TRAVIS A. MEREDITH
Table Of Contents
CLINICAL PRESENTATION AND CHARACTERISTICS
PRINCIPLES OF DIAGNOSIS
PRINCIPLES OF THERAPY
MANAGEMENT STRATEGIES AND OUTCOMES
OUTCOME BY ORGANISM
|Infectious bacterial endophthalmitis occurs when bacteria gain entrance
to the interior of the eye, replicate, and produce an inflammatory response
that may eventually involve tissues throughout the eye. Most infections
occur after a breech of the external coats of the eye by surgery
or trauma. Direct invasion of the interior of the eye through the
cornea or sclera following local abscess can occur but is relatively uncommon. Endogenous
endophthalmitis occurs when an infected source elsewhere
in the body leads to metastatic intraocular infection. There are
many classifications of endophthalmitis, but the most commonly recognized
categories are: (a) Postsurgical: acute, and delayed or
chronic; (b) posttraumatic; (c) bleb-related; and (d) endogenous: fungal, bacterial, and other. |
Eyes with apparent infection may not grow out bacteria when appropriately subjected to culture. These cases are commonly called sterile endophthalmitis and accounted for 31% of the post-cataract surgery infected eyes entered into the Endophthalmitis Vitrectomy Study (EVS).1,2 Polymerase chain reaction (PCR) evidence indicates that many of these cases may be the result of bacteria present in the eye that do not grow on culture.3,4 Another syndrome, commonly referred to as a Toxic Anterior Segment Syndrome, is a reaction to intraocular foreign material, such as intraocular lenses or toxic irrigating solutions.5 The term pseudo-endophthalmitis also has been used, probably incorrectly, to designate accumulation of injected corticosteroid in the anterior chamber producing a white layering called a pseudohypopyon.
|In most mixed clinical series, approximately two-thirds of cases of endophthalmitis
are caused by postoperative infection.6–9 The majority of cases of endophthalmitis occur after cataract extraction
because of the frequency of the procedure. The causative organisms
are presumed in most cases to be introduced during surgery, and genetic
studies have demonstrated that after cataract surgery the organisms
isolated are genetically identical to the patient's own flora.10,11 There is some difference in incidence based on the epidemiologic approach. National
registries in Sweden12 and Norway13 found rates of 0.1% and 0.11% to 0.16%, respectively. In
hospital-based surveys, Allen reported from the Massachusetts
Eye and Ear Infirmary for the 13-year period ending in 1977 that the incidence
of endophthalmitis was 0.057%.14 A review from the Bascom Palmer Eye Institute, of 23,625 cases of cataract
extraction, revealed an incidence of 0.072%,15 but a more recent study, from 1995 to 2001, noted that there was a drop
in incidence to 0.05%.16 A review found that a freestanding (ASC) experience documented
a 0.3% incidence.6 The incidence after pars vitrectomy is similar.16,17 |
In cases of endophthalmitis after cataract extraction, a significant relationship to operative and postoperative complications has been identified. Capsular rupture, vitreous loss, wound leaks, postoperative filtering blebs, and vitreous wick have been identified as associated with the postoperative infection.18 The type of incision may contribute to the incidence of postoperative infection. In a recent case-control study from an ASC, a threefold greater risk of and optimize with clear corneal incisions was identified compared to scleral tunnel incisions.19 A higher incidence of infection has been noted in temporal incisions that was superior incisions in another study.20 In secondary intraocular lens (IOL) implantation, a case-control study showed an association of infection with diabetes mellitus, propylene haptics, scleral suture fixation of posterior chamber IOLs, preoperative eyelid abnormalities, postoperative wound defects, and re-entry of the eye through a previous wound.21
Intraocular injections of corticosteroid and antibiotics have become increasingly common and will undoubtedly increase in the future, because intraocular injections are being introduced for treatment of macular degeneration. A review of 4,382 eyes that received 14,866 injections documented 38 cases of endophthalmitis for risk of 0.2% per injection.22
After filtration bleb surgery the immediate incidence of infection is similar to cataract surgery or higher, but the eye remains at risk as long as the bleb is present. Studies suggest that the rate of infection on a yearly basis may be as much as 1% to 7.9%.23–25 Glaucoma drainage devices also may become infected.26
Posttraumatic cases account for roughly 25% of all cases of endophthalmitis.6–9 The likelihood of infection is at least 100 to 300 times higher than after elective cataract surgery. Rates of infection after penetrating injury are approximately 2% to 3%,27 rising to 11% to 20% with retained intraocular foreign body.28–30 The National Eye Trauma Registry System surveyed 429 cases of intraocular foreign bodies and documented an incidence of endophthalmitis of 6.9%. There was no significant difference in risk depending on whether the foreign body was metallic, nonmetallic, or organic material.29 Thirty percent of injuries in rural environments in one study became infected.31
The relevant group for incidence studies for endogenous endophthalmitis is not clear. Patients with known fungal septicemia frequently develop white dots in the choroid and retina that respond as a systemic disease is treated. Frank endogenous infection is relatively uncommon.
