Blog
August 9, 2026
The word “stage” can imply a straight line in which each eye moves predictably from one level to the next. Diabetic retinopathy is more variable. One eye may remain at mild NPDR for years; another progresses quickly during pregnancy, poor systemic control or kidney disease. Treatment can make visible lesions regress without erasing the underlying lifetime risk.
Staging is still valuable because it converts a complex retinal examination into a risk category. It helps determine follow-up, imaging and urgency. This article explains what changes across the four commonly taught stages without duplicating Netra Eye Institute’s diabetic retinopathy condition page.
The retina is neural tissue lining the back of the eye. Photoreceptors capture light; multiple neural layers process signals; retinal ganglion-cell axons form the optic nerve. This tissue has a high metabolic demand and depends on an organized microvascular network.
The inner retina is supplied by branches of the central retinal artery. The outer retina, including photoreceptors, relies heavily on the choroidal circulation beneath the retinal pigment epithelium. The blood–retinal barrier regulates fluid, proteins and cells crossing into neural tissue.
Chronic hyperglycemia, metabolic stress, inflammation, oxidative injury and abnormal signaling damage endothelial cells and pericytes, thicken basement membranes and disrupt autoregulation. Capillaries leak or close. The resulting combination of edema and ischemia produces the visible lesions used for staging.
The retina is also neural, not merely plumbing. Functional and neurodegenerative changes can precede or accompany visible microvascular lesions. Current routine clinical staging, however, remains based primarily on vascular examination and imaging because those features have validated prognostic value.
Mild NPDR is defined by microaneurysms without the more advanced lesions required for later stages. A microaneurysm is a small outpouching of a retinal capillary wall. On color photography it appears as a tiny red dot; fluorescein angiography may show focal hyperfluorescence.
Microaneurysms can leak plasma into the retina or thrombose and disappear. The count at one photograph is therefore dynamic. A few red dots can also represent small hemorrhages, and other diseases can produce microaneurysm-like lesions, so diagnosis includes the diabetes history and broader pattern.
Most people with mild NPDR have no symptoms and good acuity. The stage signals that diabetes has produced clinically visible retinal microvascular injury. It is not “nothing,” but it is also not an emergency by itself when the macula is dry and no other high-risk feature is present.
Management usually emphasizes systemic risk control and follow-up rather than immediate ocular treatment. The interval depends on diabetes type, glycemic and blood-pressure control, pregnancy, kidney disease, the fellow eye and access to return care.
Moderate NPDR includes more than microaneurysms but does not meet severe NPDR criteria. Findings may include dot-and-blot hemorrhages, hard exudates, cotton-wool spots, venous caliber change and intraretinal microvascular abnormalities (IRMAs).
These hemorrhages lie within deeper retinal layers and appear round or irregular. Flame-shaped hemorrhages occupy the nerve-fiber layer. Hemorrhage number and distribution contribute to severity.
Hard exudates are lipid and protein residues left after vascular leakage. They appear yellow-white with distinct borders and may cluster around leaking microaneurysms. Exudates near the fovea raise concern for macular edema, but OCT is needed to characterize retinal fluid.
Cotton-wool spots are localized nerve-fiber-layer infarcts caused by interrupted axoplasmic flow. They appear fluffy and white. Diabetes is one cause; hypertension, vascular, inflammatory and hematologic disease can also produce them.
IRMAs are abnormal intraretinal channels associated with capillary nonperfusion. They remain within the retina, unlike the new vessels of PDR that grow on the retinal surface or into the vitreous interface.
Moderate NPDR spans a wide range, so prognosis and follow-up vary. The retina specialist looks at lesion extent, macular status and change over time rather than treating the word “moderate” as a single precise risk.
Severe NPDR represents substantial ischemic burden without definite proliferative new vessels. The classic “4-2-1 rule” identifies severe disease when any one of the following is present:
If two or more criteria are present, the eye may be classified as very severe NPDR, a category with still higher short-term risk of progression to PDR.
Retinal veins develop irregular segments of dilation and narrowing, resembling a string of beads. This finding reflects widespread ischemic stress and has strong prognostic significance.
