The Slower the Better: On the Instability of Gas Jets in a Model of Pneumatic Retinopexy

Abstract

purpose. To investigate the effect of injection technique parameters on the formation of multiple gas bubbles in a porcine eye model for pneumatic retinopexy.

methods. Three hundred twenty-four adult porcine eyes were injected with 0.4 mL of C3F8 with variations in the depth of injection, speed of injection, and size of needle bore. The number of gas bubbles in the eye was assessed with indirect ophthalmoscopy.

results. Shallow injections resulted in a higher incidence of a single bubble than did deep injections (P < 0.001; Fisher exact and Wilcoxon rank sum tests). Slow injections were significantly advantageous in producing a single gas bubble during shallow as well as during deep injections (P < 0.001, Fisher exact and Wilcoxon rank sum tests). With a shallow needle insertion, the slow speed of injection produced a single bubble in 75.9% of the eyes, whereas moderately brisk injections resulted in one bubble in 50.9% of the eyes. During deep needle insertion, 44.4% of the eyes had one bubble if the gas was injected slowly and only 11.1% had a single bubble with moderately brisk gas injections. The bore of the needle did not significantly change the number of bubbles during deep or shallow injections.

conclusions. The factors that were found to be important in reducing the formation of multiple gas bubbles in the eye were shallow depth of injection and slow speed of gas delivery.

Pneumatic retinopexy (PR) has been established as a treatment modality for rhegmatogenous retinal detachment (RD). This outpatient procedure consists of transconjunctival cryopexy with intravitreal gas injection, followed by head positioning. Alternatively, laser photocoagulation can be applied around the tear instead of cryopexy when the retina is attached 1 or 2 days after the initial gas injection. 

One potential complication of PR may occur if small bubbles (“fish eggs”) are formed in the vitreous cavity during gas injection. This occurrence may facilitate migration of one of the bubbles into the subretinal space. ^{1 2} Subretinal gas in the vicinity of the tear will prevent sufficient chorioretinal scarring, leading to the persistence of the RD. ³ In addition, the subretinal gas bubble can shift and detach more areas of the retina, including the macula. ⁴ Numerous bubbles in the vitreous cavity can prevent optimal visualization of the retina in cases in which laser retinopexy is planned after the gas injection. Current literature suggests performing a moderately brisk and shallow gas injection to avoid an abundance of bubbles. ^{5 6 7}  

We designed an experiment to test the effect of injection technique parameters on the formation of multiple gas bubbles in a porcine eye model. The effect of the depth of injection, speed of injection, and the needle bore on the formation of gas bubbles was studied. 

Methods

Three hundred fifty adult porcine eyes were purchased (Clougherty Packing Co., Los Angeles, CA). Of the 350 eyes, 26 were not used because of either corneal scars (22 eyes) or ruptured globe (2 eyes). Each study eye was mounted in an artificial head, and secured with pins to prevent movement. A volume of 0.4 mL perfluoropropane (C3F8) gas (Alcon Surgical, Fort Worth, TX) was placed in a 1.0-mL tuberculin syringe, fitted with a 25-, 27-, or a 30-gauge needle (BD Biosciences, Franklin Lakes, NJ). To make the injection site uppermost, the artificial head was placed supine with a 45° tilt. The needle was then passed into the eye perpendicular to the sclera, on the temporal side 3.0 mm posterior to the limbus. Injection of the entire volume of gas was performed, and the needle was withdrawn from the eye with the plunger held down. 

The depth of the needle during gas injection was approximately 3 mm (one-fourth needle length) for the shallow injections (after initial insertion of 6 to 8 mm to penetrate the anterior hyaloid) and 12 mm for the deep injections (full needle length). The speed of injection had two variations. Fast injections were performed in a moderately brisk fashion. The injection was given smoothly and quickly but not with excessive force. Slow injections, in contrast, were performed in a period of 8 seconds (timing by counting “one-one thousand, two-one thousand… ,”). Immediately after the gas injection, an anterior chamber paracentesis was performed with a 30-gauge needle mounted on a 1.0-mL syringe without a plunger, to reduce the pressure associated corneal edema. An indirect ophthalmoscope with a 20-D lens was used to assess the number of gas bubbles in the vitreous cavity. 

