Abstract
Purpose:
We have reported that the absence of posterior vitreous detachment (PVD) is related to the onset and severity of infectious endophthalmitis, based on clinical experience. To demonstrate clinical findings in animal models, we created endophthalmitis models for the presence or absence of PVD and examined differences in severity.
Method:
We estimated a rabbit infectious eye model with and without PVD using Pseudomonas aeruginosa (PVD(+) and PVD(−) groups). After injection of bacteria inoculation for 3, 6, 12, and 24 hours, we evaluated the clinical score of the anterior chamber (n = 14). Removing the vitreous and retina from the enucleated eyeballs, the number of bacteria was counted using each specimen (n = 12). In addition, the number of inflammatory cells approximately 3 mm2 around the optic disc and histopathologic grading of intraocular inflammation was compared from histopathologic images (n = 7). Electroretinogram (ERG) was performed in experimentally infected rabbit eyes in both groups at three times after injection of the bacterial suspension.
Results:
There was no difference between the two groups in the clinical score of the anterior chamber of each time phase, but the bacterial cultures showed significantly fewer bacteria in the PVD(−) group 24 hours after bacterial inoculation (P < 0.05). Furthermore, the number of inflammatory cells was significantly less in the PVD group (P < 0.05). As a result of ERG, the decreases of a- and b-waves in amplitude were significantly greater in the PVD(−) group than in the PVD(+) group.
Conclusions:
The present study confirms using animal models that the absence of PVD contributed to the severity of bacterial endophthalmitis.
In intraocular surgeries, postoperative bacterial endophthalmitis occurs at a certain rate. However, even with the recent advances in vitreous surgery, postoperative bacterial endophthalmitis remains a disease with poor visual prognosis. Except for toxicity of the causative bacteria, the factors that aggravate postoperative bacterial endophthalmitis are still unknown. Moreover, the risk factors for aggravation in the early stage of endophthalmitis remain unclear. The most common causative microorganism of endophthalmitis after cataract surgery is coagulase-negative
Streptococcus,
1,2 followed by Gram-positive cocci. The main causative bacteria of bleb-related endophthalmitis (BRE) after glaucoma surgery are
Streptococcus species (37%) and
Haemophilus influenzae (8.9%). According to the classification of BRE by severity, stage I is defined as infection confining to the filtration bleb, stage II as infection mainly confined to inside the anterior chamber and not extending to the vitreous body, stage IIIa as infection involving the vitreous body but mild, and stage IIIb as infection involving the vitreous body and serious. In Japan, the incidence of BRE in one survey was reported to be 1.5% at 2.5-year follow-up after mitomycin C–augmented trabeculectomy, and a subsequent survey reported an incidence of 2.6% at 5-year follow-up.
3–5 In endogenous endophthalmitis caused by metastatic spread of bacteria or fungi from other organs, apart from
Staphylococcus aureus and
Streptococcus species, Gram-negative bacteria such as
Klebsiella species and fungi were also reported as causative microorganisms.
6–8
We reported the absence of posterior vitreous detachment (PVD) despite old age in 70% of patients with endogenous endophthalmitis (7 of 10 patients) and 80% of those with BRE (8 of 10 patients).
9,10 In addition, retinal necrosis occurred in many patients with endogenous endophthalmitis and those with BRE who did not have PVD. In one patient with BRE who required enucleation, histopathologic examination confirmed the absence of PVD. According to a previous report, PVD generally occurs between the ages of 54 and 57 years,
11 and the possibility that PVD may be associated with the onset and severity of endophthalmitis has been suggested. A study in rabbits has shown more vigorous bacterial growth in the vitreous than in the anterior chamber.
12
On the basis of the clinical observations and the findings to date, we hypothesized that continuity of the infection focus, the vitreous body, and the retina facilitates infection in the eye and accelerates aggravation of the disease. Therefore, this research aimed to elucidate the relationship between the presence or absence of PVD and aggravation of endophthalmitis by verifying in vivo that the presence of gel vitreous is a factor associated with the development of endophthalmitis and aggravation of the pathologic state. As the first step, an animal model of endophthalmitis using rabbits was established, and the effect of the presence or absence of PVD on the severity of endophthalmitis was investigated.
The endophthalmitis model without induced PVD was created according to a previous report.
17 After performing an anterior chamber paracentesis with a 27-gauge needle to reduce the intraocular pressure sufficiently, 100 µL
P. aeruginosa suspension was injected with a 30-gauge needle via a site 6 mm from the corneal limbus. For inoculation, a bacterial suspension adjusted to a density of 1.0 × 10
8 CFU/mL was diluted to 1.0 × 10
5 CFU/mL. Fourteen rabbits were used in this group (PVD(−) group).
The endophthalmitis model with induced PVD was created by injecting the bacterial suspension into the eyes using the same methods as described above. Fourteen rabbits were used in this group (PVD(+) group).
After intraocular inoculation of the bacterial suspension, the eyes were observed at 3, 6, 12, 24, and 48 hours after inoculation, and then the rabbits were euthanized. The eyeballs were enucleated for pathologic examination to evaluate the state of infection.
The numbers of bacterial colonies on the cultures of retinal tissue from all the rabbits in both groups were enumerated at 3, 6, 12, and 24 hours after injection of the bacterial suspension. Twelve rabbits were used in both groups.
At each time point, the eyeball was enucleated and disinfected with iodine in polyvinyl alcohol and saline. Next, the retina was removed from the ora serrata of the enucleated eyeball and mixed with saline to a total volume of 1 mL. After serial dilution by physiologic saline, the sample was spread on NAC medium and cultured at 37°C for 24 hours. Thereafter, colonies that formed on the medium were counted.
