Incidence of Presumed Silicone Oil Droplets in the Vitreous Cavity After Intravitreal Bevacizumab Injection With Insulin Syringes

Key Points

Question

What is the incidence of silicone oil droplets in the eye after intravitreal injection of bevacizumab preloaded in insulin syringes with compounding preparation?

Findings

In this observational study (60 patients), there was an increase in the incidence of presumed silicone oil droplets from May to November 2016 (1.73% [59 of 3402]) compared with the incidence from October 2015 to April 2016 (0.03% [1 of 3230]).

Meanings

These findings suggest that physicians should counsel their patients of the risk of floaters with intravitreal bevacizumab preloaded in insulin syringes.

Abstract

Importance

Intravitreal bevacizumab is a frequently used antivascular endothelial growth factor medication in the United States, but its off-label use is associated with risks associated with the compounding preparation.

Objective

To determine the incidence of presumed silicone oil droplets after intravitreal bevacizumab was prepared in insulin syringes by a compounding pharmacy.

Design, Setting, and Participants

A retrospective review was conducted of 60 patients who experienced intravitreal silicone oil droplets in the eye after intravitreal bevacizumab injections from a single specialist practice from October 1, 2015, to November 30, 2016. Bevacizumab, 1.25 mg/0.05 mL, was delivered in insulin syringes with a 31-gauge needle.

Main Outcomes and Measures

Small, round clear spheres in vitreous on dilated biomicroscopic retinal examination.

Results

Over a 14-month period involving 6632 intravitreal bevacizumab injections, 60 cases (35 [58%] women) of intravitreal silicone droplets were identified. Mean [SD] age of the patients was 80 [12] years; the population comprised 48 white, 9 Asian, and 3 Hispanic patients. The incidence of silicone oil droplet injections was 0.03% (1 of 3230) from October 2015 to April 2016 and 1.7% (59 of 3402) from May to November 2016 (Fisher exact test, P < .001; odds ratio [OR], 57; 95% CI, 9.8-2260). From May to November 2016, nonpriming the syringe before the intravitreal injection had a higher risk of intravitreal silicone oil droplets compared with priming the syringe (6.4% [47 of 739] vs 0.5% [12 of 2627]; Fisher exact test, P < .001; OR, 15.1; 95% CI, 7.9-33.4). Among the 60 cases, 41 patients (68%) were symptomatic, and the main symptom was floaters with spots of light. Among the patients with floaters, 36 (88%) improved over time (range, 2-8 months) despite the silicone droplets still being present on ophthalmoscopic examination.

Conclusions and Relevance

An increase in intravitreal silicone oil associated with bevacizumab prepared with insulin syringes was documented. Priming the syringe before injection was associated with a lower frequency of this complication. These findings suggest that physicians should counsel their patients on the risk of floaters with intravitreal bevacizumab preloaded in insulin syringes.

This observational study examines the incidence of silicone oil droplets introduced into the eye through intravitreal injection of bevacizumab.

Introduction

Intravitreal bevacizumab is a frequently used intraocular medication in the management of retinal diseases in the United States¹; however, bevacizumab is not approved by the US Food and Drug Administration for this indication. As a result, compounding pharmacies often prepare bevacizumab for its off-label use as an intravitreal injection, and there have been increased risks involving endophthalmitis outbreaks,^2,3 impurities involved with plastic syringe use,⁴ and variations in protein concentration.⁵

In August 2016, the American Society of Retina Specialists reported presumed silicone oil droplets associated with intravitreal bevacizumab injected using insulin syringes. We describe an increase in the incidence of this complication over a 7-month period observed in our practice.

Methods

This study was a retrospective review of all patients who experienced intravitreal silicone oil droplets after intravitreal bevacizumab injections from a single practice (Northern California Retina Vitreous Associates, Mountain View, California) from October 1, 2015, to November 30, 2016. The El Camino Institutional Review Board approved the study with waiver of informed consent.

Bevacizumab was aliquoted into single-use doses (1.25 mg/0.05 mL) in insulin syringes with a 31-gauge needle (BD Insulin Syringe with BD Ultra-Fine Short Needle; Becton, Dickinson and Company) by a single compounding pharmacy (California Compounding Pharmacy).

The diagnosis of presumed intravitreal silicone oil droplets was made by observation of small, round, clear spheres in the vitreous cavity with biomicroscopy on dilated retinal examination (Figure 1). In some cases, the diagnosis was supported by a highly echogenic pattern on ultrasonography. The date of onset was presumed to be the most recent injection before onset of the symptoms or first detection on clinical examination, assuming that the silicone oil occurs at 1 time instead of in small increments of oil that are added after each injection and then diagnosed with the accumulated oil surpassing a threshold of clinical detectability.⁶ After the presence of intravitreal silicone oil was documented, subsequent injections of bevacizumab in that eye were excluded from further analysis and calculation of the incidence during this time period.

Figure 1. Appearance of Presumed Silicone Oil Droplets After Intravitreal Bevacizumab.

Slitlamp photography shows small, round, clear spheres (black arrows) in the vitreous cavity that are either solitary (A) or clusters (B). Fundus photography shows a large, clear sphere near the retinal surface (C). Scanning laser ophthalmoscopy shows a larger clear sphere superiorly in the vitreous cavity (D).