We performed a total of 324 gas injections: 216 shallow injections and 108 deep injections. Of the 216 shallow injections, 108 injections were performed slowly and 108 were delivered in a moderately brisk fashion. At each injection speed, 36 injections were made with each needle size (25-, 27-, and 30-gauge). Of the 108 deep injections, 54 injections were performed slowly and 54 were given briskly. For every injection speed, 18 injections were performed with each needle size (25-, 27-, and 30-gauge). 

Results

A summary of the results is presented in Table 1 . 

Shallow Versus Deep Injections

Among 216 shallow injections, one bubble formed in 137 (63.4%) eyes, two in 65 (30.1%), and three or more in 14 (6.5%; 13 eyes with three bubbles and 1 with four). Among the 108 deep injections, one bubble formed in 30 (27.8%) eyes, two in 39 (36.1%), and three or more in 39 (36.1%; 24 eyes with three bubbles, 6 with four, 3 with five, and 6 with six). Shallow injections resulted in a significantly lower incidence of multiple bubbles than did deep injections (P < 0.001, Fisher exact test), whether the injection was slow or brisk. 

When the number of bubbles was treated as a continuous variable, the average number was 1.4 ± 0.6 in superficial injections, compared with an average number of 2.4 ± 1.3 in deep injections (P < 0.001, Wilcoxon rank sum test). Based on the Poisson regression model, deep injections had a higher risk of formation of several bubbles than did superficial injections (relative risk [RR] = 1.65, 95% confidence interval [CI] = 1.39–1.95, P < 0.001). 

Fast Versus Slow Injections

During 216 shallow injections, 108 slow injections resulted in 82 (75.9%) eyes with one bubble and 26 (24.1%) with two. The 108 moderately brisk injections resulted in 55 (50.9%) eyes with one bubble, 39 (36.1%) with two, and 14 (13.0%) with three or more (13 eyes with three bubbles and 1 eye with four). Slow injections significantly reduced the occurrence of multiple gas bubbles compared with moderately brisk injections (P < 0.001, Fisher exact test). When treated as continuous, the average number of bubbles was 1.2 ± 0.4 for slow injections, compared with an average of 1.6 ± 0.7 bubbles for moderately brisk injections (P < 0.001, Wilcoxon rank sum test). In the Poisson regression model, brisk injections had a higher risk of more bubbles than did slow injections (RR = 1.31, 95% CI = 1.05–1.64, P = 0.017). 

During 108 deep injections, 54 slow injections resulted in 24 (44.4%) eyes with one bubble, 19 (35.2%) with two, and 11 (20.4%) with three or more (6 eyes with three bubbles, 2 with four, 2 with five, and 1 with six). The 54 fast injections resulted in 6 (11.1%) eyes with one bubble, 20 (37.0%) with two, and 28 (51.9%) with three or more bubbles (18 eyes with three bubbles, 4 with four, 1 with five, and 5 with six). Again, slow injections produced significantly fewer bubbles than did moderately brisk injections, even when the injection was deep (P < 0.001, Fisher exact test). When treated as continuous, the average number of bubbles was 1.9 ± 1.2 in slow injections, compared with an average of 2.8 ± 1.3 in moderately brisk injections (P < 0.001, Wilcoxon rank sum test). Based on the Poisson regression model, brisk injections had a higher risk of multiple bubbles than did slow injections (RR = 1.45, 95% CI = 1.13–1.86, P = 0.003). 

The Effect of Needle Gauge

Although there was a small trend for the formation of more gas bubbles with a higher needle gauge (Table 2) , we did not find a statistically significant difference in the occurrence of multiple bubbles (P = 0.381, Fisher exact test) and average number of bubbles (P = 0.132, Kruskal-Wallis test) among the three gauge levels during shallow injections. Compared with the 25-gauge needles, the 27-gauge needles had similar effect on the formation of multiple bubbles (RR = 1.06, 95% CI = 0.80–1.40, P = 0.670), and 30-gauge needles had a slightly higher but nonsignificant risk of more bubbles (RR = 1.17, 95% CI = 0.89–1.53, P = 0.268). 

