Article Text

Download PDFPDF

Release of silicone oil and the off-label use of syringes in ophthalmology
  1. Gustavo Barreto Melo1,2,
  2. Geoffrey Guy Emerson3,
  3. Celso Souza Dias Jr1,
  4. Fábio Barreto Morais1,2,
  5. Acacio de Souza Lima Filho2,4,
  6. Shoko Ota5,
  7. Michel Eid Farah2,4,
  8. Eduardo Büchele Rodrigues2,4,
  9. Maurício Maia2,4,
  10. Rubens Belfort Jr2,4
  1. 1 Hospital de Olhos de Sergipe, Aracaju, Brazil
  2. 2 Ophthalmology, Federal University of São Paulo, São Paulo, Brazil
  3. 3 Retina Center of Minnesota, Minneapolis, Minnesota, USA
  4. 4 Vision Institute, IPEPO, São Paulo, Brazil
  5. 5 Chemical Analysis Laboratory, Center for Chemistry and Manufactured Goods, Institute for Technological Research, São Paulo, Brazil
  1. Correspondence to Dr Gustavo Barreto Melo, Hospital de Olhos de Sergipe, Aracaju 49020-380, Brazil; gustavobmelo{at}


Background/aims To assess silicone oil (SO) release by different brands of syringes used for intravitreal injection under different handling conditions.

Methods Eight syringes were analysed: from the USA, Terumo 0.5 mL, Becton-Dickinson (BD) Tuberculin 1 mL, BD Luer-lok 1 mL, BD Ultra-Fine 0.3 mL and Exel Insulin 0.3 mL; from Germany, Braun Omnifix-F 1 mL and Braun Injekt-F 1 mL and from Spain, BD Plastipak 1 mL. The impact of air, priming the plunger, agitation by flicking and fluid temperature on SO release were assessed by light microscopy. Fourier transform infrared spectroscopy (FTIR) was performed to identify the molecular compound in each syringe.

Results Five hundred and sixty syringes were analysed. Terumo 0.5 mL and BD Ultra-Fine 0.3 mL released more SO than all others. BD Luer-lok 1 mL, BD Plastipak and Braun Omnifix-F 1 mL released little SO; BD Tuberculin 1 mL, Exel 0.3 mL and Braun Injekt-F 1 mL released the least SO. Priming the syringe and different temperatures did not significantly affect SO release. Agitation by flicking caused a significantly higher proportion of samples to have SO droplets and an increased number of oil droplets. Air had an additive effect on the release of oil in the agitation groups. FTIR identified polysiloxane in all syringes but Injekt-F.

Conclusion Syringes commonly used for intravitreal injections frequently release SO droplets, especially when agitated by flicking. To avoid unnecessary ocular risks, syringes should not be agitated before intravitreal injection. It is desirable that syringes be manufactured specifically for ophthalmic use.

  • syringe
  • intravitreal injection
  • silicone oil droplets

Statistics from


The off-label use of syringes in ophthalmology is commonplace. Since antivascular endothelial growth factor agents are effective for treating age-related macular degeneration,1 the number of intravitreal injections and their use has increased dramatically.

The manufacturing process of most commercially available syringes relies on siliconisation of the inner surface of the syringe barrel, which facilitates reduced force to initiate the movement of the plunger and its subsequent gliding.2 3

Silicone oil (SO) droplets in the vitreous have been identified in clinical studies.4–7 Bakri and Ekdawi4 identified 15 eyes with presumed SO droplets after 1529 injections. Khurana et al 5 estimated that the incidence rates of presumed SO droplets in the vitreous cavity after intravitreal bevacizumab (Avastin, Genentech, South San Francisco, California, USA) injections using insulin syringes range from 0.03% (3230 injections) to 1.7% (3402 injections) at different time periods. Our group found SO droplets, regardless of their size and clinical impact, in the vitreous of 68% and 75% of 37 consecutive eyes treated with intravitreal injections when assessed by slit-lamp and ultrasound examination, respectively (unpublished data). We also found that the greater the number of previous injections, the higher the likelihood was of having more SO in the vitreous when the ultrasound scans were analysed using area measurement software.

The presence of SO droplets in the vitreous might become symptomatic, leading to the complaint of floaters, and sometimes require vitrectomy with associated risks.8

Considering that syringes are the most likely source of the SO droplets in the eyes of subjects undergoing intravitreal injections,2 the purpose of the current study was to assess the release of SO from different brands of syringes often used for intravitreal injections under varying handling conditions.

