Abstract
AIM
To identify the clinical features and treatment outcomes of endogenous Klebsiella pneumoniae endophthalmitis and investigate prognostic factors of poor visual outcome.
METHODS
The clinical records of all patients diagnosed with endogenous Klebsiella endophthalmitis between January 2007 to December 2018 in Prince of Wales Hospital, Hong Kong, China were retrospectively reviewed. Thorough ophthalmological examination findings were recorded in the case note, including visual acuity testing, slit-lamp examination, indirect ophthalmoscopy and B-scan ultrasonography if media opacity precluded fundus viewing.
RESULTS
A total of 18 eyes in 14 patients were identified. Bilateral involvement was noted in 4 patients (28.6%). Hepatobiliary sepsis was the source in 9 patients (64.3%). Culture of intraocular fluid was positive in 5 out of 18 eyes (27.8%). Mortality was noted in 2 patients (14.3%). Mean final visual acuity was 20/1500. Six out of 16 eyes had total loss of sight (37.5%) and 3 eyes required evisceration (18.8%). Multivariate linear regression revealed poor presenting visual acuity (P=0.031) and lack of fundus view due to vitritis (P=0.02) as prognostic factors of poor visual outcome.
CONCLUSION
Visual outcome of endogenous Klebsiella endophthalmitis is poor. Poor presenting visual acuity and lack of fundus view predict poor visual outcome. High index of suspicion for endophthalmitis is important in Klebsiella sepsis patients with complaints of ocular symptoms. Ophthalmological screening is recommended in non-communicable patients with Klebsiella sepsis.
Keywords: Klebsiella pneumoniae, endogenous endophthalmitis, screening, liver abscess, sepsis
INTRODUCTION
Endogenous endophthalmitis is a severe form of intraocular inflammation, results from haematogenous spread of pathogens from a primary site of infection into the normally sterile eye. The condition is sight-threatening with often poor visual outcome[1].
Endogenous Klebsiella pneumoniae endophthalmitis is a severe complication of systemic Klebsiella sepsis. Despite being a rarity, it is recognised as most prevalent in East Asia, where abundance of Klebsiella pyogenic liver abscess and invasive Klebsiella infection has been observed[2]–[3]. Even with increasing reports and awareness, together with improving intervention for endogenous Klebsiella endophthalmitis, visual outcomes reported by recent studies remain poor with high rates of total loss of vision and evisceration[4]–[8]. Ophthalmological screening has been advocated but benefits are uncertain.
This study aimed at identifying the clinical features and treatment outcomes of endogenous Klebsiella endophthalmitis. Prognostic factors associated with poor visual outcome were analysed. By further understanding the features of the disease entity, we sought to improve recognition of the condition among clinicians managing Klebsiella sepsis, and make recommendations on ophthalmological screening.
SUBJECTS AND METHODS
Ethical Approval
An Institutional Review Board approval was obtained from the Joint Chinese University of Hong Kong-New Territories East Cluster Clinical Research Ethics Committee (CREC Ref. No.2019.133). This study was adherent to the tenets of the Declaration of Helsinki. Informed consent was waived due to the retrospective nature of the study.
A retrospective review of all patients diagnosed with endogenous Klebsiella pneumoniae endophthalmitis between January 2007 to December 2018 by ophthalmologists in Prince of Wales Hospital, Hong Kong was performed.
Cases of endogenous Klebsiella pneumoniae endophthalmitis were retrieved from the Clinical Data Analysis and Reporting System (CDARS) of the Hospital Authority of Hong Kong with the diagnosis code of “endophthalmitis”. Simultaneous review of our vitreoretinal clinic and inpatient consultation record was also performed to identify cases of endogenous Klebsiella endophthalmitis. A case was defined as an intra-ocular inflammation in the background of systemic sepsis caused by Klebsiella pneumoniae, where positive culture was obtained from blood or the primary source of infection. Alternatively, when there was no obvious primary source, a case was defined as intraocular inflammation with positive culture of Klebsiella pneumoniae from intraocular fluid (vitreous, aqueous or intraocular content of evisceration). Cases with exogenous cause of endophthalmitis, including recent corneal ulcer, ocular trauma or intraocular surgery in the past 6mo, and history of glaucoma filtration surgery were excluded. Cases with history of non-infective intermediate or posterior uveitis, and infectious uveitis caused by organisms other than Klebsiella pneumoniae were also excluded.
