Intended for healthcare professionals

Clinical Review State of the Art Review

# How to prevent the microvascular complications of type 2 diabetes beyond glucose control

BMJ 2017; 356 (Published 17 January 2017) Cite this as: BMJ 2017;356:i6505

1. Willy Marcos Valencia, voluntary assistant professor, physician scientist, (geriatric medicine – endocrinology, diabetes and metabolism)12,
2. Hermes Florez, director, professor, chief234
1. 1Department of Public Health Sciences, University of Miami Miller School of Medicine, Miami, FL 33125, USA
2. 2Geriatric Research, Education and Clinical Center (GRECC), Miami VA Healthcare System, Miami, FL, USA
3. 3Departments of Public Health Sciences and Medicine, University of Miami Miller School of Medicine, Miami, FL, USA
4. 4Division of Epidemiology, Department of Public Health Sciences, University of Miami Miller School of Medicine, Miami, FL, USA
1. Correspondence to: W M Valencia willy.valencia-rodrigo{at}va.gov

## Abstract

Microvascular complications (retinopathy, nephropathy, and neuropathy) affect hundreds of millions of patients with type 2 diabetes. They usually affect people with longstanding or uncontrolled disease, but they can also be present at diagnosis or in those yet to have a diagnosis made. The presentation and progression of these complications can lead to loss of visual, renal, and neurologic functions, impaired mobility and cognition, poor quality of life, limitations for employment and productivity, and increased costs for the patient and society. If left uncontrolled or untreated, they lead to irreversible damage and even death. This review focuses on the primary and secondary prevention of diabetic microvascular complications in patients with type 2 diabetes, beyond glycemic control. Interventions discussed include standard of care interventions supported by guidelines from major organizations, as well as additional proposed interventions that are supported by research published in the past decade. High level evidence sources such as systematic reviews and large, multicenter randomized clinical trials have been prioritized. Smaller trials were included where high quality evidence was unavailable.

## Introduction

Diabetic microvascular complications (DMC) develop in all types of diabetes. This review focuses on DMC in patients with type 2 diabetes. Diabetic retinopathy can lead to impaired vision, increased risk of falls, and, ultimately, blindness. Diabetic nephropathy can lead to severe chronic kidney disease (CKD), need for renal replacement therapy, and possibly death. Diabetic autonomic neuropathy and diabetic peripheral neuropathy (DPN) can lead to orthostatic hypotension, dysautonomy, foot ulcers, and limb amputations.1 All three DMC impair function and quality of life, lead to disability, and decrease employment chances and productivity.1 2

National and international guidelines stress timely and intensive interventions early in the disease, individualizing targets, agents, and approach. Detecting and treating complications of diabetes are major components for these interventions.3 4 5 Strong evidence supports the need to achieve and maintain proper glycemic control to prevent the incidence (primary prevention) or progression (secondary prevention) of microvascular complications.

This review excludes the well established role of glycemic control for DMC prevention and focuses on all other potentially successful interventions. Notably, cellular damage created by hyperglycemia and other factors may be shared in the three types of DMC. These include biochemical and molecular pathways, such as increased polyol pathway flux, increased advanced glycation end products formation, activation of protein kinase C isoforms, and increased hexosamine pathway flux.6

This review focuses on evidence published in the past decade on interventions for the primary and secondary prevention of DMC in type 2 diabetes (box 1). It also discusses upcoming trials, gaps in knowledge, and emerging strategies, providing a practical framework to guide specialists and academic clinicians who treat patients with DMC.

#### Box 1: Primary and secondary prevention of microvascular complications in patients with type 2 diabetes

##### Primary prevention: to prevent or delay onset of complications
• Prevention and management of known modifiable risk factors: improving lifestyle behaviors and pharmacologic interventions for hypertension, dyslipidemia, obesity, and smoking

• Screening for disease pre-stages, leading to interventions to prevent onset*

##### Secondary prevention: to reduce progression and damage from established complications
• Screening for early stages of disease, leading to early detection and intensive interventions to slow progression

• Continuation of interventions aimed at known risk factors. Specific targets may be proposed (for example, rather than prevention of hypertension or treatment to a target of systolic blood pressure <140 mm Hg, a lower target may be desirable to prevent progression)

• Monitoring disease progression and implementing strategies to reduce the effect on patient centered outcomes, such as quality of life and independent functioning

• *Detection of genetic predisposition may further empower the interventions proposed for early screening (for example, a patient with diabetes may be further motivated to control risk factors if evidence shows genetic predisposition to a diabetic complication). Genetic screening could become part of clinical care in the future, as the detection of genetic risk (for example, towards a greater likelihood of more severe or disabling disease) could prompt more aggressive interventions (primary or secondary) to prevent or delay the progression of the disease, enhancing interventions against risk factors and potentially incorporating more targeted approaches

## Sources and selection criteria

We searched the Cochrane, PubMed, and Embase databases from 1 January 2006 until 30 September 2016 by using medical subject headings (MeSH) and combined search terms that included “diabetes”, “microvascular complications”, “diabetic retinopathy”, “diabetic nephropathy”, and “diabetic neuropathy” in combination with “prevention” and “progression”. In addition, we included individualized searches for terms corresponding to specific interventions selected during our review. Languages searched included English, Spanish, and Portuguese, unless translation from another language was available. We reviewed titles and abstracts, excluding publications in which the main intervention was intensification of glycemic control.

We prioritized systematic reviews and meta-analyses, well designed (placebo controlled, intention to treat, adjustment for covariables), large, multicenter, randomized clinical trials (RCTs). When this high quality evidence was unavailable, we included smaller or non-placebo controlled RCTs or cohort studies to highlight the lack of high level evidence and the need for future research.

For the emerging treatments section, we used a similar search approach focused on RCTs registered in www.clinicaltrials.gov. We included studies with active recruitment or listed with pending results. Otherwise, we did individualized searches in the databases described above.

## Incidence and prevalence of microvascular complications in type 2 diabetes

Type 2 diabetes is a chronic progressive disease with a global impact. In 2015, 415 million people were affected worldwide, and this is expected to increase to 642 million (10% of the total global population) by 2040.5 This growth directly affects the incidence and prevalence of microvascular complications. About one in three patients with type 2 diabetes will develop retinopathy,7 one in four will develop nephropathy,8 and one in two will develop neuropathy.9 The risk of developing DMC is greater with longer duration of the disease (or if it is left uncontrolled). Furthermore, these rates underestimate the true burden, as DMC can be present at the time of diagnosis,10 11 12 and there are millions of patients with undiagnosed diabetes worldwide.

## Pathophysiology and risk factors for DMC in type 2 diabetes

Understanding the pathophysiologic pathways and risk factors for each of the DMC in type 2 diabetes is important when considering target interventions for primary and secondary prevention. This article does not provide an in-depth review of pathophysiologic pathways and risk factors leading to end organ damage. However, figure 1 offers a summary of the evidence for diabetic retinopathy,13 14 15 16 nephropathy,17 18 19 20 and neuropathy.21 22 23 24 25

Fig 1 Pathophysiology and risk factors leading to end organ damage in microvascular complications of type 2 diabetes. AGEs=advanced glycation end products; ESRD=end stage renal disease; Na-K-ATPase=sodium-potassium adenosine triphosphatase; NOS=nitric oxide synthase; PKC=protein kinase C; RAAS=renin-angiontensin-aldosterone system; TGF-β=transformation growth factor β, VEGF=vascular endothelial growth factor

## Primary and secondary prevention of diabetic retinopathy

### Screening

Even though diabetic retinopathy can lead to progressive visual loss and ultimately blindness, its onset may not be perceived at all by the patient. Approximately 10.5% of patients with type 2 diabetes have features of diabetic retinopathy on diagnosis,7 and despite established recommendations for monitoring only about 60% have yearly screening for diabetic retinopathy according to the recent guidelines from the American Academy of Ophthalmology.26 More recent community based studies also report poor rates of screening and monitoring,27 especially in low income areas and in ethnic minorities,28 particularly if patients do not understand the need despite availability of services.29 Effective screening and follow-up are needed to detect retinal changes quickly and accurately, and this should be coupled with education for patients that highlights the need for screening.

No pre-retinopathy stages exist that can guide additional interventions for primary prevention. Thus, screening for retinopathy aims for early detection of the disease that would then lead to further interventions for secondary prevention (of progression and irreversible damage).

