Introduction
Diabetic nephropathy, a complication of diabetes, is the leading cause of kidney failure, responsible for approximately 40% of incident cases [
1].
Diabetic nephropathy is characterized by hypertension, variable levels of albuminuria and a progressive loss of kidney function [
2,
3]. The progression of histological and pathological changes in diabetic nephropathy are due to hyperglycemia [
4]. The histological and pathological changes differ between type 1 and type 2 diabetes (T1DM, T2DM, respectively). In T1DM, hyperglycemia starts earlier hence it causes pure diabetic glomerulopathy that could be evaluated at the stage of microalbuminuria. Whereas in T2DM hyperglycemia starts later in life when kidneys were already damaged due to the long-term effects of many possible promoters of kidney injury such as aging, hypertension, and dyslipidemia. Therefore, there is a heterogenous combination of pathophysiological pathways that sustain structural changes in the kidneys of T2DM patients. Regardless of the involved mechanism, the final common pathway of diabetic nephropathy is kidney fibrosis that is caused by kidney hemodynamic and ischemic abnormalities, oxidative stress and the overactivation of the renin-angiotensin aldosterone system (RAAS) [
5,
6]. Clinical practice guidelines recommend prescribing angiotensin converting enzyme (ACE) inhibitors and angiotensin receptor blockers (ARBs) which are the two major classes of renin–angiotensin aldosterone system inhibitors (RAASi) to prevent and manage diabetic nephropathy at any stage [
2,
3]. The blockade of RAAS is critical to control blood volume, systemic vascular resistance and electrolyte balance [
7,
8]. This results in RAASi protecting the kidneys from developing diabetic nephropathy, as well as slowing the progression of the disease [
9,
10]. Therefore, these RAASi are the antihypertensive class of choice recommended for the management of patients with hypertension and DM [
2,
3].
A number of meta-analyses have been published on the role of RAASi in renoprotection for patients with diabetes. The authors concluded that ACE inhibitors and ARBs are equally effective in slowing the progression of diabetic nephropathy [
9‐
12]. However, these meta-analyses of RAASi have focused on cardiovascular and kidney outcomes, and had restricted inclusion and exclusion criteria of eligible studies and limited the included clinical trials to patient populations with one type of diabetes, a specific level of albuminuria, and/or excluding patients with advanced stages of chronic kidney disease (CKD). These findings are therefore applicable to a narrow population, and may be limited in their ability to guide clinical care and decision making for a large proportion of patients with diabetes. In addition, most of the previous meta-analyses evaluated RAASi effect on binary kidney outcomes (
e.g. kidney failure, progression to albuminuria, mortality), and seldom provided an analysis of RAASi effect on continuous kidney outcomes (
e.g. creatinine clearance level, albuminuria level). To address this knowledge gap we have conducted a systematic review (SR) with broader inclusion criteria to allow for conducting sub-group analysis for different variables and hence to identify their effects on kidney and other health outcomes. The objective of this systematic review and meta-analysis is to compare the effects of ACE inhibitors/ARBs with placebo and other antihypertensives in adults with diabetes on both continuous and binary kidney outcomes.
Methods
For this systematic review, we followed the reporting guidance provided in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Supp
1 Table 1) [
13]. The SR protocol was registered and published with PROSPERO (CRD42020149133). A brief summary of the methodology is described here, and is based on PRISMA guidelines for reporting SRs [
14].