Clinical records of all retrieved cases were reviewed. Thorough ophthalmological examination findings were recorded in the case note, including visual acuity testing (Snellen chart or hand-held Snellen chart), slit-lamp examination, indirect ophthalmoscopy and B-scan ultrasonography if media opacity precluded fundus viewing. Lack of fundus view was defined as a vitreous haze of National Institutes of Health (NIH) grade 3+ or above[9]. Intraocular fluid sample (vitreous or aqueous) was taken, while systemic antibiotics and intravitreal ceftazidime (2 mg/0.1 mL) with or without intravitreal vancomycin (1 mg/0.1 mL) were given upon suspicion of endogenous endophthalmitis. Decision to repeat intravitreal antibiotics and perform pars plana vitrectomy or evisceration was made at the discretion of the treating ophthalmologist with regard to patients' ocular and systemic condition. Data variables including demographics, clinical course, medical history, laboratory results and treatment were collected and recorded.
Statistical Analysis
Statistical analysis was performed using IBM SPSS Statistics version 25.0 (Armonk, NY: IBM Corp., USA). Descriptive statistics were expressed as frequency and mean, with standard deviation. Multivariate linear regression was performed as appropriate. Regression was performed for all individual eyes, as well as single eye from bilateral cases (with right eye selected for bilateral cases by convention). Occasional missing values were replaced with mean during regression. P<0.05 was taken as significant. For statistical analysis, Snellen visual acuities were converted to the logarithm of the minimal angle of resolution (logMAR). Categorical acuities of low non-numerical vision were converted to logMAR scale based on previous works[10]–[12]. The following values were used: finger counting (CF)=1.7, hand movement (HM)=2.0, light perception (LP)=2.3 and no light perception (NLP)=3.0. Visual acuity post-evisceration was counted as no light perception during calculation.
RESULTS
Totally 18 eyes in 14 patients were identified as cases of endogenous Klebsiella pneumoniae endophthalmitis in the study period. Clinical summary of individual patients is presented in Table 1. Demographics, clinical features and outcomes are summarised in Table 2.
Table 1. Clinical summary of patients with endogenous Klebsiella endophthalmitis.
| Patient No. | Age/gender | Eye(s) involved | Presenting ocular symptoms | Ocular signs | Presenting VA | Medical comorbidities | Systemic source of infection | Time from systemic symptom onset to EKE symptoms | Time from ocular symptom to EKE diagnosis | Blood culture | Culture of intraocular fluid | Ocular treatment | Final VA/outcome | Remarks |
| 1 | 60/M | RE | BOV, redness, pain | Injection, vitritis, lack of fundus view, EOM limitation, proptosis | LP | GPH | Liver abscess | 5 | 0 | - | + | IVI×1; PPV×1 | Eviscerated | Eviscerated eye had NLP VA before operation |
| LE | BOV, redness, pain | Injection, vitritis, lack of fundus view | 10/200 | 5 | 0 | + | IVI×1; PPV×1 | NLP | ||||||
| 2 | 47/M | RE | BOV | Injection, AC fibrin, vitritis, lack of fundus view, EOM limitation | NLP | DM | Perianal abscess, liver abscess | 1 | 17 | + | - | IVI×1 | Eviscerated | Eviscerated eye had NLP VA before operation |
| LE | BOV | Injection, AC fibrin, vitritis, lack of fundus view, EOM limitation | LP | 1 | 17 | - | IVI×1; PPV×2 | NLP | ||||||
| 3 | 58/M | RE | BOV | Injection, hypopyon, vitritis, lack of fundus view | HM | CA lung | Pneumonia | 2 | 2 | - | + | IVI×1; PPV×1 | NLP | |
| 4 | 43/M | LE | BOV, floaters | Injection, subretinal abscess | 20/20 | DM | Liver abscess | 7 | 1 | + | - | IVI×1; PPV×1 | 20/200 | |
| 5 | 55/M | LE | Asymptomatic | Injection, vitritis, subretinal abscess, exudative RD | N/A | DM | Lung abscess, liver abscess | N/A | N/A | + | - | IVI×5; PPV×1 | 20/200 | Asymptomatic consult, ICU patient; Non-communicable at presentation |
| 6 | 48/M | LE | BOV, redness, floaters | Injection, hypopyon, vitritis, lack of fundus view | 20/100 | HT HL | Liver abscess | -4 | 4 | - | - | IVI×1 | LP | Ocular symptoms preceded systemic symptom of infection |