Statistical Analysis

The proportion of silicone oil droplet complications over time was analyzed with the Mann-Kendall trend test (using time series bootstrap). The Fisher exact test was used to compare the incidence over 7-month periods and differing intravitreal injection techniques. An unpaired, 2-tailed P value <.05 was considered significant. Calculations were conducted using R, version 3.2 (R Foundation for Statistical Computing).

Results

There were 60 cases (35 [58%] women) of presumed intravitreal silicone droplets identified over a 14-month period involving 6632 intravitreal bevacizumab injections. Mean (SD) age of the patients was 80 (12) years; the population comprised 48 white, 9 Asian, and 3 Hispanic patients. The incidence of the intravitreal silicone oil droplet per bevacizumab injections per month is shown in Figure 2; a rise in the incidence was observed over time (Mann-Kendall test, P = .001). The incidence of silicone oil droplet injections was 0.03% (1 of 3230) from October 2015 to April 2016, and 1.7% (59 of 3402) from May to November 2016 (Fisher exact test, P < .001; odds ratio [OR], 57; 95% CI, 9.8-2260). The mean (SD) number of prior bevacizumab treatments was 12 [9] over 21 (20) months, with 4 (7%) patients receiving their first bevacizumab injection during that period.

Figure 2. The Incidence of Silicone Oil Droplets After Intravitreal Bevacizumab From October 2015 Through November 2016.

There was a rise in the silicone oil droplets over time (Kendall test, P = .001).

Variation in the incidences among treating physicians raised the possibility that this complication could be related to injection technique. The only variation in technique that was identified was whether the syringe was primed before injection. Priming the syringe involves applying pressure to the plunger until efflux of the medication solution at the end of the needle tip is visible to ensure free flow of the medication once the injection is started. From May to November 2016, nonpriming the syringe before the intravitreal injection was associated with a higher risk of intravitreal silicone oil droplets compared with priming the syringe (6.4% [47 of 739 injections] vs 0.5% [12 of 2627]; Fisher exact test, P < .001; OR, 15.1; 95% CI, 7.9-33.4).

Among the 60 cases, 41 patients (68%) were symptomatic, and the main symptom was floaters with spots of light. Among the symptomatic patients, 36 (88%) reported improvement over time (range, 2-8 months) despite the silicone droplets still being present on ophthalmoscopic examination. No cases of associated inflammation, intraocular pressure rise, or corneal edema were observed. There were no cases of silicone oil involving 4544 and 1830 intravitreal injections of aflibercept or ranibizumab, respectively, during this same period.

Discussion

Small, spherical vitreous opacities have been reported following intravitreal injections of various medications and have been presumed to be silicone oil.^6,7,8 The source is thought to be polydimethylsiloxane, a lubricant used to reduce friction between the syringe barrel and plunger to permit smooth movement of the plunger within the barrel. Herein, we report an increase in the incidence of presumed intravitreal silicone oil droplets after intravitreal injection of bevacizumab from May to November 2016 in comparison with the previous 7 months. Other physicians nationally have reported similar findings. The compounding pharmacy denied any change in the preparation or handling of bevacizumab during the time in question. A change in the manufacturing of the insulin syringes, resulting in increased amounts or different behavior of the silicone, may explain the increased incidence that we noted. The manufacturer declined any comment when contacted.

Priming the syringe before injection was associated with a lower frequency of this complication. The needle may become clogged and priming the syringe before injection may reduce the force required to expel the contents. Without priming the syringe, more force is required, and this could also expel all of the residual silicone oil within the syringe during injection. The syringes used in this report have the staked-on design, which has been associated with a higher frequency of intravitreal silicone oil droplets than a luer cone design.⁶ It is suspected that, with the staked-on syringe design, the silicone oil is “squeegeed” from the inside of the syringe barrel and outside the plunger and is injected into the eye since there is no residual space in the syringe.⁶ Because silicone oil has a higher viscosity than the aqueous solution, greater injection force on a nonprimed syringe may increase the likelihood of silicone oil passage through the needle.

Most patients (68%) reported floaters with spots of light after experiencing this complication. Symptoms improved in many patients, with only 12% continuing to report the floaters. It is possible that the oil droplets migrated out of the visual axis or the patients adjusted to their presence over time. Longer follow-up is needed to assess the clinical significance of silicone oil toxicity, including intraocular pressure and corneal edema.

Limitations

One limitation of this retrospective study is possible underassessment; 32% of the patients who received the injections were asymptomatic. Other limitations include the possibility of the diagnosis being missed on examination, possible overestimation of the incidence since the droplets may be from earlier injections, and the lack of definite determination of the composition of the presumed silicone oil droplets.

Conclusions

We report an increase in the intravitreal silicone oil droplets associated with bevacizumab prepared with insulin syringes. Physicians should counsel their patients on the risk of floaters with intravitreal bevacizumab preloaded in insulin syringes. Using syringes that do not contain silicone oil for lubrication may minimize this complication. Further study of the source of this complication and its long-term effects is needed.