Positive bacterial culture from the conjunctiva have been detected in 13–28% of cataract surgery, even after disinfection with povidone-iodine, suggesting it is difficult to completely eliminate bacteria on the conjunctiva. Postoperative endophthalmitis after MIVS has two opportunities to develop; during and after the operation. Tominaga et al. found significantly higher bacterial cultures in vitreous collected soon after trocar insertion with a 25-G compared with a 20-G, suggesting conjunctival bacteria are introduced during trocar insertion. It is thought that indigenous bacteria on the conjunctiva penetrate into the eye causing the endophthalmitis, as the pathogenic bacteria of endophthalmitis has been reported to be identical to normal conjunctival bacteria flora.
In the present study, microsphere penetration into the eye was visually confirmed in all cases using a 25-G and 27-G cannula. With opening of the irrigation canula, spreading of microspheres around the cannula was especially evident. Furthermore, when microspheres were applied on the conjunctiva in a quantitative study, significantly higher fluorescence intensities were obtained from samples compared to controls. These indicate that contamination on the conjunctiva adheres to the trocar and cannula with trocar insertion, and that the open irrigation cannula enables dispersal into the eye.
While not all contamination introduced into the eye is bacterial, previous studies have shown endophthalmitis can develop from just a few bacteria in the vitreous body. Thus, the risk of postoperative endophthalmitis might be increased with MIVS as it inserts a trocar through the conjunctiva. While the irrigation volume could be sufficiently achieved using 20-G, the total irrigation volume will decrease due to the small size of cannula in MIVS. It has been reported that the lower flow volume through the 25-G infusion during surgery as compared with the 20-G cannula, at an approximately 7-fold lower infusion rate, may diminish the washing effect of the infusion. Furthermore, if the peripheral vitrectomy was not performed effectively to allow wound self-closure, it might result in a further reduction in irrigation volume. A report by Tominaga et al. found complete peripheral vitrectomy resulted in a 0% positive bacterial culture rate in both 20-G and 25-G groups at the end of the operation. In other words, it is possible to remove bacteria adhering to the periphery of the cannula by performing peripheral vitrectomy. Our experiments also show that the contamination from the conjunctiva possibly remains, particularly near the cannula, and that peripheral vitrectomy near the cannula is important in preventing endophthalmitis.
Shimada et al. proposed the concept of intraoperative disinfection that exploited the advantages of shorter disinfection times with 0.25% PI and fewer side effects on the ocular surfaces. With routine MIVS, they observed a bacterial detection rate of 2% from the vitreous cavity at the end of vitrectomy, but continuous ocular surface irrigation with 0.25% PI resulted in a bacterial detection rate of 0% at the completion of vitrectomy, without any occurrence of endophthalmitis. In MIVS, it is inevitable that contamination on the conjunctiva will penetrate into the eye when the trocar is inserted. However, we are able to prevent the introduction of live bacteria. Therefore, it is of great importance that the ocular surface is disinfected just before insertion of trocar, such as by applying disinfecting agents like 0.25%PI.
It is possible that intraoperative infection can be reduced by such intraoperative infection countermeasures. Additionally, postoperative infection can also be reduced by performing sclerotomies by oblique insertion and using an air tamponade to close the inner valve securely, thus lowering the chance of postoperative infection. These measures have lowered the current rate of postoperative endophthalmitis with MIVS. Our study is the first to visually demonstrate that contamination can be brought into vitreous cavity through conjunctiva, even with 27-G MIVS. It is essential that the surgeon be aware of the possibility of introducing contamination on the conjunctiva at all times during surgery, especially when they insert the trocar at the beginning of the MIVS.
The present study has limitations. First, porcine eyes were used as an alternative model for human eyes. The porcine eye has a thicker sclera than the human eye. However, a histological study revealed human and porcine scleras to have a similar histology and collagen bundle organization, indicating that the porcine sclera can serve as a good model for the human sclera. In the human eye, which has a thinner sclera, significantly more bacteria might be introduced, such that our model might underestimate the severity of contamination. Second, the number of experimental cases was small. Yet, we visually detected fluorescent microspheres in 100% of eyes, and in the experimental quantification study, the fluorescence strength averages differed significantly between the samples and controls. Therefore, we consider the sample size of this study to have been sufficient. Our experimental vitreous contamination models do not exactly reproduce the actual surgical field for vitrectomy. However, we succeeded in demonstrating that trocar insertion carries a risk of introducing contamination directly into the eyes mechanically, even when a 27-G trocar is used. Thus, disinfection of surgical fields with povidone iodine during surgery seems to be the best way to prevent endophthalmitis after MIVS. The supplementary video will be beneficial for the vitreous surgeon, providing the clear images of how bacteria are introduced into the vitreous cavity during MIVS.