| 7 | 67/M | LE | BOV, pain | Injection, hypopyon, vitritis, lack of fundus view | HM | HT, DM, CA rectum | No source identified | N/A | 1 | - | + | IVI×1; PPV×2 | LP | No other site of sepsis identified; Rectal cancer discovered by PET-CT |
| 8 | 76/M | LE | BOV | Injection, hypopyon, vitritis, lack of fundus view | 20/400 | HT DM CVA | Liver abscess | 2 | 2 | + | - | IVI×1; PPV×2 | HM | |
| 9 | 42/M | RE | BOV | Vitritis | 20/100 | DM | Lung abscess | 6 | 3 | + | - | IVI×1 | 20/16 | LE index examination normal; Found LE vitritis on the next day; No injection noted at first examination |
| LE | BOV, floaters | Vitritis | 20/60 | 4 | 4 | - | IVI×1 | 20/13 | ||||||
| 10 | 43/F | RE | Asymptomatic | Subretinal abscess | N/A | DM | Lung abscess, liver abscess | N/A | N/A | + | - | IVI×1 | Death | Asymptomatic consult, ICU patient. Non-communicable at presentation; No injection noted at first few examinations |
| 11 | 74/F | RE | BOV | Injection, hypopyon, vitritis, lack of fundus view | CF | HT | Liver abscess, brain abscess | 7 | 1 | - | - | IVI×1 | Eviscerated | Eviscerated eye had NLP VA before operation |
| 12 | 65/F | RE | BOV, redness | Injection, vitritis, subretinal abscess | 5/200 | GPH | Liver abscess | 4 | 2 | + | - | IVI×2; PPV×1 | 20/200 | |
| LE | BOV, redness | Injection, hypopyon, vitritis, lack of fundus view | LP | 4 | 2 | - | IVI×3 | HM | ||||||
| 13 | 77/F | RE | Redness, periocular swelling | Injection, AC fibrin, vitritis, lack of fundus view | N/A | DM, HT, CVA, recurrent IO | No source identified | N/A | 1 | - | + | IVI×1 | Death | Non-communicable institutionalised patient presented with eye redness; Resistant to amoxicillin/clavulanate. |
| 14 | 60/F | LE | Redness | Injection, subretinal abscess | 20/60 | ALL with neutropenic fever | Persistent bacteremia | 42 | 4 | + | - | IVI×4 | 20/80 | Multidrug resistant Klebisella with ESBL; Normal ocular exam screened by ophthalmologist 6wk prior to presentation |
AC: Anterior chamber; ALL: Acute lymphoblastic leukemia; BOV: Blurring of vision; CA: Carcinoma; CVA: Cerebrovascular accident; DM: Diabetes mellitus; EKE: Endogenous Klebsiella endophthalmitis; EOM: Extra-ocular movement; ESBL: Extended-spectrum beta-lactamases; GPH: Good past health; HL: Hyperlipidemia; HT: Hypertension; ICU: Intensive care unit; IO: Intestinal obstruction; IVI: Intravitreal injection; LE: Left eye; PET-CT: Positron emission tomography-computed tomography; PPV: Pars plana vitrectomy; RD: Retinal detachment; RE: Right eye; VA: Visual acuity.
Table 2. Demographics, clinical features and outcomes of patients with endogenous Klebsiella endophthalmitis.
| Demographics | Data |
| No. of patients (eyes) | 14 (18) |
| Unilateral, n (%) | 10 (71.4) |
| Bilateral, n (%) | 4 (28.6) |
| Age, y (mean±SD) | 58.2±12.4 |
| Gender, n (%) | |
| Male | 9 (64.3) |
| Female | 5 (35.7) |
| Ocular features, n (%) | |
| Symptomatic eyes (n=18) | 16 (88.9) |
| Presenting symptoms (n=16) | |
| Blurring of vision | 14 (87.5) |
| Redness | 7 (43.8) |
| Floaters | 3 (18.8) |
| Ocular signs, n (%) | |
| Injection | 15 (83.3) |
| Hypopyon | 6 (33.3) |
| No fundus view from vitritis | 11 (61.1) |
| Positive intraocular fluid culture, n (%) | 5 (27.8) |
| Time from systemic symptom to ocular symptom, d (mean±SD) | 6.1±10.7 |
| Time from ocular symptom to EKE diagnosis, d (mean±SD) | 3.8±5.3 |
| Systemic features, n (%) | |
| Diabetes mellitus | 8 (57.1) |
| Background malignancy | 3 (21.4) |
| Bacteremia | 7 (50.0) |
| Systemic non-ocular site of sepsis, n (%) | |
| Hepatobiliary system | 9 (64.3) |
| Respiratory system | 4 (28.6) |
| Others (brain abcess, perianal abscess) | 2 (14.3) |
| No clear source identified | 2 (14.3) |
| Clinical outcomes | |
| Final VA | |
| logMAR, mean±SD (Snellen equivalent) | 1.87±1.16 (20/1500) |
| 20/200 or worse, n (%) | 13/16 (81.3) |
| Hand movement or worse, n (%) | 10/16 (62.5) |
| No light perception or evisceration, n (%) | 6/16 (37.5) |
| Evisceration, n (%) | 3/16 (18.8) |
| Mortality, n (%) | 2 (14.3) |
EKE: Endogenous Klebsiella endophthalmitis; SD: Standard deviation; VA: Visual acuity.
