Subretinal abscess: causative pathogens, clinical features and management

Abstract

Purpose

To review the literature on endogenous subretinal abscess (SRA).

Methods

We searched in the literature for the terms ‘subretinal abscess’, ‘chorio-retinal abscess’ and ‘choroidal abscess’.

Results

A total of 122 patients were identified, of whom 20 patients (22 eyes) had no identified systemic infective foci (group 1) and 102 (120 eyes) had systemic infective foci (group 2). The mean age for group 1 was 44.6 years (range 2 weeks-82 years) and for group 2 was 43.2 years (range 1–89 years). The responsible pathogen was identified in 90% and 95% of cases, respectively. In group 1 the most frequent causative agents were Aspergillus and Nocardia, while in group 2 were Nocardia, Mycobacterium Tuberculosis and Klebsiella. In both groups the most common symptoms were reduced vision (70% and 72.5%, respectively), pain (65% and 29.4%, respectively) and redness (35% and 17.6%, respectively). For group 1 there was no difference between mean initial and final visual acuity (1.7 logMAR, range 0–3 logMAR), while for group 2 mean initial and final visual acuities were 0.8 logMAR and 0.6 logMAR, respectively. Final visual acuity was significantly better in group 2 (p = 0.003). Anterior segment inflammation was seen in 77.3% of cases of group 1 and 66.7% of cases of group 2. In both groups the abscess most common locations were posterior pole (45.4% and 32.5%, respectively) and temporal periphery (13.6% and 13.3%, respectively). Clinical features included hemorrhages (76.5% and 76.3%, respectively) and subretinal fluid (75% in both groups). Diabetes mellitus (20% and 25.5%) and immunosuppressive drug intake (35% and 23.5%) were the main predisposing factors for SRA. Combination of systemic and intravitreal antibiotics/antifungals and vitrectomy was the main therapeutic strategy for both groups. Systemic treatment alone was used mainly for cases of tubercular etiology. The timing of vitrectomy differed between the two groups, as it more commonly followed the use of systemic and intravitreal antibiotics in the forms associated with systemic infective foci. Additional abscess drainage or intralesional antibiotics were performed in 23.8% of cases.

Conclusion

At present no guideline exists for the treatment of subretinal abscess. Systemic broad-spectrum antibiotic treatment is of primary importance and should be used in all cases unless contraindicated. Combination of systemic and local treatment is the most frequently adopted strategy.

Introduction

Subretinal abscess (SRA) is a rare and sight-threatening manifestation of endogenous endophthalmitis, where a pathogen reaches the choroid via the bloodstream from another site of the body and crosses the blood-retinal barrier, invading the retina and potentially the vitreous cavity [1]. In 1983, Wilmarth was the first to describe a case of SRA associated with endogenous endophthalmitis caused by Aspergillus fumigatus in an intravenous (IV) drug user [2]. Subsequently, several cases of SRA [3,4,5], intraretinal abscess [6], septic retinal cyst [7], choroidal or chorio-retinal abscess [8, 9] were reported, but the distinction between these entities remains unclear. Given the rarity, there are no data on the incidence of SRA. In the British Ophthalmological Surveillance Unit (BOSU) study on 62 cases of endogenous endophthalmitis, SRA was the second most common retinal finding (6.5%) after retinitis (31%) and followed by Roth’s spots (4.8%) [10]. Predisposing conditions of SRA are the same of endogenous endophthalmitis [1], including diabetes, immunosuppression, extraocular foci of infection, sepsis [4], IV drug use, blood malignancies and autoimmune diseases. The etiology is most frequently bacterial, the commonest being Nocardia followed by Pseudomonas Aeruginosa, Streptococcus Viridans and Klebsiella Pneumoniae. Aspergillus is the most frequent fungal agent [3]. Rarely the etiology is protozoan [11] or mixed [12]. The visual prognosis of SRA is often very poor due to the aggressive course despite treatment, with the most severe cases complicated by subretinal pseudo-hypopyon [11] and exudative or rhegmatogenous retinal detachment, requiring enucleation or evisceration if not timely treated.

