The macula is the small central patch of retina you read with. It is what you point at things. Age-related macular degeneration destroys it, and because peripheral vision is usually spared, the result is a particular and cruel pattern: a person can walk across a room unaided but cannot see the face of the person they are walking towards.
That is the exact inverse of glaucoma, which takes the periphery first and leaves central vision intact until late. Two diseases of the same organ, eroding vision from opposite ends. AMD is the leading cause of central vision loss in older adults, and a systematic review and meta-analysis projected its global prevalence and burden to 2040 [1], with US prevalence estimated separately [2] and fifteen-year cumulative incidence measured in a population cohort [3] [4].
And "AMD" names two conditions that behave differently enough to be considered separately.
Dry (non-neovascular) AMD is the common form and the slow one. Deposits called drusen accumulate under the retina, the retinal pigment epithelium and photoreceptors above them degenerate, and in the advanced form — geographic atrophy — patches of retina die outright and expand over years, leaving permanent blind spots.
Wet (neovascular) AMD is the fast one. Abnormal new vessels grow from the choroid beneath the retina, leak fluid and blood, and can destroy central vision in weeks to months. It is much less common than dry AMD but historically caused most of the severe vision loss.
The reason the distinction matters is that one of them was solved and the other was not. Wet AMD went from a rapidly blinding condition to a treatable one. That is the subject of the figure below.
Start here: drusen, complement, and a genetics result that changed the field
Drusen are not simply debris. An integrated hypothesis proposed in 2001 reframed them as biomarkers of immune-mediated processes at the retinal pigment epithelium–Bruch's membrane interface [5], and the proteins found in drusen turned out to overlap with those in deposits elsewhere in the body associated with chronic inflammation [6]. Oxidative stress is a further contributor [7] [8], acting on a retina that is uniquely exposed — high oxygen consumption, constant light, and photoreceptor outer segments turned over daily by the pigment epithelium [9]. Destroying the pigment epithelium causes the underlying choriocapillaris to atrophy [10], which is part of why the damage propagates.
Then genetics confirmed the immune hypothesis in the most direct way available. In 2005 a genome-wide screen of just 96 cases and 50 controls across 116,204 SNPs found a common variant in complement factor H strongly associated with AMD (P < 10⁻⁷), with homozygotes for the risk allele having a 7.4-fold increased likelihood of the disease (95% CI 2.9–19). The associated polymorphism changes tyrosine to histidine at amino acid 402, in the region of the protein that binds heparin and C-reactive protein [11]. Three other groups reported the same association essentially simultaneously [12] [13] [14].
A 7.4-fold effect from a common variant is enormous for a complex disease, and it was found in 146 people. Complement factor H is the major inhibitor of the alternative complement pathway — it restrains complement activation — so a variant that impairs it means unchecked complement activity at the retina. And it is not acting at a distance: factor H accumulates within drusen and is synthesised by the retinal pigment epithelium itself [14], its synthesis by those cells is down-regulated by oxidised photoreceptor outer segments [15], and manipulating Cfh dosage in mice alters AMD-like pathology and Bruch's-membrane lipoprotein binding [16]. Genetics, histology and oxidative stress meet at the same protein. Further complement genes followed, including a C3 variant [17], along with HTRA1 [18], and a large GWAS later mapped contributions of both rare and common variants [19]. Complement activation has been demonstrated in the retina itself [20], with a protective role for C3 in the ageing mouse retina complicating the picture [21].
The modifiable risk factor is smoking, and its effect persists after quitting. In the Nurses' Health Study, 31,843 women followed over 556,338 person-years produced 215 incident cases of AMD with visual loss. Current smokers of 25 or more cigarettes a day had a relative risk of 2.4 (95% CI 1.4–4.0) against never-smokers — and past smokers of that amount still had 2.0 (1.2–3.4) [22]. Reading those together: of the 1.4 units of excess risk carried by current heavy smokers, 1.0 remains after quitting — about 70 percent of the excess persists, which the paper reports but does not compute. Quitting is still worth it; it is not a reset. Diet has been examined extensively, with carotenoid and omega-3 intake studied against risk [23] [24] [25] [26], and risk factors for progression reviewed [27].
Pillar 1: measurement and diagnosis
The workhorse is OCT
Optical coherence tomography images the retina in cross-section, non-invasively, in seconds. It shows drusen, the fluid that defines wet AMD, and the atrophy that defines advanced dry AMD, and it is what determines whether an eye needs an injection this month. OCT angiography extended it to imaging blood flow without dye [28]. Fundus photography and dye angiography remain in use for characterising neovascular lesions [29].
Automated grading of AMD from colour fundus images using deep learning has been demonstrated [30], part of the same wave that produced validated systems for diabetic retinopathy [31] [32].
