Purpose
To evaluate the efficacy of quantitative real-time polymerase chain reaction (qPCR) in the diagnosis of postoperative bacterial endophthalmitis among patients who underwent cataract surgery at a tertiary care center.
Design
Prospective experimental study.
Methods
This was a single-center study of 64 eyes of 64 patients presenting with clinical signs and symptoms of endophthalmitis within 1 year of cataract surgery. Patients with glaucoma filtering or cornea surgery in the past year, postoperative trauma, fungal endophthalmitis, or preoperative inflammatory conditions were excluded. Vitreous samples were obtained during vitreous tap or vitrectomy and sent for both culture and qPCR with sequencing. Vitreous samples obtained from 50 patients undergoing vitrectomy for noninflammatory indications served as controls. The main outcome measures were the sensitivity of qPCR compared to culture and concordance of results of pathogen identification with sequencing vs phenotypic speciation.
Results
qPCR detected 16s bacterial DNA in 37 patients (66%), compared to 19 (34%) with traditional culture. Only 1 patient had a positive result by culture ( Nocardia species) but negative result by qPCR. For the 18 samples positive by both qPCR and culture, there was a 100% concordance in pathogen identification between sequencing and phenotypic speciation.
Conclusion
In cases of suspected bacterial endophthalmitis, qPCR offers an improved diagnostic yield and may be a useful adjunct to traditional culture. Further large-scale clinical studies are needed to elucidate the full clinical utility of qPCR.
Endophthalmitis is an inflammatory reaction to pathogens in the posterior segment of the eye. It is a rare complication of cataract surgery, occurring in 0.02% to 0.09% of patients. Despite its low incidence, postoperative endophthalmitis has severe visual consequences, thus making rapid pathogen identification and treatment vital for optimizing recovery. Although broad-spectrum empiric antibiotics are initiated immediately upon suspicion of endophthalmitis, they should be adjusted to more narrow-spectrum agents once the offending pathogen is known. This is both to ensure that the pathogen is susceptible to the chosen antibiotic regimen and to prevent the development of antibiotic resistance. Historically, the gold-standard technique for pathogen identification has been Gram stain and microbial culture, since the vast majority of pathogens responsible for endophthalmitis are bacterial. However, microbiological techniques are time-consuming and have relatively low sensitivity for bacterial detection in aqueous and vitreous humor.
More recently, eubacterial polymerase chain reaction (PCR) has been employed as a fast, accurate, and sensitive method for detecting the etiologic agent in bacterial endophthalmitis. The extreme sensitivity of traditional PCR has 2 well-described potential pitfalls. First, even very small amounts of contamination with bacterial DNA (for example, from unaffected adjacent tissue harboring commensal organisms or from laboratory cross-contamination) may lead to false-positive results. Second, traditional PCR results are qualitative and generally reported as positive or negative, depending on if an amplified product is present after electrophoresis. No quantitative information regarding pathogen load is obtained from results of routine nested PCR.
Real-time PCR is a variant of PCR in which the amplification reaction is performed using fluorescent probes or DNA intercalating dyes that fluoresce in proportion to the degree of accumulation of double-stranded PCR product. Each PCR reaction is monitored by means of fiber-optic fluorimetry during the thermal cycling process. The resulting curve of fluorescence accumulation can be used to quantify the amount of pathogen in the sample. The real-time PCR assay has many advantages over conventional PCR methods including rapidity, quantitative measurement, lower contamination rate, high sensitivity and specificity, and ease of standardization. Real-time PCR techniques have previously been used as an aid in the diagnosis of‘ sepsis, infectious arthritis, nosocomial urinary tract infections, infections in intensive care unit patients, and viral diseases of the eye.
More recently, real-time PCR has been described as a diagnostic tool for bacterial endophthalmitis in vitreous and aqueous humor. Studies by Bispo and associates and Sugita and associates affirm that real-time PCR of bacterial 16s rDNA region is a sensitive, rapid test for the diagnosis of bacterial endophthalmitis. The main objective of the present study is to evaluate the usefulness of quantitative real-time PCR (qPCR) in the diagnosis of postoperative bacterial endophthalmitis among patients undergoing cataract surgery at a tertiary care center and to further characterize the use of a universal probe and primers in the speciation of pathogenic bacteria.
