriction interventions. Results Eighteen studies (11 randomized and seven nonrandomized controlled trials) with 911 participants were included. The meta-analyses showed that compared with a low-energy diet (LED) and mild energy restriction (MER), VLED is superior in the reduction of body weight (mean difference (MD) MDLED = −2.77, 95% confidence interval (CI) CILED = −4.81 to − 0.72, PLED = 0.008; MDMER = −6.72, 95%CIMER = −10.05 to − 3.39, PLED = 0.008; MDMER = −6.72, 95%CIMER = −10.05 to − 3.39, PLED = 0.008; MDMER = −6.72, 95%CIMER = −10.05 to − 3.39, PLED = 0.008; MDMER = −6.72, 95%CIMER = −10.05 to − 3.39, PLED = 0.008; MDMER = −6.72, 95%CIMER = −10.05 to − 3.39, PLED = 0.008; MDMER = −6.72, 95%CIMER = −10.05 to − 3.39, PLED = 0.008; MDMER = −6.72, 95%CIMER = −10.05 to − 3.39, I2 = 0%) and TG level (MD = −0.25, 95%CI = −0.55 to 0.06, PLED = 0.008; MDMER = −6.72, 95%CIMER = −10.05 to − 3.39, I2 = 0%) and TG level (MD = −0.25, 95%CI = −0.55 to 0.06, PLED = 0.008; MDMER = −6.72, 95%CIMER = −10.05 to − 3.39, PLED = 0.008; MDMER = −6.72, 95%CIMER = −10.05 to − 3.39, PLED = 0.008; MDMER = −6.72, 95%CIMER = −10.05 to − 3.39, PLED = 0.008; MDMER = −6.72, 95%CIMER = −10.05 to − 3.39, PLED = 0.008; MDMER = −6.72, 95%CIMER = −10.05 to − 3.39, Conclusion Dietary intervention through VLEDs is an effective therapy for rapid weight loss, glycemic control, and improved lipid metabolism in overweight and obese individuals with T2DM. Thus, VLEDs should be encouraged in overweight and obese individuals with T2DM who urgently need weight loss and are unsuitable or unwilling to undergo surgery. As all outcome indicators have low or extremely low quality after GRADE evaluation, further clinical trials that focus on the remission effect of VLEDs on T2DM are needed.


Introduction
It is well known that obesity is a major risk factor for type 2 diabetes mellitus (T2DM) [1] and the majority of patients with T2DM are overweight or obese [2]. Obesity management is confirmed as an effective strategy in the prevention and remission of T2DM [3].
Multiple strategies including diet, physical activity, behavioural therapy, pharmacologic therapy, and bariatric surgery are recommended for obesity management [3]. In previous evidence-based clinical guidelines, dietary modification is recommended as a fundamental aspect of diabetes care, based on its benefits on glycemia and HbA1c levels [4]. Recently, several studies suggest that short-term and more extreme dietary energy restriction aiming on intensive weight loss can even reverse some cases of T2DM [5][6][7]. Very low-energy diet (VLED) has been confirmed as an effective and safe option for weight loss in obese individuals [8]. There is no standard definition of a VLED programme across different countries and continents [9][10][11]. However, a VLED is generally defined as a very low total energy intake (≤800 kcal/day) [8,10]. Recently, a growing body of studies focus on the efficacy and acceptability of VLEDs in patients with T2DM who are overweight or obese [12][13][14] and propose that VLEDs may be an underutilized therapy for patients with T2DM. Intermittent VLED is an alternative strategy of continuous VLEDs for T2DM, which typically involves periods of VLEDs interchanged by periods of ad libitum energy intake or mild energy restriction (MER, a slight diet intervention method which provides energy less than ad libitum energy intake but more than 1600 kcal/day) [15,16]. The efficacy of both intermittent and continuous VLED should be considered.
A low-energy diet (LED) containing 800-1600 kcal/day is also considered an option of clinical obesity management of patients with T2DM [17,18], but the difference in efficacy and safety between VLEDs and LEDs is rarely discussed. Bariatric surgery is recommended for obese patients (body mass index (BMI), 35.0-39.9 kg/m 2 ) with T2DM who did not achieve durable weight loss and improvement in comorbidities with reasonable surgical methods [3]. For example, Roux-en-Y gastric bypass (RYGB), as currently one of the most effective types of bariatric surgery, achieves energy limitation by reducing stomach capacity and reducing dietary intake. However, bariatric surgeries have more adverse effects and complications compared with energy restriction strategy. Moreover, VLEDs may produce a similar effect on glycemic control, β-cell function, and insulin sensitivity as bariatric surgeries. Thus, it is necessary to evaluate the efficacy of VLEDs compared with other methods of energy restriction in overweight and obese individuals with T2DM.
A previous systematic review among overweight and obesity individuals with T2DM found that VLED has benefits of weight loss and glycemic control [19]. However, the systematic review included a small number of participants, and the long-term effect of VLEDs is unclear. Another recently published systematic review found that VLED programmes in children and adolescents with obesity induce short-to medium-term weight loss and also demonstrated significant improvements in diabetic outcomes, such as HbA1c and glucose levels [10]. Recently, several clinical studies have been conducted to compare VLEDs with other energy restriction methods. Thus, it is necessary to investigate the efficacy of VLEDs in overweight and obese adult individuals with T2DM. Our systematic review and meta-analyses are aimed at clarifying the effect of VLEDs on weight loss, glycemic control, and blood lipid levels in overweight and obese individuals with T2DM and further exploring the longterm efficacy of VLEDs to provide more substantial evidence in the clinical application of VLEDs.