We reviewed all cases of SRA with and without systemic infective foci published between 1967 and 2021, focusing on epidemiology, causative pathogen and method of identification, symptoms and signs at presentation, systemic predisposing factors and treatment approach.

Methods

We identified published studies from Pubmed (National Library of Medicine), EMBASE (Embase.com) and Scopus (Elsevier) from inception to January 2021. The publication period ranges between September 1967 and January 2021.

There were no language restrictions. To ensure appropriate study inclusion, the search terms were ‘subretinal abscess’, ‘chorio-retinal abscess’ and ‘choroidal abscess’. The titles and abstracts were screened and full-texts were obtained for inclusion and data collection. A total of 105 articles were preliminarily enrolled and after a full-text review a total of 96 articles were chosen for inclusion, with 122 patients in total. For the rarity of the condition all chosen articles were case reports (84) and case series (12), and were considered of sufficient quality for inclusion if documented: 1) presenting clinical features, responsible pathogen and method of identification; 2) treatment strategies; 3) final outcomes. Articles lacking detailed information and cases of exogenous SRA developing after ocular surgery or ocular trauma were excluded. For the purpose of our study SRA cases without identified systemic infective foci (group 1) and with systemic infective foci (group 2) were analyzed separately. Snellen visual acuity (VA) was converted to logarithm of minimal angle of resolution (logMAR) and was analyzed as a continuous variable. Continuous variables (e.g. VA) were compared using an unpaired t-test. A loss of 0.3 logMAR of VA or more from baseline was considered a worsening, while a gain of 0.3 logMAR or more was considered an improvement.

Results

Group 1: SRA without identified systemic infective foci

Demographics

Our literature search identified 20 patients (22 eyes) with a mean age at presentation of 44.6 years (range 2 weeks-82 years). There were 11 males and 9 females. The majority of reports were from the United States (10 patients) and India (5 patients). Mean follow-up duration was 9.8 (median 6.5 , range 0.3–48 ) months.

General health was reported as unremarkable in 5 patients (25%), whereas 7 patients (35%) were on systemic steroids and/or immunosuppressive drugs, 4 (20%) were type 2 diabetic, 2 (10%) had a history of IV drug use and 2 (10%) were human immunodeficiency virus (HIV) positive. In all patients an active systemic infective process was not identified, but one suffered with gastro-enteritis 2 weeks before the onset of ocular symptoms, one had an urinary tract infection a month before and one had fevers of unknown origin during a recent exotic travel.

Clinical features at presentation

At presentation SRA was isolated, without vitreous involvement, in 4 patients (4 eyes) and associated with endophthalmitis in 16 patients (18 eyes). There was no difference in laterality (9 right eyes, 9 left eyes, 2 bilateral). The three most frequent symptoms were reduced vision (14 patients, 70%), pain (13 patients, 65%) and redness (7 patients, 35%). Mean symptom duration, specified for 17 patients, was 18.1 (median 7, range 1–180) days.

VA at presentation, available for 21 eyes, was 20/40 or better in 7/22 eyes (31.8%), and 14 eyes (63.6%) had a VA of 20/200 or worse. In one eye (4.6%) of a newborn baby the initial VA was not specified.

No difference was observed between initial and final VA, as the mean VA at presentation was 1.7 logMAR (median 1.9 logMAR, range: 0–3 logMAR) and the mean final VA, available for 17 eyes, was 1.7 logMAR (median 1.9 logMAR, range: 0–3 logMAR). Four eyes were enucleated.

Anterior segment involvement was observed in 17 eyes (77.3%), and signs consisted in cellularity in the anterior chamber (15 eyes), conjunctival injection (10 eyes) and hypopyon (6 eyes). In one case the anterior segment description was missing.

The most frequent location of SRA was the posterior pole (10 eyes, 45.5%), followed by the temporal periphery (3 eyes, 13.6%). Where described, hemorrhages were associated with SRA in 13 out of 17 eyes (76.5%), and subretinal fluid in 9 out of 12 eyes (75%).