|CLINICAL PRESENTATION AND CHARACTERISTICS|
ACUTE POSTOPERATIVE ENDOPHTHALMITIS
Acute postoperative endophthalmitis may follow any intraocular surgery. Because cataract surgery is by far the most common ophthalmologic procedure, infections following lens extraction and IOL implantation are most often encountered. Infection may follow minor manipulations such as cutting sutures in corneal grafts or cataract incisions.32 Infection usually begins within the first 6 weeks, often within the first 1 to 2 weeks. More virulent organisms generally produce symptoms earlier. There is also a subgroup, termed delayed or chronic endophthalmitis, in which the onset of infectious symptoms may be delayed for 2 to 4 months.33,34
The EVS was a National Eye Institute supported multicenter randomized clinical trial that evaluated four treatment regimens in post-cataract endophthalmitis. The study is important in understanding the current demographics and clinical parameters of post-cataract infection in United States. The study also served to validate proper diagnostic techniques and establish certain treatment regimens as most likely to be successful.2
Patients typically notice increased pain, sometimes beginning with a scratchy or irritated sensation. This usually is accompanied by blurred vision. Patients may notice increased exudates as well. In the EVS approximately 75% of patients had pain as a presenting symptom.35 The time of onset of symptoms is correlated with the organism's virulence. In a large study of Staphylococcus epidermis endophthalmitis by Omerod36,37 58% of the cases had the onset within the first week after surgery, whereas 42% were delayed into the second week or later. In the EVS there was a tendency for infections produced by more virulent organisms such as Staphylococcus aureus, Streptococcus, and gram-negative infections to present in the first two days after surgery.38 The eye has varying degrees of redness, exudate, and lid edema. Slit-lamp examination may reveal wound complications or wound infiltrate, although in the phacoemulsification era these are relatively less common than identified previously. Increased anterior chamber flare and cell is almost always present; as the condition becomes more severe, variable degrees of fibrin are noted. Hypopyon is considered the hallmark of infection but it was only seen at presentation in 75% of the EVS patients.35 In early cases when the fundus may still be seen, vasculitis has been described as one of the earliest findings.39 Vitreous inflammation, however, is an invariable part of the presenting picture and usually is severe enough to obscure a clear fundus detail. Eyes with more virulent organisms are more likely to present with swollen eyelids, have light perception-only vision, demonstrate an afferent papillary defect, a corneal infiltrate, and loss of the red reflex.38
In the EVS, 69% of the eyes were culture positive, but 31% were culture-negative. Of culture-positive eyes, 70% grew coagulase-negative staphylococci. Only 6% of eyes grew gram-negative organisms. Polymicrobial infections were identified in 9% of the cases.2,40
After most intraocular surgery the major differential is between infection and an exaggerated inflammatory reaction to the surgery. When there is significant retained lens material, inflammation usually is greater than expected.41 Retained cortical material may produce an increased inflammation, which usually is evident from seeing pieces of cortex in the peripheral portion of the capsular bag or displaced into the vitreous or anterior chamber. Retained nuclear material usually creates less inflammation.
DELAYED-ONSET POSTOPERATIVE ENDOPHTHALMITIS
Endophthalmitis may present some weeks after the initial surgical procedure. Less virulent bacteria and fungi usually are suspected. Propionibacterium acnes produce a particular syndrome that may be suspected from the clinical presentation.42–44 In the initial description of this disorder the average interval between cataract surgery and microbiological confirmation of the infection was 12 months. Most cases present 2 months or later after the initial surgery. The majority of patients present with granulomatous anterior uveitis, although nongranulomatous reactions also are observed, and some patients can even have hypopyon. A finding that is almost pathognomonic is a white plaque found on the capsule. These may be observed to increase in size under serial observation.45,46 Some relatively nonvirulent bacteria can produce a chronic inflammatory response as do fungi. There is no other constellation of findings that can distinguish between fungal infection and less virulent bacteria. Culture and stains are necessary to make the diagnosis in these cases. The laboratory must be notified that the specimen must be specifically analyzed for fungi; for those in which P. acnes is suspected that the culture must be held for 2 weeks because the organism is slow growing.
Most cases of bleb-related endophthalmitis occur months to years after the original procedure. Both intentional filtration blebs and traumatic blebs are susceptible to infection. Patients may have a prodrome of browache and/or headache. There often is an antecedent conjunctivitis manifested as redness of the eye with some degree of exudation.23–25,47,48 Patients who have inferior bleb and those in whom antifibrotic agents have been used are more susceptible to infection.23–25 Although some blebs are Seidel-positive, intact blebs also may become infected. Thin blebs may be at higher risk. A recent report documented that bleb manipulations and bleb revisions are significant risk factors for the creation of bleb-related endophthalmitis.49
The clinical presentation is classically described as “white-on-red” because the bleb is filled with white or yellow material in contrast with the surrounding conjunctival erythema. Examination findings of the cornea, anterior chamber, iris, and vitreous are similar to that in postoperative endophthalmitis.
An important differential diagnosis is that of blebitis versus bleb-related endophthalmitis. Absence of vitreous involvement indicates the diagnosis of blebitis.48,50 Bleb-related infections have a very different bacterial spectrum than those from endophthalmitis following cataract surgery Streptococcus sp. are identified in 31% to 57% of eyes.49 More recently, Staphylococcus and Enterococcus have been reported increasingly as the causative organisms, and gram-negative species also are found.49,51
In traumatic endophthalmitis bacteria are introduced into the eye by traumatic penetration of the sclera or cornea. Studies have shown that not all patients who are culture-positive at vitrectomy for trauma develop endophthalmitis, possibly as the result of removal of the organism by the vitrectomy.30,52 The differential diagnosis is between posttraumatic uveitis and true infection; severe inflammatory signs from the initial injury may make the diagnosis of endophthalmitis more difficult.
Typically patients have increased pain and redness postoperatively. Gram-positive organisms are responsible for approximately two-thirds of postoperative infections. Gram-negative organisms may cause 10% to 15% of the infections. However, several series have indicated that Bacillus infections are a responsible for 20% to 25% of posttraumatic infections, and most of these are associated with intraocular foreign bodies.28,31,53–55
The rapidity of onset of symptoms may offer some clue to bacteriologic diagnosis, because Bacillus spp. produce a rapidly progressive infection. Findings in Bacillus infection include marked chemosis and exudates, a ring infiltrate of the cornea, diffuse opacification of the cornea, and severe intraocular inflammation. In other cases of endophthalmitis, a hypopyon appears and the vitreous may become more clouded after otherwise successful surgery.
Fungal infections may have a late onset and are seen in as many as 15% of cases of traumatic endophthalmitis. In the clinical presentation of these infections there may be the appearance of yellow or white fluffy material in the anterior chamber and/or vitreous cavity. Other cases are indolent and have progressively increased inflammatory changes in the anterior chamber and vitreous cavity.
