The retina is a highly metabolically active tissue and has the highest oxygen consumption per gram of tissue of all the organs of the human body. To supply this oxygen and nutrient demand, the retinal neurons that provide vision are heavily dependent on an adequate blood supply. Breakdown of this blood supply is the hallmark of a multitude of retinal vascular diseases that occur concurrently with damage to retinal neurons (78). Poor metabolic control over extended periods of time can lead to microvascular damage to the retinal circulation mediated by pericyte loss, basement membrane thickening, and endothelial cell dysfunction, along with nerve cell dysfunction and damage. These pathologic processes lead to capillary occlusion and progressive retinal nonperfusion, with subsequent local ischemia and disruption of the neurovascular unit as a whole (4). Vision impairment ultimately results from damage to the neurons, particularly the ganglion cells and their axons. Thus, diabetes-related retinopathy (DR) can be considered a form of diabetes-related sensory neuropathy, analogous to the diabetes-related peripheral sensory neuropathies (79).

There are many directly visible, ophthalmic examination findings associated with DR. The most common retinal findings include microaneurysms, intra-retinal hemorrhages, hard exudates, edema or thickening of the retina, venous beading, “cotton wool spots” (i.e., small yellowish-white deposits in the retina), intra-retina microvascular abnormalities, and pre-retinal neovascular tissue (Figures 57). By contrast, retinal nerve cells are transparent and not visible by ophthalmic or photographic examinations. Layers of retinal neurons and their synapses can be seen by optical coherence tomography (OCT) and reveal atrophy and disorganization of the retinal layers in early stages of DR (80,81).


Fundus photograph of a right eye with severe NPDR. Microaneurysms, intraretinal hemorrhages, hard exudates, cotton wool spots, and intraretinal microvascular abnormalities are visible.


Fundus photograph of a right eye with severe NPDR. Microaneurysms, intraretinal hemorrhages, hard exudates, cotton wool spots, and intraretinal microvascular abnormalities are visible.

Close modal

A) Fundus photograph of a left eye with PDR and DME. Microaneurysms, intraretinal hemorrhages, and hard exudates are visible. B) Wide-field fluorescein angiogram illustrating zones of retinal nonperfusion (*), diffuse microvascular leakage consistent with breakdown of the blood-retinal barrier, and prominent leakage temporally associated with pathologic neovascularization. C) OCT map showing multifocal zones of retinal thickening (highlighted red) consistent with DME. D) OCT line scan through the center of the macula showing central involvement of the DME with multiple intraretinal cysts (*).


A) Fundus photograph of a left eye with PDR and DME. Microaneurysms, intraretinal hemorrhages, and hard exudates are visible. B) Wide-field fluorescein angiogram illustrating zones of retinal nonperfusion (*), diffuse microvascular leakage consistent with breakdown of the blood-retinal barrier, and prominent leakage temporally associated with pathologic neovascularization. C) OCT map showing multifocal zones of retinal thickening (highlighted red) consistent with DME. D) OCT line scan through the center of the macula showing central involvement of the DME with multiple intraretinal cysts (*).

Close modal

Fundus photograph of a right eye with PDR and tractional retinal detachment, a common end-stage manifestation of untreated PDR. Extensive vitreous hemorrhage, pathologic vessels (neovascularization of the optic disc and neovascularization elsewhere), and a tractional retinal detachment predominately involving the superior and nasal retina with sparing of the macula are visible.


Fundus photograph of a right eye with PDR and tractional retinal detachment, a common end-stage manifestation of untreated PDR. Extensive vitreous hemorrhage, pathologic vessels (neovascularization of the optic disc and neovascularization elsewhere), and a tractional retinal detachment predominately involving the superior and nasal retina with sparing of the macula are visible.