The eye has not yet formed proliferative neovascularization, but it is close to that threshold. Retinal nonperfusion increases hypoxia-driven signals, including vascular endothelial growth factor (VEGF). Follow-up becomes more frequent, and the retina specialist discusses what would happen if PDR or center-involved DME develops.
Preventive anti-VEGF injections can reduce the occurrence of PDR or vision-reducing DME while treatment continues. DRCR Protocol W found fewer anatomic complications with scheduled aflibercept in high-risk NPDR, but no average visual-acuity advantage at four years compared with observation and treatment if complications developed. For many patients, routine preventive injections are therefore not automatically warranted. Follow-up reliability, fellow-eye status, access, systemic risk and individual preference matter.
Panretinal photocoagulation is not routinely applied to every severe NPDR eye, but it may be considered in selected high-risk circumstances, particularly when return care is uncertain or progression risk is unusually high.
PDR begins when abnormal new vessels grow at the optic disc, elsewhere on the retinal surface or on the iris or angle. The vessels arise in response to retinal ischemia and VEGF signaling. They are fragile and unsupported, so they leak and bleed easily.
New vessels on the disc are called NVD; those elsewhere are NVE. Careful examination, widefield photography, OCT angiography or fluorescein angiography may help distinguish them from IRMA and normal vascular variation.
“Proliferative” refers to new vessel and fibrovascular tissue proliferation, not to a tumor. The stage is sight-threatening even when central acuity remains good.
New vessels grow along the posterior hyaloid and retinal surface. Vitreous movement can tear them, releasing blood in front of or into the vitreous. A small hemorrhage may cause floaters or cobwebs; a dense hemorrhage can reduce vision to hand motion.
Blood can clear, but the ischemic drive remains. A new shower of floaters, dark haze or sudden loss in diabetes requires urgent retinal evaluation.
New vessels are accompanied by fibrous tissue. As it contracts, it pulls on the retina. Traction can distort the macula, create a tractional retinal detachment or combine with a retinal break to produce a more complex detachment.
Surgery is considered when hemorrhage does not clear, traction threatens or involves the macula, or other complications require mechanical repair. Anti-VEGF may be given before vitrectomy to reduce vascularity, but timing matters because rapid fibrovascular contraction can occasionally worsen traction.
Severe ischemia can drive new vessels on the iris and in the drainage angle. Fibrovascular tissue can block aqueous outflow and cause neovascular glaucoma, a painful high-pressure emergency. Retinal treatment and pressure control are both required.
The Diabetic Retinopathy Study defined high-risk features based on the extent and location of neovascularization and hemorrhage. These eyes have substantial risk of severe vision loss without treatment.
Modern imaging and anti-VEGF therapy have changed management choices, but the principle remains: active proliferative disease requires timely treatment and reliable follow-up. Good acuity is not a reason to wait for bleeding.
Diabetic macular edema is retinal thickening from leakage in or near the macula. It can occur with mild NPDR, severe NPDR or PDR. A four-stage label therefore does not fully describe vision risk.
Center-involved DME affects the central subfield on OCT. It may reduce acuity, contrast and reading, but some eyes retain 20/25 or better vision. DRCR Protocol V showed that initial observation with aflibercept only if acuity worsened was reasonable for many eyes with center-involved DME and good vision under structured follow-up.
Vision-reducing center-involved DME is commonly treated with intravitreal anti-VEGF. Steroid implants or injections and focal/grid laser have selected roles based on lens status, pressure risk, edema pattern, treatment response and follow-up.
Learn more in the dedicated article on diabetic macular edema.
Stereoscopic or high-quality color photographs document microaneurysms, hemorrhages, exudates, cotton-wool spots, venous change and neovascularization. The ETDRS photographic scale provides detailed research grading; clinical scales simplify it for practice.
Ultra-widefield imaging captures peripheral lesions beyond standard fields. Peripheral findings can influence risk assessment, but image quality and eyelid artifact matter. Photography complements rather than universally replaces a dilated examination.