Endogenous endophthalmitis occurs primarily in patients with systemic illness. Many of these individuals are immune compromised; others may have systemic illness including endocarditis, urinary tract infections, recent gastrointestinal surgery, or hepatobiliary infections.56–59 In a large series of endogenous fungal endophthalmitis, important predisposing factors were intravenous hyperalimentation, a history of cancer, and a fever of unknown origin.60 The presenting symptoms and signs depend on the series being reviewed. Illness may be detected in an outpatient setting where ocular symptoms are the first presenting signs of the generalized illness. Others present as a complication in patients with known systemic sepsis or immunosuppression. In two series most initial contact was made with an ophthalmologist, whereas in another series most of the patients were hospitalized at diagnosis. Initial symptoms may be mild with decreased vision and redness, pain, and photophobia. Because patients may be quite ill, sometimes these symptoms are not adequately stressed to the caregivers on an acute medical service.59 The initial misdiagnosis rate approaches 50% in some series.56,58 Initial incorrect diagnoses included iritis and uveitis, conjunctivitis, glaucoma, and cellulitis. Therefore, diagnosis is frequently delayed until the disease has become more advanced. Bilateral presentation is fairly common and is reported in 15% to 26% of cases.56–59
When endogenous endophthalmitis is considered, an internist or infectious disease specialist usually is consulted to search for the systemic source of infection. Blood cultures are frequently critical in establishing the diagnosis. A wide range of organisms are causative for this infection and 50% to 62% of the cases have been reported owing to fungal agents. Candida spp. are the most common isolate in some series, whereas Aspergillus has been noted by other authors. Gram-positive organisms are more prevalent than gram-negative organisms.56–59
|PRINCIPLES OF DIAGNOSIS|
|The clinician must be alerted to the possibility of the diagnosis of endophthalmitis
by the clinical findings. In eyes with greater than expected
intraocular inflammation based on the clinical setting, a high degree
of suspicion is important. Failure to appreciate the potential severity
of the problem may delay diagnosis. If endophthalmitis is suspected
frequent observation versus active intervention must be chosen. In
some instances it may be appropriate to initiate anti-inflammatory therapy
and monitor the patient two to three times daily until the clinical
course in becomes clear.|
For unequivocal diagnosis samples of the aqueous and vitreous must be obtained for culture, sensitivity, and stain using the techniques described in the following. Samples from the vitreous are more often positive that are samples from aqueous.61,62 Vitreous samples are positive as frequently from a tap/biopsy as they are after vitrectomy to obtain the specimen.22 Undiluted vitreous and aqueous may be placed on the following media for culture: enriched thioglycolate liquid medium, chocolate agar, Sabouraud's agar. Anaerobic cultures typically use either thioglycolate enriched broth or blood agar enriched with hemin and vitamin K. In some institutions material from the vitreous cassette after vitrectomy is filtered through a 0.45μ membrane filter. The filter is subsequently divided into three pieces under sterile conditions and used for culture.2
Stains are prepared from anterior chamber and vitreous specimens. In the EVS, the findings of a positive gram stain were associated with significantly worse final media clarity and visual acuity. The gram stain result did not reveal any subgroups in which vitrectomy had a beneficial value and therefore was of little consequence in making initial therapeutic decisions.63 In the EVS there was no difference between the positive rate for culture between samples obtained by tap and those which were obtained by vitrectomy. There was no significant difference in operative complication between the two methods.
|PRINCIPLES OF THERAPY|
|In modern endophthalmitis therapy, intravitreal antimicrobial injections
are the mainstay of treatment.64 Pars plana vitrectomy has an important role in management of several clinical
presentations and stages of the disease. Subconjunctival antibiotic
injection and topical antibiotic injection were previously considered
major routes for antibiotic administration, but now has a less important
role.65 Corticosteroid therapy is important in improving the final outcome, but
there is debate over the proper routes of administration.|
Choice of Antimicrobial Agent
Because endophthalmitis therapy must be initiated emergently, the identity of the organism is unlikely to be known at the time antimicrobial agents must be chosen. The vast majority of endophthalmitis infections are gram-positive in origin.2,66,67 As noted, the clinical setting often determines which organisms are more likely to be present. Because it is almost impossible initially to rule out gram-negative organisms as the cause of most infections on clinical examination, broad-spectrum antibiotic coverage usually is chosen.38 In practice, this usually means empiric treatment with two separate antimicrobial agents as the initial choice.
Desirable characteristics for antimicrobial agents for treatment of bacterial endophthalmitis include the following:68
The highest possible initial dose is chosen whenever possible so that the drug remains above the MIC for the longest period of time. Drugs with longer half-lives are also preferred because the effective dose of drug is thought to be eliminated from the drug by the expiration of five half-lives. The upper limit of drug concentration is defined by the toxicity, usually in the form of retinal deterioration. If other characteristics are equivalent, then choice of the drug exceeding the MIC by the greatest degree is preferable. Some authors have suggested that concentrations of 10 to 30 times the MIC are necessary to effectively treat infections elsewhere in the body.
The standard of care for most cases of endophthalmitis is intraocular injection of antimicrobial agents. Antibiotics injected into the vitreous cavity are eliminated from the eye by either an anterior or posterior route. β-Lactam antimicrobials were removed posteriorly, whereas the aminoglycosides exit through the anterior (trabecular meshwork) route.78,79 The removal of the vitreous shortens the half-life of antimicrobials, whereas lens removal also can shorten the half-life of those using an anterior route of elimination. Inflammation breaks down blood–retinal barriers and also decreases the half-life of anteriorly eliminated drugs.80–83
Although some controversy exists, there is a short half-life of most injected antimicrobials, so that effective duration of action of many antibiotics may be somewhere between 36 and 72 hours. Toxicity has been demonstrated in the form of retinal vascular shutdown by intravitreal aminoglycosides84–87 and retinal necrosis in the case of other antibiotics.70,71 Combinations of antimicrobials such as vancomycin and amikacin may increase their toxicities.70,71 The antimicrobials usually chosen for intraocular injections at this time are the following agents:
Systemic antimicrobials were once considered crucial for the treatment of endophthalmitis, but now their efficacy is debated. Because inflammation breaks down the blood–eye barriers, greater penetration into inflamed eyes occurs after systemic administration.68,72,74–76 Vitrectomy enhances the penetration of some medications (cefazolin,75 vancomycin,76 and ceftazidime68,74) into the eye. Repeat intravenous dosing likewise contributes to progressive increase in intraocular concentrations.68,72,74–76
Antimicrobials may be administered either orally or intravenously. Although studies have demonstrated response in staphylococcus endophthalmitis and other more virulent organisms to intravenous antibiotics only,94 the EVS did not demonstrate any improvement in visual outcome in patients given intravenous therapy along with intraocular therapy as compared to patients given intraocular therapy only.2 However, in that study the intravenous antibiotic given for gram-positive coverage was amikacin, which has very poor penetration into the vitreous cavity after intravenous injection and would not be expected to provide a significant effect.73 Newer data indicates that fluoroquinolones, particularly fourth-generation agents, may have improved intraocular penetration due to their ability to cross the blood–eye barriers.95,96 Most antimicrobials are more likely to be able to achieve intravitreal levels if the eye is inflamed because of a breakdown of the blood–ocular barrier. Specific antimicrobials often considered for systemic therapy include vancomycin, fluoroquinolones, and cephalosporins.