Close modal

At present, DR severity is quantified using feature-based, structured grading of color images of the retina, referred to as fundus images, and allows the designation of an eye to a category on the Early Treatment Diabetic Retinopathy Study (ETDRS) DR severity scale (DRSS), which was defined and refined through the 1970s and 1980s and is still widely used today (30). From a clinical perspective, the DRSS is complicated and not intuitive; there are at least 12 distinct steps, each with discrete sub-levels. Simplified, however, the DRSS can be used to classify eyes into one of two clinically relevant DR categories: eyes with nonproliferative DR (NPDR; Figure 5) and those with proliferative DR (PDR; Figure 6).

The pathologic changes underlying DR lead to visual impairment through three inter-related mechanisms. First, retinal nonperfusion (Figure 6) leading to ischemia can directly impair retinal function. In response to local ischemia and breakdown of the blood-retinal barrier, a local inflammatory response is mounted that includes upregulation of multiple cytokines, including vascular endothelial growth factor-A (VEGF). Elevated levels of VEGF then directly contribute to visual impairment through two additional pathologic processes related to an aberrant wound-healing response. First, pathologic levels of VEGF further impair normal retinal vascular integrity, leading to breakdown of the blood-retinal barrier and exudation into the retina of fluid, lipids, and proteins that are normally contained intravascularly. This process leads to retinal thickening, referred to as diabetes-related macular edema (DME) (Figure 6), and dysfunction of the neuronal signaling needed for optimal visual function. Second, pathologic levels of VEGF can drive abnormal angiogenesis (the development of new blood vessels), a process termed PDR. These pathologic new vessels classically sprout from retinal veins and extend into the vitreous cavity along the collagen network composing the optically clear vitreous gel. At first, these vessels are isolated, but over time they recruit fibrotic components. These friable vessels can bleed into the vitreous cavity causing vitreous hemorrhage, and the fibrotic components can contract and cause retinal detachment (Figure 7).

Although the retina can be visualized directly using specific lenses in combination with an appropriate light source, management decisions often rely heavily on the use of exceptionally sensitive imaging techniques, so DR management often employs multimodal imaging.

Noninvasive OCT is a cornerstone of DME management and is used to confirm the presence of DME, quantify retinal thickness, and evaluate the effectiveness of intervention. Patients with any DR typically undergo OCT imaging at most, if not all, clinic encounters. In addition, OCT can be used to visualize individual retinal layers, which can be a helpful prognostic tool.

Color fundus photography, fluorescein angiography (FA), ultrawide-field (UW) imaging, and OCT angio- graphy (OCTA) are adjunct modalities that can also provide valuable information. Physicians can compare images longitudinally to look for DR progression and use images as a tool for patient education. Angiography can be used to evaluate retinal perfusion and vascular leakage. Beyond the macula and posterior pole, physicians often use UW imaging to more completely understand the severity of retinopathy across the entire retina and to guide treatment. For example, UW FA may demonstrate neovascularization and nonperfusion in the periphery that are not otherwise apparent (Figure 6), and this finding may alter classification and prognostication compared to clinical exam or more limited posterior pole fundus photography alone. OCTA provides high-resolution noninvasive images of the retinal vasculature without the use of the intravascular dye injection needed for traditional FA and is being used with increasing frequency in both clinical trials and daily practice.

Diabetes is the most common chronic systemic disease seen by ophthalmologists and is unique in that patients assume primary responsibility for their care, in contrast to cancer or infectious diseases. Therefore, the fundamental treatment of all forms of DR centers on the treatment of the underlying systemic risk factors, as detailed in the first chapter of this compendium. Factors such as diabetes duration, sex, and genetic susceptibility obviously cannot be altered. However, metabolic control, blood pressure, renal function, lipid levels, and nonhealing ulcers are modifiable and should be addressed in a comprehensive manner coordinated with patients, their physicians, and ophthalmologists (82). This intensive care is vital because DR severity can improve in 18% of patients with type 1 diabetes whose A1C declines by 1 percentage point (83). Moreover, the outcome of treatment with panretinal photocoagulation (PRP) for PDR is significantly better in patients with a baseline A1C <8% compared to those with an A1C ≥8% (84).