Materials and methods

Eight models of syringes were analysed for the release of SO. From the USA, The Terumo 0.5 mL Insulin Syringe (Terumo Medical Corporation, Elkton, Maryland, USA, lot #RN1837, ref #SS*05M3010M) (herein referred to as Terumo), the Becton-Dickinson (BD) 1 mL Tuberculin Slip Tip Syringe (Becton-Dickinson and Co., Franklin Lakes, New Jersey, USA, lot #8213918, ref #309659) (herein referred to as BD Tuberculin), the BD 1 mL Luer-lok Tip Syringe (lot #8173961, ref #309628) (herein referred to as BD Luer-lok), the BD Ultra-Fine 0.3 mL Short Needle with half-unit-scale (lot #7100846, ref #328440) (herein referred to as BD Ultra-Fine) and the Exel Comfort Point 0.3 mL Insulin Syringe (Exelint International Co., Redondo Beach, California, USA, lot #170905, ref #26010) (herein referred to as Exel); from Germany, the Braun Omnifix-F 1 mL Luer Solo Syringe (B Braun Melsungen AG, Melsungen, Germany, lot #18B12C8, ref #9161406V) (herein referred to as Omnifix-F) and the Braun Injekt-F 1 mL Luer Solo Syringe (B Braun Melsungen AG, lot #18G30C8, ref #9166017V) SO-free (herein referred to as Injekt-F) and from Spain, the BD Plastipak 1 mL Syringe (Becton-Dickinson SA, Madrid, Spain, lot #1802003, ref #303172) (herein referred to as BD Plastipak).

The Terumo, BD Ultra-Fine and Exel syringes have 31, 31 and 29-gauge staked-in needles, respectively. All others were tested with an attachable 26-gauge BD PrecisionGlide needle (Becton-Dickinson Ind Cirur LTDA, Curitiba, Brazil, lot #7234884, ref #300110). The impact of air, priming the plunger, agitation by flicking and temperature of the fluid (25°C in most samples and 8°C when stated) on SO release and the difference between the first and last drops ejected from the syringe were assessed by light microscopy.

All syringes were tested under seven different conditions: positive control (fluid with addition of SO) with agitation (group 1, n=10); fluid only (group 2, n=10); fluid, air and priming (group 3, n=10); fluid and agitation (group 4, n=10); fluid, air and agitation (group 5, n=10); fluid, air, priming and agitation (group 6, n=10) and fluid at 8°C, air, priming and agitation (group 7, n=10).

Syringe preparation

The syringes generally were prefilled with distilled water to 0.06 mL (backfilled either via their detachable needle or their own preattached needle). Air, when stated, also was aspirated up to 0.04 mL to facilitate injection of the entire amount of fluid, which prevents fluid retention in the dead space of the syringe.

The samples in groups 2 and 4 required an additional volume of water to facilitate ejection of 0.06 mL of fluid. The volume of dead space varied among the different syringe models. In these groups, a total of 0.14–0.16 mL was drawn and no air was allowed.

After drawing the fluid and air, when indicated, the syringe initially was kept upright in all groups with the needle side facing up. In groups 1, 4, 5, 6 and 7, the syringes were agitated with five consecutive flicks (ie, sudden release of a bent finger, causing a sharp motion). Subsequently, the syringe was turned 180 degrees so that the needle faced down. The syringes then were flicked 10 consecutive times. In groups 2 and 3, only gentle movement was applied whenever needed to allow for complete fluid ejection. Priming, whenever stated, comprised a forth and back movement of the plunger in order to supposedly allow for a better gliding. One investigator (GBM) performed the previous steps in all syringes to avoid interexaminer variability.

In group 7, fluid was drawn at a controlled temperature (8°C as measured by a mercury thermometer). For the other groups, fluid was drawn at room temperature (25°C).

Microscopic examination of syringe fluid

Ten drops of fluid were ejected sequentially from each syringe onto glass slides (Yancheng Huida Medical Instruments Co., Jiangsu, China). The slides were viewed using light microscopy (Nikon Eclipse 200, Shanghai, China) and examined for oil droplets and photographed at 100× optical magnification (iPhone 6S mounted on the microscopy stage) and an additional 50% digital magnification. One image that captured the largest number of SO droplets from the first and last drops of fluid per syringe was obtained at the same magnification, yielding two images per sample.