#### Frequency

All major guidelines (American Diabetes Association (ADA),4 American Academy of Ophthalmology,26 UK National Institute for Health and Care Excellence (NICE) 30) recommend retinopathy screening for all patients newly diagnosed as having type 2 diabetes. They concur on the need for a dilated pupil and comprehensive examination by a specialist. ADA and NICE agree that further screening frequency should be based on findings. If retinopathy is present, further examination should be repeated at least annually (ADA), potentially with earlier review by an ophthalmologist (NICE). If retinopathy is absent, further examination could be considered every two years. If retinopathy is present, recent work suggests that patients with stable evaluation may not need yearly reassessment (NICE). This would be further supported if future genetic assessment could determine which patients have a low risk of progression of diabetic retinopathy.

#### Measurement

ADA guidelines state that a retinal photograph does not substitute for a comprehensive eye examination, although it may serve as a screening tool.4 NICE guidelines emphasize that a quality assured digital retinal photography program should be used by appropriately trained staff.30

Although the gold standard for retinal evaluation is by a retinal specialist under mydriasis, concerns exist about access, costs, and patients’ safety.1 In parallel, technological progress in digital retinal photography and telehealth has led to consideration of alternate approaches that could mitigate these concerns.

A study in 2016 proposed the use of single field fundus photography to initially screen for diabetic retinopathy before referral for formal specialized ophthalmologic evaluation.31 This small pilot study recruited 99 patients in India and used a fundus camera designed as a device that has slots to fit a smartphone (built-in camera and flash) and 20-D lens. Researchers reported that they could obtain high quality fundus videos with easy extraction of retinal images. They proposed that the approach could help patients to communicate with their providers, understand their disease, and engage in treatment and could potentially be an inexpensive tool for mass screening.32 33

Measurement of retinal vascular geometry and caliber has been proposed as a marker of progression of diabetic retinopathy in type 2 diabetes.34 Furthermore, measurements could be obtained before retinopathy develops and could be used as a tool to predict onset or progression to severe proliferative diabetic retinopathy.35 36 The largest retinal data repository in a prospective population cohort is included in the UK Biobank.37 More than 11 500 retinal images of patients with and without diabetic retinopathy have been collected to engage researchers in clinical trials and to improve detection and outcomes of diabetic retinopathy.38

New technologies are yet to be tested in large multicenter RCTs that span diverse populations and settings, and this is needed before they can be validated and incorporated into recommended retinal screening programs.39 Cost effectiveness studies are needed in parallel to assess their financial feasibility. A recent Japanese study modeled the cost-utility of yearly screening using current technology and found that despite expected increases in cost ($64.6 (£51.2; €60.9) for each of the proposed 5147 patients), improvements in quality of life occurred (incremental cost effectiveness ratio$11 857 per quality adjusted life year), as well as a 16% reduction in blindness for adults aged 40 years and older.40

### Blood pressure control

A systematic review and meta-analysis published in 2015 reviewed the evidence for blood pressure interventions to reduce diabetic retinopathy.41 It included 15 RCTs, with 9512 patients with type 2 diabetes, with or without hypertension, who received standard or more aggressive blood pressure interventions. Notably, the authors highlighted the differences between blood pressure targets in the older studies and current standards. Older studies considered intensive therapy for a systolic blood pressure (SBP) target below 150 mm Hg and conservative therapy for a target below 180 mm Hg. More recently, intensive therapy has been considered for a target below 120 mm Hg and standard therapy for below 140 mm Hg SBP. The authors reported a reduction in the four to five year incidence of diabetic retinopathy (risk ratio 0.80, 95% confidence interval 0.71 to 0.92) with more intensive blood pressure control. However, no significant reduction in the four to five year diabetic retinopathy progression occurred (risk ratio 0.88, 0.73 to 1.05). The baseline blood pressure (normal versus elevated) did not affect the results.

A second systematic review and meta-analysis, also published in 2015, included 21 RCTs (13 823 patients) comparing angiotensin converting enzyme (ACE) inhibitor or angiotensin receptor blocker (ARB) monotherapy with other antihypertensive agents.42 The results were similar: renin-angiotensin-aldosterone system (RAAS) inhibitors limited the onset but not the progression of diabetic retinopathy. The benefits were greater in normotensive patients and in those receiving ACE inhibitor: decreased risk of retinopathy progression (0.84, 0.75 to 0.94; P=0.002) and increased possibility of regression (1.50, 1.2 to 1.86; P<0.001). With ARBs, the possibility of regression was increased (1.32, 1.07 to 1.61; P=0.008) and no effect on disease progression was seen.

A third systematic review and meta-analysis published in 2015 assessed the effect of any blood pressure lowering agents on multiple outcomes, including diabetic retinopathy.43 Of the 45 selected RCTs, 16 looked at diabetic retinopathy. The review found significant benefits from lowering SBP by 10 mm Hg (risk ratio 0.87, 0.76 to 0.99). No additional benefit was related to baseline blood pressure, and no significant differences were seen between the classes of antihypertensive drugs.

#### Degree of blood pressure control

The Action to Control Cardiovascular Risk in Diabetes (ACCORD) study was a large RCT of 10 251 participants with type 2 diabetes at high risk for cardiovascular disease.44 The ACCORD BP trial was a non-blinded RCT within ACCORD that studied 4733 participants who did not have dyslipidemia, with mean age 62.2 (SD 6.9) years and baseline SBP 139.2 (SD 15.8) mm Hg.45 Patients were randomly assigned to intensive blood pressure control (target <120 mm Hg) versus standard control (target <140 mm Hg). Although no cardiovascular benefits were seen (primary outcome), intensive blood pressure control slowed the rate of progression of diabetic retinopathy. A systematic review and meta-analysis published in 2016 included 49 RCTs (each of ≥100 participants, most with type 2 diabetes, and ≥12 months’ duration) that compared antihypertensive agents against placebo.46 If baseline SBP was higher than 150 mm Hg or between 140 and 150 mm Hg, lowering SBP improved cardiovascular risk. In both situations, it decreased all cause mortality and myocardial infarction. For baseline SBP above 150 mm Hg, it also reduced cardiovascular mortality, stroke, and end stage renal disease (ESRD). For baseline SBP 140-150 mm Hg, it also reduced heart failure. However, for patients with baseline SBP less than 140 mm Hg, there were trends to increased risk for cardiovascular mortality (relative risk 1.15, 1.00 to 1.32) and all cause mortality (1.05, 0.95 to 1.16), without additional benefits on microvascular complications including nephropathy and progression to ESRD.

### Peroxisome proliferator activated receptor α agonist: fenofibrate

The Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) RCT evaluated treatment with fenofibrate (200 mg/day) in 9795 patients with well controlled (baseline glycated hemoglobin (HbA1c) 6.9%) type 2 diabetes of short duration (median five years). Most participants’ (84%) diabetes was controlled with diet alone or diet plus one antihyperglycemic agent, and most did not have retinopathy (prevalence of 8% at baseline) or significant dyslipidemia.47 Fenofibrate treatment was associated with a decrease in the progression of diabetic retinopathy (3.1% v 14.6%; P=0.004) and less need for laser intervention (3.4% v 4.9%; hazard ratio 0.69, 95% confidence interval 0.56 to 0.84).48

The ACCORD-EYE study included 2856 patients within ACCORD who had been randomly assigned to the combination of simvastatin plus fenofibrate (160 mg/day) or placebo, to evaluate the progression of diabetic retinopathy.49 Participants had type 2 diabetes (mean duration 10 years), mean age 61.6 (SD 6.3) years, and baseline absent or present diabetic retinopathy (50%), without proliferative retinopathy treated with laser photocoagulation or vitrectomy. Patients receiving fibrate had significantly reduced progression of retinopathy compared with those who received placebo (odds ratio 0.60, 0.42 to 0.87; P=0.006). Of note, the design of the ACCORD study had previously selected 5518 participants with dyslipidemia and randomly assigned them to simvastatin.

The benefits from fenofibrate in FIELD and ACCORD did not seem to be related to improvements in lipid profile,50 but they were greater in patients with mixed dyslipidemia, increased triglycerides, or low high density lipoprotein cholesterol. More recently, a large British retrospective cohort study compared the progression to diabetic retinopathy in patients with type 2 diabetes who were exposed or not exposed to fenofibrate.51 The study included 5038 patients from 700 practices followed for five years and showed a reduction of diabetic retinopathy in those exposed to fibrates (33.4 v 40.4 events/1000 person years; P=0.002). Limitations were related to the retrospective design, but results supported a role for fenofibrate in primary prevention.