Demographics
The mean age of the patients was 58.2±12.4y. Male predominance was noted (n=9, 64.3%). All patients were Chinese. The commonest underlying medical condition was diabetes mellitus (n=8, 57.1%). Background malignancy was noted in 3 patients (21.4%).
Ocular Symptoms and Signs
The majority of cases were unilateral (n=10, 71.4%). The 16 out of 18 eyes (88.9%) were symptomatic before the time of diagnosis. Common presenting symptoms included blurring of vision (87.5%), eye redness (43.8%) and floaters (18.8%). Mean time from systemic symptom onset to ocular symptom onset was 6.1±10.7d, while mean time from ocular symptom onset to clinical diagnosis of endogenous Klebsiella endophthalmitis was 3.8±5.3d. Mean presenting visual acuity was poor at logMAR 1.48±0.86 (Snellen equivalent 20/600). Conjunctival injection was noted in 83.3% of eyes. Hypopyon was seen in 33.3%. No fundus view could be obtained from 61.1% of eyes due to severe vitritis. Subretinal abscesses were seen in 5 out of the 7 eyes with visualisable fundi.
Systemic Sources and Microbiological Profile
Hepatobiliary system was the commonest source of infection (64.3%), followed by respiratory system (28.6%). Other non-ocular sites of infection accounted for 14.3% of patients. No clear source could be identified in 2 cases (14.3%). Blood culture was positive in 7 patients (50.0%). Positive intraocular fluid (aqueous or vitreous) culture was obtained in 5 eyes (27.8%). The majority of patients (85.7%) had infection caused by non-resistant strain of Klebsiella pneumoniae, which was sensitive to amoxicillin/clavulanate, cefuroxime, ciprofloxacin and gentamicin. Two patients (14.3%) had infection caused by resistant strains, one of them was resistant to amoxicillin/clavulanate and the other one was extended spectrum beta-lactamases producing.
Treatment and Clinical Outcomes
All patients received intravenous broad-spectrum antibiotics before or at the diagnosis of endophthalmitis. Intraocular fluid was sampled (vitreous or aqueous if dry vitreous tap) and intravitreal ceftazidime with or without vancomycin was injected in all affected eyes at clinical diagnosis of endophthalmitis. Four eyes (22.2%) received more than one intravitreal injections and 9 eyes (50%) underwent pars plana vitrectomy as indicated. Mortality occurred in 2 patients (14.3%). Mean final visual acuity was logMAR 1.87±1.16 (Snellen equivalent 20/1500). Visual outcome was unfavourable with 13 eyes (81.3%) having vision of 20/200 or worse. Six out of 16 eyes (37.5%) had complete loss of vision with no light perception and 3 eyes (18.8%) required evisceration. All eviscerated eye had visual acuity of no light perception before the procedure.
Statistical Analyses and Prognostic Factors
Poor prognostic factors of final visual acuity are summarised in Table 3. Multivariate linear regression showed that unilateral involvement (P=0.035), poor presenting visual acuity (P=0.031) and lack of fundus view (P=0.02) were prognostic factors for poor visual outcome when each eye were analysed. Supplementary analysis using one eye model for bilateral cases, where right eye was analysed by convention, showed similar findings for poor presenting visual acuity (P=0.028) and lack of fundus view (P=0.042). But the effect of unilateral involvement lost statistical significance (P=0.08).