Causative pathogens and methods of identification

Causative agents were identified in 18/20 patients (90%), and the most frequent were Aspergillus (Fumigatus, Nidulans, Terreus and Flavus species) and Nocardia, detected in 4 and 3 patients, respectively. Given the absence of a systemic infective focus, pathogen identification was from ocular samples, the commonest being SRA (8 eyes from 8 patients), followed by vitreous (7 eyes from 7 patients). In the 3 patients (3 eyes) with isolated SRA where the vitreous analysis was performed it did not allow the pathogen identification, while of the 16 patients (18 eyes) with SRA associated with vitritis the pathogen was isolated from SRA drainage or biopsy in 7 eyes (7 patients) and from vitreous in 7 eyes (7 patients). In the 8 eyes (8 patients) where the vitreous culture failed in yielding a growth the pathogen was identified by direct drainage of the SRA.

Clinical features, pathogen and method of identification of SRA without systemic foci are summarized in Table 1.

Table 1 Summary of study details, clinical features, pathogen and method of identification of SRA without systemic foci

Therapeutic approaches

The most common therapeutic strategy was the combination of systemic and intravitreal antibiotics or antifungals followed by vitrectomy (13 patients, 13 eyes; 65%). In the majority of these patients (9 patients, 9 eyes, 69.2%) vitrectomy was a second line treatment performed for progression despite systemic and intravitreal antibiotics/antifungals, and in 4 patients (4 eyes, 30.8%) was first line strategy with the dual purpose of diagnosis and treatment. Additional SRA drainage was performed in 5 patients (5 eyes) and intralesional antibiotics or antifungals were administered in 3 patients (3 eyes). In one case (one eye) enucleation was performed due to failure of these treatments.

Vitrectomy and SRA drainage without systemic treatment was adopted only in one eye of a case caused by Aspergillus where intravitreal agents were not deemed necessary for minimal vitreous involvement [13].

Systemic treatment and vitrectomy without intravitreal antibiotics was performed in 2 cases (2 eyes).

Of the 17 patients undergoing vitrectomy, in 8 it was first line treatment while in the remainder 9 it was performed as second line treatment.

Systemic treatment was the only treatment strategy in one eye of a case of Tubercular etiology [14].

Enucleation was the first line treatment for 2 paediatric cases (2 eyes) where retinoblastoma could not be clinically excluded [15] and for one case (one eye) refusing other treatments where the SRA progressed to pan-ophthalmitis [16]. Systemic and intravitreal agents were the strategy adopted for the eye of a patient with bilateral involvement by Candida where the fellow eye underwent vitrectomy [17] and in one case of bilateral SRA caused by Nocardia and Mycobacterium TB where the fellow eye underwent vitrectomy [18].

Steroids were used in four cases of bacterial etiology: one receiving intravitreal dexamethasone at the end of vitrectomy [19] and three receiving oral steroids, of whom one was also on systemic anti-tubercular treatment. Intravitreal dexamethasone was given in a diabetic patient with SRA caused by S. aureus where the final VA improved.

Responses to treatment, secondary complications and their treatment

Of the 17 eyes where the final VA was available, in 3 eyes (17.6%) VA was 20/40 or better, in 2 eyes (11.8%) was between 20/40 and 20/200, and in 12 eyes (70.6%) was 20/200 or worse.

Comparing initial and final VA, in 6 eyes (35.3%) VA improved, in 8 eyes (47.1%) remained stable and in 3 eyes (17.6%) worsened. VA was not specified for one eye and 4 eyes were enucleated.

In 8/13 cases multiple surgeries were necessary either because the lesion expanded despite previous vitrectomy or for development of complications such as retinal detachment or retinal traction. One case caused by Aspergillus Fumigatus, progressed despite the initial treatment with systemic and intravitreal amphotericin, vitrectomy and lensectomy, rendering enucleation necessary [2]. In 2 paediatric cases enucleation was performed as retinoblastoma could not be ruled out.

Baseline and final VA, treatment interventions and final outcomes of SRA without systemic foci are summarized in Table 2.

Table 2 Summary of initial and final visual acuity, treatment interventions and final outcomes of SRA without systemic foci

Group 2: SRA with identified systemic infective foci

Demographics

One hundred and two patients (120 eyes) were identified. There were 69 males and 33 females, with a mean age of 43.2 (median 44; range 1–89) years specified for 101 patients. The mean duration of follow-up, available for 96 patients, was 8.3 (median 6, range 0.2–48) months.