Considering benefits and risks, including combination with statin therapy in those who need it, fenofibrate could be an adjunct therapy for diabetic retinopathy.52 In vitro studies suggest that the mechanisms of action include effects on inflammation, angiogenesis, and apoptosis, and animal models suggest benefits in retinal outcomes.53

In 2013 Australia became the first country to approve a fenofibrate indication for secondary prevention of diabetic retinopathy. The 2013 Canadian retinopathy guidelines strongly recommended it.54 The 2016 ADA guidelines mention that diabetic retinopathy progression may be slowed by fenofibrate but do not consider it a strong recommendation.4

A forthcoming study will assess the effect of fenofibrate on circulating endothelial progenitor cells,55 which may enhance our knowledge on its potential role for primary and secondary prevention of diabetic retinopathy. The study was estimated to be completed by December 2015, but results have not yet been published.

### Laser photocoagulation for non-center involving diabetic macular edema

Laser photocoagulation is a procedure implemented in severe diabetic retinopathy, consisting of light energy applied to the retina which stops neovascularization, limits the progression of retinopathy, and preserves visual function. A systematic review in 2014 assessed the effects of laser photocoagulation for diabetic retinopathy compared with no treatment or deferred treatment.56 Five RCTs were included, four in proliferative diabetic retinopathy and one in non-proliferative diabetic retinopathy. Laser photocoagulation significantly reduced the risk of progression (risk ratio 0.49, 037 to 0.64) and severe visual loss at 12 months (0.46, 0.24 to 0.86). None of the included studies reported patient relevant outcomes such as quality of life, pain, loss of driving license, or retinal detachment, which would be preferable in future studies.

### Intravitreal agents for center-involving diabetic macular edema

Vascular endothelial growth factor (VEGF) is a potent angiogenic and vaso-permeability factor believed to be important in the pathophysiology of diabetic retinopathy. Drugs that directly inhibit VEGF include pegaptanib (a VEGF aptamer), ranibizumab (a monoclonal antibody fragment), and bevacizumab (a full length antibody).57 Anti-VEGF agents are injected as office procedures, under topical anesthesia. These therapies may be replacing laser photocoagulation as first line therapy for severe diabetic retinopathy; however, their availability may vary and laser photocoagulation may still be needed to control disease progression.58

Two RCTs confirmed the short term benefits (at 12 months) of ranibizumab for diabetic macular edema (DME). The first one in 2010 compared 0.5 mg ranibizumab plus prompt (within one week) or deferred (≥24 weeks) laser therapy with sham injection or triamcinolone plus prompt laser therapy (768 eyes of patients with type 2 diabetic).59 It found that ranibizumab with laser (prompt or deferred) offered greater improvements in visual acuity compared with groups that did not receive ranibizumab.59 The authors reported that 50% of eyes treated with ranibizumab experienced 10 letter gain from baseline and 30% gained 15 letters (three lines on the eye chart) , with improvements in letter score visual acuity at one year with prompt (+9 (SD 11); P<0.001) or deferred (+9 (SD 12; P<0.001) laser therapy. This was not observed in the triamcinolone group with prompt laser (+4 (SD 13); P=0.31).

The other study, published in 2011, was the Ranibizumab Monotherapy or Combined with Laser versus Laser Monotherapy for Diabetic Macular Edema (RESTORE) study.60 Patients (302 with type 2 diabetes) were randomized to ranibizumab (0.5 mg) plus sham laser, ranibizumab plus laser, or sham ranibizumab plus laser therapy. Visual acuityimproved from baseline to month 1 (+6.1 (SD 6.4); P<0.001) and month 12 (+5.9 (SD 7.9); P<0.001) in patients treated with ranibizumab, but not in those who did not receive it (+0.8 (SD 8.6)). In addition, this study included assessment of health related quality of life, assessed with the National Eye Institute Visual Function Questionnaire. Results indicated significant improvements for composite score and vision related subscales.60

A multicenter RCT in 2013 compared the effects of different lengths of follow-up and treatment with ranibizumab (24 or 36 months) in the Ranibizumab for Edema of the Macula in Diabetes (READ-2) study.61 The study was undertaken to assess whether more frequent injections (monthly instead of every two months) for an additional 12 months (month 24 to 36) further improved visual outcomes (visual acuity and foveal thickness) and ultimately resolved macular edema. The study included 33 patients treated with ranibizumab, 34 treated with laser, and 34 with ranibizumab plus laser therapy. However, the authors reported benefits in individuals within each group, suggesting heterogeneity among patients with diabetic macular edema. Some patients improved with more frequent injections and others with less frequent injections; others, especially those receiving laser or ranibizumab plus laser therapy, achieved resolution of DME.

An RCT in 2015 evaluated changes in vision related quality of life in 70 patients with severe diabetic retinopathy (clinically significant DME).62 Participants were randomly allocated to focal/grid laser or intravitreal ranibizumab applied to the better seeing eye. Baseline retinal parameters were similar in the two groups. After six months, outcomes (distance visual acuity, near visual acuity, central retinal thickness; all with P<0.01) and vision related quality of life (P=0.03) were better in the ranibizumab group. Despite limitations in duration, sample size, and study design, the results support anti-VEGF as an intervention for severe diabetic retinopathy with DME. Future research using similar patient relevant outcomes is needed.

An ongoing multicenter RCT will randomly assign 322 patients to the anti-VEGF aflibercept, photocoagulation (focal/grind and panretinal), or a prompt sham injection.63 Primary outcomes for secondary prevention of proliferative diabetic retinopathy will include development of neovascularization, procedures for diabetic retinopathy, development of DME, and non-topical treatments for DME. Completion is scheduled for January 2022, with primary outcome measures estimated for January 2020.

### Protein kinase C inhibitor: ruboxistaurin

Ruboxistaurin is a specific inhibitor of protein kinase C (PKC) β, suggested to reduce visual loss in diabetic retinopathy through anti-angiogenic effects reported in vitro and in vivo.64 However, the benefits seemed to be limited unless more severe retinopathy was present at baseline or the higher dosage of 32 mg/day was used.65 66 67 A follow-up study in 685 patients with severe non-proliferative diabetic retinopathy compared ruboxistaurin 32 mg/day versus placebo (PKC-DRS2).68 Results included better vision (improvement of ≥15 letters, 4.9% v 2.4%; P=0.027) and reduced events of vision loss (5.5% v 9.1%; P=0.034), DME progression (50% v 68%; P=0.003), and need for laser treatment (26% less frequent; P=0.008).

Subsequently, an RCT in 2007 compared ruboxistaurin (4, 16, or 32 mg/day) with placebo for 30 months in 686 patients with non-center DME (PKC-DMES).69 Ruboxistaurin did not reduce the need for focal/grid laser photocoagulation. Despite that, it significantly delayed the progression of sight threatening DME, defined by retinal thickening (hazard ratio 0.66, 0.47 to 0.93; P=0.02). An open label extension of PKC-DRS2 recruited 203 participants from the previous study (103 received intervention, 100 placebo), who were off study medications for approximately one year.70 Then, all participants received ruboxistaurin 32 mg/day for another two years. After six years, those who had had the greatest exposure (five years) experienced less sustained visual loss compared with those treated for only two years (8% v 26%). Despite the limitations of this study design, the results suggested a potential role for ruboxistaurin in delaying the progression of severe diabetic retinopathy. We did not find subsequent trials that confirmed these findings. Furthermore, this agent may no longer be available for further research or clinical use.

### Other agents

Our literature search retrieved reports on Qi ming granule, calcium dobesilate, and Chinese green tea,71 72 73 74 as agents that helped to decrease the progression of diabetic retinopathy. However, the level of evidence was considered too poor for inclusion.

### Bariatric (metabolic) surgery

Recent observational studies indicate that bariatric surgery may provide benefits to DMC beyond its effect in glycemic control, owing to improvements in risk factors for diabetic retinopathy (components of the metabolic syndrome: obesity, hypertension, dyslipidemia) subsequent to the weight loss.75 76 Nevertheless, other observational studies indicated a post-surgical worsening or new onset of diabetic retinopathy, despite glycemic improvements.77 78

A systematic review and meta-analysis in 2015 assessed the effects of bariatric surgery on diabetic retinopathy.79 No RCTs were found, but four non-randomized case series studies were included. From the 80 patients without preoperative diabetic retinopathy, only 7.5% (SD 7.4%) went on to develop it, and the rest remained disease free. Of the 68 patients with preoperative diabetic retinopathy, only 19.2% (SD 12.9%) showed improvement, 57.4% (18.5%) had no change, and 23.5% (18.7%) had progression.

A more recent, small observational study assessed the need for more frequent ophthalmologic evaluations and perioperative monitoring for diabetic retinopathy in patients with type 2 diabetes undergoing bariatric surgery.80 The authors found no evidence to support a change to the standard screening intervals.