Frequent subconjunctival injections were previously advocated as a mainstay of therapy, but the entry of antimicrobials into the eye after subconjunctival injection is marginal.104–108 Vitreous levels of antimicrobial after subconjunctival injection rarely reached therapeutic levels and are essentially insignificant by comparison to the concentration produced by intravitreal injections. Subconjunctival injections may release antimicrobial onto the surface of the eye and therefore may be useful for any external aspect of the infection such as bleb infection or wound infection.
Topical antimicrobials have the same limitations as do subconjunctival injections. Attempts are made to overcome this problem by producing highly concentrated formulations and administering the drops frequently during the day. Some authors recommend topical vancomycin (25 mg/mL), aminoglycosides (9 to 14 mg/mL) or ceftazidime (50 mg/mL). Like subconjunctival injections, these may have some benefit for external aspects of the infection but the amount of drug delivered to the vitreous cavity is marginal
Two basic procedures are used for initial treatment of endophthalmitis: a) vitreous tap with injection of intraocular antibiotics, and b) pars plana vitrectomy.
Tap and Inject
A vitreous tap is usually done through the pars plana because most cases of endophthalmitis occur in eyes that have a posterior chamber intraocular lens (PCIOL) in position or that are phakic. The anterior chamber may be tapped to obtain aqueous culture, or may be the entrance site for a tap to sample the vitreous cavity in an aphakic eye. To tap the anterior chamber, topical anesthesia is usually employed. The eye may be stabilized with a Q-tip or forceps in one hand and entered with a 30-gauge needle at the limbus with the operative hand. A tuberculin syringe often is used for convenience, and as much as 0.2 cc of fluid may be withdrawn. If there is significant fibrin in the anterior chamber, the small-bore needle may not be sufficient to achieve a good sample, and an entry must be made with a 25- or 27-gauge needle. A small puncture incision with a blade may facilitate use of the larger gauge needles.
Tap of the vitreous cavity usually is made through the pars plana 3.5 to 4 mm posterior to the limbus. Depending on the amount of inflammation and the pain tolerance, this may be done with a conjunctival injection or retrobulbar injection. A topical anesthetic of choice may be placed in the eye and supplemented if desired with a pledget of lidocaine placed over the area for injection for a few moments. A bleb may be raised with subconjunctival lidocaine 2% in the area if the topical anesthesia does not seem to produce sufficient analgesia. In the case of very inflamed eyes, this may not be sufficient, and retrobulbar or peribulbar injection of lidocaine 2% to 4% may be chosen.
Topical povidone iodine 5% is instilled in the conjunctival sac for 1 to 2 minutes prior to the injection. A lid spectrum usually is helpful to prevent blinking. The eye then is stabilized with forceps or a Q-tip and a position 3.5 mm posterior to the limbus is identified. A 22- to 27-gauge needle may be used for the initial penetration and inserted with a slight screwing motion. Once in the eye, it should be advanced far enough to be clear of the vitreous base and then suction should be applied. If the medium is clear enough, it is best to visualize the position of the needle in the vitreous cavity before suctioning. Suction without clear visualization is dangerous and should be avoided if possible. The goal should be to obtain 0.2 to 0.3 cc of fluid for gram stain, culture, and sensitivity. Needles as large as 20 to 22 gauge sometimes are necessary because of the viscous nature of an inflamed vitreous. The tap should be aborted if liquid vitreous is not obtained. A single port vitrectomy then may be used to obtain material for analysis. Once the needle is withdrawn, pressure is placed over the sclerotomy site with a cotton tip. Material is submitted immediately to the laboratory. After the tap has been done, injection is performed often through the same incision site if possible. Typically, 0.1 cc of each of two antimicrobials is injected into the vitreous cavity. The usual antibiotic injections are vancomycin 1 mg in 0.1 cc and either amikacin 200 γg in 0.1 cc or ceftazidime 2.25 mg in 0.1 cc.
Pars Plana Vitrectomy
Pars plana vitrectomy was adopted early for treatment of intraocular infection shortly after its introduction and acceptance as a surgical procedure.109,110 Pars plana vitrectomy allows application of the principles of incision and drainage of an abscess. This intervention allows for the removal of the infected organisms, removal of opacified vitreous and debris, and potentially reduction of toxins.
Pars plana vitrectomy for infection is performed under local or general anesthesia, although it may be somewhat difficult to achieve adequate local anesthesia in a very inflamed eye. Standard conjunctival cutdowns are performed and an infusion port placed into the inferotemporal sclera 3.5 mm posterior to the limbus. It is important to visualize the port inside before turning on infusion.
The anterior chamber may be quite opaque. In these cases pars plana infusion is put in position but not turned on. Another infusion is carried through the limbus and the vitrectomy cutting probe introduced into the anterior chamber to clear debris. It is often necessary to strip away fibrin membrane from the surface of the iris and intraocular lens. In some instances iris retractors may be beneficial if the pupil will not dilate properly. Material obtained from the anterior chamber is submitted for culture and sensitivity.
Once the vitreous cavity may be examined, the infusion is identified and turned on so that a light source and cutting probe may be introduced into the posterior segment. Before turning on the posterior infusion the vitrectomy probe is hooked up to a small syringe and an undiluted sample is taken, cutting the opaque vitreous and sucking the material into the syringe: 0.5 to 1 cc may be safely removed in this manner and submitted for culture and stains.
Vitrectomy is carried further into the vitreous cavity using simultaneous fluid infusion. Once the central cavity is clear, whitish material on the surface of the retina may be noted. It is important to be careful not to apply traction either with suction or direct pulling on areas of necrotic retina. Sometimes white material represents white cells layered on the surface of the retina which can be vacuumed free. White material trapped in membrane often clears within the first few days without being stripped away if the infection has been adequately treated. In cases of Propionibacterium acnes there may be small plaque of white material sequestered in the capsule that is thought to be the nidus of infection. It is then necessary to remove this plaque in most eyes in order to cure the infection. Removal of the entire lens capsule and even removal of the lens and/or IOL exchange is sometimes necessary to effect a cure in these cases.42–44,111
One sclerotomy is closed and then a stitch put into the second sclerotomy. Intraocular antibiotics are injected into the eye through the second sclerotomy before closure.