Table 1
Characteristics of Studies Included in the Meta-Analysis
Melbourne Diabetic Nephropathy Study Group 1991 | Australia | RAAS inhibitors; Other Anti-HTN | Perindopril; Nifedipine | Outpatient clinic | Pharmaceutical | Microalbuminuria | Mixed (38% type 1) | Stage 1 |
Chan 1992 | China | RAAS inhibitors; Other Anti-HTN | Enalapril; Nifedipine | Outpatient clinic | Pharmaceutical | Mixed (normoalbuminuria, microalbuminuria and macroalbuminuria) | Type2 | Up to stage 4 |
Lacourciere 1993 | Canada | RAAS inhibitors; Other Anti-HTN | Captopril; Metoprolol or HCZ | Others | Not mentioned | Mixed (normoalbuminuria and microalbuminuria) | Type2 | Stage 1 |
Lewis 1993 | USA | RAAS inhibitors; Placebo | Captopril; Placebo | Hospital | Both pharmaceutical and governmental | Not mentioned | Type1 | Generally abnormal GFR |
Ravid 1993 and 1995 | Israel | RAAS inhibitors; Placebo | Enalapril; Placebo | Outpatient clinic | Governmental | Microalbuminuria | Type2 | Stage 1 |
Lebovitz 1994 | USA | RAAS inhibitors; Placebo | Enalapril; Placebo | Not mentioned | Both pharmaceutical and governmental | Mixed (normoalbuminuria, microalbuminuria and macroalbuminuria) | Type2 | Stage 3 |
Viberti 1994 | International | RAAS inhibitors; Placebo | Captopril; Placebo | Hospital | Pharmaceutical | Microalbuminuria | Type1 | Stage 1 |
Agardh 1996 | International | RAAS inhibitors; Other Anti-HTN | Lisinopril; Nifedipine | Not mentioned | Not mentioned | Microalbuminuria | Type2 | Up to stage 3 |
Bakris 1996 | USA | RAAS inhibitors; Other Anti-HTN | Lisinopril; Verapamil or Diltiazem; Atenolol | Outpatient clinic | Governmental | Macroalbuminuria | Type2 | Generally abnormal GFR |
Schnack 1996 | Austria | RAAS inhibitors; Other Anti-HTN | Ramipril; Atenolol | Outpatient clinic | Not mentioned | Microalbuminuria | Type2 | Stage 1 |
Ahmad 1997 | India | RAAS inhibitors; Placebo | Enalapril; Placebo | Outpatient clinic | Governmental | Microalbuminuria | Type2 | Stage 1 |
Chaturvedi 1997 | Europe | RAAS inhibitors; Placebo | Lisinopril; Placebo | Outpatient clinic | Pharmaceutical | Mixed (normoalbuminuria, microalbuminuria and macroalbuminuria) | Type1 | Not clear |
Fogari 1997 | Italy | RAAS inhibitors; Other Anti-HTN | Enalapril; Amlodipine | Not mentioned | Not mentioned | Microalbuminuria | Type2 | Stage 1 |
Crepaldi 1998 | Italy | RAAS inhibitors; Other Anti-HTN; Placebo | Lisinopril; Nifedipine; Placebo | Hospital and clinic | Pharmaceutical | Microalbuminuria | Type1 | Stage 1 |
Ravid 1998 | Israel | RAAS inhibitors; Placebo | Enalapril; Placebo | Outpatient clinic | Governmental | Normoalbuminuria | Type2 | Stage 1 |
UKPDS 1998 | UK | RAAS inhibitors; Other Anti-HTN | Captopril; Atenolol | Outpatient clinic | Both pharmaceutical and governmental | Mixed (normoalbuminuria, microalbuminuria and macroalbuminuria) | Type2 | Stage 1 |
Fogari 1999 | Italy | RAAS inhibitors; Other Anti-HTN | Ramipril; Nitrendipine | Not mentioned | Not mentioned | Macroalbuminuria | Type2 | Generally abnormal GFR |
Muirhead 1999 | Canada | RAAS inhibitors; Placebo | Valsartan; Captopril; Placebo | Outpatient clinic | Pharmaceutical | Microalbuminuria | Type2 | Stage 1 |
HOPE 2000 | International | RAAS inhibitors; Placebo | Ramipril; Placebo | Hospital and clinic | Both pharmaceutical and governmental | Mixed (normoalbuminuria and microalbuminuria) | Mixed (2.3% Type 1) | Stage 1 |