Subsequently, a large retrospective study with 10 years’ follow-up assessed the risk of DMC in 4683 patients who underwent bariatric surgery.81 The study focused on new onset disease after the procedure, according to remission and relapse of type 2 diabetes. A 29% reduction in new onset microvascular complications was seen in those who experienced remission of type 2 diabetes compared with those who did not. Even in those whose diabetes initially remitted but then relapsed, DMC was reduced (hazard ratio decreased by 19% for each year in remission), suggesting a legacy of effect from bariatric surgery. Certainly, long term prospective trials are warranted to monitor the outcomes from bariatric surgery in terms of diabetic retinopathy and the other DMC. It is noteworthy that treating obesity is an integral component of type 2 diabetes management, and given the obesity epidemic we can expect more people with type 2 diabetes and diabetic retinopathy who will need bariatric surgery.

## Primary and secondary prevention of diabetic nephropathy

### Screening

The ADA and NICE recommend measuring spot urinary albumin to creatinine ratio (UACR) and glomerular filtration rate (GFR) on diagnosis of type 2 diabetes and then yearly.4 82 Because of the known intra-person variability, guidelines recommend collection of at least three samples within a six month period and define diabetic nephropathy as when two of the three are abnormal. NICE guidelines provide details on stratification of albuminuria (A1 <3 mg/mmol, A2 3-30 mg/mmol, A3 >30 mg/mmol),82 and the ADA guidelines offer further detailed description of considerations for screening and monitoring (exercise, fever, congestive heart failure, marked hyperglycemia, menstruation, and marked hypertension may elevate UACR).4

### Blood pressure control

#### Target and monitoring

The ADA and NICE guidelines recommend monitoring blood pressure at least annually, unless hypertension or renal disease is already present.4 82 Both recommended a blood pressure target of below 140/90 mm Hg for patients without diabetic nephropathy and below 130/80 mm Hg for those with diabetic retinopathy, diabetic nephropathy, or cerebrovascular damage, highlighted by NICE. In addition, the ADA recommends against lowering diastolic blood pressure (DBP) below 70 mm Hg (or below 60 mm Hg in older adults) owing to potential adverse outcomes.

Notably, no RCTs have been specifically designed to determine whether intensive lowering of SBP prevents the onset of DMC. On the other hand, consistently lowering SBP without lowering DBP may prove challenging and even risky, especially in older adults with arterial stiffness.83 In the setting of diabetes and absence of diabetic nephropathy (primary prevention) the role of lowering SBP below 130 mm Hg has been considered, especially in minorities at higher risk for DMC, such as Asian,84 African American,85 and Hispanic populations.86 Furthermore, the results of the Systolic Blood Pressure Intervention Trial (SPRINT), which recruited 9361 patients without diabetes,87 triggered a conversation at several scientific meetings in 2016 as better outcomes were observed in the group targeting an SBP of 120 mm Hg compared with the 140 mm Hg group. Patients with type 2 diabetes without DMC could benefit from blood pressure closer to 120 mm Hg for primary prevention.

#### Renin-angiotensin-aldosterone system

A 2006 systematic review evaluated the benefits and harms of ACE inhibitors and ARBs for secondary prevention of diabetic nephropathy.88 It included 38 studies that compared ACE inhibitors with placebo (vast majority of participants with type 2 diabetes) and four studies that compared ARBs with placebo (all with type 2 diabetes). Both agents seemed to provide similar beneficial effects (remission or prevention of progression of albuminuria and ESRD, decreased doubling of creatinine) when compared with placebo. The review included seven studies comparing ACE inhibitors and ARBs (557 patients), six of them with type 2 diabetes cases and one with combined cases. Analysis showed that ACE inhibitors used at maximum tolerable dose offered greater additional survival benefits (five studies, 2034 patients; risk ratio 0.78, 0.61 to 0.98), whereas ARBs did not reduce all cause mortality. As ACE inhibitors cost less, they were suggested as the first line option.

ROADMAP is a large, multicenter RCT in Europe that recruited 4447 patients with type 2 diabetes, with baseline blood pressure above 130/80 mm Hg or taking antihypertensive drugs, and without albuminuria, to compare the ARB olmesartan 40 mg/day with placebo.89 90 91 After 3.2 years, olmesartan delayed the time to onset of albuminuria by 25% (hazard ratio 0.75, 0.61 to 0.92; P=0.007), independent of baseline and follow-up SBP. Study limitations included within and between patient variation in albuminuria, the short follow-up period, and the difficulty of separating drug specific effects and those related to blood pressure lowering. A follow-up report from the ROADMAP researchers in 2016 indicated that patients taking olmesartan had a sustained reduction of albuminuria.92

Another sub-analysis of ROADMAP in 2014 indicated that heart rate was associated with diabetic nephropathy outcomes. Using the heart rate recorded during visits, it found that baseline low heart rate reduced the risk of albuminuria only in the group taking olmesartan (P=0.002).93 However, a major limitation of this analysis was the concomitant use of negative chronotropic agents.

A 2012 systematic review looked at the role of antihypertensive agents.94 It included 26 RCTs (61 264 patients with or without hypertension and kidney disease) that compared any antihypertensive drug with placebo or another agent. Eight studies (11 906 participants) looked at the effect of ACE inhibitor (compared with placebo) in reducing the risk for new onset nephropathy. ACE inhibitors significantly reduced diabetic nephropathy in patients with or without hypertension (risk ratio 0.71, 0.56 to 0.89). Five studies (1253 participants) compared ACE inhibitors with calcium channel blockers (CCBs) and found similar results (risk ratio 0.60, 0.42 to 0.85). Conversely, another five studies (7653 participants) looked at ARBs, without finding a significant reduction in new onset nephropathy (risk ratio 0.90, 0.68 to 1.19).

Another 2012 systematic review assessed renal outcomes from ACE inhibitors and ARBs compared with other antihypertensive agents (18 studies) or placebo (13 studies) in patients with type 2 diabetes.95 Compared with placebo, ACE inhibitor/ARB reduced progression to ESRD (risk ratio 0.8, 0.69 to 0.93), decreased doubling of serum creatinine (0.76, 0.69 to 0.84), and reduced albuminuria greater than 300 mg/day (0.67, 0.54 to 0.83), albeit with no difference for albuminuria between 30 and 300 mg/day. Compared with other antihypertensive drugs, there were no significant differences in progression to ESRD and albuminuria, but other patient relevant benefits were greater with ACE inhibitor/ARB, such as reduction in serum creatinine doubling (risk ratio 0.66, 0.53 to 0.83) and trends for better regression of albuminuria greater than 300 mg/day (0.71, 0.5 to 1.0). The review highlighted the limitation of not having data on blood pressure.

A few studies have assessed the use of ARBs in specific subpopulations. A Japanese/Chinese trial randomized 557 patients with albuminuria to olmesartan or placebo.96 More than 90% were taking at least one other antihypertensive agent (73% an ACE inhibitor). After a mean follow-up of 3.2 years, no additional renal benefits were seen with olmesartan. Another small Japanese study used telmisartan in 68 patients with early diabetic nephropathy with albuminuria,97 which decreased after 12 months of intervention (P<0.001). The authors suggested that telmisartan could be more potent that other previously tested ARBs (olmesartan, candesartan, losartan).

An RCT of 169 patients in 2013 assessed the effect of losartan copmpared with placebo in native Americans with type 2 diabetes, with follow-up for six years.98 It measured both functional (GFR) and structural (mesangial fractional volume through kidney biopsy) renal outcomes. Losartan improved structural outcomes only in patients with albuminuria at baseline (reduction in mesangial fractional volume, P=0.02; increased filtration surface area density, P=0.03); no significant difference in these parameters was seen in patients with normoalbuminuria at baseline. Limitations included the use of a specific subpopulation, the confounding occurrence of pregnancy in 19% of participants, and the use of other antihypertensive drugs. A follow-up report indicated that morphometrically determined structural variables predicted loss of renal function in native Americans.99 The results indicate a potential role for baseline structural parameters to predict the effect of RAAS interventions in the prevention of diabetic nephropathy.

In general, patients with type 2 diabetes and indication for antihypertensive treatment are given RAAS blockers as first line treatment, with evidence favoring ACE inhibitors over ARBs and over other alternate agents.