There are a variety of indications for pars plana vitrectomy in the treatment of endophthalmitis. The EVS demonstrated that in eyes with light perception there is a 33% chance of achieving 20/40 vision at the initial intervention with vitrectomy and intraocular antibiotic injection versus only 11% likelihood if the initial intervention was a vitreous tap and injection only. For eyes with hand motion or better vision, there was no difference in visual outcome between these two initial strategies.2 Subsequent analysis of the data demonstrated that eyes of patients with diabetes mellitus had a tendency to fare better with a vitrectomy as an initial intervention, but the trend did not reach statistical significance.112 This study was an analysis of a factor identified retrospectively which was not part of the randomization scheme.
Vitrectomy also may be elected for eyes with acute infections not responding to the initial strategy of tap and inject.113 In the EVS, over 10% of eyes were subjected to additional surgery within a week of the initial therapeutic intervention. Early secondary interventions were undertaken in 8% of eyes undergoing initial vitrectomy and in 14% of eyes treated with tap and inject. Surgeons should consider a vitrectomy in eyes that do not respond to initial therapeutic intervention, because it is clear from laboratory113–117 and clinical studies113,118,119 that a single injection of antimicrobials does not cure all endophthalmitis. Furthermore, in animal studies the strategy of initial vitrectomy is more effective in sterilizing the vitreous cavity as an initial strategy than is an injection of antibiotics alone.114,115 In the EVS cultures obtained in 33 eyes operated for worsening inflammation were positive 42% of the time. Eyes initially treated with vitrectomy were positive in only 13% of the cases, whereas eyes treated with tap and inject initially were positive 71% of the time.113
In eyes that have apparent cure of the infection some vitreous opacities may persist for several months interfering with the patient's ability to see. Only a few of these eyes eventually require a vitrectomy to clear the media.113
Vitrectomy and intraocular antibiotic injection may be elected as initial intervention in other circumstances when vision is better than light perception. In cases with severe bleb infections, for example, the incidence of virulent organisms is high and the onset may be rapid. In such instances the suspicion of rapid progression of a more severe infection may tip the balance toward vitrectomy as an initial intervention. Vitrectomy also may be elected in situations where there is an indication for vitreous surgery in addition to the infection. These might include infection associated with retained intraocular foreign body or retained lens material.
Chronic indolent cases of endophthalmitis have been demonstrated to respond better to vitrectomy than initial antibiotic injection. This is particularly true for Propionibacterium acnes infections that have a very low rate of cure with injection of antibiotic injection only.42–44,111 Fungal cases also may require vitrectomy with an intraocular injection of amphotericin B.120
In most instances, an eye treated with vitrectomy shows as much or more inflammation the following day than it did at presentation because of the addition of surgical trauma. Aggressive anti-inflammatory therapy is often indicated. In the immediate postoperative period, the cornea is often edematous if the inflammation has been severe. Persistent epithelial defects occur in some patients and sometimes require management with a bandage contact lens.
Intraocular pressure may be elevated in the postoperative period usually responding to medical management. If there has been severe inflammation, hypotony may occur, associated with persistent inflammation, and lead to a downhill course. A leaking wound site may be suspected, but in general the cause of this complication is ciliary body shutdown.
In eyes that are phakic at the time of infection and therapy, cataract often develops in the postoperative period. Usually time is allowed to pass for the inflammation to clear before cataract surgery is undertaken. If there is apparent vitreous opacity on clinical examination or ultrasound, pars plana lensectomy combined with vitrectomy may be elected.
The most feared complication of endophthalmitis with or without vitrectomy is retinal detachment. In the EVS, retinal detachment occurred in 8.3% of the eyes; there was no significant difference in incidence between eyes treated with initial vitrectomy and those treated with tap and inject as the first intervention.121 Retinal detachment may occur because of breakdown of necrotic retina or because of traction on the vitreous or vitreous base during vitrectomy or vitreous tap. Detachments are a major cause of failure in many series,122 but in the EVS series, 78% of cases were successfully repaired. There was a correlation of retinal detachment and poor visual outcome.121
In eyes of the retinal detachment, proliferative retinopathy is a significant complication. Sympathetic ophthalmia has been reported.
In laboratory studies concomitant corticosteroid use almost always leads to better clinical outcomes than controls not treated with corticosteroids. The proper route of administration remains a source of some controversy.
In the EVS patients were routinely treated with prednisone 30 mg PO twice daily for 5 days following surgical intervention.2 In animal studies similar treatment has proven effective in many models of gram-positive infections.114,123
Intraocular corticosteroids are advocated by some authors.124–127 In many animal models, this route of administration is associated with better clinical outcomes than no corticosteroid treatment at all. However, neither laboratory nor clinical studies demonstrate a clear-cut advantage for intraocular injection over systemic administration. Retrospective clinical studies have provided conflicting results. Das and co-workers, in a prospective randomized trial, demonstrated an earlier reduction of intraocular inflammation after injection intraocular corticosteroids, but there was no lasting benefit on visual outcome compared to controls.128 Two retrospective studies reached differing conclusions. Shah and co-workers reported that those patients receiving intravitreal corticosteroids had a reduced chance of achieving a three-line improvement of visual acuity in the postoperative period.129 All patients had treatment with intraocular antibiotic. At another institution a retrospective analysis of 42 patients demonstrated no difference between those treated with intraocular corticosteroids and those without intraocular corticosteroids.130 None of these studies, however, is a randomized trial, so there is no data to indicate clearly the superiority of this technique over systemic administration. A single laboratory evaluation has demonstrated worse outcomes in an animal model of Staphylococcus aureus given intraocular corticosteroids when compared to systemic administration of corticosteroids.114
|MANAGEMENT STRATEGIES AND OUTCOMES|
|Initial management strategy depends on the type of endophthalmitis, the
presenting findings, systemic conditions (including drug allergies, systemic
diagnosis, and systemic fitness for surgery), and the
possibility of local infection elsewhere as the cause for infection. The
outcome depends on the clinical characteristics of the infection
and is largely dependent on the bacterial or fungal species creating the
ACUTE POSTOPERATIVE INFECTION
In the EVS, four separate initial management strategies were evaluated. All eyes received intravitreal vancomycin 1 mg and amikacin 0.4 mg. Eyes were initially randomized to either a pars plana vitrectomy or tap and injection of antibiotics. The two groups were then randomized into those receiving intravenous amikacin, and those of receiving no intravenous antimicrobial. Visual outcome was evaluated at 9 to 12 months. Fifty-three percent of patients achieved visual acuity 20/40 or better, whereas 74% achieved vision of 20/100 or better. Only 15% had visual acuity worse than 5/200.2