Three Clinically Relevant DME Sub-Categories

Historically, the threshold for treating DME with laser photocoagulation was set by the ETDRS in the 1980s as clinically significant macular edema. However, this definition has become outdated. More clinically relevant is classification of DME as either central-involved DME (CIDME) or non-CIDME.

In the setting of non-CIDME, the results of the ETDRS trial remain relevant to current practice, and laser photocoagulation appropriately applied to the macula remains a validated option for treatment. Specifically, laser photocoagulation significantly reduces the risk of moderate visual loss by approximately 50%, a protective effect that is independent of baseline visual acuity (85). Despite its potential value in slowing the progressive visual loss from DME, however, laser photocoagulation has limitations and possible untoward effects. For example, photocoagulation for DME has shown limited effectiveness in improving visual acuity and can rarely cause blind spots in the central visual field.

CIDME is most accurately determined by OCT and is defined by thickening that affects the 1-mm-diameter central sub-field of the macula. In the setting of CIDME, pharmacological management of DME is now first-line therapy in most clinical situations. Seven pharmaceutical agents encompassing two mechanistic classes are used for the treatment of DME. All are given by direct injection into the eye in an in-office procedure called an intravitreal (or intravitreous) injection. Engineered proteins, including the U.S. Food and Drug Administration (FDA)-approved ranibizumab (86) and aflibercept (87), as well as the non– FDA-approved bevacizumab (88), block the activity of VEGF. Alternatively, the FDA-approved dexamethasone (89) and fluocinolone acetonide (90) implants, as well as two formulations of triamcinolone acetonide—an FDA-approved preservative-free version and a non–FDA- approved preserved version—are corticosteroid agents.

In cases of CIDME, there are two clinically relevant sub-categories: those with preserved visual function and those with associated visual loss. Although there are limited prospective data to guide treatment of CIDME with preserved visual function, the field continues to move toward earlier intervention. Supporting earlier pharmacologic treatment, better visual function at the time of initiation of intravitreal pharmaceutical therapy for DME is associated with better long-term visual outcomes, a correlation that has demonstrated remarkable consistency across many exudative retinal diseases, including neovascular age-related macular degeneration and retinal venous occlusive disease (86,91,92).

In eyes with CIDME and visual loss, intravitreal pharmaceutical agent delivery is usually first-line therapy. Numerous well-designed, phase 3 clinical trials have demonstrated significant benefit with intravitreal pharmaceutical treatment compared to observation or macular laser therapy (93,94). Most of these trials enrolled patients with 20/40 or worse visual acuity in a randomized, double-blinded manner (meaning that neither the patients nor the treating physicians knew which treatment specific patients were receiving). In these trials, patients treated with fixed dosing regimens of ranibizumab or aflibercept through 3 years gained an average of 10 or more ETDRS letters, also referred to as two lines of visual acuity (since five letters represent one line), or the equivalent of an eye improving from 20/60 to 20/40 visual acuity. In comparison, patients treated with macular laser therapy or observation gained little or no visual acuity over the same time period.

A single phase 3 trial, primarily sponsored by the National Institutes of Health and known as the Diabetic Retinopathy Clinical Research Network Protocol T (DRCR.Network Protocol T), compared the three available anti-VEGF agents through 2 years of DME management (95). Among better-seeing eyes, all three agents (bevacizumab, ranibizumab, and aflibercept) achieved similar visual benefit, whereas anatomic benefit was superior with both aflibercept and ranibizumab compared to bevacizumab. Among worse-seeing eyes (baseline vision of 20/50 or worse), although all three medications achieved robust visual acuity gains, aflibercept achieved the greatest visual and anatomic gains compared to bevacizumab, with similar ultimate visual and anatomic outcomes compared to ranibizumab at the 2-year endpoint.