Methods
We undertook an experimental study of patients presenting to a single institution between August 2008 and August 2010 with signs and symptoms suspicious for endophthalmitis within 1 year of cataract extraction. These signs and symptoms included decreased visual acuity as well as anterior and posterior segment inflammation confirmed by clinical examination and ultrasonography. Patients were excluded from the study if they had an insufficient quantity of vitreous sample for performing both culture and qPCR, if they had a concurrent glaucoma filtering or corneal surgery in the past year, or if they had any history of trauma in the affected eye in the postoperative period. Patients with a preoperative intraocular inflammatory condition or suspicion for fungal endophthalmitis were also excluded. Based on the above criteria, 64 consecutive patients with clinically suspected acute postoperative bacterial endophthalmitis were enrolled. For all 64 patients, uncontaminated nondiluted vitreous fluid samples (0.5–1.0 mL) were collected during pars plana vitrectomy (PPV) or vitreous tap. All patients received topical antibiotics only, but no intravitreal or systemic antibiotics prior to vitreous sampling. However, after sampling, all patients received intravitreal administration of vancomycin, ceftazidime, and dexamethasone.
We also analyzed vitreous samples from a control group of 50 patients undergoing vitrectomy. The indications for vitrectomy included retinal detachment (n = 20), idiopathic macular hole (n = 15), and idiopathic epiretinal membrane (n = 15). No patients in the control group had evidence of preoperative inflammation or endophthalmitis. The control group patients received preoperative topical antibiotics only, but no intravitreal or systemic antibiotics prior to sampling.
Immediately after collection, vitreous samples from both the control and experimental groups were divided into 2 aliquots. One aliquot was sent for microbiological identification with Gram stain and culture within 30 minutes of collection. Samples were inoculated on blood agar, chocolate agar, thioglycolate broth, brain heart infusion broth, and potato dextrose agar. The plates were incubated for 7 to 14 days in aerobic and anaerobic conditions at 37 C with 5% carbon dioxide. The bacterial colonies cultured on agar plates were identified by the Analytical Profile Index strip method.
The remaining aliquot was immediately transferred aseptically into a presterilized microfuge tube and stored at −20 C for qPCR. DNA was isolated from 100 μL of vitreous fluid using a QIAamp DNA Mini Kit (Qiagen Inc, Valencia, California, USA) according to the manufacturer’s instructions. The DNA was eluted in elution buffer and stored at −20 C until use.
The qPCR assay was performed on the Applied Biosystems Fast Real Time PCR 7900HT platform (Applied Biosystems, Foster City, California, USA). Priority was given to probes that could facilitate the choice of a suitable antibiotic therapy. We chose the universal Taqman probe in order to differentiate a wide range of gram-positive and gram-negative bacteria. The primers RW01 5′-AACTGGAGGAAGGTGGGGAT-3′ and DG74 5′-AGGAGGTGATCCAACCGCA-3′ and the universal bacterial Taqman probe 5′- (6FAM) GTAAGGGCCATGAGGACTTGACGT-(TAMRA)-3′ have been previously described. The 380-bp product resides within the highly conserved bacterial 16s rDNA target. The primers and probe were synthesized at Applied Biosystems. The primers and probe were tested for specificity and cross-reactions. The reaction volume included 2× reaction mixtures of Taqman (Applied Biosystems), 150 nM concentration of forward and reverse primer, 100 nM concentration of probe, and 2.5 μL of template DNA. For each run, a negative control (2.5 μL of distilled water as a template) was included. The total reaction volume was 25.0 μL. The samples were run in duplicates in 96-well microtiter plates followed by the PCR. The initial stage of denaturation used a thermocycler at 95 C for 10 minutes, followed by amplification in the same machine for 35 cycles at 95 C for 10 seconds, 60 C for 30 seconds, and 72 C for 45 seconds.