O'Hare 2000 | UK and Ireland | RAAS inhibitors; Placebo | Ramipril; Placebo | Outpatient clinic | Pharmaceutical | Microalbuminuria | Type1 | Stage 1 |
Schrier 2000 | USA | RAAS inhibitors; Other Anti-HTN | Enalapril; Nisoldipine | Not mentioned | Not mentioned | Macroalbuminuria | Type2 | Up to stage 3 |
Tarnow 2000 | Denmark | RAAS inhibitors; Other Anti-HTN | Lisinopril; Nisoldipine | Outpatient clinic | Pharmaceutical | Macroalbuminuria | Type2 | Up to stage 3 |
Lewis 2001 | International | RAAS inhibitors; Other Anti-HTN; Placebo | Irbesartan; Amlodipine; Placebo | Outpatient clinic | Pharmaceutical | Macroalbuminuria | Type2 | Up to stage 5 |
Baines 2001 | UK and Italy | RAAS inhibitors; Other Anti-HTN; Placebo | Enalapril; Nifedipine; Placebo | Hospital | Both pharmaceutical and governmental | Mixed (microalbuminuria and macroalbuminuria) | Type1 | Stage 1 |
Bojestig 2001 | Sweden | RAAS inhibitors; Placebo | Lisinopril; Placebo | Outpatient clinic | Pharmaceutical | Microalbuminuria | Type1 | Not mentioned |
Brenner 2001 | International | RAAS inhibitors; Placebo | Losartan; Placebo | Outpatient clinic | Pharmaceutical | Macroalbuminuria | Type2 | Generally abnormal GFR |
Parving 2001 | International | RAAS inhibitors; Placebo | Irbesartan; Placebo | Not mentioned | Pharmaceutical | Microalbuminuria | Type2 | Stage 1 |
Kvetny 2001 | Denmark | RAAS inhibitors; Placebo | Perindopril; Placebo | Not mentioned | Pharmaceutical | Normoalbuminuria | Type1 | Stage 1 |
Baba 2001 | Japan | RAAS inhibitors; Other Anti-HTN | Enalapril; Nifedipine | Outpatient clinic | Not mentioned | Mixed (normoalbuminuria and microalbuminuria) | Type2 | Up to stage 3 |
Katayama 2002 | Japan | RAAS inhibitors; Placebo | Imidapril; Captopril; Placebo | Not mentioned | Governmental | Mixed (microalbuminuria and macroalbuminuria) | Mixed (97.5% Type 1) | Stage 1 |
Fogari 2002 | Italy | RAAS inhibitors; Other Anti-HTN | Fosinopril; Amlodipine | Not mentioned | Not mentioned | Microalbuminuria | Type2 | Stage 1 |
Schrier 2002 | USA | RAAS inhibitors; Other Anti-HTN | Enalapril; Nisoldipine | Not mentioned | Both pharmaceutical and governmental | Mixed (normoalbuminuria, microalbuminuria and macroalbuminuria) | Type2 | Up to stage 3 |
Ahmad 2003 | India | RAAS inhibitors; Placebo | Enalapril; Placebo | Outpatient clinic | Not mentioned | Microalbuminuria | Mixed (85.8% Type 1) | Stage 1 |
Marre 2004 | International | RAAS inhibitors; Other Anti-HTN | Enalapril; Indapamide | Hospital | Pharmaceutical | Microalbuminuria | Type2 | Stage 1 |
Jerums 2004 | Australia | RAAS inhibitors; Other Anti-HTN; Placebo | Perindopril; Nifedipine; Placebo | Hospital | Both pharmaceutical and governmental | Microalbuminuria | Type2 | Up to stage 2 |
Ruggenenti 2004 | Italy | RAAS inhibitors; Other Anti-HTN; Placebo | Trandolapril; Verapamil; Placebo | Not mentioned | Both pharmaceutical and governmental | Normoalbuminuria | Type2 | Stage 1 |
DallaVestra 2004 | Italy | RAAS inhibitors; Other Anti-HTN | Ramipril; Lercanidpine | Others | Not mentioned | Microalbuminuria | Type2 | Stage 1 |
Fogari 2005 | Italy | RAAS inhibitors; Other Anti-HTN | Lisinopril; Manidipine | Outpatient clinic | Governmental | Microalbuminuria | Type2 | Stage 1 |
Ogawa 2007 | Japan | RAAS inhibitors; Other Anti-HTN | Temocapril; Candesartan; Nifedipine | Outpatient clinic | Not mentioned | Microalbuminuria | Type2 | Stage 1 |
Makino 2008 | Japan | RAAS inhibitors; Placebo | Telmisartan; Placebo | Hospital and clinic | Pharmaceutical | Microalbuminuria | Type2 | Stage 1 |