In clinical practice, treatment of DME in most patients is initiated with monthly anti-VEGF intravitreal injections. When optimal visual and anatomic outcomes are achieved, the anti-VEGF dosing frequency is often then reduced. Such a reduction in treatment frequency is often achieved in one of two ways. First, patients may continue to receive treatments separated by increasing time intervals, a process known as “treat and extend.” Alternatively, treatment may be stopped and only reinitiated when recurrence of DME is observed, a management approach known as “pro re nata” (PRN).

Although anti-VEGF pharmacotherapy is typically considered to be first-line treatment for DME, corticosteroids can play an important role in management. Some eyes treated with anti-VEGF monotherapy will have an incomplete response to adequate anti-VEGF dosing. In prospective studies, this proportion of patients ranges from approximately 30 to 68% (96). In these cases, incorporation of an intravitreal steroid agent can lead to better DME control (97). Mechanistically, this is likely related to the observation that numerous inflammatory pathways are active in DR and DME that are not influenced by anti-VEGF monotherapies. In comparison, corticosteroids are capable of modulating a multitude of inflammatory pathways, including blockade of VEGF.

Despite their potential extended durability compared to anti-VEGF monotherapies, corticosteroids are typically not used as first-line therapy because of the risks inherent to intravitreal steroid delivery, including cataract acceleration and increased intraocular pressure (IOP). Although cataract is a readily treatable pathology and the IOP elevation observed in approximately one-third of eyes treated with intravitreal steroids is generally manageable and reversible, these factors often place corticosteroids as a second-line option.

Long-Term Dosing Requirements in DME Management

The management burden for patients with CIDME through 2 years is substantial and should be directly communicated with patients and their caregivers. For example, regardless of treatment, through 2 years of the DRCR.Network Protocol T trial, patients underwent a mean of 23 clinical visits and received a mean of 15–16 intravitreal injections (95).

Fortunately, after initially intensive anti-VEGF therapy for CIDME, several analyses have suggested that less-frequent-than-monthly anti-VEGF dosing may be effective in maintaining visual and anatomic gains in most patients and that a clinically meaningful proportion of patients can maintain quiescent disease without ongoing treatment through at least 2 additional years of follow-up. For example, after fixed dosing through 3 years in three phase 3 trials, approximately one-fourth of patients received no additional anti-VEGF dosing, with no DME recurrence, whereas an average of 3–4 intravitreal injections were given annually to the entire population for DME control through the fourth and fifth years of management (98,99).

Diseases relatively confined to the macula, including most cases of DME and age-related macular degeneration, may cause substantial visual impairment, with loss of the capacity to perform visual activities requiring detailed vision such as reading, driving, and recognizing faces. However, even in their severe form, such diseases often allow affected individuals to retain the gross visual function to recognize large objects and ambulate. In comparison, the natural history of diseases that routinely affect both the macula and the peripheral retina, such as PDR, can and often do lead to more complete loss of vision, with more profound impairment in performance of activities of daily living.

PDR can be treated with either PRP or intravitreally delivered pharmaceuticals that inhibit VEGF. Each of these treatments has unique benefits and challenges. PRP is analogous to radiation therapy for cancers, whereas intravitreal anti-VEGF or corticosteroid administration is analogous to chemotherapy. The two approaches are often used in combination. The advantages and disadvantages of each approach must be weighed carefully by patients, their families, and their physicians.

In the 1970s, the pivotal Diabetic Retinopathy Study defined PRP as the cornerstone of treatment for PDR by demonstrating a dramatic reduction in blindness with treatment compared to observation (100). Although PRP remains a mainstay of PDR management today, it is an inherently destructive treatment. Retinal tissue is selectively ablated, and a scar is created. Although this process often stops the progression of the proliferative process inherent to PDR, it has limitations and can lead to untoward effects. First, PRP is not a cure in many eyes. For example, a 5-year study comparing PRP to anti-VEGF pharmacotherapy (the DRCR.Network Protocol S) found that slightly greater than half of eyes treated with PRP at baseline required additional PRP through 5 years of follow-up (101). Second, PRP can lead to peripheral visual field defects, night vision loss, and loss of contrast sensitivity. The key benefit of PRP is that the treatment is permanent. PRP creates a lasting scar where laser energy is applied. Thus, some eyes can be adequately managed with one or at least a limited number of PRP applications. In the DRCR.Network Protocol S, just 15% of eyes randomized to PRP received additional laser application cumulatively in years 3, 4, and 5.