Visual acuity at presentation had the most important impact on the strategy that achieved the best outcome, and on the final visual acuity. Eyes with light perception only vision at presentation had a three times greater chance of achieving 20/40 vision with vitrectomy compared to tap and inject (33% versus 11%). These eyes also had a 50% decrease in the frequency of severe visual loss (20% versus 47%) when treated with vitrectomy compared to those eyes treated without vitrectomy. More favorable outcomes were achieved in eyes with hand motion or better vision at presentation, and there was no statistically significant difference between initial tap/ biopsy or vitrectomy in the effect on outcome. Thus for these eyes an initial strategy of vitrectomy is not usually indicated. Almost two-thirds of these eyes achieved 20/40 or better vision and 85% achieved a 20/100 or better. The risk for severe visual loss was approximately 4%.2
Additional procedures were required in 10% of eyes within the first week of a surgical intervention. Thirteen percent of eyes undergoing initial tap and 8% of the vitrectomy eyes underwent additional procedures. Most were operated for indications of worsening inflammation. Cultures obtained from these eyes were positive in 42% of the cases. These secondary cultures were positive for eyes having an initial tap 71% of the time, but only 13% of the time in eyes which had an initial vitrectomy. Twenty-seven percent of patients required additional procedures after 7 days. These included posterior capsulectomy, vitrectomy, retinopexy, scleral buckling, and epiretinal membrane peeling.113
The type of bacteria isolated from the culture had an impact on visual acuity. Rates of achieving final visual acuity of 20/100 or better were: gram-positive, coagulase-negative micrococci, 84%; Staphylococcus aureus, 50%; streptococci, 30%; enterococci, 14%; and gram-negative organisms, 50%.63 Administration of intravenous antibiotics did not change the outcome of treatment and therefore been abandoned for routine administration.2
The most common cause for final visual impairment was abnormality of the macula, accounting for approximately 50% of patients with vision less than 20/40. Risk factors predicting a poor visual outcome were worse initial vision, small pupil size after maximal dilatation, presence of rubeosis irides, and absence of a red reflex.2 Other clinical factors predicting decreased final visual acuity included an open posterior capsule, corneal infiltrate or ring ulcer, history of diabetes, older age, and abnormal intraocular pressure.
DELAYED-ONSET POSTOPERATIVE INFECTION
Eyes with delay of the onset of postoperative infection for 6 weeks to longer should be suspected of harboring P. acnes infection or fungal infection. Fungal infections also can present acutely.131 An initial tap may be elected for identification of the organism so that the surgical intervention may be staged. For eyes with P. acnes infection, vancomycin is the preferred initial antibiotic choice. Eyes treated only with intraocular antibiotic, however, have a very high rate of recurrence. Two studies have demonstrated that an initial strategy of pars plane vitrectomy and intraocular antibiotics also fails to eradicate infection in half the cases. Thus, for initial therapy, a vitrectomy with removal of capsular plaque (and in selected cases, the capsule) is the preferred intervention. IOL exchange is a safe approach to treat the primary infection in a single procedure, unlike treatment of other forms of bacterial endophthalmitis. Fifty percent of patients achieved vision equal to or better than 20/40, and 78% were at least 20/400 in one series, whereas in another the average vision was 20/30 to 20/40.42–44,111
A recent survey of members of the American Glaucoma Society indicated that approximately half of the respondents chose initial treatment with topical quinolones only.132 Systemic microbial therapy has been reported to be successful in treating blebitis before the vitreous cavity becomes involved.50 Because moderate levels of aqueous humor drug concentration can be achieved by systemic administration, systemic antimicrobials may be more effective. Such eyes should be examined frequently because extension of the infection to the vitreous cavity may herald a rapid downhill course. In studies of blebitis, good outcomes have been reported by Poulsen,48 Brown,50 and Ciulla.133
Bleb-related endophthalmitis usually is a late onset disease after the bleb has been present for some time. Because virulent organisms are frequently the causative agents, outcomes are poor. Pars plana vitrectomy should be considered as one option for early intervention even in the absence of visual decline to the light perception level. Although not effective in the EVS, systemic antibiotics also may be considered because of the presence of infection in the bleb and anterior chamber. Furthermore, a modest vitreous cavity concentration may also be achieved.
Review of five series of treatment of bleb-related endophthalmitis demonstrated that only 41% of eyes achieved vision equal to or better than 20/400. This was especially influenced by the number of eyes with streptococcal infection; only 33% of these eyes achieved vision at that level. Streptococcus infections accounted for 45% of the combined series.49 In the series by Song 22% of patients underwent enucleation or evisceration secondary to pain, and/or poor vision. Of the other eyes, pressure was poorly controlled in 11%. Eyes treated with vitrectomy had worse outcomes than those treated with tap and inject, presumably because selection bias. Worse eyes usually were chosen for vitrectomy as initial management.49
In traumatic endophthalmitis early mild disease may be approached with a tap and injection strategy if it is clear there is no retained intraocular foreign body or significant amount of lens material in the eye. If these conditions are present, vitrectomy may be adopted as the initial strategy in order to rid the eye of these materials.134,135 Early onset severe endophthalmitis may herald a mixed infection or infection with a severe virulent organism such as Streptococcus or Bacillus. If such an infection is suspected, vitrectomy may be chosen because of its demonstrated ability to clear the infection more quickly.134,135 Systemic antimicrobials also should be considered if there an organism other than coagulase-negative Staphylococcus is suspected. Studies of trauma have demonstrated that not all eyes with positive cultures at the time surgery will develop and infection.30,136 In a study of deadly weapon, open 11 injuries, 10% became infected. Only 21% of the infected eyes of had visual outcomes 5/200 or better. Slightly more than half of the globes were to develop phthisis, were enucleated, or eviscerated.137 Outcomes in this study were better if they were infected with less virulent bacteria. Eyes with retinal breaks or retinal it detachment have worse outcomes.28,134 One study examined eyes in which the culture was positive at the time of initial repair of trauma, and separated the eyes into two groups: those who developed endophthalmitis (44%) and those who did not (56%). There was no statistically significant difference in the visual outcomes between the two groups, and the median vision was less than 20/200 in both. The eyes had better outcomes if they: (a) had been better initial visual acuity, (b) were infected with nonvirulent bacteria, (c) did not have retinal detachment, (d) did not develop endophthalmitis, (e) had shorter initial wound length.52
Management of endogenous endophthalmitis involves early recognition of the clinical picture. Consultation with an infectious disease specialist in the early stages may be very helpful if the patient is not already known to be septic. A search for systemic sepsis should be undertaken in cases of endophthalmitis, because the systemic condition might be life threatening.