Materials and Methods
2.1. Search Strategy. We comprehensively searched PubMed, EMBASE, Cochrane Library, Web of Science, SINOMED, China National Knowledge Infrastructure, WanFang, and Chongqing VIP Information databases from inception until July 2019 for clinical trials investigating intermittent or continuous VLEDs for overweight and obese adults with T2DM. Additional studies were searched in the reference lists of all identified publications, including relevant meta-analyses and systematic reviews.

Inclusion Criteria.
Published and unpublished randomized controlled trials (RCTs) and non-RCTs, which are clinical controlled studies evaluating the efficacy of intermittent or continuous VLEDs and qualitative studies exploring the acceptability of, barriers to, and facilitators of VLEDs, were considered for inclusion in this review.
We included clinical studies that satisfied the following criteria: (1) participants in the included studies were overweight or obese (mean BMI ≥ 30 kg/m 2 or ≥10% above the ideal body weight based on the Metropolitan Life Insurance Company's tables); (2) adults (aged ≥18 years) had T2DM in older studies using a different measure of obesity; (3) studies used intermittent or continuous VLEDs comprising ≤800 kcal/day in at least one intervention arm; and (4) studies also had to include a control arm receiving other energy control methods, including LEDs (800-1600 kcal/day), bariatric surgery, and MER. We excluded clinical studies with the following features: (1) both the intervention and comparator arms received VLED treatment (except VLEDs after surgical treatment) and (2) the intervention is VLED combined with other weight loss drugs. If a study compared three or more arms, VLED arms were considered to be the intervention and other energy control methods the comparators.

Data Extraction.
Two reviewers (YS Huang and XW Fu) independently extracted data from original trial reports using a standardized form. Data extracted included study characteristics (first author, publication year, single center or multicenter, sample size, intervention and control, period of treatment, and follow-up duration), characteristics of 2 Journal of Diabetes Research patients (inclusion criteria, background treatments, mean age, proportion of men, baseline weight, and baseline glucose levels), reported outcomes (weight, fasting plasma glucose levels, and adverse events), and information on methodology. We contacted the study authors when we needed to obtain additional information that was unavailable in the online publications or supplementary materials.
2.4. Quality Assessment. Risk of bias of RCTs was assessed using the Cochrane Collaboration's tool [20]. We evaluated non-RCTs according to the Risk Of Bias In Nonrandomised Studies of Interventions (ROBINS-I) tool [21]. Two investigators independently completed the assessments, and discrepancies were discussed with a third party and resolved by consensus. Additionally, the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) framework was used to assess the quality of evidence contributing to each network estimate, which characterizes the quality of a body of evidence on the basis of the study limitations, imprecision, inconsistency, indirectness, and publication bias for the primary outcomes [22].