Self-monitoring, because wet AMD is an emergency
Dry AMD progresses over years. Wet AMD can destroy central vision in weeks, and the treatment works better the earlier it starts — so patients with dry AMD are taught to self-monitor for the sudden distortion that signals conversion, classically with an Amsler grid. Whether structured home monitoring outperforms this was tested directly in a randomised trial of a home monitoring system for early detection of choroidal neovascularisation [33], and mobile hyperacuity self-monitoring apps have since been assessed [34].
Measuring geographic atrophy
Because dry AMD has no fluid to resolve, its trials need a different endpoint, and the one used is lesion area growth. Change in geographic atrophy area was characterised in AREDS [35] and progression studied since [36]. Hold that thought — the choice of endpoint is what makes the newer dry-AMD drugs hard to interpret.
Centerpiece: what anti-VEGF did
Vascular endothelial growth factor was identified as a secreted angiogenic mitogen in 1989 [37] and found at high concentration in the ocular fluid of patients with neovascular eye disease in 1994 [38] [39]. The therapeutic logic was direct: the vessels are driven by VEGF, so block VEGF in the eye.
The first agent to work was pegaptanib, an aptamer: in 1,186 patients, 70 percent given 0.3 mg lost fewer than 15 letters at 54 weeks against 55 percent of controls, and severe loss of 30 or more letters fell from 22 percent to 10 percent [40]. Real, and modest — most patients still lost vision.
Then came ranibizumab, and two trials that are best read together.
MARINA randomised 716 patients with minimally classic or occult lesions to 24 monthly intravitreal ranibizumab or sham injections. At 12 months, 94.6 percent on 0.5 mg lost fewer than 15 letters against 62.2 percent of sham; 33.8 percent gained 15 or more letters against 5.0 percent; and mean acuity rose 7.2 letters against a fall of 10.4 [41].
ANCHOR randomised 423 patients with predominantly classic lesions to monthly ranibizumab or verteporfin photodynamic therapy — the standard of care at the time [42]. Mean acuity rose 11.3 letters against a fall of 9.5 [43].
The check the arithmetic was never given. These are separate trials, with different lesion types and different comparators, and neither compares itself to the other. One control arm received nothing at all; the other received the era's approved treatment. They lost 10.4 and 9.5 letters — within 0.9 letters of each other. Photodynamic therapy, on mean acuity, did approximately what sham did. That comparison is not made anywhere in either paper, and it is the clearest available statement of why anti-VEGF registered as a break rather than an increment: it was not improving on a partially effective therapy, it was the first thing to change the direction of the line.
A second finding, about endpoints rather than drugs. MARINA's primary endpoint — proportion losing fewer than 15 letters — separated the two ranibizumab doses by 0.1 percentage points (94.5 versus 94.6). Its secondary endpoint, proportion gaining 15 or more, separated them by 9.0 (24.8 versus 33.8). The primary endpoint had saturated: nearly everyone on either dose met it, so it could not resolve a difference that was plainly there. Same patients, same trial, a ninety-fold difference in resolving power. The dose-response is visible in four comparisons across the two trials — 0.5 mg beat 0.3 mg on mean letters and on the 15-letter gain, in MARINA and in ANCHOR — and the primary endpoint saw none of it.
And the number that matters to a patient. In MARINA, 37.8 percent of sham patients lost 15 or more letters at one year, against 5.4 percent on 0.5 mg — a sevenfold reduction in substantial vision loss [41].
CATT then asked whether the far cheaper off-label bevacizumab would do the same, randomising 1,208 patients: monthly bevacizumab gained 8.0 letters against ranibizumab's 8.5, and as-needed dosing 5.9 against 6.8 [44]. Equivalent on vision. Serious systemic adverse events were more common with bevacizumab (24.1 versus 19.0 percent, risk ratio 1.29, 95% CI 1.01–1.66), and the paper reports that plainly.
The teaching point, and the catch. Anti-VEGF turned wet AMD from a cause of rapid central blindness into a disease where vision is preserved and often improved. But look at what MARINA required: 24 monthly injections into the eye over two years. The benefit is a function of continued treatment, not a cure — the vessels are suppressed, not removed. CATT's as-needed arms did slightly worse than monthly, which is the injection-burden problem in miniature, and treat-and-extend regimens exist to find each patient's longest tolerable interval [45] [46].
Three honest limits. The lines in the figure join baseline to the 12-month mean; intermediate timepoints are not plotted because the substrate does not carry them, and real trajectories are not straight. MARINA and ANCHOR enrolled different lesion types, so their treated arms are not directly comparable to each other — only each to its own control. And these are one-year outcomes under monthly injection in a trial; real-world outcomes with less frequent treatment are generally worse.