OCT provides cross-sectional retinal anatomy and is central to DME assessment. It shows intraretinal cysts, subretinal fluid, thickening, vitreomacular traction and structural damage. OCT can be normal in a patient with advanced peripheral PDR; it does not replace examination of the whole retina.
Injected dye circulates through retinal vessels. Sequential images show leakage, microaneurysms, nonperfusion and neovascularization. Nausea is common; allergic reactions are uncommon but possible. Angiography is ordered when the result will guide diagnosis or treatment, not for every routine screening visit.
OCTA detects motion of blood cells without dye and maps capillary plexuses and neovascular networks. It cannot show leakage directly and is sensitive to motion and segmentation artifact. It is a useful adjunct, not a universal replacement for fluorescein angiography.
Peripheral hemorrhages and new vessels can be extensive without central symptoms. Conversely, a small amount of center-involved DME can blur reading with only mild NPDR.
Vitreous hemorrhage causes sudden floaters or loss, while traction may distort vision. Cataract, glaucoma, dry eye and glucose-related lens shifts can create symptoms unrelated to the stage. Each eye needs separate acuity, OCT and examination.
The four-stage international scale is designed for practical communication. Large trials often use the more granular Early Treatment Diabetic Retinopathy Study scale, which assigns levels based on standardized photographic fields. Those levels distinguish subtle changes within moderate, severe and proliferative disease and allow researchers to define two-step improvement or worsening.
A clinic note may therefore say “moderate NPDR” while a reading center reports an ETDRS level. These are not contradictory systems; they use different resolution. Treatment decisions should not depend on a patient trying to convert one number online without knowing the imaging protocol.
Screening programs may use categories such as no retinopathy, mild background, referable retinopathy or sight-threatening disease. An AI system may output “more-than-mild diabetic retinopathy.” Those categories are designed to route referral, not replace the retina specialist’s staging, DME assessment or treatment plan.
There is no single timetable. Duration and baseline stage are powerful predictors, but progression is probabilistic. An eye with isolated microaneurysms can remain stable for years, while severe NPDR can convert to PDR over months.
Rapid change is more likely with very high or rapidly improving glucose, uncontrolled hypertension, nephropathy, pregnancy, puberty and severe disease in the fellow eye. Cataract can make the retina harder to grade, creating the appearance of sudden discovery rather than sudden biological onset.
Treatment also changes the visible stage. Anti-VEGF can cause rapid regression of neovascularization, but effect can wear off. PRP reduces ischemic drive more durably, though additional laser may be needed. Vitrectomy removes blood and traction but does not remove diabetes from remaining retinal vessels.
The safest interpretation of stage includes date and trend: stable mild NPDR for three visits means something different from newly severe NPDR after a long gap in care.
The Diabetes Control and Complications Trial in type 1 diabetes established that intensive glycemic therapy reduces the onset and progression of retinopathy, with long-lasting benefit during follow-up despite later convergence of A1C between groups. This “metabolic memory” supports early durable control.
The UK Prospective Diabetes Study showed that improved glucose and blood-pressure management in type 2 diabetes reduces microvascular risk. These landmark findings do not create one correct target for every patient. Hypoglycemia, age, kidney disease, cardiovascular status, pregnancy and treatment burden shape individualized goals.
Fenofibrate reduced retinopathy progression or laser need in major trials among selected people with type 2 diabetes, with effects not fully explained by lipid change. Its role differs by country and patient, and it has kidney, liver and interaction considerations. A retina clinician should communicate advanced retinopathy, while diabetes and primary-care clinicians decide systemic therapy.
Blood-pressure control is particularly important because hypertension increases leakage and vascular stress. Excessively rapid or poorly tolerated reduction is not the goal. The patient’s medical team balances retinal, kidney, brain and heart protection.
Annual examination is a common baseline message, but an eye with moderate, severe or proliferative disease may need visits every few months or more often. DME treatment can require monthly intervals initially. Pregnancy produces its own schedule.
Conversely, some well-controlled adults with repeated normal examinations may be assigned longer intervals under specific guidelines and clinician judgment. The patient should never lengthen the interval simply because vision feels normal.
