Endophthalmitis: Infection Inside the Eye and the Intravitreal Route

Why oral and topical antibiotics fail

The eye protects itself so well that it defeats its own rescue. The inside of the eye is one of the body's privileged sanctuaries. A blood–retinal barrier — tight junctions in the retinal vessels and the retinal pigment epithelium — keeps the vitreous chemically sealed off from the bloodstream, exactly as the blood–brain barrier guards the brain. This is wonderful for keeping toxins out, but disastrous during infection: swallow or infuse an antibiotic and only a trickle crosses into the vitreous, usually far below the concentration needed to kill bacteria. From the other direction, a topical drop faces a different wall. Drops penetrate the cornea and reach the front of the eye — the anterior segment — reasonably well, which is why they cure conjunctivitis and most corneal ulcers. But they cannot diffuse backwards across the whole globe to the vitreous in the posterior segment. So the two routes we reach for by reflex — a tablet and a drop — are precisely the two that cannot deliver a killing dose to where the infection actually is. The blood–ocular barrier is discussed in full in the Foundations chapter on ocular drug delivery.

The answer: injecting straight into the vitreous

If the drug cannot get in from the blood or across the cornea, you bypass the barrier entirely: you put the antibiotic where the bacteria are, by passing a fine needle through the sclera — a few millimetres behind the limbus, through the pars plana — and injecting a tiny volume (about 0.1 mL) directly into the vitreous cavity. This is the intravitreal route, and it changes the whole arithmetic. A dose small enough to be measured in micrograms produces a local concentration in the vitreous that oral therapy could never approach, without flooding the rest of the body. Before injecting antibiotic, the surgeon usually performs a vitreous tap — aspirating a small sample of vitreous for Gram stain and culture — so the empiric treatment can later be narrowed to the organism actually grown. Tap and inject, in the same sitting: sample first, then treat.

Where it comes from: the routes of infection

Endophthalmitis is classified by how the organism got in, and the route predicts both the likely bug and its aggressiveness. Post-operative (exogenous) endophthalmitis is the commonest, most often after cataract surgery; the culprit is usually a low-virulence skin commensal — coagulase-negative staphylococci such as Staphylococcus epidermidis — carried in from the patient's own eyelids and conjunctiva. Post-intravitreal-injection endophthalmitis is the same mechanism on a smaller scale: every anti-VEGF or steroid injection carries a small but real risk of introducing organisms, which is why sterile technique for those injections is obsessive. Post-traumatic endophthalmitis follows a penetrating injury, especially with a retained foreign body or soil/vegetable contamination; here Bacillus cereus is the feared organism — a strikingly aggressive infection that can destroy an eye within a day or two. Bleb-related endophthalmitis arises years after glaucoma filtration surgery, through a thin filtering bleb, often with streptococci. And endogenous endophthalmitis is the outlier: the infection arrives by the bloodstream from a distant source, in a septic or immunocompromised patient — classically Candida in an intravenous drug user or a patient on long-term IV access, or Klebsiella seeding the eye from a liver abscess.

What goes into the eye: the empiric regimen

You treat before the culture returns, so you cover both sides of the Gram stain at once. Because a delay of hours matters, treatment is empiric — broad enough to cover the likely organisms before any culture result. The standard is a pair of intravitreal antibiotics chosen to cover both Gram-positive and Gram-negative bacteria. For Gram-positive cover (including the resistant staphylococci and streptococci that dominate post-operative cases), vancomycin. For Gram-negative cover, including Pseudomonas, either ceftazidime (a third-generation cephalosporin, now generally preferred) or amikacin (an aminoglycoside, effective but carrying a risk of retinal toxicity — macular infarction — which is why ceftazidime has largely replaced it). The two are injected separately, in tiny volumes, into the vitreous. Note how directly this maps onto the systemic Antimicrobials chapter: the same vancomycin, the same anti-pseudomonal cephalosporin and aminoglycoside you learned for sepsis, redeployed here by a different route. If a fungal cause is suspected — the endogenous Candida picture, or an indolent infection after vegetable-matter trauma — the injection is an intravitreal antifungal instead: amphotericin B or voriconazole. The role of intravitreal dexamethasone, given alongside the antibiotic to dampen the destructive inflammation, remains genuinely debated — some trials suggest benefit, others none, and practice varies.

When injection is not enough: vitrectomy

For the most severe cases, the surgeon does more than inject: a pars plana vitrectomy physically removes the infected vitreous — the culture medium, the bacteria and the inflammatory debris all at once — and lets antibiotic distribute in the cleared cavity. The landmark Endophthalmitis Vitrectomy Study (EVS) shaped how this decision is made. Its central, much-quoted lesson was that vision severity guides the choice: eyes that still had reasonable vision at presentation did just as well with a tap-and-inject alone, whereas eyes that had already fallen to perception of light only did better with immediate vitrectomy. In other words, reserve the bigger operation for the worst eyes. The EVS also found no benefit from adding systemic intravenous antibiotics to the intravitreal ones for the drugs it tested — a direct confirmation of this whole article's premise that the systemic route barely reaches the vitreous. (Practice has since evolved, and newer agents with better ocular penetration have reopened the question, but the principle that intravitreal delivery is the mainstay still holds.)

Better than any cure: prophylaxis at surgery

Because established endophthalmitis is so blinding, the greatest gains have come from preventing it. Two evidence-based measures dominate. First, pre-operative povidone-iodine antisepsis: painting the conjunctival sac and lid margins with dilute povidone-iodine before surgery is the single best-supported step, drastically lowering the bacterial load that would otherwise be carried into the eye. Second, and transformative, intracameral antibiotic at the end of cataract surgery: injecting a small dose of an antibiotic — classically cefuroxime, with moxifloxacin a common alternative — directly into the anterior chamber as the operation finishes. A large European (ESCRS) randomised trial showed intracameral cefuroxime cut post-cataract endophthalmitis rates several-fold, and it is now standard in most cataract practice. Note the recurring theme: even in prevention, the winning move is to place the drug inside the eye rather than trusting drops or tablets to get there.