Bilous 2009 | International | RAAS inhibitors; Placebo | Candesartan; Placebo | Secondary care facility | Pharmaceutical | Normoalbuminuria | Mixed (63.6% Type 1) | Stage 1 |
Mauer 2009 | USA and Canada | RAAS inhibitors; Placebo | Enalapril; Losartan; Placebo | Outpatient clinic | Both pharmaceutical and governmental | Normoalbuminuria | Type1 | Stage 1 |
Haller 2011 | International | RAAS inhibitors; Placebo | Olmesartan; Placebo | Secondary care facility | Pharmaceutical | Normoalbuminuria | Type2 | Up to stage 3 |
Ruggenenti 2011 | Italy and Slovenia | RAAS inhibitors; Placebo | Delapril; Placebo | Outpatient clinic | Both pharmaceutical and governmental | Mixed (normoalbuminuria and microalbuminuria) | Type2 | Stage 1 |
Weil 2013 | USA | RAAS inhibitors; Placebo | Losartan; Placebo | Not mentioned | Both pharmaceutical and governmental | Mixed (normoalbuminuria and microalbuminuria) | Type2 | Stage 1 |
Fuchs 2016 | Brazil | RAAS inhibitors; Other Anti-HTN | Losartan; Chlorthalidone/Amiloride | Secondary care facility | Governmental | Not mentioned | Type2 | Not mentioned |
Research question
The clinical question of this systematic review was: In an adult who is diagnosed with T1DM or T2DM, what is the efficacy of RAASi compared with other antihypertensive medications, or with placebo on continuous kidney outcomes including eGFR, SrCr, and albuminuria levels?.
Literature search
As there is a large number of randomized controlled trials (RCTs) and SRs on this topic, a staged approach to identify eligible RCTs was used. This approach was used successfully by other researchers [
11,
15]. First, we conducted a search of relevant SRs and meta-analyses in PubMed and the Cochrane Database of Systematic Reviews. The identified SRs and meta-analyses were used to provide lists of relevant RCTs to identify studies that fit the inclusion criteria.
Next, we performed a systematic search to identify other RCTs published since the date of publication of the SRs and meta-analyses identified above. Most of the relevant meta-analyses were published around 2010, [
9‐
12] therefore, the date limit of our systematic search was from 2010 to Jan 28, 2020. Electronic searches were conducted with the help of a medical librarian in MEDLINE, CINAHL, EMBASE, Cochrane library and the clinical trials registry at clinicaltrials.gov, with no language restriction. Search terms included generic names and Medical Subject Headings of all RAASi (including ACEIs and ARBs) combined with diabetic nephropathy and other relevant keywords as identified by the librarian (Supp
1 Table 2). Manual search of references included in relevant reviews, clinical trials and clinical practice guidelines was also conducted.
Inclusion criteria
Studies satisfying the following criteria were included: Randomized parallel-group controlled trials of a duration of 12 months or more that recruited more than 50 adult participants (18 years or older) with T1DM or T2DM with any stage of CKD and proteinuria. The RCTs had to study the effects of RAASi on the progression of albuminuria and the progression of CKD. Progression of diabetic nephropathy was examined using the incidents of albuminuria and regression of albuminuria endpoints, and changes in urine albumin excretion levels. Progression of CKD was assessed using doubling of serum creatinine (SrCr), changes in SrCr and estimated glomerular filtration rate (eGFR).