The syringes were challenged with 1 mm droplets of 1300-cS SO (RS-Oil ECS, Alchimia, Ponte San Nicolò, Italy, lot #P17090022, ref #RSO 010–00) added to the tip of the syringe plunger to serve as positive controls. The grader also was instructed on identification of the SO and air. The former usually appears as a transparent, round shape with a thin border; the latter, which commonly is seen as a cluster, usually appears as a round shape with a double grey ring that is easily distinguishable from the former (figure 1).

Figure 1

Silicone oil (SO) droplets and air bubbles (arrow) in an agitated sample from a Terumo syringe. (A) Clustered SO droplets with a thin border; (B) seven air bubbles surrounded by oil droplets and (C) SO droplets at a different depth (out of focus).

One masked grader evaluated the samples (first and last drops from each syringe) microscopically for the presence or absence of SO, based on the criteria described previously. Whenever SO was identified, the number of drops was counted on each image. If any image from the syringe had SO droplets, regardless of the number, it was considered as a positive result for that syringe.

Fourier-transform infrared spectroscopy (FTIR)

All eight models of disposable, individually and hermetically packed syringes were studied. After the plunger was removed, the rubber on the tip was rubbed gently onto a cesium iodide crystal to transfer the possible oil content from one to another. Since the Injekt-F syringe does not have a rubber tip, the inner tip of the plunger was rubbed onto the crystal. Identification was performed using the FTIR Nicolet iS10 (Thermo Electron Scientific Instruments LLC, Madison, Wisconsin, USA), with bands ranging from 4000/cm to 400/cm, resolution of 4/cm and 32 scans.2 9

Statistical analysis

Statistical analyses were performed using SPSS V.20.0 and STATA V.12 software. The data are expressed as the percentages of syringes detected with SO. The mean numbers of SO droplets and SD also are reported. For quantitative analysis, one image of the first and one of the last drops were used for each syringe. The χ2 and Fisher’s exact tests compared the nominal variable, that is, the presence of oil, among the different brands of syringes. A logistic regression model was applied for the presence of oil droplets (dependent variable) versus conditions (ie, groups) and syringe models (predictive variables), yielding ORs. The Wald test was performed to directly compare the ORs of the two groups. The mean numbers of SO droplets among the groups and syringes were assessed using the Kruskal-Wallis non-parametric test. The Dunn-Bonferroni’s post hoc test was performed on each pair of groups. Poisson’s regression model was used to adjust for the simultaneous effect of both the syringe model and condition on the number of oil droplets. To compare the number of SO droplets in the first and last drops, the Wilcoxon test was applied, and a multiple linear regression model was used to adjust for the negative values in this comparison. A p value of 0.05 was considered statistically significant.


A total of 560 syringes were assessed; 70 from each of eight syringe models and 10 samples for each of the seven conditions tested.

Qualitative analysis by light microscopy

The presence of SO by syringe type and condition is shown in table 1. The Terumo syringes released oil in 95.7% of 70 samples, regardless of the conditions (p=0.367). The BD Ultra-Fine, BD Luer-lok, BD Plastipak and Omnifix-F syringes were 100% positive under the following conditions: groups 1 (positive control), 5 (fluid, air and agitation), 6 (fluid, air, priming and agitation) and 7 (fluid at 8°C, air, priming and agitation). The Exel, BD Tuberculin and Injekt-F syringes had 100% positivity only for syringes from the positive control group. Excluding the positive controls (group 1), the Injekt-F, Exel and BD Tuberculin syringes released oil in a low percentage of samples that was similar across all conditions (p=0.412, p=0.533 and p=0.063, respectively).

Table 1

Qualitative data for each condition and syringe

The results of the logistic regression model for the presence of oil droplets as a dependent variable versus condition (groups) and syringe models as predictive variables are shown in table 2 and both were significant (p<0.001). With group 2 as a reference (fluid only), the OR for the presence of oil in group 4 (fluid and agitation) was 20.33 (95% CI 6.89 to 59.99), after being adjusted for the syringe model. The ORs were 86.58 (95% CI 26.34 to 284.53), 61.28 (95% CI 19.13 to 196.34) and 61.28 (95% CI 19.13 to 196.34) for groups 5, 6 and 7, respectively. No significant difference was found for group 3 (p=0.793). As expected, the likelihood of finding oil in the positive control samples was significantly (p<0.001) higher than in the reference group. The Wald test showed that fluid, air and priming (group 3) had a lower likelihood of being positive compared with fluid, air, priming and agitation (group 6, p<0.001). Moreover, the likelihood was greater with fluid, air and agitation (group 5) than with fluid and agitation (group 4), indicating that the presence of air might play an additive role in the release of SO by the syringes. However, no differences were seen between groups 6 and 7, which assessed the effect of room temperature versus 8°C (p=1.000).