Pillar 2: treatment
Dry AMD and early disease
The AREDS formulation slows progression. The Age-Related Eye Disease Study randomised 3,640 participants aged 55–80 with drusen or advanced disease in one eye to antioxidants (vitamin C, vitamin E, beta carotene), zinc with copper, both, or placebo, and followed them an average of 6.3 years, with progression to advanced AMD and loss of 15 or more letters as the outcomes [47].
AREDS2 refined it. 4,203 participants at risk of progression, in a 2×2 factorial design, testing whether adding lutein and zeaxanthin, or DHA and EPA, or both, further reduced risk — and simultaneously testing whether beta carotene could be removed and zinc lowered [48].
That second question turned out to matter, and the trial answered it from its own data. There was no apparent effect of eliminating beta carotene, or of lower-dose zinc, on progression to advanced AMD — and more lung cancers occurred in the beta-carotene group than without it: 23 (2.0 percent) against 11 (0.9 percent), nominal P = 0.04, mostly in former smokers [48]. A component that contributed nothing detectable to the benefit and roughly doubled observed lung cancers in a population enriched for smokers [22] could be dropped at no cost. That is what a factorial design is for.
Note what this treatment is and is not. It slows progression in people who already have drusen. It is not a treatment for established geographic atrophy, and it is not a preventive for people with healthy retinas.
Complement inhibition for geographic atrophy is the genuinely new development, following directly from the factor H genetics [11] [17]. This substrate contains the complement genetics densely but not the pegcetacoplan or avacincaptad trials, so this review states no efficacy figures for them. What can be said from what is here is the interpretive problem: those drugs are assessed on lesion area growth [35] [36], and slowing the expansion of an atrophic patch is not the same as preserving the vision that patch would have taken. Whether a change in lesion growth translates into a change in what patients can see is the open question, and it is open precisely because the endpoint is structural.
Wet AMD
Intravitreal anti-VEGF is the treatment, delivered by injection into the vitreous, typically monthly at first and then extended. Ranibizumab [41] [43] and bevacizumab, used off-label at a fraction of the cost [44], are the agents this substrate documents in AMD; aflibercept is also used for macular degeneration [49], though its pivotal AMD trials are not in this substrate, which carries aflibercept only in diabetic macular oedema [50] [51]. Pegaptanib is of historical interest [40], and photodynamic therapy has been largely displaced [42].
Faricimab is a bispecific antibody targeting both VEGF-A and angiopoietin-2, designed to allow longer intervals between injections. The trial in this substrate is in diabetic macular oedema, not AMD — YOSEMITE and RHINE, showing extended dosing up to every 16 weeks [52] — so this review does not state AMD results for it.
The burden is the real-world limiting factor. Monthly injections into an eye, indefinitely, in elderly patients who must attend every visit. Treat-and-extend was developed to reduce it [45] [46]. Longer-acting formulations, refillable implants and gene therapy are the strategies being pursued to reduce it further; none of those are represented in this substrate and no results are quoted for them here.
Pillar 3: what is unresolved
Turning the complement result into a treatment that preserves vision. The genetics have been solid for twenty years [11] [12] [13] [19]. Converting that into a drug that changes what patients see, rather than how fast a lesion expands [36] [35], is where dry AMD stands.
Reducing injection burden. Every strategy in wet AMD now aims at fewer visits for the same effect [45] [52].
Catching conversion early. Wet AMD treated late does worse, so detecting the switch from dry to wet quickly has real value — the case for home monitoring [33] [34] and for automated screening [30].
Prevention. Smoking is the clear modifiable factor and its effect largely persists after quitting [22]. Diet has been studied extensively without producing anything as decisive [23] [24] [25] [53] [26].
And the biology under the drusen. Whether complement activation, oxidative stress and pigment-epithelium failure are one process or several remains unsettled [5] [7] [9] [8].
Dig deeper in lmmol
Glaucoma is the direct sibling and the instructive contrast: it takes peripheral vision while AMD takes central, it is silent while AMD announces itself with distortion, and its treatment lowers a pressure the patient cannot feel while AMD's requires an injection into the eye. Both are diseases where imaging now leads the diagnosis. Type 2 diabetes and glycemic control is the other great retinal disease, and the overlap here is unusually concrete — the same anti-VEGF drugs treat diabetic macular oedema [51] [50], the faricimab evidence in this substrate is in diabetic eyes [52], and the deep-learning screening systems were built for diabetic retinopathy first [31] [32]. Lung cancer connects through a specific and awkward fact: beta carotene raises lung-cancer risk in smokers, which is why AREDS2 tested removing it from a formulation given to a population enriched for smokers [48] [22]. And Alzheimer's disease shares both the complement-and-innate-immunity theme and the deposit-as-biomarker problem — drusen were reframed as immune markers [5] in much the way amyloid plaques were. The full collection is at health.