Twenty-three (22.5%) patients were healthy, 26 patients (25.5%) had diabetes mellitus, 24 (23.5%) were on immunosuppressive medications or oral steroids, and 3 (2.9%) were HIV-positive. The most common infective focus was the respiratory system (25 patients) followed by disseminated infection (16 patients) and soft tissue infection (13 patients).

Clinical features at presentation

SRA was isolated in 13 patients (14 eyes), associated with endophthalmitis in 84 patients (100 eyes), and not better described in 5 patients (6 eyes). There were 30 right eyes, 40 left eyes, 18 bilateral and 14 eyes unilateral with unspecified laterality.

The three most frequent symptoms were reduced vision (74 patients, 72.5%), sudden in 8 patients and gradual in 66 patients, pain (30 patients, 29.4%) and redness (18 patients, 17.6%), followed by floaters (9 patients, 8.8%) and photophobia (5 patients, 4.9%).

Mean symptom duration, specified for 67 patients, was 16.3 days (median 3 days, range 0–365 days).

The mean visual acuity at presentation, available for 105 eyes, was 1.0 (median 1, range: 0–3) logMAR, and the mean final visual acuity, available for 99 eyes, was 0.6 (median 0, range: 0–3 ) logMAR.

VA at presentation, available for 105 eyes, was 20/40 or better in 15 eyes (14.3%), between 20/40 and 20/200 in 25 eyes (23.8%), and 20/200 or worse in 65 eyes (61.9%).

Anterior segment involvement was observed in 76 eyes from 68 patients (66.7%), with the most frequent signs being anterior chamber cells (63 eyes), conjunctival hyperemia/chemosis (33 eyes) and flare (23 eyes).

The most frequently observed location of SRA was the posterior pole (39 eyes, 32.5%) followed by the temporal periphery (16 eyes, 13.3%). Hemorrhages were present in 58/76 eyes (76.3%), and subretinal fluid in 33/44 eyes (75%).

Causative pathogens and methods of identification

The causative pathogen was identified in 97 patients (95%) and not identified in 5 patients.

Nocardia was the most frequent pathogen - detected in 24 patients (23.5%) - followed by Mycobacterium Tuberculosis (18 patients, 17.6%), Klebsiella (18 patients, 17.6%) and S. aureus (14 patients, 13.7%). Mycobacterium Tuberculosis and Nocardia spp. were the two main responsible agents of respiratory infections (14 and 5 patients, respectively) and systemic infections (3 and 14 patients, respectively).

Pathogens were identified from ocular tissues in 42 patients (43.3%) and from extra-ocular tissues in 49 patients (50.5%), of whom 27 from blood cultures. In six patients Mycobacterium Tuberculosis was not isolated but tubercular disease was diagnosed based on the typical pulmonary findings on chest X-ray or CT and a positive Mantoux test and/or QuantiFERON-TB gold.

The vitreous was the main ocular source for pathogen identification (20 patients, 47.6%, 20 eyes), followed by SRA (14 patients, 33.3%, 14 eyes) and aqueous (7 patients, 16.7%, 7 eyes).

In all cases of isolated SRA (13 patients, 14 eyes) the pathogen was identified, but ocular sampling was performed only in 7/13 (53.8%), with a yielding rate for vitreous of 20% (1/5 patients) and for SRA of 100% (3/3 patients).

In cases of SRA and endophthalmitis (84 patients, 100 eyes), vitreous sampling was performed in 50 cases and the pathogen yielding rate for vitreous was 38% (19 cases), while SRA biopsy allowed pathogen identification in 20% of cases (10 cases).

Nocardia was by far the most common pathogen in immunosuppressed patients (18/24 patients), Klebsiella and S. aureus were the main causative agents in diabetic patients, detected in 9/26 cases and 7/26 cases, respectively; Mycobacterium Tuberculosis was the most common pathogen in healthy subjects (12/23 patients).

Clinical features, pathogen and method of identification of SRA with systemic foci are summarized in Table 3.