Table 2

Estimates of the logistic regression model

When the syringe models, adjusted by the condition, were appraised, the chances of finding SO in Exel (OR 4.78; 95% CI 1.32 to 17.33) and BD Tuberculin (OR 5.37; 95% CI 1.49 to 19.32) syringes was similar (p=0.810) but higher than in the oil-free Injekt-F syringe (reference category). The BD Plastipak, BD Luer-lok and Omnifix-F syringes had a similar (p=0.310) and higher likelihood of having SO. The Terumo syringe had the highest OR for the presence of SO, followed by the BD Ultra-Fine syringe. As expected, the SO-free Injekt-F syringe had the lowest ORs for the presence of SO among the tested models.

Quantitative analysis by light microscopy

The descriptive quantitative data for all syringes and groups are shown in table 3. Although some variations occurred in the positive control among the syringes, the mean number was within the expected range. Even though most syringes had droplets in most conditions, it is notable that only the BD Ultra-Fine and Terumo syringes had high mean numbers of droplets in the agitation groups.

Table 3

Descriptive quantitative data presented for each condition and syringe

The quantitative analysis also compared the differences between the first and last drops ejected from the syringes. Table 4 shows that the mean number of oil droplets was significantly higher in the last drops than in the first drops for half of the syringes and conditions, except for the Injekt-F (p=0.414), BD Tuberculin (p=0.507), BD Plastipak (p=0.432) and Exel (p=1.000) syringes. When Poisson’s regression model was used, similar results were found (data not shown), indicating that it is more likely to find SO at the end of the ejection procedure.

Table 4

Quantitative data of the first and last drops of fluid ejected onto the slide*

Fourier-transform infrared spectrometry

Analysis of the material extracted from the inner tip of the plunger indicated the presence of polysiloxane (SO) in all syringes, except for the Injekt-F syringe (figure 2). The following characteristic bands were found in those syringes: 2960/cm, stretching vibrations of CH3 in Si–CH3; 1260/cm, bending vibrations of CH3 in Si–CH3; 1090/cm and 1020/cm, asymmetric stretching vibrations of Si–O–Si; 800/cm and stretching vibrations of Si–C in Si–CH3.9

Figure 2

Fourier-transform infrared spectroscopy graph shows the characteristic bands of polysiloxane (silicone oil) in all syringes, except for the oil-free Injekt-F syringe. (A) Becton-Dickinson (BD) Plastipak (Spain); (B) BD Tuberculin (USA); (C) BD Luer-lok (USA); (D) BD Ultra-Fine (USA); (E) Terumo (USA); (F) Exel (USA); (G) Omnifix-F (Germany) and (H) Injekt-F (Germany).

The Exel syringe had an additional band in the 1740/cm region, which is typical of the stretching vibrations of C=O of the esters of carboxylic acid at small proportions. No bands were found in the oil-free Injekt-F syringe, meaning that no SO coating was detected on its inner surface.10


Case series have reported the presence of persistent droplets in the vitreous of subjects after intravitreal injections and postulated the droplets to be SO lubricant from the syringe used in the procedure.4–7 This current study identified SO in and released by various syringes used for intravitreal injections and highlighted the impact of agitation by flicking on SO release. Additionally, priming the plunger, different temperatures and the presence of air bubbles (without agitation) had little impact on the release of SO by the syringes. However, agitation by flicking had a tremendous impact. In the agitation groups, the presence of air also showed an incremental effect on the release rate. Some syringes tested, including the BD Tuberculin 1 mL, Exel 0.3 mL and Braun Injekt-F 1 mL syringes, released less SO compared with other models; however, we did not verify whether these differences were consistent across multiple lots, that is, there could have been lot-to-lot variations in the safety of these syringes.