A safe interval considers:
If follow-up is likely to be interrupted, treatment strategy may change. Durable PRP can be favored over injection-only PDR management for some patients because missing repeated injections allows recurrence. Shared decision-making must include the real life in which the plan will operate.
A patient with type 2 diabetes has a few microaneurysms, no DME and 20/20 acuity. The retina does not need injection. The useful response is not complacency or panic: document the baseline, improve systemic risk where safe and return at the recommended interval.
Photography shows exudates near the macula, but OCT demonstrates that fluid does not involve the center and acuity is good. Follow-up may be closer, with systemic control and possible focal treatment discussion depending on leakage and progression. The yellow deposits themselves are not removed surgically.
An eye meets the 4-2-1 rule, the fellow eye has PDR and the patient lives hours from care. Observation remains possible, but the risk of missed progression is high. Preventive anti-VEGF or PRP may be discussed more seriously than for a patient able to return promptly, with Protocol W’s lack of average four-year acuity benefit explained honestly.
New vessels are visible on the disc, but central acuity remains 20/20 and OCT is dry. Treatment is still indicated because bleeding and traction risk arise from proliferation, not current acuity. Anti-VEGF, PRP or combination is selected based on adherence, field, DME, pregnancy and traction.
Vision suddenly falls behind dark blood. Ultrasound may be used when the retina cannot be seen. Anti-VEGF, laser when the view clears or vitrectomy is considered according to hemorrhage duration, traction, fellow-eye status and functional need. NRT cannot safely wait for blood to “detox.”
Autonomous AI and remote photography can expand detection in primary-care or diabetes settings. A camera captures retinal images, and software or a reading center determines whether referral is needed. Real-world studies show useful sensitivity, but ungradable images and false positives or negatives remain.
Screening is not the same as a comprehensive eye examination. It may miss glaucoma, peripheral tears, cataract, corneal disease or DME when OCT is unavailable. An ungradable result requires referral, not reassurance. A positive screen needs completion of the referral pathway; detecting disease without access to treatment does not prevent vision loss.
AI categories also require equity monitoring across pupil size, media opacity, pigmentation, camera and clinic workflow. The tool supports access but does not determine an individualized injection, laser or pregnancy plan.
Patients sometimes hear “stage 3” and assume they are one step from inevitable blindness. Others hear “background changes” and conclude that follow-up is optional. Both interpretations are harmful.
Clinicians should translate stage into absolute actions: current macular status, near-term progression risk, next visit, systemic priorities and urgent symptoms. Visual prognosis depends on treatment timing, baseline structure, adherence and other eye disease—not the label alone.
People should not be blamed for progression. Biology, duration, socioeconomic access, medication cost, food security, stress and comorbidity all influence control. A supportive plan identifies barriers and resources rather than treating A1C as a moral grade.
Diabetes is systemic, but retinal findings are often asymmetric. One eye may have moderate NPDR while the other has PDR after a vein occlusion, prior surgery or different local vascular history. Follow-up and treatment are therefore documented by eye.
The worse eye can influence the schedule for both, but treatment is not automatically identical. One eye may receive anti-VEGF for DME while the fellow eye is observed; one may need PRP while the other has no proliferation. Patients should know which eye received which procedure and carry that information if care is transferred.
Binocular vision can mask unilateral change. Testing one eye at a time at home occasionally may reveal new blur, but it does not replace examination. Sudden asymmetry is a reason to contact the retina team.
Clear right-eye and left-eye documentation also prevents confusion about injection intervals, laser completion, driving function and the prognosis discussed at each visit.
That clarity supports safer continuity of care.
Duration of diabetes is a major nonmodifiable risk. Glycemic exposure, blood pressure, lipids, kidney disease, anemia, pregnancy, puberty, smoking and access to care influence risk.
Rapid improvement in very high glucose can temporarily worsen retinopathy in some patients, particularly with advanced baseline disease. This does not mean glucose should remain high. It means systemic improvement and retinal monitoring should be coordinated rather than avoided.
Pregnancy can accelerate retinopathy in preexisting type 1 or type 2 diabetes. Baseline stage is a strong predictor. Examination before conception or early pregnancy and stage-specific follow-up are important. Gestational diabetes alone does not carry the same retinopathy screening recommendation.