A decade ago it was postulated that the combination of ACE inhibitors and ARBs could be more beneficial than either agent alone.100 The Ongoing Telmisartan Alone and in Combination with Ramipril Global End Point Trial (ONTARGET) was a large RCT that randomized 25 620 patients to ramipril (8586), telmisartan (8542), or both (8502).101 Although ARB combined with ACE inhibitor reduced proteinuria more than ACE inhibitor alone (P<0.001), it worsened the composite primary renal outcome of dialysis, serum creatinine doubling, and death (hazard ratio 1.09, 1.01 to 1.18; P=0.037). Moreover, the events of hypotension and discontinuation of treatment due to hypotension were also more frequent in the combination group. Although this was an important study, it did not focus on type 2 diabetes nephropathy. A small Brazilian study evaluated the combination of enalapril and losartan in 56 patients with albuminuria and found no additional benefit from adding the ARB.102

Subsequently, the Veterans Affairs Nephropathy in Diabetes (VA NEPHRON-D) study was designed to test the efficacy of losartan versus the combination of losartan and lisinopril.103 This study recruited 1448 patients with type 2 diabetes, estimated GFR of 30-89.9 mL/min/1.73 m2, and UACR greater than 300 mg/mmol, treated with losartan and randomly allocated to add lisinopril or placebo. The study was stopped early owing to safety concerns. The combination group had a significantly higher incidence of adverse reactions: acute kidney injury (AKI) (hazard ratio 1.7, 1.3 to 2.2, P<0.001) and hyperkalemia (2.8, 1.8 to 4.3; P<0.001).

This was followed in 2015 by a network meta-analysis of 157 studies including 43 256 patients, most with type 2 diabetes and CKD, aimed at investigating the benefits and harms of combining multiple antihypertensive drugs.104 The combination of ACE inhibitor and ARB was the most effective against ESRD (network odds ratio 0.62, 0.43 to 0.90), without significantly increasing AKI (2.69, 0.98 to 7.38) or hyperkalemia (2.69, 0.97 to 7.47). The authors suggested that combining ACE inhibitor and ARB therapy in 1000 adults with type 2 diabetes and CKD for one year would avoid ESRD in 14 patients, foster albuminuria regression in 208, at the expense of AKI in 55 and hyperkalemia in 135. Problems with design and data availability are limitations. However, the study highlighted the potential role of dual RAAS blockade in selected patients.

Most recently, in 2016, another network meta-analysis with a relatively similar design assessed renal outcomes from combined RAAS blockade and other antihypertensive drugs in patients with type 2 diabetes.105 Once again, ACE inhibitor plus ARB combined therapy was not superior to either agent alone, with the same risks as above.106

#### Mineralocorticoid receptor antagonists

The use of steroidal (spironolactone and eplerenone) and non-steroidal (finerenone) mineralocorticoid receptor antagonists (MRAs) has been studied in the past decade. A multicenter RCT in 2006 evaluated different dosages (50 and 100 mg/day) of the selective MRA eplerenone (n=91 and n=85, respectively), or placebo control (n=91), added to enalapril 20 mg/day.107 After 12 weeks, eplerenone significantly reduced albuminuria (41% reduction from baseline in the 50 mg group and 48.4% in the 100 mg group, compared with 7.4% in the placebo group; P<0.001), without increased hyperkalemia.

A small RCT in 2006 evaluated the effects of adding the non-selective MRA spironolactone (n=29) versus placebo (n=30) in patients taking an ACE inhibitor or ARB.108 The researchers described improved antiproteinuric effects when adding spironolactone compared with placebo, as albuminuria reduced by 40.6% (95% confidence interval 23.4% to 57.8%) but did not change with placebo (P<0.001). A few other studies followed.

A systematic review and meta-analysis was published in 2014.109 It aimed to evaluate the effect of eplerenone and spironolactone, alone or in combination with ACE inhibitor or ARB, in adults with CKD and proteinuria. Of the 27 selected studies (1549 participants), 13 included patients with diabetic nephropathy, mostly from type 2 diabetes. No significant differences were seen in death, cardiovascular events, ESRD, or worsening GFR. Regarding proteinuria, the results favored spironolactone combined with ACE inhibitor or ARB versus ACE inhibitor, ARB, or both (standardized mean difference −0.61, 95% confidence interval −1.08 to −0.13). Nonetheless, greater hyperkalemia (risk ratio 2.00, 1.25 to 3.20) and gynecomastia (5.14, 1.14 to 23.23) were observed. The analysis on six studies including patients with diabetic nephropathy showed less proteinuria with spironolactone (standardized mean difference −0.52, −0.82 to −0.23). Regarding eplerenone, combined with ACE inhibitor or ARB, compared with ACE inhibitor, ARB, or both, no significant difference in hyperkalemia was seen (two studies: risk ratio 1.62, 0.66 to 3.95). For proteinuria, only the results from single studies were presented (not in meta-analysis). The five included studies showed reductions in albuminuria/proteinuria (two described highly significant differences for lower albumin excretion; P<0.001) in patients treated with eplerenone plus ACE inhibitor or ARB, compared with those taking ACE inhibitor, ARB, or both. The results favored the potential role of aldosterone antagonists.

Most recently, in 2016, another systematic review aimed to balance the benefits of MRA (spironolactone) with the risk of hyperkalemia in CKD.110 Nineteen trials (with 1644 participants) were included, most in patients with CKD due to diabetic nephropathy. Similarly to the above, adding MRA to ACE inhibitor or ARB significantly improved blood pressure and albuminuria control but increased hyperkalemia.

#### Other antihypertensive agents

Older, smaller studies suggested benefit with the non-dihydropyridine CCB verapamil in reducing albuminuria. A large multicenter RCT (BENEDICT study) randomized 1204 patients with type 2 diabetes, normoalbuminuria, and hypertension to trandolapril plus verapamil, each agent alone, or placebo.111 Albuminuria, defined as two consecutive values greater than 20 μg/min of overnight albumin excretion, occurred less frequently in patients treated with the ACE inhibitor alone (6%) or CCB plus ACE inhibitor combination (5.7%) than those taking CCB alone (11.9%) or placebo (10%). The authors concluded that the combination of CCB and ACE inhibitor reduced the onset of albuminuria, even after adjustment for SBP (hazard ratio 0.46, 0.22 to 0.93; P=0.03) and DBP (0.46, 0.22 to 0.95; P=0.04). The researchers followed a subset of 281 patients who had developed albuminuria (BENEDICT-B study), and randomized them to CCB plus ACE inhibitor versus ACE inhibitor alone. After two years of follow-up, addition of CCB to ACE inhibitor did not offer additional renal improvements.112

### Early referral to nephrologist and better quality of life

A 2014 systematic review evaluated the benefits in terms of quality of life of early versus late referral to nephrology services.113 Only cohort studies (40, with 63 887 patients) were included as no RCTs were found. Patients referred early were more likely to receive renal replacement therapy with peritoneal dialysis (risk ratio 1.74, 1.64 to 1.84), earlier placement of arteriovenous fistula (3.22, 2.92 to 3.55), and reduced mortality (0.61, 0.55 to 0.67). Although the quality of the studies was only low to moderate, and analysis showed high heterogeneity, the findings were consistent, indicating benefits for early referral.

### Fenofibrate

The effects of peroxisome proliferator activated receptor α agonists in diabetic nephropathy were previously considered in older, small scale studies, with inconsistent results.114 115 116 The FIELD study, described above in the diabetic retinopathy section, evaluated diabetic nephropathy as a tertiary pre-specified outcome, assessing the progression to ESRD, dialysis, renal transplant, or death from renal disease. The study randomly allocated 9795 patients aged 50 to 75 with type 2 diabetes and estimated GFR 30 mL/min/1.73 m2 or above to fenofibrate 200 mg or placebo. Although no harm from long term fenofibrate use was seen in moderate CKD, no benefits were seen either.117 No newer studies have assessed whether these agents can slow progression of diabetic nephropathy.

### Sodium/glucose cotransporter 2 inhibitors

A potential role of sodium/glucose cotransporter 2 (SGLT-2) inhibitors in diabetic nephropathy, beyond the expected benefits from glycemic control, has been proposed in the past few years. An initial study examined the effect of the SGLT-2 inhibitor dapagliflozin on paramaters of renal glucose reabsorption. The researchers recruited 24 patients (12 with diabetes, 12 healthy controls) and found changes in the threshold of renal glucose reabsorption, which is abnormal in diabetes.118

Additional improvements in GFR were observed in another small RCT in which 75 patients with type 2 diabetes without diabetic nephropathy were randomly allocated to SGLT-2 inhibitor, hydrochlorothiazide, or placebo.119 The favorable GFR changes were related to better SBP control and decreased circulating volume.