Statistical Analyses.
The data entry and analysis were conducted using Microsoft Excel 2016 and Review Manager software version 5.3, respectively. Risk ratio and standard mean difference with 95% confidence interval (CI) of the outcomes were calculated as effect measure. The I 2 -statistic was calculated for heterogeneity, as a measure of the proportion of the overall variation that is attributable to between-study heterogeneity. A fixed-effects (FE) model was used if I 2 < 50%; otherwise, the random-effects model was used.
To assess whether the results were influenced by study characteristics (effect modifiers), a subgroup analysis was conducted according to the study duration (<12 or ≥12 months). Additionally, sensitivity analyses were performed before combining RCTs and non-RCTs in the metaanalyses to determine possible additional sources of heterogeneity and changes in effect sizes.
Publication bias was tested by visual inspection of the funnel plots. When few studies are included in the analysis, the power of the tests is too low; therefore, publication bias was only examined if >10 study comparisons were included in the analysis [23].

3.2.
Evaluation of the Risk of Bias of the Selected Studies. The risk of bias for the included RCTs was assessed using the Cochrane risk of bias tool. None of the RCTs had an overall low risk of bias. Most RCTs had unclear risk of bias for sequence generation, allocation concealment, blinding of participants, blinding of outcome, and selective reporting because no detailed information was provided. However, three studies had high risk of bias for blinding of participants and blinding of outcome assessment, and one study had high risk of bias for allocation concealment because it could not be performed. Moreover, there is incomplete outcome data that most studies had a low risk of bias. Risk of bias assessment of included trials is shown in Figure 2.
The risk of bias for the included non-RCTs according to the ROBINS-I tool is presented in Figure 3. None of the studies had a low or moderate risk of bias, six (Jackness et al. [24], Lips et al. [32], Plum et al. [33], and Paisey et al. [36][37][38]) had signs of serious bias, and one (Cinkajzlova et al. [34]) had critical bias. The domain "bias due to confounding" was a main source of critical or serious risk of bias. The domain "bias in selection of participants into the study" had moderate or serious risk of bias in all studies. Risk of bias assessment is shown in Figure 3.
(2) VLEDs versus Bariatric Surgery. Four studies [24,32,33,35] analyzed the weight loss between the VLED and surgery groups, including 84 participants. Moreover, the surgical methods used in these four studies were RYGB as comparator arms. The merged data with no evidence of interstudy heterogeneity (I 2 = 0%), according to the DerSimonian-Laird FE model, revealed that the VLEDs and RYGB have (3) VLEDs versus MER. Four studies [37][38][39][40] analyzed the weight loss when a VLED (n = 88) was compared with MER (n = 88). Three studies provided data at the end of the intervention, and one provided data for long-term follow-up (5 years). In particular, the study of Williams et al. [40] contains two types of VLED interventions, and that of Paisey et al. [38] contains data for two endpoints.
According to the results of the subgroup analysis, the data at the end of the intervention showed that VLED was significantly better than MER in weight loss (MD = −6:72; 95%CI = −10:05,−3:39; P < 0:0001), with evidence of moderate heterogeneity (I 2 = 55%; P heterogeneity = 0:06). Sensitivity analysis showed that the heterogeneity was 0% when "Paisey et al. [38]" was removed, and the effect of VLEDs on weight loss was still significantly better than that of the control (MD = −5:19; 95%CI = −7:6,−2:78; P < 0:0001). However, when followed up for 5 years, similar to the result of the "Paisey et al. [37]" study, MER was better   Journal of Diabetes Research   were seen intensively to wean them back onto a low-fat diet. A standardized programme of low-fat, low-refined carbohydrate foods was introduced over a 6-week period as patients transferred from VLED to normal eating patterns. They were advised to continue low-fat nutrition in the long term with three main meals daily MER: low-fat, low-sugar, and high-fibre intake advised; 5-day self-report food records were collected and discussed individually, repeated every 6-8 weeks.
Aerobic exercises with encouragement performed at each visit followed by a group discussion on nutrition All antidiabetic medication was stopped on day one, and insulin dosage halved.