Our previous study also showed similar results when two syringes produced in Brazil (SR, Manaus, Brazil and BD Plastipak, Curitiba, Brazil) and one US-manufactured syringe (BD SafetyGlide, Holdrege, Nebraska, USA) were tested. Both the SR and BD SafetyGlide syringes had an incremental effect on the release of SO when agitated by flicking. Only the BD Plastipak syringe did not release oil. However, the sample size was smaller than the one in the current study.11

It should be remarked that the syringes were chosen for the current analysis because anecdotal information led the authors to believe they comprised some of the most commonly used models in some European countries and in North America. BD Tuberculin, BD Ultra-Fine, Exel and Braun Injekt-F are often sold with bevacizumab by compounding pharmacies in the USA. BD Luer-Lok is distributed in the same package as aflibercept both in Europe and in the USA. Braun Omnifix-F is distributed in the same package as ranibizumab in Brazil. Both the BD Ultra-Fine and Terumo syringes, although USA-made, are largely found and used in Brazil as well. BD Plastipak is largely available in Europe.

FTIR confirmed that there was no evidence of SO on the plunger tip of the oil-free Braun Injekt-F syringe (figure 2). However, we identified SO droplets in three of 60 samples (excluding the positive controls) from the oil-free syringe (table 1), which was unexpected and prompted additional tests. We reasoned that the BD PrecisionGlide needle (used for the entire study, except for those syringes with staked-in needles) was the potential source of SO. We submitted the 100 BD PrecisionGlide needles to FTIR, which showed an average of 20 µg of SO in each sample, which is a small amount relative to the siliconised syringes in this study but sufficient to explain the positive samples for our oil-free syringe (Braun Injekt-F).12

Although SO droplets in the vitreous are often harmless, when the increasing number of intravitreal injections is considered, the number of subjects with clinically relevant SO droplets also may increase. Further complications are possible when vitrectomy is needed to remove the SO droplets.

In addition, retina specialists worldwide are concerned about inflammation after intravitreal injections of antiangiogenics without a clear aetiology. Non-infectious vitritis has been reported in 0.10% after 66 356 bevacizumab injections, 0.02% after 26 161 ranibizumab (Lucentis, Genentech) injections and 0.16% after 8071 aflibercept injections (Eylea, Regeneron Pharmaceuticals, Tarrytown, New York, USA).13 The American Society of Retina Specialists Therapeutic Surveillance Committee also reported notifications of cases of aflibercept-related sterile inflammation. However, the authors did not suggest an explanation.14 More recently, our group identified six cases of inflammation after aflibercept injections with a new syringe in use at the clinical practice.15 The syringe was the SR, which also has been shown to release SO with or without agitation. Since one author flicked this syringe to dissociate the fluid from the air and other retina specialists at the same setting did not and had no cases of inflammation, it was speculated that release of SO droplets may contribute to the inflammatory reaction. In agreement with this hypothesis, some studies have reported more intense protein aggregation and insoluble molecules resulting from agitation in the presence of the SO from the syringes.16–18 Finally, SO droplets also have been reported to act as immunologic adjuvants in dermatologic procedures in the subcutaneous space in mice.19 20 We believe the risks of SO droplets introduced into the eye or other areas of the body are relevant to ophthalmologists and clinicians in other medical fields.

In conclusion, syringes commonly used for intravitreal injections frequently release SO droplets, especially when agitated by flicking. To avoid unnecessary ocular risks, we recommend that the syringes not be agitated at the time of intravitreal injection. In addition, since their use in ophthalmology is off-label, syringes for specific ophthalmic use should be manufactured.


The authors are grateful to Policlínica Santa Maria for providing the light microscope used in this study. They also acknowledge Pine Pharmaceuticals for donating the Exel syringe used in the study.



  • Contributors GBM had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. GBM, GGE, ASL, MM, EBR contributed to the study concept and design. RBJr, CSDJr, MEF, GBM, FBM, SO, GGE and ASL contributed to the acquisition, analysis or interpretation of data. GBM and GGE drafted the manuscript. All authors contributed to the critical revision of the manuscript for important intellectual content and the final approval.

  • Funding This study was supported by EyePharma (São Paulo, Brazil), FAPESP (São Paulo, Brazil), CNPq (Brasília, Brazil) and Pan-American Association of Ophthalmology/Pan-American Ophthalmological Foundation, Paul Kayser/Retina Research Foundation Global Award (Pan-American Association of Ophthalmology/Pan-American Ophthalmological Foundation, Arlington, Texas, USA). None of the funders/sponsors had any role in design and conduct of the study; collection, management, analysis and interpretation of the data; preparation, review or approval of the manuscript and decision to submit the manuscript for publication.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • At a glance
    Keith Barton Jost B Jonas James Chodosh