Experimental and mice models supported this hypothesis,120 121 but rat models did not find renoprotection.122 Although the direct mechanism in type 2 diabetes is not fully understood, small, short duration studies have suggested that SGLT-2 inhibitors enhance markers of arterial stiffness and vascular resistance and perhaps have direct renovascular effects.123 The potential interaction with the RAAS and the suggested independent action on renal tubular function could offer synergistic effect with ACE inhibitors/ARBs.124 125 126

The recently published Empagliflozin, Cardiovascular Outcomes, and Mortality in Type 2 Diabetes trial (EMPA-REG OUTCOME) was a multicenter RCT with 7020 patients, assigned to empagliflozin 10 or 25 mg/day or placebo. The primary outcome was a composite of cardiovascular events, but the study had pre-specified microvascular outcomes, including diabetic nephropathy related outcomes.127 After a mean follow-up of 3.1 years, lower rates of incident or worsening diabetic nephropathy were seen in patients who received empagliflozin (10 or 25 mg/day) (525/4124, 12.7%; hazard ratio 0.61, 0.53 to 0.70), compared with those in the placebo group (388/2061, 18.8%). The additional beneficial effects on blood pressure and HbA1c reduction make this class of agents appealing for future research, contrasted with their safety profile and risk for adverse reactions. Further studies are likely to follow to increase understanding of their role in new onset or short duration type 2 diabetes, with and without diabetic nephropathy.

### Pentoxifylline

Use of pentoxifylline for diabetic nephropathy started more than 30 years ago, given its anti-inflammatory, anti-fibrotic, and anti-proliferative properties that may aid the secondary prevention of diabetic nephropathy.128 129 A Korean multicenter RCT evaluated the effect of pentoxifylline 1200 mg/day or placebo in 174 patients with diabetic nephropathy who had albuminuria (>30 mg/g creatinine) while taking an ACE inhibitor or ARB for six months.130 Proteinuria decreased significantly in the pentoxifylline group (23% v 4%; P=0.012), but no difference was seen in estimated GFR between the groups. An older and smaller Mexican RCT had evaluated the same dose for a longer duration (12 months) in 37 patients, 20 with early and 17 with moderate diabetic nephropathy.131 Despite the small sample size and lack of intention to treat analysis after three patients withdrew, the results were similar to those of the Korean study. A more recent single center Spanish RCT assessed pentoxifylline in 169 patients with CKD stage 3-4, treated with ACE inhibitor or ARB.132 Limited resources prevented a placebo control. The results showed reduction in albuminuria and slowing in GFR decline in patients treated with pentoxifylline. Taking into consideration the limitations in study design and sample size, these three pilot studies, in three different subpopulations, presented consistent results favoring pentoxifylline.

More recently, two systematic reviews looked at the efficacy and safety of combining pentoxifylline with ACE inhibitors and ARBs and its usefulness in slowing down the progression of nephropathy in patients with or without diabetes.133 134 The first review (2015) included eight RCTs (587 patients with type 2 diabetes), with a median duration of five months. Pentoxifylline contributed to the antiproteinuric effect independently of changes in blood pressure. Five studies showed decreased urinary protein excretion (standard mean difference 0.76, 0.52 to 0.99; P<0.001), and the other three showed decreased albuminuria (0.36, 0.12 to 0.59; P=0.003).133 The second review (2016) included 26 randomized and quasi-randomized trials, 24 of which focused on diabetic nephropathy (1049 patients with type 2 diabetes). Results showed that albuminuria was reduced in subgroup analysis of patients with higher (≥500 mg/day) baseline proteinuria (mean difference −0.38 g/day, −0.63 to −0.14) and in studies comparing pentoxifylline with RAAS blockers (−0.42, −0.76 to −0.08). There was also a significant improvement in renal function (estimated GFR/creatinine clearance), with a mean difference of 3.42 (2.10 to 4.75) mL/min (P<0.001). No benefits were found for total urinary albumin excretion and serum creatinine.134 Both systematic reviews were limited by the quality and duration of the included studies, and the authors called for studies that assess hard outcomes rather than surrogate outcomes.

### Bardoxolone methyl

This oral antioxidant and inflammation modulator is proposed to prevent the progression of advanced diabetic nephropathy.135 136 A 2011 phase II, double blind RCT studied 227 adults with diabetes and CKD (GFR 20-45 mL/min/1.73 m2).137 Patients were allocated to placebo or increasing dosages of bardoxolone methyl (25, 75, or 150 mg/day). Despite increases in estimated GFR for the groups receiving the antioxidant (8.2 (SD 1.5), 11.4 (1.5), and 10.4 (1.5) mL/min/1.73 m2, respectively, for the three doses; P<0.001), these were not dose dependent. A 2015 review described the termination of the phase III Bardoxolone Methyl Evaluation in patients with CKD and type 2 diabetes (BEACON) trial, halted at nine months owing to excessive adverse outcomes (heart failure, non-fatal myocardial infarction or stroke, death from cardiovascular disease).138 This agent is unlikely to be considered for future research. We did not find newer reports.

### PKC inhibitor ruboxistaurin mesylate

Ruboxistaurin is a highly specific inhibitor of PKC-β isoforms, which are associated with increased concentrations of transformation growth factor β, fibronectin, and collagens, which lead to diabetic nephropathy damage (glomerular hypertrophy, extracellular accumulation, oxidative stress, and albuminuria).139 Human and animal studies suggested nephroprotective effects, with several limitations.140 One study randomly allocated 707 patients to ruboxistaurin or placebo, for a mean duration of 2.7 years.141 142 Although the nephropathy outcomes (changes in UACR and estimated GFR) were reported as favorable trends by the authors and supporters, the results did not show significant differences between the groups. We have not found further research using this agent for diabetic nephropathy.

### Hyperuricemia

In the past decade, reports suggest that asymptomatic hyperuricemia is an independent risk factor for new onset and progression of diabetic nephropathy.143 144 145 Unfortunately, the evidence to support benefit for diabetic nephropathy outcomes from hyperuricemia treatment in type 2 diabetes is not strong, and only a few clinical trials have been conducted. We did not find RCTs, but a Chinese randomized, open, parallel controlled study in 176 patients with type 2 diabetes and asymptomatic hyperuricemia was reported in 2015.146 The researchers allocated 82 patients to receive allopurinol (mean uricemia 433 (SD 11) μmol/L) and 70 patients to the control group, without placebo (mean uricemia 432 (9) μmol/L). After three years of follow-up, the group receiving allopurinol had improved GFR and UACR.

### Alcohol consumption

A 2014 sub-analysis of the Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified-release Controlled Evaluation (ADVANCE) study explored the effects from alcohol use.147 After adjustment for confounders, moderate alcohol consumption (≤21 drinks/week in men, ≤14 drinks/week in women) was associated with lower incidence of worsening of diabetic nephropathy and diabetic retinopathy compared with no alcohol consumption. Remarkably, the difference was barely significant (adjusted hazard ratio 0.85, 0.73 to 0.99; P=0.03). Given the limitations of retrospective design, potential bias, and limited results, the evidence does not support moderate consumption of alcohol and does not discourage recommendations to limit or abstain from it.

### Bariatric surgery and diabetic nephropathy

A systematic review in 2015 explored renal benefits in patients with type 2 diabetes undergoing bariatric surgery.148 Eleven studies (in 530 patients) were included in the meta-analysis. Only one of the studies had a control group. Analysis compared UACR and albuminuria/day between before and after surgery. Improvements were seen in UACR (mean difference −6.6 (standard error 1.32); P<0.01) and albuminuria/day −55.76 (18.55); P<0.01), with no evidence for renal harm. Long term studies for reversibility of diabetic nephropathy, including patient relevant outcomes such as lifelong risk for renal replacement therapy, are needed.

The upcoming Prevention of Microvascular Complications in Overweight Diabetics With Surgery or Best Medicine (PROMISE) is a multicenter RCT that aims to recruit 100 patients with type 2 diabetes, increased weight (body mass index 28-35) and impaired renal function (estimated GFR 30-60 mL/min/1.73 m2 and albuminuria).149 All patients will receive optimal medical treatment, and 50 will be randomly assigned to gastric bypass surgery (this would be outside current standard body mass index criteria for surgery). After three years of intervention, the primary outcome will be renal function measured by iohexol clearance. The completion date for the study is estimated for December 2017.