Comparisons accepted in this SR and meta-analysis were between either ACEI or ARBs versus placebo or other antihypertensives including calcium channel blockers (CCBs), beta blockers (BBs), or diuretics or their combinations.
Prespecified outcome measures
The primary outcomes of interest were continuous kidney outcomes including eGFR, SrCr levels, and albuminuria levels. Secondary outcomes were binary kidney outcomes including kidney failure, doubling of SrCr, and regression of albuminuria. Secondary outcomes also included all-cause mortality, blood pressure (BP) outcomes (diastolic and systolic BP, mean arterial BP [MAP] and the need for additional antihypertensives to control BP), and safety outcomes (the incidence of any adverse drug reactions, acute kidney injury, hyperkalemia, disruptive cough, and reasons for patients’ withdrawal from the RCTs). A list of definitions of each outcome measure is appended (Supp
1 Table 3).
Screening and data extraction
Studies were screened against the inclusion criteria by two reviewers independently (NA and ML) using the Covidence web-based application (Veritas Health Innovation, Melbourne, Australia) [
16]. Any disagreements were resolved by discussion between the two reviewers or referred to a third reviewer (JPL) when no consensus could be reached. A data extraction form was used to extract data from the eligible studies, including study and participant characteristics (sample size, age, sex, albuminuria stage, type and duration of diabetes, presence of hypertension and cardiovascular disease, smoking status, body mass index (BMI), race, and history of recent use of antihypertensives), interventions used, mean or median follow-up and outcome data. The data extraction tool was piloted on a small sample (10%) of studies by the same reviewers. One reviewer (NA) was responsible for extracting the data and the other reviewer (ML) was responsible for double-checking the entered data for accuracy. Discrepancies were resolved by consensus.
Data were extracted from studies' tables and texts reported in the main study manuscript or supplementary materials. In cases where important baseline and outcome data were not reported in tables or text, we extracted data from figures and graphs using WebPlotDigitizer which is a validated web-based application to extract numerical data from plot images [
17]. Additionally, corresponding authors were contacted to seek missing or incomplete data from their studies.
Some studies provided the medians and interquartile range, and we used estimated mean of the sample equation from Luo et al. (2017), and estimated standard deviation (SD) of the sample equation from Wan et al. (2014) to calculate the means and SD, respectively using Hozo et al
. method [
18]. We calculated the effect size from the reported events numbers.
Risk of bias assessment
Included studies' risk of bias was assessed using the Cochrane Collaboration’s risk of bias scale that addresses six domains: sequence generation, allocation concealment, blinding of participants/outcome assessors, incomplete outcome, selective outcome reporting and the source of funding [
19]. Two investigators (NA and ML) were responsible for completing the assessment using the Covidence web-based application [
16].
Statistical analysis
We collectively assessed the effects of RAASi by the use of either ACEIs or ARBs on kidney outcomes by assigning trial arms of ACEIs or ARBs as the intervention group. Comparator groups were trial arms that used placebo or other antihypertensives (CCBs, BBs, diuretics or their combinations). Studies that included more than one arm including two agents of the same medication group were merged together for all outcomes.
Weighted kappa statistics were used to assess the agreement between the two reviewers for study selection. We generated descriptive statistics to provide a representation of patients included in the selected studies. We used the random effect approaches for meta-analysis of outcomes, with DerSimonian-Laird estimator for variance, to calculate the pooled effect size for each outcome because of known clinical and methodological heterogeneity of the studies. We reported the results as odds ratios (OR) using forest plots and tables.
We assessed heterogeneity between studies using I2 statistics with a 50% significance threshold. We used a funnel plot and Egger test to assess publication bias and Abbé plot to visually identify extreme, influential or outlier studies. We conducted different sensitivity analyses to evaluate the effect on the pooled estimate by removing the low-quality studies and removing extreme studies.