Mild and moderate NPDR without vision-threatening DME are commonly observed with systemic management. Severe NPDR requires closer follow-up and individualized preventive discussion. PDR requires anti-VEGF, panretinal laser or a combination, with surgery for hemorrhage and traction.
Protocol S showed that ranibizumab and PRP were both viable PDR strategies over five years among trial participants. Anti-VEGF produced less peripheral field loss and less DME, but required ongoing visits and injections. PRP offers durable regression without indefinite injection dependence but can affect peripheral and night vision and exacerbate DME.
The real-world decision depends on DME, pregnancy, traction, adherence, transportation, cost and fellow-eye status. Missed visits after anti-VEGF monotherapy can be dangerous because neovascularization may recur.
Lesions and neovascularization can regress after systemic improvement, anti-VEGF or laser. Photography may show a lower apparent severity. The retina does not become a never-diabetic retina, and recurrence risk remains.
Laser scars persist. Capillary nonperfusion may persist even when new vessels collapse. OCT thickness can normalize while photoreceptor or inner-retinal damage limits vision. “Regression” is a treatment outcome, not proof of cure.
Netra Restoration Therapy (NRT) is adjunctive integrative care. It may support whole-person goals involving diet quality, sleep, stress, activity, medication adherence and coordination of systemic risk management. These factors matter to diabetes health, but NRT does not replace endocrinology or retinal treatment.
Traditional Chinese medicine studies for NPDR report changes in retinal lesions, vision or surrogate outcomes, but a 2025 evidence map found substantial methodological limitations across the review literature. Trials often vary in formulations, comparators, staging, masking and reporting. Mechanistic research on oxidative stress and inflammation is biologically interesting but does not establish that acupuncture or herbs stop progression to PDR.
An evidence-bounded NRT plan never delays dilation, OCT, angiography, injection, PRP or vitrectomy. It does not advise stopping insulin, GLP-1 therapy, blood-pressure medicine, statins or fenofibrate. Oral herbs can alter glucose, bleeding, liver metabolism and perioperative risk, so every product must be disclosed.
No nonsterile preparation should be placed in the eye. Sudden floaters, haze, loss or pain is routed to urgent retinal or glaucoma care, not to an NRT visit.
Learn about Netra Restoration Therapy for diabetic retinopathy, Netra Eye Institute’s approach, preventing diabetic vision loss and how to request an appointment.
No. PDR is high risk, but many patients retain useful vision with timely treatment. Hemorrhage, traction, DME and optic or macular damage determine function.
No. Some eyes remain stable or regress, especially with risk control. Duration and systemic and ocular factors change probability.
No. They are retinal findings. Symptoms arise if edema, hemorrhage, traction or another disease affects vision.
It is not usually an emergency in an asymptomatic stable eye, but it requires timely retina follow-up because progression risk is high. New loss or hemorrhage symptoms are urgent.
Yes. Proliferation and macular edema are separate axes. An eye can have one, both or neither.
NRT cannot be relied on to regress retinal neovascularization or validated lesions. Stage change is assessed through retinal examination and imaging under established medical care.
The four stages describe increasing retinal vascular severity: microaneurysms; broader nonproliferative lesions; severe ischemic warning features; and proliferative new vessels. DME runs alongside that ladder and can threaten central vision at any point.
Staging should lead to action, not fear. Know the macular status, follow-up interval, systemic priorities and treatment threshold. NRT may support the whole person, but injections, laser, surgery and metabolic care remain the tools that protect against established sight-threatening disease.
Medical Disclaimer: This article provides general education and is not medical advice, diagnosis or a personal follow-up schedule. Diabetic retinopathy may progress without symptoms. Sudden floaters, dark haze, vision loss, distortion, a curtain, eye pain or redness requires urgent eye care. Do not change insulin, glucose-lowering drugs, blood-pressure medicine, lipid therapy, antiplatelets or retinal treatment without the responsible clinicians. Netra Restoration Therapy is adjunctive and must not delay or replace dilated examination, OCT, anti-VEGF injection, laser, surgery, pregnancy monitoring or emergency care.