Table 3 Summary of study details, clinical features, pathogen and method of identification of SRA with systemic foci

Therapeutic approaches

The combination of systemic and intravitreal antibiotics/antifungals with vitrectomy was the most common therapeutic approach, performed in 28 patients (27.5%), with additional drainage of SRA and/or intralesional antibiotics or antifungals performed in most of cases (21 patients and 1 patient, respectively).

Vitrectomy was the first line strategy in 6 patients, all caused by Klebsiella or Nocardia, while for the remainder (33 patients) it was performed as second line treatment for failure of other treatment modalities. In 10 cases multiple surgeries were necessary to address SRA progression or complications (retinal detachment or tractional complications).

Combined systemic and intravitreal antibiotics were the second most common treatment strategy (20 patients). In a limited number of cases (3 cases) only vitrectomy and intravitreal antibiotics were adopted. Intravitreal dexamethasone was used in 9 eyes, of whom 7 eyes had SRA caused by Klebsiella, one eye caused by Mycobacterium TB and in one eye the pathogen was not isolated. Of these, in 6 eyes the VA improved, in one eye the VA remained the same, in one eye VA not specified and in one eye evisceration was necessary.

Systemic treatment without other treatment modalities was used in 18 patients, of whom 6 had TB, 5 had S. aureus-related and 4 had Nocardia-related infections.

Combined systemic and topical antibiotics or steroids were adopted in 18 cases, the majority of whom (13/18) had TB.

In one SRA case caused by Klebsiella treatment consisted only in intravitreal antibiotics, and one case caused by Nocardia received systemic treatment and intravitreal anti-VEGF.

In 13 cases enucleation (7 cases), evisceration (4) or exenteration (2 cases) were performed due to failure of other treatments.

Responses to treatment, secondary complications and their treatment

Of the 99 eyes where the final VA was available, in 29 eyes (29.3%) was 20/40 or better, in 30 eyes (30.3%) was between 20/40 and 20/200, and in 40 eyes (40.4%) was 20/200 or worse. Comparing initial and final VA of 89 eyes, in 49 eyes (55.1%) VA improved, in 23 eyes (25.8%) remained stable and in 17 eyes (19.1%) worsened. Final VA was not specified in 9 eyes. Six eyes were enucleated, 5 were eviscerated and 1 was exenterated.

In 10 cases multiple surgeries were necessary either because the lesion expanded despite the previous surgery or for development of complications such as retinal detachment or retinal traction. Twelve cases were non responsive to treatment and therefore enucleation, evisceration or exenteration were necessary, and the most common pathogens associated with these were Nocardia (one evisceration and two enucleations) and Pseudomonas (one evisceration and two enucleations).

Baseline and final VA, treatment interventions and final outcomes of SRA with systemic foci are summarized in Table 4.

Table 4 Summary of initial and final visual acuity, treatment interventions and final outcomes of SRA with systemic foci

Discussion

Our review of the literature showed that Nocardia was the most frequent causative pathogen of SRA associated with systemic infective foci, while for SRA in absence of systemic foci Aspergillus was seen with a higher frequency. In absolute numbers Nocardia was the most frequent causative agent.

In SRA without systemic infective foci the pathogen was more commonly isolated from SRA if there was no vitreous involvement, while for forms with vitreous involvement a similar yielding rate from vitreous and from SRA was observed. By contrast, in presence of a systemic infective focus the pathogen was isolated mainly from extra-ocular sites, and when ocular sampling was performed in cases of SRA with no vitreous involvement the observed yielding rate for vitreous was 20% and for SRA was 100%. In the forms with vitritis where ocular sampling was done the vitreous was the commonest ocular site of pathogen identification. However, failure of vitreous sampling in isolating the pathogen has been described by many authors, who were subsequently able to isolate it by direct drainage of the lesion [11, 19]. Despite vitritis being observed with a similar frequency in both groups (81.8% versus 83%) the vitreous yielding rate was higher in the forms without systemic infective foci (43.8% versus 35.7%).