In our analysis of the continuous outcomes, we considered the difference of change from baseline between the arms of the study as the effect size for our meta-analysis except for albuminuria level, where we used standardized mean difference (SMD) as our effect size. SMD is used as an estimate of effect size when different studies measure the same outcome but in different units. Albuminuria was reported using different ways of reporting. This made SMD difficult to interpret; therefore, we used the following parameters for interpreting the size of the SMD: small, SMD = 0.2; medium, SMD = 0.5; and large, SMD = 0.8 [
20]. For eGFR we considered the raw mean difference (end of the study to baseline) and the raw difference between the annual change of eGFR. All statistical assumptions used in this SR can be found in Supp
1 Table 4.
We conducted subgroup analyses to explore the effect of relevant factors for the following groups: Age groups, type and duration of diabetes, hypertension status, stage of CKD, stage of albuminuria (normoalbuminuria, microalbuminuria, macroalbuminuria), BMI category, and study duration, sample size, and year of publication on pooled effect sizes of the study, and we conducted sensitivity analysis by excluding outlier studies. We stratified the included studies based on the study duration to account for the variation of studies’ mean follow-up periods.
Discussion
In this meta-analysis, evidence was drawn from 26,551 patients with diabetes from 46 studies on the effect of RAASi on continuous and binary kidney outcomes. This meta-analysis provided interesting findings on the effects of RAASi on some important continuous outcomes in comparison with placebo or other antihypertensives. RAASi were better than placebo in reducing SrCr and albuminuria. Estimated GFR was slightly increased by RAASi compared to placebo after performing a sensitivity analysis. When compared to active treatments, RAASi resulted in a modest increase of SrCr, led to eGFR decline, and resulted in modest reduction of albuminuria levels. Our analysis shows that the RAASi class was superior to placebo in reducing the risks of kidney failure and doubling of SrCr levels, but not in promoting the regression of albuminuria. RAASi, however, were not superior to other antihypertensive agents in reducing the risks of these kidney outcomes or all-cause mortality. Despite some key differences in the selection criteria (Supp
1 Table 6), the findings of our meta-analysis are consistent with previously published meta-analyses.
What distinguishes our meta-analyses from earlier reviews is the inclusion criteria of clinical trials, as well as our analysis of continuous kidney outcomes.(Supp
1 Table 6) We excluded RCTs with sample size of less than 50 participants to exclude small-size effects on the analysis. We performed subgroup analysis based on sample size to further isolate small-size effect of the studies of less than 100 participants. A duration of 12 months or more was a key inclusion criterion to help study the long-term effects of RAASi. Unlike some earlier meta-analyses, we did not limit our analysis to one type of diabetes or to a specific degree of kidney function or albuminuria, which allowed us to perform a variety of subgroup analyses.
This meta-analysis provided noteworthy findings about the effect of RAASi in each subgroup of patients. We have studied RAASi effects on each kidney outcome in different subgroups of study participants. The analysis showed that specific subgroups of patients had better outcomes with RAASi. Patients with T2DM, macroalbuminuria and an average duration of diabetes more than or equal 10 years had less risk to develop kidney failure in placebo-controlled trials, while longer duration of diabetes, normal kidney function, and hypertension increased the probability to achieve regression of albuminuria in active-controlled trials. Type 1 diabetes and hypertensive patients had higher rates of regression of albuminuria in placebo-controlled trials. As these findings suggest, the type and the duration of diabetes as well as other characteristics can influence the response to interventions on some kidney outcomes, which highlights the importance to consider each patient’s medical history when deciding on starting a treatment for them. These findings point to the need to direct more research initiatives on exploring patients’ characteristics that can predict who would benefit most from each intervention, including the broadening of inclusion criteria in studies, and conducting studies powered to look at divergent subgroups. The interpretation of subgroup analyses should be performed with caution, due to some inherited limitations, majorly because they are observational in nature although being derived from randomized trials [
60].