## Primary and secondary prevention of diabetic neuropathy

Diabetic peripheral sensorimotor polyneuropathy (DPN) is the most common and more studied type of diabetic neuropathy. Several other types exist (autonomic neuropathy: hypoglycemia unawareness, resting tachycardia, orthostatic hypotension, gastroparesis, constipation or diarrhea, fecal incontinence, erectile dysfunction, neurogenic bladder, and sudomotor dysfunction with hyperhidrosis), for which the evidence for primary and secondary prevention is scarce. To aid the evaluation of diabetic neuropathy through history and examination, several specific tools exist which are described in detail elsewhere.150 151 This review will focus on DPN.

### Screening

All other potential causes (alcohol, chemotherapy, vitamin B12 deficiency, hypothyroidism, renal disease, paraneoplastic syndromes (multiple myeloma, bronchogenic carcinoma), HIV infection, chronic inflammatory demyelinating neuropathy, inherited neuropathies, and vasculitis) must be investigated and ruled out before making the diagnosis of DPN. As highlighted by the ADA guidelines,4 it is a diagnosis of exclusion. Screening includes 10 g monofilament testing and at least one of pinprick, temperature, or vibration sensation at onset of type 2 diabetes and then annually.4

### α-lipoic acid

α-lipoic acid is a potent antioxidant introduced in Germany, where most of the research has been conducted. A meta-analysis in 2004 of four RCTs combining data from 1258 patients (mostly with type 2 diabetes for >10 years, with DPN for 2.5 years) suggested benefits from α-lipoic acid (600 mg, infused over 30 minutes, Monday to Friday for two weeks, followed by Monday to Thursday for the third week) compared with placebo. Measurements included the total symptom score, which measures neuropathic symptoms (pain, burning, paresthesia, numbness) from 0 (no symptoms) to a maximum of 14.64 points (severe and continuous symptoms), and the neuropathy impairment score, evaluating weakness, reflexes, touch, vibration, and proprioception. The analysis found greater benefits in both scales for patients receiving α-lipoic acid (P<0.05).152

A newer (2009) multicenter, double blind trial randomized 460 patients with mild to moderate DPN to 600 mg of α-lipoic acid (n=233) versus placebo (n=227) for four years.153 It found no significant changes in the primary endpoints (neuropathy impairment score and neurophysiologic tests). However, the authors reported clinically meaningful DPN outcomes, defined as a change of 2 points in neuropathy impairment score, and fewer cases with DPN progression. Their analysis indicated that a greater percentage of patients in the intervention group showed clinically meaningful improvements, whereas a greater percentage in the control group showed progression of the disease. In spite of these findings, no more recent studies have been published. α-lipoic acid was approved in Germany, country of source for these publications; to our knowledge, it is not approved in other parts of the world.

### Aldose reductase inhibitors

These agents inhibit sorbitol and fructose accumulation through the polyol pathway, which includes the aldose reductase enzyme. Older pilot studies were conducted by Japanese and American researchers.154 155 A 2006 Japanese open label, multicenter RCT allocated 289 patients to the aldose reductase inhibitor (ARI) epalrestat and 305 to placebo.156 After three years of intervention, improvements were seen in DPN assessed with median motor nerve conduction velocity (improvement by 1.6 m/s in the epalrestat group compared with placebo; P<0.001). No benefits for cardiovascular autonomic neuropathy were seen. Another Japanese study of similar intervention duration albeit smaller size reported decreased symptoms of numbness or hyperesthesia.157 In 2007 a systematic review assessed the effects of ARIs on the progression of functional disability related to DPN.158 Thirty two RCTs were included, of which only 13 (879 participants in intervention group, 909 in the control group) had enough data for analysis, and no significant differences were observed between ARI and placebo.

Fidarestat has been used to reduce oxidative stress induced inflammatory disorders such as colon, breast, prostate, and lung cancer and asthma.159 Researchers suggested that ARIs could offer benefits in diabetes, given the inflammatory and pro-vasculoproliferative pathways. Basic science studies focused in the molecular characterization of ARIs, with efforts to increase their effectiveness and safety profile.160 161 More recently, an animal model compared ranirestat with epalrestat,162 finding that both agents enhanced DPN outcomes (nerve conduction velocity studies), whereas only ranirestat prevented cataracts. Nonetheless, at the moment, epalrestat is the only ARI approved in Japan for DPN.163

### PKC inhibitor ruboxistaurin

A systematic review in 2013 aimed to assess the efficacy and safety of ruboxistaurin in patients with DPN.164 It included six RCTs (285 participants) with at least six months of intervention that measured clinical changes (neurological total symptom score and vibration detection threshold) as primary outcomes, and included quality of life among the secondary outcomes. No significant difference was reported in the primary sensory outcomes. The secondary outcome of quality of life was significantly improved in two studies. The first was a small, randomized, six month pilot study in 2007 that compared ruboxistaurin 32 mg/day with placebo in 40 patients with DPN (83% with type 2 diabetes).165 Quality of life was assessed with the Norfolk Quality-Of-Life Questionnaire for Diabetic Neuropathy (QOL-DN). Ruboxistaurin significantly improved this measure compared with placebo (endpoint score compared with baseline −41.2% v −4.0%; P=0.04). The second was another small study comparing ruboxistaurin, topiramate, and placebo, administered for 18 weeks, in 54 patients with DPN (all with type 2 diabetes).166 Patients taking ruboxistaurin had improvements in the neuropathy total symptom score-6, from 4.38 (SD 0.75) to 1.49 (1.38) (P<0.01), whereas the topiramate and placebo groups did not show improvement. However, no significant difference was seen in the components and total neuropathy impairment score and neuropathy symptom score. Remarkably, improvements in QOL-DN were present with ruboxistaurin (−9.56 (SD 4.13); P<0.04) but were greater with topiramate (−12.22 (2.76); P<0.01). Authors called for longer duration studies, but despite the initial evidence we did not find newer studies.

### Exercise

A 2004 systematic review concluded that inadequate evidence existed to evaluate the effect of exercise on functional ability in people with DPN. Regardless, exercise is an integral component of the comprehensive individualized management of diabetes, especially for type 2 diabetes. Hence, further research is warranted.

A small subsequent study randomized 78 patients without baseline PDN to a supervised aerobic exercise intervention group (brisk walking) or control.167 After four years of specialized testing (vibration perception threshold, nerve distal latency, nerve conduction velocity, nerve action potential amplitude), results favored the exercise group, as much greater percentages of patients in the control group developed motor neuropathy (0% v 17%; P<0.05) and sensory neuropathy (6.45% v 29.8%; P<0.05). The authors acknowledged several limitations related to confounders (glycemic control, among others), although their results highlighted the effect of exercise training in the natural history of PDN. More recently, a small pilot study evaluated 17 patients with DPN before and after a short aerobic and resistance exercise intervention and reported improvements in cutaneous re-innervation and pain severity.168 A separate report indicated beneficial effects from exercise in a pilot study of patients with pre-diabetic neuropathy.169 Further research may increase our understanding of the effect of exercise on non-glycemic related factors for the primary and secondary prevention of DPN.

### Pulse infrared light therapy

Infrared laser therapy had been proposed as an intervention to enhance wound healing and peripheral sensitivity of patients with DPN.170 A small pilot study in 2006 of 22 Spanish patients with diabetes and chronic peripheral neuropathy (20 with type 2 diabetes) randomly assigned one foot to the intervention and the other to control.171 Pulse infrared light therapy was applied for 30 minutes a day, three days a week, for eight weeks. For the analysis, the researchers divided cases by degree of baseline peripheral protective sensation, Peripheral sensitivity improved significantly (P<0.05) at all tested sites (great toe and first, third, and fifth metatarsal). The authors suggested that the mechanism might be related to foot perfusion, although they did not find increased blood concentrations of nitric oxide after pulsed infrared light therapy.172

A retrospective cohort study of 252 patients reported that those who received infrared phototherapy had reversal of DPN after one year.173 Despite the limitations of the study design, it found benefits in patient relevant outcomes such as decrease in geriatric syndromes of falls and fear of falling, which can lead to morbidity and impaired quality of life. A small pilot study compared the effects of infrared light therapy with sham treatment in 24 patients (30 feet).174 After 12 treatments, no changes in perception threshold were seen.

Another small pilot study evaluated the efficacy of low level laser therapy, administered for 10 days, in 19 patients with painful DPN.175 It reported improvements in pain visual analog scale and vibratory perception threshold. More research is needed to confirm the role of infrared laser therapy in secondary prevention of DPN, considering the clinical utility for patient relevant outcomes, including pain, quality of life, mobility, and ability to walk and perform activities of daily living.