Blood Glucose and Changes in Medication
(1) VLEDs versus LEDs. Four studies [25,28,30,31] analyzed the blood glucose levels between the VLED and LED groups, and all of them provided data at the end of the intervention. Simultaneously, two provided data for long-term follow-up (≥1 year). A significant difference in weight change in favor of the intervention arm was noted at both the end of the intervention (MD = −1:18; 95%CI = −2:05,−0:30; P = 0:008, <0.05) and follow-up (MD = −1:43; 95%CI = −2:65,−0:20; P = 0:02, <0.05), and both of them had no evidence of interstudy heterogeneity (I 2 = 0%). Regarding the use of hypoglycemic drugs, Carter et al. reported that although medication dose decreased with time, all participants using medication at baseline were also using medication at the end of the study. At 2 years, one study (Wing et al. [31]) reported that fewer participants in the VLED group required medication (45% vs. 69% in the VLED and LED groups, respectively) ( Figure 7).
(2) VLEDs versus Bariatric Surgery. Five studies [24,[32][33][34][35] analyzed the blood glucose levels between the VLED (n = 69) and bariatric surgery groups (n = 55). The merged data with no evidence of interstudy heterogeneity (I 2 = 49%), according to the DerSimonian-Laird FE model, revealed that VLEDs and surgery have similar effects on weight loss, and there is no significant difference between them (MD = 0:37; 95%CI = −0:22, 0:96; P = 0:22, >0.05) (Figure 8). In the use of hypoglycemic drugs, one study [33] showed that all hypoglycemic drugs were discontinued in the RYGB arm and decreased by 55% in the VLED arm after the intervention. In another study [32], metformin was reintroduced in 4/15 participants in the RYGB arm and 2/12 participants in the VLED arm after the intervention, and the difference was not significant.
(3) VLEDs versus MER. Five studies [36,37,[39][40][41] analyzed the blood glucose levels between the VLED (n = 86) and MER groups (n = 84). Results from the subgroup analyses showed that VLED was significantly better than MER in lowering blood glucose levels (MD = −6:72; 95%CI = −10:05,−3:39; P < 0:0001) at the end of the intervention, with evidence   Journal of Diabetes Research of low heterogeneity (I 2 = 48%; P heterogeneity = 0:17). However, at the 5-year follow-up, only one study by "Paisey et al. [37]" reported that the difference in blood glucose levels compared with controls was not significant (MD = −1; 95%CI = −4:62, 2:62; P = 0:59). In the use of hypoglycemic drugs at the end of the intervention, the study of Paisey et al. showed that, at 6 months (all patients who underwent VLEDs had reverted to normal food for at least two weeks), the patients in the VLED group discontinued insulin, sulphonylureas, or hypolipidemic agents, while patients in the MER group were not able to discontinue their antidiabetic or hypolipidemic therapies. At 1 year, 14 of 15 patients in the VLED group, but none in the conventional diet group, had discontinued insulin and any oral hypoglycemic medication (Figure 9).

Dropout.
Comparing the VLED and bariatric surgery groups, no loss of patients was reported. However, most studies on VLEDs compared with those on LEDs or MER reported increased dropout rate.
Andorson's study showed that frequently reported side effects during the weight loss phase included constipation, diarrhea, dizziness, and fatigue. The adverse reactions reported in Li et al.'s study were slight headache and dizziness during energy restriction. None of these studies reported significant differences in side effects between the VLED and control groups (see Table 2 for details).  Only three studies mentioned a rebound in body weight, blood glucose level, and other indicators after energy restriction therapy. One study [28] reported that at 24 months, in the completer analysis of 84 participants at follow-up, 44 (52%) regained weight (>1 kg weight gain) and participants regained 33% of their weight losses between 12 and 24 months. In this follow-up study, HbA1c level had increased by 0.3% (3.3 mmol/mol) from baseline at 24 months. Paisey et al. [37] found that weight loss was slower in the intensive conventional diet group than in the VLED group but better maintained at 5 years: group 1, 4:8 ± 6 kg, and group 2, 8:9 ± 4 kg. Wing et al. [30,31] reported that, although initial weight losses were greater in the VLED group, these participants regained significantly more weight than those in the behavioural therapy group in 1 year of follow-up. Moreover, at one-year assessment, the measures of glycemic control had returned to baseline, and no differences were observed between treatment groups.

Publication Bias.
All outcome indicators were analyzed in <10 studies, so publication bias was not examined.

GRADE for the Outcomes.
We evaluated all outcome indicators by GRADEprofiler 3.6 from the following aspects: (1) downgrade quality of evidence, risk of bias, inconsistency, indirectness, imprecision, and publication bias and (2) upgrade quality of evidence, large effect, plausible confounding changing the effect, and dose-response gradient.
After a comprehensive analysis, the evidentiary body was formed and found that all outcome indicators had low quality or extremely low quality (see Tables 3-5 for details).