The main benefit of anti-VEGF pharmacotherapy is that it avoids the destructive nature of PRP. This translates into less severe, even though progressive, visual field loss with anti-VEGF pharmacotherapy compared to PRP. The downside to anti-VEGF pharmacotherapy is the transient biological effect related to the medication pharmacokinetics, which translates into an increased visit and treatment burden. Through 5 years of the DRCR.Network Protocol S, eyes randomized to anti-VEGF treatment required a median of 43 clinical visits, compared to 21 clinical visits among the patients randomized to PRP. Furthermore, although the treatment burden from anti-VEGF therapy decreased after the first year, a majority (63–75%) of eyes still required repeated dosing annually through year 5, with 43% still requiring four or more injections in the fifth year of management (101).

For the past four decades, treatment of DR has been reactionary, typically initiated only once the potentially blinding pathologies of PDR and DME are manifest. However, accumulating data indicate that there may be tremendous value in initiating ocular-specific pharmacological treatment for DR at earlier stages. Specifically, prevention of progression to PDR and development of DME may represent a tremendous public health opportunity. The clinical rationale for doing so is twofold. First, worsening NPDR severity carries prognostic information indicating an increased risk of disease progression and visual loss. Second, as NPDR severity worsens even in the absence of DME, health-related quality of life related to vision-dependent functions such as driving ability progressively decline.

Improving DR Severity with Pharmacotherapy

Historically, DR as assessed by the DRSS was generally accepted clinically as a one-way track, with progressive accumulation of retina damage associated with DR over time. More recently, the field has realized that position on this scale can be modified with pharmaceutical treatment. Although the phase 3 trials leading to FDA approval of aflibercept and ranibizumab for DME treatment revealed remarkable efficacy at improving retinal anatomy and function, they also demonstrated that VEGF blockade can affect far more than just macular edema.

First, just as PRP of eyes with severe NPDR can delay progression to PDR, pharmacotherapy can significantly blunt progression from NPDR to PDR (102,103). Second, anti-VEGF therapy not only slows progression of DR, but also has the added benefit of improving DR severity in a substantial proportion of eyes. In phase 3 trials focused on DME management, approximately one-third of anti-VEGF–treated eyes experienced a clinically meaningful improvement in DR severity, defined as a two-step or greater DRSS improvement, compared to 5–16% of sham-treated eyes (86,87). Anti-VEGF treatments can also reduce the development of PDR in eyes with moderately severe and severe NPDR. Within this high-risk population, a much larger proportion of eyes with DME, more than 75%, experienced a clinically meaningful DRSS improvement with ranibizumab treatment (104). Third, VEGF blockade appears to have a significant impact on the underlying retinal vasculature itself, slowing progressive capillary loss (105), suggesting that pharmacotherapy may be able to achieve fundamental disease modification.

Pharmacological Treatment of DR in Individuals with NPDR But No DME

Although the value of pharmacological therapy for improving DR severity has been documented in multiple prospective studies enrolling individuals with DME or PDR, the value of such therapy in those with NPDR but no DME, representing a larger patient population who currently remain largely untreated, is under active investigation. Intravitreal injections do carry risk, especially when considering the cumulative risk of many years of repeated treatments. These risks include infection (referred to as “endophthalmitis”), retinal tear, retinal detachment, cataract, and potentially even systemic side effects. The costs of treatment also include time off from work for patients and their families. To better define the risk-benefit ratio in high-risk NPDR eyes without DME (DRSS levels of 47 and 53), two independent, large, randomized, phase 3 trials were initiated in 2016 comparing sham injections to VEGF blockade with aflibercept injections.