Table 3. Multivariate linear regression using final visual acuity in logMAR scale as dependent variable.
| Factors | Unstandardised coefficients beta | 95%CI | Standardised coefficients beta | P |
| Using individual eyes as unit of analysis | ||||
| Constant | 1.35 | -1.145 to 3.846 | - | 0.241 |
| Age | -0.028 | -0.073 to 0.017 | -0.306 | 0.189 |
| Gender (male) | 0.185 | -0.693 to 1.062 | 0.082 | 0.634 |
| Background DM | -0.609 | -1.413 to 0.195 | -0.286 | 0.116 |
| Positive blood culture | 0.012 | -1.448 to 1.473 | 0.006 | 0.985 |
| Positive intraocular fluid culture | -0.501 | -1.972 to 0.971 | -0.212 | 0.448 |
| Unilateral involvement | 0.991 | 0.094 to 1.888 | 0.466 | 0.035a |
| Time lapsed from ocular symptom to EKE diagnosis | -0.052 | -0.19 to 0.086 | -0.237 | 0.405 |
| Presenting VA in logMAR | 0.771 | 0.092 to 1.449 | 0.552 | 0.031a |
| Hypopyon | -1.002 | -2.219 to 0.216 | -0.447 | 0.093 |
| Lack of fundus view | 2.119 | 0.452 to 3.785 | 0.977 | 0.02a |
| Supplementary analysis using one eye from each patient (right eye taken for bilateral cases by convention) | ||||
| Constant | 1.842 | -0.972 to 4.657 | - | 0.129 |
| Age | -0.046 | -0.101 to 0.009 | -0.563 | 0.078 |
| Gender (male) | -0.247 | -1.373 to 0.88 | -0.122 | 0.536 |
| Background DM | -0.541 | -1.436 to 0.355 | -0.275 | 0.151 |
| Positive blood culture | 0.648 | -1.414 to 2.71 | 0.33 | 0.391 |
| Positive intraocular fluid culture | -0.497 | -2.036 to 1.042 | -0.231 | 0.379 |
| Unilateral involvement | 0.93 | -0.203 to 2.062 | 0.432 | 0.08 |
| Time lapsed from ocular symptom to EKE diagnosis | -0.119 | -0.299 to 0.061 | -0.491 | 0.126 |
| Presenting VA in logMAR | 0.983 | 0.201 to 1.766 | 0.763 | 0.028a |
| Hypopyon | -0.504 | -1.919 to 0.911 | -0.248 | 0.339 |
| Lack of fundus view | 2.406 | 0.17 to 4.642 | 1.224 | 0.042a |
CI: Confidence interval; DM: Diabetes mellitus; EKE: Endogenous Klebisella endophthalmitis; VA: Visual acuity. aP<0.05 was taken as significant.
Cases Highlights
No clear septic source could be identified in 2 cases (Table 1, No.7 and No.13) after extensive investigations. The diagnosis of endogenous Klebsiella endophthalmitis in the 2 cases was confirmed by positive culture of intraocular fluid only. Klebsiella is a normal flora in the gastrointestinal tract. It was unlikely a coincidence that both cases shared the similarity of having gastrointestinal pathologies (rectal cancer in No.7 and recurrent intestinal obstruction in No.13). In a separate report[13], it was postulated that bacteria gained access into the bloodstream via mucosal defect of the tumour in case No.7 (Figure 1). Therefore, it is advisable to offer gastrointestinal workup for cases of endogenous Klebsiella endophthalmitis without apparent septic source.
Figure 1. Case of Klebsiella endophthalmitis with rectal cancer (Case No.7).
A: Slit lamp photo showing rapid development of hypopyon after first pars plana vitrectomy; B: PET-CT revealed hypermetabolic lesion at rectum, compatible with rectal cancer.
Clinical signs of endophthalmitis include conjunctival injection, hypopyon, vitritis and subretinal abscess. However, 2 cases of the series (Table 1, No. 9 and 10) were documented with no conjunctival injection at initial examinations. Therefore, absence of conjunctival injection does not preclude the possibility of endophthalmitis.
One patient with background acute leukaemia (Table 1, No.14) had neutropenic fever and persistent Klebsiella bacteraemia for 6wk before developing endophthalmitis. She was seen by ophthalmologist as an asymptomatic consult 6wk prior to ocular involvement. Initial ocular examination was normal. It was until the development of eye redness when ophthalmologist was consulted once again and endogenous endophthalmitis was diagnosed (Figure 2). Despite blood culture yielded extended spectrum beta-lactamases producing Klebsiella pneumoniae, her vision remained relatively good after repeated intravitreal antibiotics injection. This case highlighted the importance of interval re-examination and clinical vigilance towards symptoms of endophthalmitis.
Figure 2. Subretinal abscesses in patient with persistent Klebisella bacteraemia documented with smart phone photography in isolation ward.