### Actovegin

Actovegin is a molecule produced from calf blood that stimulates oxygen absorption and use, as well as cellular energy metabolism,176 thus potentially counteracting the negative effect of glucotoxicity in cells. A 2009 multicenter RCT evaluated the effect of actovegin versus placebo in 567 patients with type 2 diabetes and symptomatic DPN, stratifying randomization according to insulin treatment.177 Actovegin 2000 mg/day intravenous infusion (or placebo), was followed by 1800 mg orally three times a day (or placebo) for 160 days. Results favored the actovegin treatment arm, with improvements in the neuropathy impairment score (−0.68 (SD 6.44); P=0.028) and the neuropathy symptoms and change score for weakness (−0.04 (0.26); P<0.05) and for severity (−0.05 (0.39); P<0.05). Limitations included the absence of an objective assessment of sensation.

A 2012 animal model reported that actovegin improved sensory nerve conduction velocity and intraepidermal nerve fiber density, suggesting a potential pathway to become a future disease modifying therapeutic agent.178 More research is warranted.

### Complex interventions for preventing diabetic foot ulceration

A systematic review in 2015 assessed the effectiveness of comprehensive integrative approach (patient education, podiatry care, foot ulceration risk assessment, and motivational coaching to reinforce self care behaviors, thus being on at least two different levels of care: the patient, the provider, and the healthcare system) for the prevention of foot ulcers most likely from DPN.179 It did not find high quality evidence and was unable to show a benefit of a combined approach compared with single interventions.

### Patient education for prevention of foot ulceration

Whereas monitoring for retinopathy and nephropathy relies mostly on office procedures and laboratory tests, monitoring for neuropathy needs more involvement from the patient. Patient education should be included in the management of DPN,180 and the prevention of foot ulcers and amputations are major outcomes of interest and targets for secondary prevention of damage from diabetic neuropathy.

A 2014 systematic review assessed the effects of education on the prevention of foot ulcers in diabetic patients.181 Twelve clinical trials were selected, of which only five reported primary endpoints from patient education. Owing to the limited quality of the studies, which were mostly short term and/or not sufficiently powered, the review concluded that insufficient evidence existed to support patient education alone as an effective intervention to reduce ulcers and amputations from diabetic neuropathy.

### Special considerations

Despite the biologic plausibility and suggested targeting of pathophysiologic pathways, even with reasonably designed clinical trials, many of the agents discussed in this review have not been approved by the US Food and Drug Administration or the European Medicines Agency. This was highlighted in a 2015 review,182 which offered an additional description of DPN.

## Emerging knowledge and strategies for DMC in type 2 diabetes

Newer well designed and powered studies with specific focus on microvascular endpoints as primary objectives and patient relevant outcomes are needed. This section describes registered clinical trials that are under way, the results of which may be useful to further review the utility of interventions described in this review and established newer targets for scientific research.

### New screening modality for diabetic retinopathy

Adherence to annual screening with dilated pupil photographs for the detection and monitoring of diabetic retinopathy has limitations. A Canadian RCT will use a special camera—non-mydriatic ultra-widefield retinal imaging—to avoid the need for dilated examinations. The study will randomly assign 740 patients to either usual screening or the experimental imaging procedure. Completion is estimated for March 2019.183

### α-lipoic acid for diabetic retinopathy

A double blind, single group assignment RCT will recruit 200 patients, to evaluate the role of α-lipoic acid in patients with moderate non-proliferative diabetic retinopathy, to test whether α lipoic acid can decrease the progression of diabetic retinopathy and preserve visual acuity. Completion is scheduled for January 2017.184

### Prediction of albuminuria and role of spironolactone in diabetic nephropathy

A multicenter RCT will recruit 3500 patients with normoalbuminuria from various European countries. Patients will be assigned to spironolactone, placebo, or standard care, stratified according to their urinary protein pattern (low risk or high risk). This study will assess the clinical utility of spironolactone reduce DPN progression, and it will also examine the role of urinary protein profile (within normoalbuminuria range) to identify patients at high risk of developing albuminuria. Completion is estimated for December 2017.185

### Early detection of foot ulcers in DPN

This Welsh phase I study will recruit 100 patients without history of diabetes or neuropathy and will evaluate a non-invasive infrared thermal camera that would aid in the detection of foot ulcers in DPN. Estimated completion is September 2017.186

### Home assessment of foot ulcers for DPN

An open label diagnostic study recruited 132 patients with history of diabetic neuropathy and previous foot ulcer to investigate the accuracy of a device, called smart foot mat, to detect signals associated with diabetic foot ulcers in patients at high risk. The study is listed as completed.187

## Conclusion

Understanding the multiple risk factors and pathophysiologic pathways for DMC in type 2 diabetes highlights the need to intensify anti-hyperglycemic and non-glycemic related interventions aimed at their primary and secondary prevention. Notably, many of these risk factors and pathophysiologic pathways are related, and although our review describes them individually we strongly recommend a patient centered, not single disease centered, approach.

The strongest evidence supports blood pressure control, although more research will help us to discern specific blood pressure targets, especially in patients without baseline hypertension and at low cardiovascular risk, as well as for primary prevention of DMC. The evidence behind RAAS blockade is strong, especially when albuminuria is present.

Evidence supports fenofibrate for secondary prevention of diabetic retinopathy, and future studies may assess its role in primary prevention. Laser photocoagulation and intravitreal anti-VEGF are used for therapeutic purposes and may slow progression of diabetic retinopathy. Bariatric surgery benefits diabetic nephropathy but requires careful screening and counseling for patients with known diabetic retinopathy. The pharmacologic interventions for DPN may require further confirmation with bigger prospective studies. Interventions such as exercise and patient education are justified as standard general interventions, even though only modestly supported by evidence for specific DMC. Nevertheless, they are an integral part of the care of patients with diabetes, as well as the assessment and active management of obesity. Nutritional interventions were not included in this review; in general, lifestyle interventions still need cost effectiveness analysis.

Given the substantial interest in patient centered outcomes, newer strategies should focus on early detection and timely prevention of further damage, functional outcomes, and quality of life. Precision medicine and genetic characterization may also play a role in enhancing and tailoring strategies for primary and secondary prevention. Further research into pathophysiologic pathways, risk factors including genetic risks, and precision medicine should shift the focus toward either preventing the development of complications or delaying their progression.

#### Glossary of abbreviations

• ACE—angiotensin converting enzyme

• AKI—acute kidney injury

• ARB—angiotensin receptor blocker

• ARI—aldose reductase inhibitor

• CCB—calcium channel blocker

• CKD—chronic kidney disease

• DBP—diastolic blood pressure

• DMC—diabetic microvascular complications

• DME—diabetic macular edema

• DPN—diabetic peripheral neuropathy

• ESRD—end stage renal disease

• GFR—glomerular filtration rate

• MRAs—mineralocorticoid receptor antagonists

• NICE—National Institute for Health and Care Excellence

• PKC—protein kinase C

• RAAS—renin-angiotensin-aldosterone system

• RCT—randomized clinical trial

• SBP—systolic blood pressure

• SGLT-2—sodium/glucose cotransporter 2

• UACR—urinary albumin to creatinine ratio

• VEGF—vascular endothelial growth factor

#### Questions for future research

• What is the role of genetic characterization of risk for the primary and secondary prevention of diabetic microvascular complications?

• Will there be a role for pharmacogenomics, and expanded use of both well established and currently off-label drugs, for the primary and secondary prevention of diabetic microvascular complications?

• Will studies with a multiple intervention design be preferred or superior to single intervention approaches, to deliver the maximum therapeutic effect for the primary and secondary prevention of diabetic microvascular complications?

We discussed individually with 10 patients their understanding of and interest in the prevention of microvascular complications. Patients were mostly interested in glycemic control and, therefore, were asked to comment on specific alternate interventions. Most were interested in screening, blood pressure, and pain control and were unaware of many of the suggested treatments found through our literature review. When asked about them, they suggested that we should be sure about the benefits before offering them. The patients expressed greater interest and priority for those with the strongest evidence. Of note, special interest was placed on bariatric surgery as a means of preventing complications. Our patients agreed on the need for a summary and a patient education sheet on such alternate interventions. The results of this review motivated us to conduct a quality improvement project to expand primary and secondary prevention of microvascular complications in our clinical practice

## Footnotes

• Contributors: WMV and HF engaged with BMJ editors for the planning, outline, and scope of this review. Both authors participated in the search and selection of publications, analysis and interpretation of data, and submission of the manuscript. WMV modified the manuscript following suggested by reviewers, with input from HF. WMV is the guarantor.

• Competing interests: We have read and understood the BMJ policy on declaration of interests and declare the following interests: none.

• Provenance and peer review: Commissioned; externally peer reviewed.

View Abstract