PCR results were analyzed using Sequence Detection Systems V2.3 software (Applied Biosystems). Samples were considered positive when they gave a cycle threshold (Ct) value of less than 29.0. The scientist responsible for the qPCR assays was masked to the culture results. To assess the sensitivity of the assay, 10-fold dilutions of 24-hour cultures of Staphylococcus aureus were used as the positive control. The positive control DNA that was amplified was separated on agarose gels, purified using a kit (EZ-10spin column DNA extraction kit, Bio Basic Inc, Canada), and quantified using nanospectrometry (ND-1000; Nano Drop Technologies Inc, Wilmington, Delaware, USA). The dilution series of purified DNA in the range of 1 × 10 1 to 1 × 10 13 copies/mL were generated and the standard curve was obtained. Each 10-fold dilution series and the samples were run in duplicate and the experiment was repeated twice.
To avoid bacterial contamination, the preparation of master mix and sample addition was performed in separate laminar flow benches and employed single-use aliquots of reagents and dedicated pipettes. Microfuge tubes and milliQ water were carefully sterilized prior to use.
The specificity of the universal probe and primer was determined after correct amplification of reference strain DNA. To exclude cross-reactions, amplification was performed with fungal, viral, and parasitic DNA. In order to sequence the amplified products, the DNA was purified using an EZ-10spin column DNA extraction kit (Bio Basic Inc) and dissolved in 10 μL of elution buffer. Cyclic sequencing was carried out using Bigdye terminator 3.0 kits (Applied Biosystems). The resulting cycle sequencing product was purified using ethylenediamine-tetraacetic acid and ethanol/sodium acetate. The air-dried product was dissolved in formamide and separated on an ABI prism 3130 Genetic Analyzer (Applied Biosystems). The sequences were aligned using Chromas2 software and finally blasted with BLAST NCBI software ( http://blast.ncbi.nlm.nih.gov/ ) to identify species and sequence similarity.
Clinical information collected included type of cataract surgery performed, age at presentation, time elapsed between cataract surgery and onset of endophthalmitis symptoms, visual acuity at presentation, and visual acuity at 3-month follow-up.
Results
The limits of detection for the 16s rDNA real-time PCR assay were determined from the 10-fold dilution series of Staphylococcus aureus . The slope and r value of the reference 10-fold dilution series were −3.52 and 0.995 respectively. In addition, the universal probe and primer did not detect any fungal, viral, or parasitic DNA, thus showing no cross-reactivity with other organisms.
Out of the 64 patients with clinically suspected bacterial endophthalmitis, the presence of 16s rDNA was detected in 37 (66%). In all 37 of these patients, high copy numbers of 16s rDNA were present, ranging from 1.42 × 10 5 to 3.64 × 10 7 copies/mL, indicating the presence of a large bacterial burden consistent with infection. In contrast, conventional bacterial cultures were positive in 19 out of 64 samples (34%). Of these 19 culture-positive samples, 16s rDNA was detected by qPCR in 18 cases (95%). For the single patient who had a culture-positive sample and a negative PCR result, the identified pathogen was a Nocardia species. For the remaining 18 samples, molecular identification using the sequenced products by BLAST analysis agreed with phenotypic species determination ( Table 1 ). BLAST analysis is only possible with high copy numbers of bacterial 16s rDNA.
| Culture Findings | Number (n) | Sequencing Result |
|---|---|---|
| Coagulase-negative Staphylococci | 10 | 10 of 10 positive |
| Pseudomonas aeuroginosa | 4 | 4 of 4 positive |
| Staphylococcus aureus | 1 | 1 of 1 positive |
| Streptococcus pneumoniae | 1 | 1 of 1 positive |
| Klebsiella pneumoniae | 1 | 1 of 1 positive |
| Pseudomonas stutzeri | 1 | 1 of 1 positive |
In the 45 culture-negative samples, qPCR was also negative in 26 cases. In the remaining 19 samples, qPCR successfully detected high copy numbers of 16s bacterial rDNA, indicating the presence of bacterial infection. The positive samples had Ct values in the range of 19.0 to 28.09. Sequenced amplification products of samples from these 19 patients with culture-negative, qPCR-positive results showed 100% concordance between the subject sequence and the query sequence.