DISCUSSION
General Characteristics of Endogenous Klebsiella Endophthalmitis
Despite increased awareness of the condition, visual outcome of endogenous Klebsiella endophthalmitis remains poor. This is echoed by the study of another centre in Hong Kong, which reported an even higher evisceration rate of 60%[7].
Emergence of drug-resistant Klebsiella pneumoniae had been reported in Asia[14]–[16]. Of note, the majority of patients in this series (85.7%) were infected with non-resistant strains of Klebsiella pneumoniae, suggesting that the standard management of intravitreal ceftazidime remains effective against the cases in this study region.
Klebsiella liver abscess is well recognised as a cause of endogenous endophthalmitis. In contrast to the vast majority of liver abscess (94.7%) in a previous series of endogenous Klebsiella endophthalmitis[7], only 64.3% of patients in this series had liver abscess. This should alarm doctors treating patients with Klebsiella sepsis that endophthalmitis can still occur without liver abscess. Moreover, ocular symptoms may precede systemic symptoms in some patients (Table 1, No.6). Therefore, systemic sources should be extensively looked for if patient presents with an acute onset of endophthalmitis.
Prognostic Factors
Poor presenting visual acuity and lack of fundus view were identified by multivariate linear regression as poor prognostic factors of visual outcome in this study. Despite using individual eye as a unit of measure in regression is a valid analysis[17], caution was exercised to consider unilateral involvement as poor prognostic factor because supplementary analysis using one eye only in bilateral cases failed to identify it as a significant factor. This method of analysis might avoid the potential magnification of effect of shared systemic factors in bilateral cases. Nevertheless, some existing reports did find unilateral involvement a poor prognostic factor[5],[8]. The postulation was that bilateral cases tend to present earlier than those with unilateral involvement. However, this study did not find time delay from ocular symptom to endophthalmitis diagnosis a statistically significant factor for poor visual outcome.
Lack of fundus view signifies a greater degree of inflammation in the vitreous cavity at the time of diagnosis, which in turn serves as a marker of infection severity. Poor presenting visual acuity had previously been reported as predictors of poor visual outcome[4],[8]. Poor presenting visual acuity may be a result of media opacity due to vitritis, which again correlates with the degree of inflammation. At the same time, extensive retinal involvement of the posterior pole of the eye by direct inoculation of Klebsiella bacteria through the vascular choroid can cause severe macular dysfunction, leading to a poor presenting visual acuity. Therefore, lack of fundus view and poor presenting visual acuity translate severe degree of infection to poor final visual acuity.
Ophthalmological Screening Recommendations
Controversies exist in asymptomatic ophthalmological screening for patients with Klebsiella sepsis. Some suggested universal ophthalmological screening for all cases of Klebsiella sepsis and supported the argument with a lower evisceration rate after implementation of a regional universal screening program after the year 2000[5]. However, there was no difference in final visual acuity in their study. The lower evisceration rate could have been the result of increased clinical awareness, improved systemic infection control and better instrumentation in pars plana vitrectomy. Others recommended screening only for cases of Klebsiella liver abscess[18]. On the other hand, a substantial proportion of cases would be missed as hepatobiliary system infection only accounted for 64.3% in our series.
In general, the occurrence of endophthalmitis in Klebsiella sepsis is rare. Our study identified only 18 eyes in 14 patients within a period of 12 years in a major regional tertiary centre which serves a population of over 1 million. The reported endophthalmitis rate of Klebsiella liver abscess in Asia ranges from 3% to 7%[4],[19]–[20]. Only 3.8% of patients with Klebsiella bacteraemia was found with endophthalmitis[21]. As shown in our study where majority of eyes were symptomatic (88.9%), the cost-effectiveness of asymptomatic screening is questionable, especially in a region with tight resources. The universal poor visual outcome reported across studies[4]–[8] and the lack of standardised treatment beyond initial intravitreal antibiotics injection, have further weakened the argument for asymptomatic ophthalmological screening. By far, no study has successfully shown a delay in recognition of Klebsiella endophthalmitis has a significant effect on visual outcome. Our study also fails to show via multivariate linear regression the significant effect on final visual acuity by a delay from ocular symptom onset to endophthalmitis diagnosis.
Considering the heterogeneous nature of primary infection sites, it is difficult to provide a recommendation on targets for asymptomatic screening. Exemplified by 2 of the cases in this study (Table 1, No.9 and 14), patients can develop endogenous endophthalmitis even after normal initial ophthalmological examinations. Therefore, a normal examination during asymptomatic screening might create a false sense of security to the clinicians managing the Klebsiella sepsis. Moreover, asymptomatic screening might not be feasible in some units, where inpatient ophthalmological service is not available.