Management depends on the severity of the presenting infection and systemic condition. Essentially all patients require systemic therapy. Patients who present with a chorioretinitis with modest vitreous inflammation may be successfully managed with intravenous antimicrobial therapy alone. If there is any suggestion that this may not be sufficient then intravitreal injection and/or pars plana vitrectomy with intraocular antibiotic injection are chosen.
Outcomes are variable depending on the infection organism and the systemic health of the patient. Patients with bacterial endophthalmitis may experience a reasonably good outcome depending on the organism and severity of the initial infection. In two series the final vision was 20/400 or better in approximately half the patients.56,59 On the other hand, a very poor outcome is not unusual. One series reported an outcome of no light perception in 37.5% of cases.58 Others note evisceration or enucleation as the outcome in a number of eyes.58,59 In fungal cases, Candida may present as a mild peripheral infection and respond well to the systemic medication being used for systemic therapy. Aspergillosis, on the other hand, has a tendency to present as a posterior segment infection involving the macula, and often destroys central vision.59 It should be remembered that these cases are often harbingers of a severe systemic condition. Because these patients are systemically ill, the death rate in published series ranges between 7% and 29%.56–59
|OUTCOME BY ORGANISM|
Gram-positive coagulase negative micrococci accounted for 69% of the culture positive infections in the EVS. In older reports, these organisms were frequently grouped together and reported as Staphylococcus epidermidis. In a large series of 90 cases of gram-positive coagulase negative staphylococcal endophthalmitis, nonepidermis staphylococci counted for 28% of the infections. S. warneri, S. hominis, S. haemolyticus, and S. capitis were the most common nonepidermidis isolates.36 In laboratory studies, Staphylococcus epidermis may self-sterilize after injection into the vitreous cavity.138,139 This suggests that some cases of culture negative endophthalmitis may be caused by these organisms, a speculation supported by PCR data.3,4There is frequently a significant delay between the inciting surgical or traumatic event and the onset of symptoms. In two studies by Omerod, approximately one-third of the cases presented after the first week.36,37,140 Only a small minority presented with more florid physical signs such as marked lid swelling, chemosis, or purulent external ocular inflammation.36 Infection with these organisms often can be treated with a good outcome.
After cataract surgery, more than half of all cases infected with these organisms achieve 20/50 vision or better.66,94,141,142 In the EVS, 84% coagulase negative micrococci infected eyes achieved 20/100 or better vision, and 58% achieved 20/40 or better.
Staphylococcus aureus endophthalmitis usually occurs after cataract surgery but can be found after endogenous spread or after penetrating trauma.66,143 Most cases have the onset within the first week after surgery.143 Eyes with Staphylococcus aureus endophthalmitis do not have as good outcome as those with gram-positive, coagulase-negative staphylococci. In the EVS, 50% of eyes achieved 20/100 or better and 36.7% achieved 20/40 or better.63 In laboratory studies Staphylococcus aureus is not always eradicated with a single injection of antimicrobial.114 This persistence of infection also has been noted by authors of clinical series including those of the EVS who found 47% of eyes re-tapped after staphylococcus aureus or Streptococcus infection remained culture positive.113,118,119 They noted that persistence of infection may be a feature of infection with more virulent organisms.113
Streptococcal species produce a particular virulent form of endophthalmitis. The genus Streptococcus is subdivided by several different classifications and comprises over 20 organisms that may cause intraocular infection. Virulence varies from one organism to the next and is often conferred by toxin production.144 Streptococcal species are the predominant organisms in bleb related endophthalmitis, but it may also occur in acute postoperative endophthalmitis.47,49,145 In the EVS they accounted for 7% of post-cataract infections. The onset of the infection after surgery tends to be more rapid than after coagulase-negative Staphylococcus infections and produce worse initial symptoms.2,38 Visual outcomes also tend to be worse than after coagulase-negative Staphylococcus infections. In the EVS, Streptococcus and Staphylococcus aureus infections were grouped together and demonstrated that the prognosis was less favorable.63
A large series, from a teaching hospital of 48 patients, demonstrated that 50% of the cases were produced by Streptococcus viridans. In this series, only 10% of eyes achieved vision equal to or better than 20/50 and only 31% were 20/400 or better.143 A later series of 27 eyes of Streptococcus pneumoniae, from the same institution, noted that the visual results were approximately the same as the larger mixed series. Thirty-seven percent had no light perception.146
Pseudomonas aeruginosa is perhaps the most common gram-negative intraocular infection. Pseudomonas is a relatively uncommon cause of post-cataract extraction endophthalmitis and is perhaps more commonly seen in mixed infections and after trauma. In one laboratory study, infection persisted despite appropriate antimicrobial therapy with a variety of agents after single injection if the infection had been established for 48 hours or longer.116 In the EVS, 70% of the infections with gram-negative organisms that were re-cultured remained culture positive.113 The largest reported series consists of 28 eyes that developed Pseudomonas endophthalmitis associated with various clinical situations. Twenty-five percent required enucleation or evisceration at presentation because of no light perception vision and intractable pain. Sixty-eight percent of eyes had a final visual acuity of no light perception, and only 7% of eyes had a final visual acuity of 5/200 or better.147
Infection with Bacillus species confers an extremely poor prognosis. The majority of cases occur after trauma and often are associated with intraocular foreign bodies.31,148 Most endophthalmitis is caused by Bacillus cereus.148 Infections are so severe that the patient may present both with fever and leukocytosis. Injection of some of the toxins produced by Bacillus produces severe intraocular inflammation.149 In the literature, very few eyes have final vision better than no light perception53,148,150,151 with the exception of one outbreak of epidemic of Bacillus spp. in a series of cataract extraction patients.152