The latest version of the Kidney Disease Improving Global Outcomes (KDIGO) clinical practice guidelines for diabetes management in CKD patients [
61] were published in late 2020, and it recommends using RAASi as first-line antihypertensives for patients with diabetes, hypertension and albuminuria. These recommendations were based on four placebo-controlled landmark trials of ARBs on patients with different levels of albuminuria [
28,
29,
52,
62]. This recommendation is concordant with those of other guidelines, [
28,
29,
62] yet it raises some doubts about the robustness of the evidence behind it. The KDIGO guideline supports its recommendation with evidence from trials of ARBs only. Nevertheless, our analysis on continuous outcomes provides moderate evidence on RAASi’s ability to reduce albuminuria levels more than active-treatments. Cativo et al
. [
63] reached a similar conclusion, and highlighted that although the effect is statistically significant, the clinical effect is small. In summary, the evidence behind promoting RAASi as the leading class in protecting the diabetic kidneys may not be as robust as commonly believed. The findings of this study suggest that the most important factor for preventing and managing diabetic nephropathy is lowering BP levels, which could be of higher significance than the class of the antihypertensive used to lower BP.
Protecting the diabetic kidneys is not exclusive to antihypertensives, as some novel classes claimed their positions in the competition towards protecting diabetic patients from kidney disease using different mechanisms. For example, the mineralocorticoid receptor antagonist (finerenone) is being evaluated in a large RCT (FIDELIO-DKD), with some preliminary promising results [
64]. The new antihyperglycemic agents from the sodium‐glucose co‐transporter‐2 (SGLT2) inhibitors class have also shown protective effects against progression of CKD, with reductions of mortality rates when used in combination with RAASi [
65]. Sacubitril/valsartan have shown preservation effects of kidney function in older patients with heart failure, and its role in the management of diabetic nephropathy is to be evaluated [
66]. Taking the collective adverse events of these agents into consideration, the prescriber today has more options to consider to reduce the progression of diabetic kidney disease. A prescription that combines these agents with a proper antihypertensive could be viewed as the recipe of kidney protection in patients with diabetes. Nevertheless, more research studies need to be carried out to prove the safety and efficacy of such combinations.
This meta-analysis sheds light on the full spectrum of RAASi effects on kidney outcomes in patients with diabetes, by studying its efficacy on both continuous and binary outcomes. It provides a comprehensive review of the class effect on several subgroups of study participants, which was facilitated by the broad inclusion criteria. While we were attempting to answer a research question on RAASi comparative efficacy, our study raised a challenging question on the role of RAASi in preventing and managing diabetic nephropathy and whether it deserves its place as a first-line therapy in the clinical practice guidelines. The analysis' protocol was initially designed to include studies that reported other relevant kidney outcomes including urinary albumin concentration, albumin creatinine ratio, fractional albumin excretion, and kidney deaths. Therefore, the initial number of included studies were 53 trials. However, there was a very limited number of studies that collectively reported these outcomes, which lead to insufficient data to perform meaningful analysis of these outcomes. Therefore, the final number of included studied was 46.
A few limitations of this meta-analysis should be considered when interpreting and applying its findings. The analysis of the continuous outcomes was performed using a number of statistical assumptions.(Supp
1 Table 4) Another limitation is the degree of heterogeneity between the RCTs that were used to determine the change in the continuous outcomes. These RCTs were published across more than 20 years with variable methodological approaches and reporting qualities that resulted in methodological heterogeneity. The included studies shared a wide range of participant characteristics due to our broad inclusion criteria which resulted in population heterogeneity. The performance of subgroup analysis and sensitivity analysis helped mitigate the effect of this type of heterogeneity. The mean follow-up of included studies ranged between 12 and 72 months. Therefore, we conducted subgroup analyses for each outcome to account the difference in the duration of follow-up between studies. (Supp
2, tables 3–18).
It is noteworthy to mention that most of the RCTs were not powered to detect the changes in the continuous kidney outcomes because these were not primary outcomes. We have calculated the effect size from the reported events numbers because of the heterogeneity in reporting effect sizes between studies. We did not analyze data presented in composite outcomes, because of the inconsistency of the trials in reporting these outcomes as the same composite. However, we analyzed data for each single outcome separately.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.