Out of the total 37 qPCR positive samples, the etiologic agent was gram-positive in 20 cases (54.1%) and gram-negative in 17 (45.9%). Of these gram-positive organisms, Staphylococcus epidermidis accounted for 8 samples, Staphylococcus lugdunesis for 6, Staphylococcus aureus for 4, and Streptococcus pneumoniae for 2. Among gram-negative organisms, Sphingomonas paucimobilis accounted for 7 samples, Pseudomonas species for 5, Ralstonia pickettii for 3, Klebsiella pneumoniae for 1, and Escherichia coli for 1. The pathogens were identified via qPCR followed by sequencing and BLAST analysis. The bacterial 16s rDNA amplification using universal probe and universal primer by real-time PCR assays was not detected in any of the 50 vitreous humor samples from the control group. The bacterial cultures obtained from these samples were also negative.
The sensitivity of microbial culture for bacterial detection in clinical samples was 34%. With the use of real-time PCR assays, the sensitivity of bacterial detection by the amplification of 16s rDNA was increased to 66%. The clinical sensitivity and specificity of real-time PCR was 95% and 58%, respectively. The positive predictive value and negative predictive value with 95% confidence intervals are shown in Table 2 .
| Quantitative Real-time Polymerase Chain Reaction a | Culture | Total | |
|---|---|---|---|
| Positive | Negative | ||
| Positive | 18 | 19 | 37 |
| Negative | 1 | 26 | 27 |
| Total | 19 | 45 | 64 |
a Sensitivity = 0.9474 (95% confidence interval [CI]: 0.7189–0.9972), specificity = 0.5778 (95% CI: 0.4224–0.7201), positive predictive value = 0.4865 (95% CI: 0.3224–0.6533), negative predictive value = 0.9630 (95% CI: 0.7911–0.9981).
Clinical data were available for 61 patients. Of these, 62% were male and the average age at presentation was 60.0 years (range 41 to 77 years). The average delay between surgery and presentation with signs and symptoms of endophthalmitis was 25.2 days (range 1–365 days). For patients with a gram-positive pathogen identified by PCR or culture, the mean interval between surgery and presentation was 7.7 days (range 1–27 days; median 7.7 days). For patients with gram-negative pathogens, the mean interval was 14.6 days (range 1–60 days, median 20.7 days). There was no statistically significant difference in the median interval between the gram-positive and gram-negative groups (7.7 days vs 20.7 days, P = 0.57 by Mann-Whitney U test).Three different types of cataract surgery were performed: extracapsular cataract extraction with intraocular lens (IOL) placement (n = 1), manual small-incision cataract surgery with IOL (n = 33), and phacoemulsification with IOL (n = 27).
Average preoperative visual acuity in logMAR units was 1.91, roughly equivalent to a Snellen fraction of 20/1600. The following estimations were used for patients presenting with vision below the threshold of Snellen acuity: count fingers vision was assigned a logMAR value of 1.90, hand motion was assigned 2.30, light perception was assigned 2.70, and no light perception was assigned 3.00. The average postoperative visual acuity for all patients improved to 0.80 (approximately 20/120).
For the 18 patients whose samples were positive by both culture and qPCR, 12 (67%) experienced improvement in visual acuity, whereas vision remained the same or worse in 6 ( Table 3 ). The average preoperative vision of these 18 patients improved from 2.05 to 1.22 postoperatively. For the 19 patients whose samples were positive only by PCR, clinical data were available for 16. Twelve patients (75%) achieved improvement in their visual acuity at 3-month follow-up, whereas visual acuity remained the same or worsened in 4 patients.
| Variable | Sample Size, n | Improvement, n (%) | Same or worse, n (%) | P Value |
|---|---|---|---|---|
| Both culture- and qPCR-positive | 18 | 12 (67) | 6 (33) | .022 (Fisher exact test) |
| Only qPCR-positive | 16 | 12 (75) | 4 (25) | |
| Both culture- and qPCR-negative | 26 | 25 (96) | 1 (4) | |
| Total | 60 | 49 (82) | 11 (18) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