Nevertheless, screening should be considered in patients who cannot express their ocular symptoms, for example cases in intensive care unit or with cognitive impairment, as recommended by some authorities[6]. As shown in this series, lack of conjunctival injection does not preclude the possibility of endophthalmitis. It might be difficult to spot early signs of endophthalmitis in non-communicable patients. Excluding the exceptional case of persistent Klebsiella bacteraemia due to acute leukaemia (Table 1, No.14), where a long-time lapse of 42d from systemic symptom onset to ocular symptom had occurred, mean time from systemic symptom onset to ocular symptom in our study was 3.38±3.01d. Therefore, an ophthalmic examination performed within 1wk of systemic infection symptom onset, as recommended in a study[21], will theoretically identify most of the cases of endogenous Klebsiella endophthalmitis.
Strengths and Limitations
This is a study in an endemic region with a long study period of 12y. The study was conducted in a major tertiary hospital in the region where on-site ophthalmological consultation and assessment was readily available. Statistical analysis was performed using a multivariate model, which took the effect of multiple factors and variables into consideration. By performing linear regression with final visual acuity as the dependent variable, this study avoided arbitrarily dividing the continuous nature of visual outcome into dichotomous nature of good or poor visual acuity as previous reports did.
Limitations of our study include the retrospective nature and the relatively small number of cases. Missing data was inevitable due to the retrospective nature. Lack of standardised treatment algorithm complicated meaningful analysis of the effect of ocular treatment on visual outcome.
In conclusion, visual outcome of endogenous Klebsiella endophthalmitis remains poor. Poor presenting visual acuity and lack of fundus view predict poor final visual acuity. High index of suspicion for endophthalmitis is important in Klebsiella sepsis patients complaining of ocular symptoms. Ophthalmological screening is recommended in non-communicable patients with Klebsiella sepsis, and the suggested time frame would be within one week of systemic infection symptom onset. Interval re-examination should also be considered in cases where systemic infection is not yet controlled.
Acknowledgments
Conflicts of Interest: Mak CY, None; Ho M, None; Iu LP, None; Sin HP, None; Chen LJ, None; Lui G, None; Brelen ME, None; Young AL, None.
REFERENCES
- 1.Jackson TL, Paraskevopoulos T, Georgalas I. Systematic review of 342 cases of endogenous bacterial endophthalmitis. Surv Ophthalmol. 2014;59(6):627–635. doi: 10.1016/j.survophthal.2014.06.002. [DOI] [PubMed] [Google Scholar]
- 2.Wong JS, Chan TK, Lee HM, Chee SP. Endogenous bacterial endophthalmitis: an east Asian experience and a reappraisal of a severe ocular affliction. Ophthalmology. 2000;107(8):1483–1491. doi: 10.1016/s0161-6420(00)00216-5. [DOI] [PubMed] [Google Scholar]
- 3.Siu LK, Yeh KM, Lin JC, Fung CP, Chang FY. Klebsiella pneumoniae liver abscess: a new invasive syndrome. Lancet Infect Dis. 2012;12(11):881–887. doi: 10.1016/S1473-3099(12)70205-0. [DOI] [PubMed] [Google Scholar]
- 4.Sheu SJ, Kung YH, Wu TT, Chang FP, Horng YH. Risk factors for endogenous endophthalmitis secondary to klebsiella pneumoniae liver abscess: 20-year experience in Southern Taiwan. Retina. 2011;31(10):2026–2031. doi: 10.1097/IAE.0b013e31820d3f9e. [DOI] [PubMed] [Google Scholar]
- 5.Ang M, Jap A, Chee SP. Prognostic factors and outcomes in endogenous Klebsiella pneumoniae endophthalmitis. Am J Ophthalmol. 2011;151(2):338–344.e2. doi: 10.1016/j.ajo.2010.08.036. [DOI] [PubMed] [Google Scholar]