Discussion
Our systematic review provides evidence based on current clinical trials on the efficacy of continuous and intermittent VLEDs in overweight and obese individuals with T2DM by comparison to other methods of energy restriction. First, during the intervention period, a VLED is superior in the reduction of body weight and blood glucose and TG levels to LEDs and MER. After long-term follow-up, there is no obvious difference in weight loss between VLEDs and LEDs, but glycemic control is still more effective in VLEDs. Second, VLEDs offer beneficial effects on weight loss, glycemic control, and improvement of insulin resistance comparable to bariatric surgery.
Increasing evidence suggested that modest and sustained weight loss improved glycemic control in overweight and obese individuals with T2DM [3]. Furthermore, recent studies reported that intentional weight losses by low-calorie diets, usually >15 kg, could reverse T2DM into a nondiabetic state [5,42]. Based on the current studies, our study concluded that more extreme dietary energy restriction with

Journal of Diabetes Research
VLEDs is an effective method to achieve intensive weight loss in a short term and improve glycemic control more effectively compared with LEDs and MER. This conclusion supported the recommendation of the American Diabetes Association (ADA) Standards of Medical Care in Diabetes that high-intensity diet intervention, physical activity, and behavioural therapies to achieve a 500-750 kcal/day energy deficit and maintain >5% weight loss should be prescribed for patients with type 2 diabetes who are overweight or obese and ready to achieve weight loss. Furthermore, previous studies showed that rapid weight loss by VLEDs is inevitably followed by weight regain [42], but recent studies with at least 1-year follow-up found that VLEDs might present a longer effect on weight maintenance [5,43]. Another study showed that even though weight was regained, the short-term weight loss had long-lasting benefits on glycemic control and prevention of cardiovascular effects in T2DM [44]. Our results are in line with this study. After further analysis of the effects of long-term follow-up (1-5 years), we found no obvious difference in weight loss between VLEDs and LEDs, but VLEDs still maintained better glycemic control. The lasting effects of VLEDs may be attributed to improved insulin sensitivity remaining from weight loss [45], "metabolic memory" from the treatment period [46], and "legacy effect" by lifestyle intervention [47]. It is reported that dyslipidemia, especially hypertriglyceridemia, is an independent risk factor in predicting the development of diabetes, which is partially mediated by insulin resistance and obesity [48]. Several prospective studies have demonstrated that weight loss induced decreases in pancreatic and liver TG levels in T2DM, which was associated with the recovery of insulin secretory function [6,49]. However, the effect of weight loss by VLEDs on the plasma TG level is rarely discussed. Our meta-analyses found that VLEDs reduced the plasma TG level in T2DM more effectively compared to LEDs and MER and had an equivalent effect with bariatric surgery, which may have potential effect on preventing the development of T2DM.
Bariatric surgery is confirmed to have superior effect in T2DM [50] and has been proposed as a first-line therapy for obese patients with T2DM [3]. Bariatric surgery can restore normal liver insulin sensitivity within days and decrease plasma glucose and TG levels within weeks [51]. In this context, some studies determined whether the effects of bariatric surgery are primarily due to negative energy balance or unique to the surgical procedure [24,51]. Our study shows that VLEDs are as effective as bariatric surgery (mainly RYGB) in terms of weight loss, glycemic control, insulin resistance improvement, and plasma TG level reduction. Additionally, VLEDs have lower costs and lesser adverse effects compared with bariatric surgery. Thus, VLEDs may be a considerable therapy when patients could not or would not wish to undergo surgical treatments.
VLEDs were found to be acceptable as indicated by the low dropout rate in both this and a previous study. The main reason may be that rapid weight loss increases patient's confidence, and hunger of patients after VLED intervention is more inhibited. A study shows that attrition was lower when Hypoglycemia (n = 2) Hyperglycemia (n = 3) Headache (n = 2) Hypoglycemia (n = 6) Hyperglycemia (n = 7) Paisey 1995 Severe hypoglycemic attack (n = 1) Myocardial infarction (n = 1)

Wing 1991
Coldness, constipation, dry skin, diarrhea, dizziness, vomiting, or weakness-commonly reported side effects of VLEDs. There were no significant differences over time in any of these symptoms and no significant difference between subjects in the LED and VLED groups. However, uric acid increased significantly in the VLED group Wing 1994 Common side effects included cold intolerance, constipation, and hair loss, which all resolved when the VLED was terminated Unclear Andorson 1994 Frequently reported side effects during the weight loss phase included constipation (56% of subjects), diarrhea (31%), dizziness (31%), fatigue (31%), flu/sore throat (13%), headache (10%), vomiting (10%), blurred vision (10%), muscle cramps (8%), and syncope (5%). None of these side effects required treatment alteration.