Number of Inflammatory Cells Around the Optic Disc and Histopathologic Grading of Intraocular Inflammation
No obvious signs of infection were observed at 3 hours and 6 hours after bacteria inoculation in both the PVD(−) and PVD(+) groups. After 12 hours, conjunctival hyperemia and discharge appeared gradually. At 24 hours, the eyes in both groups showed findings suggestive of endophthalmitis, including conjunctival edema and hypopyon in the anterior chamber. In many eyes, the fundus could not be visualized due to opacity of the anterior chamber. At 48 hours, the inflammatory findings became more severe, and the enucleated eyeballs showed necrosis of the retina, suggesting that intraocular infection was established.
In this study, animal models were created, which validated the findings related to endophthalmitis observed in clinical practice. In all the experiments, various outcome measures were evaluated at 3, 6, 12, and 24 hours after inoculation of the bacteria. No significant differences in the anterior segment grading and in the histopathologic grading of retinal and vitreous tissues were observed between the PVD(−) and PVD(+) groups at all the observation time points. In contrast, significant differences in the number of bacteria in the retina, the number of inflammatory cells in the retina and vitreous body, and electroretinographic results were observed between the two groups at 24 hours after bacteria inoculation, and all these outcome measures were more severe in the PVD(−) group. The result of the anterior segment grading in the animal model in this study may not contradict with the experience in clinical practice, because the clinical assessment of severity in the anterior segment often does not match the fundus findings in patients.
In the eyes without induced PVD, the retina is adhered to the vitreous body, and hence there is continuity between the infected site and the retina. In other words, the infection focus, the vitreous body, and the retina are connected. Under this condition, it is possible that infection of the retina may be established earlier and aggravation of the pathologic condition may be accelerated. On the other hand, induction of PVD breaks the continuity between the retina and the vitreous body, resulting in the loss of continuity between the infection site and the retina, and as a result, it may take a longer time to complete the infection process in the retina. This probably gave rise to the significant differences in the number of bacteria in the retina, the number of inflammatory cells in the retina and vitreous body, and the electroretinographic result between PVD(−) and PVD(+) eyes.
While indirect experiments examining the involvement of bacteria produced significant results, our experiment demonstrated that despite the significantly larger number of inflammatory cells in the retina and vitreous in the PVD(−) group, there was no significant difference in the histopathologic grading of the retina and vitreous between the two groups. One possible explanation is that in both the PVD(−) and PVD(+) groups, few eyes showed a state of full-thickness retinal destruction, and in most cases, destruction did not extend beyond the inner nuclear layer. It is possible that a significant difference in histopathologic grading could be detected by using a more detailed scoring scale.
In the present study, in all evaluations that showed a significant difference between the PVD(−) and PVD(+) groups, there were no significant differences between the two groups before 24 hours following bacteria inoculation. Destruction of retina was mild, and the bacterial load and number of inflammatory cells in the retina were low up to 12 hours after inoculation. On the other hand, the number of bacteria in the retina and the number of infiltrated inflammatory cells increased dramatically over the period from 12 to 24 hours, accompanied by progression of retinal tissue damage. These findings may suggest that intraocular infection is established after 12 hours following inoculation of the bacteria.
The above results suggest that endophthalmitis progresses more rapidly in the eyes without PVD. Postoperative endophthalmitis may cause serious visual impairment, and a multifaceted approach is required, which includes appropriate prevention, rapid diagnosis, and proper treatment.
22 Currently, the mainstay of endophthalmitis management is identification of the causative organism, proper use of antibacterial agents, and vitrectomy.
23,24 Based on the results of the current study, it may be possible to propose a guideline that if the presence or absence of PVD can be diagnosed before surgery by direct slit-lamp examination and/or ultrasonography, early vitrectomy is recommended for eyes without PVD.
In addition, the P. aeruginosa strain used in this experiment has one flagella at one end of the cell and moves actively. Even though infection was caused by a motile P. aeruginosa strain, intraocular inflammation and the degree of tissue destruction were mild in the PVD(+) endophthalmitis model animals. This finding may support the relationship between the presence or absence of PVD and the severity of postoperative endophthalmitis.
One on the limitations of the present study was that only P. aeruginosa was used that aimed to examine the aggravating factors of postoperative endophthalmitis and BRE. Other causative microorganisms of bacterial endophthalmitis, including staphylococci and streptococci, have not been studied. Further research using these bacterial species is required in the future. This experiment is a proof of concept that focuses on ensuring the establishment of infection and demonstrating the relationship between posterior vitreous detachment and bacterial infection. Our previous clinical report has shown many cases of endophthalmitis caused by bacteria with low virulence, such as coagulase-negative staphylococci. Conducting experiments using these bacteria would simulate conditions closer to the actual clinical practice. In addition, since the number of model rabbits used in each experiment was small, the present results should be verified by conducting experiments using a larger number of animals.
To the best of our knowledge, there is no other report demonstrating the relationship between PVD and endophthalmitis using animal models as in this study. Using animal models, the present study was able to validate that the presence or absence of PVD is involved in the aggravation of endophthalmitis, as is known from clinical experience; in other words, the absence of PVD is one of the factors that aggravate endophthalmitis. Although P. aeruginosa is a rare cause of endophthalmitis, it does cause severe sight-threatening endophthalmitis. Here we use it in our experimental model because it consistently caused endophthalmitis in our model. Future work will involve repeating these studies with Gram-positive bacteria that induce endophthalmitis.