PANORAMA was a 2-year trial that randomized 402 patients to either sham treatment or anti-VEGF dosing with aflibercept (106). The primary outcome was met, with 55–62% and 65–80% of anti-VEGF–treated patients achieving at least two steps of DR severity improvement at 6 and 12 months compared to 6 and 15% of sham-treated eyes, respectively. More clinically relevant, however, was that 41% of sham-treated eyes had developed either DME or PDR by 12 months, compared to 11% of anti-VEGF–treated eyes. A distinct 4-year trial with a similar sham-controlled design and a 2-year primary endpoint is ongoing (107). At the time of the completion of this compendium, ranibizumab has been approved by the FDA for the treatment of all forms of DR (with or without DME), and aflibercept has been approved for the treatment of DR in patients with DME.

In the prospective trials evaluating current-generation anti-VEGF agents for DME and DR management, frequent visits and regular treatments are typically employed. However, recommendations based on clinical trial protocols can be challenging to implement in routine clinical settings. Patients with DR are often in poor health and require complex medical care—so much so that many DR patients have difficulty adhering to frequent office visits, especially given that this disease often manifests within a working-age population. In a recent health care claims database analysis, patients with DME averaged 25.5 heath care visit days annually, of which 4.4 visits were attributed to ophthalmic care (108).

Multiple real-world analyses have concluded that anti-VEGF dosing in the real world appears to be substantially less than that given during registration trials on a population basis (108,109). Because visual gains across multiple DME trials have been positively correlated with the number of injections, especially in the first year of treatment, overall consistent anti-VEGF dosing until maximal visual and anatomic improvement have been achieved in the setting of DME management is generally recommended.

In the setting of PDR, challenges with patient compliance are especially common. In the prospective DRCR.Network Protocol S, just 66% of living patients completed the 5-year endpoint (101). In most real-world clinical settings, noncompliance can be even more dramatic, with one analysis of more than 2,000 PDR patients followed over a 4-year period reporting that approximately 25% were lost to follow-up for more than 12 months. Age, race, and regional average adjusted gross income were found to be key risk factors associated with loss to follow-up (110).

The core tenet of DR management is that all patients with diabetes need regular ophthalmic examinations over the long term. The primary reason for this is because patients could have substantial DR and yet remain asymptomatic. Highly effective, ocular-specific treatments are widely available and are often used even when patients have no or limited symptoms. Furthermore, accumulating data from many perspectives indicate that earlier intervention leads to better outcomes, likely with less intensive treatment. If patients receive appropriate screening and follow-up care, much of the visual impairment associated with diabetes and DR could be reduced or prevented.

The opinions expressed are those of the authors and do not necessarily reflect those of Genentech or the American Diabetes Association. The content was developed by the authors and does not represent the policy or position of the American Diabetes Association, any of its boards or committees, or any of its journals or their editors or editorial boards.

To request permission to reuse or reproduce any portion of this publication, please contact

Dualities of Interest

T.W.G. has received research support from Zebra Biologics and consulting fees from Novo Nordisk.

C.C.W. has received research support from Adverum, Allergan, Apellis, Clearside, Genentech, Roche, Neurotech, Novartis, Opthea, Regeneron, Regenxbio, Samsung, and Santen; is a consultant for Adverum, Alimera Sciences, Allegro, Allergan, Apellis, Bayer, Clearside, EyePoint, Genentech, Kodiak, Novartis, Regeneron, Regenxbio, and Roche; and is a speaker for Regeneron.

B.A.C. has been a speaker for Novo Nordisk and served on an advisory board for Regeneron.

No other potential conflicts of interest relevant to this compendium were reported.


Editorial and project management services were provided by Debbie Kendall of Kendall Editorial in Richmond, VA.

Author Contributions

All authors researched and wrote their respective sections. Lead author T.W.G. reviewed all content and is the guarantor of this work.

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. See for details.