- 6.Odouard C, Ong D, Shah PR, Gin T, Allen PJ, Downie J, Lim LL, McCluskey P. Rising trends of endogenous Klebsiella pneumoniae endophthalmitis in Australia. Clin Exp Ophthalmol. 2017;45(2):135–142. doi: 10.1111/ceo.12827. [DOI] [PubMed] [Google Scholar]
- 7.Chung CY, Wong ES, Liu CCH, Wong MOM, Li KKW. Clinical features and prognostic factors of Klebsiella endophthalmitis-10-year experience in an endemic region. Eye (Lond) 2017;31(11):1569–1575. doi: 10.1038/eye.2017.92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Chen YH, Li YH, Lin YJ, Chen YP, Wang NK, Chao AN, Liu L, Wu WC, Lai CC, Chen TL, Chen KJ. Prognostic factors and visual outcomes of pyogenic liver abscess-related endogenous klebsiella pneumoniae endophthalmitis: a 20-year retrospective review. Sci Rep. 2019;9(1):1071. doi: 10.1038/s41598-018-37643-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Nussenblatt RB, Palestine AG, Chan CC, Roberge F. Standardization of vitreal inflammatory activity in intermediate and posterior uveitis. Ophthalmology. 1985;92(4):467–471. doi: 10.1016/s0161-6420(85)34001-0. [DOI] [PubMed] [Google Scholar]
- 10.Schulze-Bonsel K, Feltgen N, Burau H, Hansen L, Bach M. Visual acuities “hand motion” and “counting fingers” can be quantified with the Freiburg visual acuity test. Invest Ophthalmol Vis Sci. 2006;47(3):1236–1240. doi: 10.1167/iovs.05-0981. [DOI] [PubMed] [Google Scholar]
- 11.Lange C, Feltgen N, Junker B, Schulze-Bonsel K, Bach M. Resolving the clinical acuity categories “hand motion” and “counting fingers” using the Freiburg Visual Acuity Test (FrACT) Graefes Arch Clin Exp Ophthalmol. 2009;247(1):137–142. doi: 10.1007/s00417-008-0926-0. [DOI] [PubMed] [Google Scholar]
- 12.Holladay JT. Proper method for calculating average visual acuity. J Refract Surg. 1997;13(4):388–391. doi: 10.3928/1081-597X-19970701-16. [DOI] [PubMed] [Google Scholar]
- 13.Mak CY, Sin HP, Chan VC, Young A. Klebsiella endophthalmitis as the herald of occult colorectal cancer. BMJ Case Rep. 2018;2018:bcr-2017-223400. doi: 10.1136/bcr-2017-223400. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Hawkey PM. Prevalence and clonality of extended-spectrum beta-lactamases in Asia. Clin Microbiol Infect. 2008;14(Suppl 1):159–165. doi: 10.1111/j.1469-0691.2007.01855.x. [DOI] [PubMed] [Google Scholar]
- 15.Wong MHY, Shum HP, Chen JHK, Man MY, Wu A, Chan EW, Yuen KY, Chen S. Emergence of carbapenem-resistant hypervirulent Klebsiella pneumoniae. Lancet Infect Dis. 2018;18(1):24. doi: 10.1016/S1473-3099(17)30629-1. [DOI] [PubMed] [Google Scholar]
- 16.Zhang SQ, Zhang XC, Wu Q, Zheng XK, Dong GF, Fang RC, Zhang YZ, Cao JM, Zhou TL. Clinical, microbiological, and molecular epidemiological characteristics of Klebsiella pneumoniae-induced pyogenic liver abscess in southeastern China. Antimicrob Resist Infect Control. 2019;8:166. doi: 10.1186/s13756-019-0615-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Glynn RJ, Rosner B. Regression methods when the eye is the unit of analysis. Ophthalmic Epidemiol. 2012;19(3):159–165. doi: 10.3109/09286586.2012.674614. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Chew KL, Lin RTP, Teo JWP. Klebsiella pneumoniae in Singapore: hypervirulent infections and the carbapenemase threat. Front Cell Infect Microbiol. 2017;7:515. doi: 10.3389/fcimb.2017.00515. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Chou FF, Kou HK. Endogenous endophthalmitis associated with pyogenic hepatic abscess. J Am Coll Surg. 1996;182(1):33–36. [PubMed] [Google Scholar]
- 20.Tan YM, Chee SP, Soo KC, Chow P. Ocular manifestations and complications of pyogenic liver abscess. World J Surg. 2004;28(1):38–42. doi: 10.1007/s00268-003-6963-2. [DOI] [PubMed] [Google Scholar]
- 21.Sng CC, Jap A, Chan YH, Chee SP. Risk factors for endogenous Klebsiella endophthalmitis in patients with Klebsiella bacteraemia: a case-control study. Br J Ophthalmol. 2008;92(5):673–677. doi: 10.1136/bjo.2007.132522. [DOI] [PubMed] [Google Scholar]