The most common systemic predisposing conditions of SRA were immunosuppression and diabetes mellitus, the former being more frequent in SRA without systemic foci (35% of patients) and the latter being more frequent in SRA associated with infective foci (25.5% of patients). Isolated SRA was more frequently observed in cases without systemic infection (18.2% of eyes versus 10.8% of eyes of group 2). A higher frequency of bilateral involvement was observed in the forms with systemic foci, where it was detected in 17.6% of patients (30% of eyes) compared to 10% of patients (18.2% of eyes) without systemic foci.

Reduced vision was observed with a similar frequency in both groups (70% and 72.5%). Baseline visual acuity did not show a significant difference between the groups, but final visual acuity was better in the group associated with systemic foci (p = 0.003).

Pseudomonas, Nocardia and Aspergillus were the microorganisms related to a worse prognosis requiring enucleation or evisceration. Some of the cases with poor prognosis may be related to delay in diagnosis and management or to systemic factors as immunosuppressive medication intake that may have an impact on the natural history of the disease.

No standard approach exists for the management of SRA because, unlike endophthalmitis, no guidelines are available at present and there is no consensus on the various proposed therapeutic approaches. Systemic and intravitreal antibiotics/antifungals and vitrectomy are the mainstay treatment in the majority of cases, but there is no consensus on the timing of vitrectomy, which in fact differed between the two groups: in group 1 vitrectomy was performed with similar frequency as first or second line treatment, while for group 2 it was mostly performed when previous non-surgical treatments failed. While for endogenous endophthalmitis the standard of care includes vitreous biopsy and intravitreal antibiotics combined with systemic antibiotics and oral steroids (once fungal infection has been ruled out), for SRA there is no universal approach. In case of no vitreous involvement a prompt systemic treatment can achieve an excellent prognosis, but a close follow-up is essential to identify the potential progression of SRA into the vitreous cavity, needing immediate revision of the therapeutic strategy. The effectiveness of intravitreal antibiotics, including vancomycin for Gram positive and ceftazidime for Gram negative bacteria, is controversial as they may not fully penetrate into the subretinal space. Surgical treatment of SRA, including pars plana vitrectomy and abscess drainage, is considered a second-line therapy when conservative treatments are not effective. However, for very aggressive pathogens some authors advocate an early surgical intervention. Some authors adopted the technique of intralesional antibiotics, namely injection of antibiotics into the subretinal space through a small retinotomy. Compared to abscess drainage, the technique has the advantage of carrying a lower risk of retinal detachment [20, 21], but in isolated SRA it could favor the invasion of the vitreous cavity by the pathogen. Tsai and Peng suggested that if the SRA is smaller than four disc areas, pars plana vitrectomy with intravitreal antibiotic injection could be successful, whereas, in larger lesions, vitrectomy with retinectomy to remove the abscess should be considered [22]. Internal drainage of the SRA leads to resolution of the infection, but carries the risk of postoperative retinal detachment due to proliferative vitreoretinopathy and therefore should be considered in cases that fail to respond to conventional therapy. Eschle-Meniconi et al. suggested fluorescein angiography as a guide for the management strategy as it helps understanding which layer is affected by the infection and identify a potential early invasion of the vitreous: if at presentation a late leakage of the lesion is observed, it means that the retinal pigmented epithelium is disrupted and the vitreous is affected and then vitrectomy and subretinal biopsy should be performed in the first instance. If no late leakage is seen, the infection is at an initial stage and limited to the subretinal space and a trans-vitreal fine needle biopsy or vitreous tap, intravitreal antibiotics and systemic treatment are the preferred options to start with. If the patient is already under treatment for an infection site elsewhere, then a late leakage on angiography would indicate the need for intravitreal antibiotics in case of small and peripheral SRA or for vitrectomy for larger ones [23]. However, vitreous involvement in practice can be identified by clinical assessment and multimodal imaging with no need for invasive tests.

Conclusion

Although SRA can develop even in the absence of clinically detectable systemic infectious foci it is of primary importance to perform a prompt physical examination and systemic investigations in order to identify or rule out a source of infection elsewhere or masquerading conditions. Our review showed that no universal approach exists for SRA. Systemic broad-spectrum antibiotics are of primary importance and should be used in all cases of SRA, even in the absence of vitreous involvement and of identifiable infective foci, given the high risk of an undiagnosed underlying systemic infection.