Even though not an acute infection, P. acnes infections can result in visual loss in some patients. The overall recovery rate is approximately the same as in Staphylococcus epidermis infections. Fifty percent of patients achieved vision equal to or better than 20/40, and 78% were at least 20/400 in one recent study.111 In another report the average visual acuity was 20/30 to 20/40 depending on the group examined.42
The outcome of fungal infections depends on the series reported and the types of infection. In a review of endogenous fungal endophthalmitis from the United States, 13 of 17 eyes with Candida infection achieved vision of 20/400 or better. Aspergillus endophthalmitis occurred far less frequently but had a more unfavorable prognosis.120 In a large series of endogenous fungal endophthalmitis from Japan, at least one-third of the eyes saw 20/200 or less after therapy. Vitreous surgery was performed in 26 eyes; six of these ended up totally blind.60 In a series of post-cataract cases from India, Aspergillus spp. were the most common isolates. Only one-third of the eyes achieved visual acuity of 3/60 or better. Initial corneal involvement was a poor prognostic sign.131
|Endophthalmitis may be classified by the clinical stetting and subdivided by time of onset. Classifications are important to understand the prognosis, which is closely related to the expected and identified pathogens in these clinical cases. The EVS established guidelines for acute postoperative infections. In those eyes presenting with light perception vision, a combination of immediate pars plana vitrectomy, intraocular antibiotic injection of vancomycin and amikacin, and systemic corticosteroids provided a better outcome than initial treatment with injection of antibiotics alone for a high percentage of patients. For those eyes with vision of hand motion or better, initial tap and injection proved as efficacious as did the initial strategy of vitrectomy, and patients had significantly better results than those presenting only with light-perception– vision. Retrospective analysis demonstrated that patients with diabetes also had a tendency to have better outcomes, even with vision of hand motion or better. Vitrectomy may be considered as initial management in post-cataract surgery eyes with rapid onset, indicating possible virulent organisms, in traumatic endophthalmitis, in chronic endophthalmitis particularly resulting from suspected P. acnes, and in bleb-related endophthalmitis. Intravenous medication was not efficacious in the EVS for eyes treated as outlined. However, intravenous antimicrobial therapy has been shown to be effective in association with intensive topical steroid therapy in patients with blebitis. By extension, there may be some role for intravenous antimicrobial therapy in eyes with infected filtering blebs. Retinochoroiditis with severe intravitreal inflammatory reaction was also commonly responsive to intravenous antimicrobial therapy alone. The most effective route of corticosteroid administration has not yet been determined in randomized trials.|
1. Endophthalmitis Vitrectomy Study Group: Results of the Endophthalmitis Vitrectomy Study: A randomized trial of immediate vitrectomy and of intravenous antibiotics for the treatment of postoperative bacterial endophthalmitis. Arch Ophthalmol 113:1479, 1995
2. Endophthalmitis Vitrectomy Study Group: Results of the Endophthalmitis Vitrectomy Study. A randomized trial of immediate vitrectomy and of intravenous antibiotics for the treatment of postoperative bacterial endophthalmitis. Endophthalmitis Vitrectomy Study Group. Arch Ophthalmol 113:1479, 1995
10. Bannerman TL, Rhoden DL, McAllister SK, Miller JM, Wilson LA: The source of coagulase-negative staphylococci in the Endophthalmitis Vitrectomy Study. A comparison of eyelid and intraocular isolates using pulsed-field gel electrophoresis. Arch Ophthalmol 115:357, 1997
14. Allen HF, Mangiaracine AB: Bacterial endophthalmitis after cataract extraction. II. Incidence in 36,000 consecutive operations with special reference to preoperative topical antibiotics. Trans Am Acad Ophthalmol Otolaryngol 77:581, 1973
35. Wisniewski SR, Capone A, Kelsey SF, et al: Characteristics after cataract extraction or secondary lens implantation among patients screened for the Endophthalmitis Vitrectomy Study. Ophthalmology 107:1274, 2000
60. Tanaka M, Kobayashi Y, Takebayashi H, Kiyokawa M, Qiu H: Analysis of predisposing clinical and laboratory findings for the development of endogenous fungal endophthalmitis. A retrospective 12-year study of 79 eyes of 46 patients. Retina 21:203, 2001
67. Rowsey JJ, Stonecipher KG, Jensen H, Krueger R, Parmley VC: Culture and sensitivities of infectious endophthalmitis: A microbiological analysis of 302 intravitreal biopsies. Invest Ophthalmol Vis Sci 33(Suppl):745, 1992
70. Oum B, D'Amico DJ, Kwak HW, Wong KW: Intravitreal antibiotic therapy with vancomycin and aminoglycoside: Examination of the retinal toxicity of repetitive injections after vitreous and lens surgery. Graefe's Arch Clin Exp Ophthalmol 230:56, 1992
89. Ferencz JR, Assia EI, Diamantstein L, Rubinstein E: Vancomycin concentration in the vitreous after intravenous and intravitreal administration for postoperative endophthalmitis. Arch Ophthalmol 117:1023, 1999
116. Davey PG, Barza M, Stuart M: Dose response of experimental Pseudomonas endophthalmitis to ciprofloxacin, gentamicin, and imipenem: evidence of resistance to “late” treatment of infections. J Infect Dis 155:518, 1987
137. Sabaci G, Bayer A, Mutlu FM, Karagul S, Yildirim E: Endophthalmitis after deadly-weapon-related open-globe injuries: Risk factors, value of prophylactic antibiotics, and visual outcomes. Am J Ophthalmol 133:62, 2002
139. Smith MA, Sorenson JA, D'Aversa G, et al: Treatment of experimental methicillin-resistant Staphylococcus epidermidis endophthalmitis with intravitreal vancomycin and intravitreal dexamethasone. J Infect Dis 175:462, 1997
144. Jett BD, Jensen HG, Atkuri RV, Gilmore MS: Evaluation of therapeutic measures for treating endophthalmits caused by isogenic toxin-producing and toxin-nonproducing Enterococcus faecalis strains. Invest Ophthalmol Vis Sci 36:9, 1995
149. Beecher DJ, Pulido JS, Barney NP, Wong ACL: Extracellular virulence factors in Bacillus cereus endophthalmitis: Methods and implication of involvement of hemolysin BL. Infect Immunol 63:632, 1995