Li 2017
No serious adverse effects: slight headache (n = 3); slight dizziness (n = 1) No serious adverse effects VLED: very low-energy diet; LED: low-energy diet.  Some studies were nonrandomized controlled trials. 2 The ratio of 95% CI to the effect is more than 50%. 95% CI is wider and its accuracy is poor, so it decreases one level. 3 Case-control. 4 The I 2 value of the combined results is larger, and there is statistical heterogeneity, so it falls by one grade. Some studies were nonrandomized controlled trials. 2 There are studies that do not account for specific stochastic methods, so they are reduced by one level. 3 The I 2 value of the combined results is larger and there is statistical heterogeneity, so it falls by one grade. 4 Interventions include continuous VLEDs and intermittent VLEDs, which differ to some extent, so they are reduced by one level. 5 After merger, the effect is large and the accuracy is high, so it goes up one level. 6 Failure to adequately control confounding factors. 7 The ratio of 95% CI to the effect is more than 50%. 95% CI is wider and its accuracy is poor, so it decreases one level. 8 Case-control. 18 Journal of Diabetes Research weight loss was undertaken rapidly rather than gradually, because rapid weight loss might motivate participants [52]. Moreover, ketosis suppresses appetite and increases the satiety hormone cholecystokinin, which increases the possibility that participants with rapid weight loss might have been less hungry during the weight loss phase than those following the gradual diet [53][54][55][56]. Of note, the experience of healthcare professionals involved in the trial in obesity treatment also had a significant impact on attrition. While the short-term efficacy of VLEDs is evident and patient compliance is acceptable according to our analysis, the reports of adverse reactions in the studies are incomplete, limiting the use of this method.

Journal of Diabetes Research
In a previous systematic review, Rehackova et al. [19] revealed that VLEDs led to considerable weight loss and blood glucose control via small sample or qualitative studies. However, evidence on the long-term efficacy of VLEDs with regard to weight loss in individuals with T2DM is lacking. Our study has expanded the sample size and further analyzed the follow-up results between VLEDs and LEDs. After the follow-up (1-5 years), VLEDs present a more effective glycemic control effect, but there is no obvious difference in weight loss between VLEDs and LEDs. Some included studies [28,30,31,37] also showed that, at the end of VLED intervention, the decrease in body weight, blood glucose level, and other indicators would rebound to varying degrees. This shows that adherence to a VLED regimen is crucial in maximizing intervention effects. It has been shown that greater initial weight loss facilitates weight maintenance if followed by an effective weight loss maintenance programme [57]. Further exploring a strategy to suppress hunger after rapid weight loss and prevent weight regain of VLEDs is greatly important in the popularization of this method.
This meta-analysis provides some objective evidence for the application of VLEDs in obese individuals with T2DM, but there are still many limitations in the study. First, both non-RCTs and RCTs were combined in the meta-analyses, which increased the heterogeneity and risk of bias. Therefore, the results of this study still need to be confirmed by higherquality research. Second, some high-quality research in this field has been conducted by a small number of research groups, resulting in insufficient representation of data. Thus, more extensive studies are needed to clarify the practicability of VLEDs in different ethnic groups. Third, most included studies did not mention the use of hypoglycemic drugs in participants. When VLEDs are used to intervene with obese patients with T2DM, determination of hypoglycemic drugs is difficult. In the future, the standardized research of this area should be strengthened. Lastly, only a few included studies that recorded follow-up results, which led to insufficient convincing evidence. Moreover, the longest follow-up duration in the included studies was only 5 years, so the longterm effect of VLEDs needs further study.

Conclusions
Dietary intervention through VLEDs is more effective in rapid weight loss and glycemic control and improved lipid metabolism in overweight and obese individuals with T2DM than LEDs and MER, although they have similar long-term effects. Moreover, VLEDs have similar efficacy and acceptability with bariatric surgery, which shows that VLEDs have considerable curative effect for remission of T2DM. However, after GRADE, it was found that all outcome indicators had low quality or base quality, so the results of this study still need to be further confirmed by high-quality research.

DM:
Diabetes mellitus VLED: Very low-energy diet LED: Low-energy diet MER: Mild energy restriction RYGB: Roux-en-Y gastric bypass CI: Confidence interval MD: Mean difference.