During deep injections, we did not find statistically significant differences in the incidence of multiple bubbles (P = 0.448, Fisher exact test) and average number of bubbles (P = 0.130, Kruskal-Wallis test) among the three gauge levels. Compared to 25-gauge, 27-gauge needles had nearly the same effect on multiple bubbles (RR = 1.00, 95% CI = 0.73–1.37, P = 1.00), and 30-gauge needles had a slightly higher but nonsignificant risk of forming more bubbles (RR = 1.23, 95% CI = 0.91–1.65, P = 0.176). 

Discussion

The technique for PR has been discussed by several authors. Text book teachings instruct that a moderately brisk and shallow gas injection be performed ^{5 6} with a 30-gauge needle. ⁵ Aylward and Lyons ⁷ showed that deep injections were more likely to produce multiple gas bubbles in a glycerol-filled chamber. In a Newtonian mathematical model, they proposed that a faster injection is preferable for achieving a single bubble. 

Our results confirm that deep injections significantly increase the risk of formation of multiple bubbles. This phenomenon can be understood if we consider the buoyancy force that is working to lift the bubble away from the tip of the needle. The forces that counteract the buoyancy of the gas are mainly the surrounding forces of the vitreous gel and inner ocular surface. ⁷ During shallow injections, the top surface of the bubble quickly reaches the inner surface of the eye, keeping it immobile around the needle’s tip. If the needle is deep within the vitreous, the forming bubble will be more likely to reach a point where the buoyancy of the gas overcomes the forces that keep it in contact with the needle. In this case, the bubble will disconnect and float to the surface, allowing additional bubbles to form during the continuing injection process. 

In contrast to the literature, our results showed that faster injections were more likely to produce an abundance of bubbles. This statistically significant difference was evident in both shallow and deep injections (P < 0.001, Fisher exact and Wilcoxon rank sum tests). We believe that fluid dynamics holds the explanation of our findings. 

The flow of one fluid in another fluid (where the fluids can be liquid or gas) has been studied for more than a century. In 1879 Lord Rayleigh published his work on the phenomenon of the capillary instability of jets. ⁸ He explained that a liquid injected through a capillary breaks into bubbles by the action of a capillary force (Fig. 1A) . More recently, Funada and Joseph ⁹ provided a comprehensive analysis for capillary instability. To calculate the capillary force, their solution accounts for the fluid’s velocity and viscosity and the capillary’s radius. The source of this force is in perturbations that form at the tip of the capillary (needle), and it works to break up the column of gas into small partitions during fast injections (Fig. 1B) . This phenomenon plays a role in many common processes (e.g., ink-jet printing, jet cutting, and fuel injection). 

Another form of instability that plays a role during brisk gas injections into fluid is known as the Kelvin-Helmholtz instability. This velocity-driven disturbance is created by sheering forces along the interface between the gas and the liquid, leading to turbulent flow. ¹⁰ An example of this phenomenon is seen when wind blowing over water causes the water surface to undulate. Kelvin-Helmholtz instability creates turbulence that forms a vortex in the ambient fluid during brisk injections. ¹¹ When the vortex is given a small perturbation, the disturbances amplify to nonlinear high-amplitude waves that break up the column of gas into small partitions (Fig. 2) . ¹²  

We found a slight, but nonsignificant, trend for the formation of more gas bubbles with a higher needle gauge. It is logical to assume that forcing a constant volume through a thinner needle would produce a faster jet and thus increase the likelihood of more bubbles. However, the moderately brisk injections were not performed with force. Therefore, it is also possible that slight accommodation of injection pressure by the surgeon plays a role in decreasing the effect of smaller needle bore on the formation of more bubbles. 

Our model is not an ideal simulation of pneumatic retinopexy because the porcine vitreous is not identical with that in the human eye. First, there is no posterior vitreous detachment in the porcine eye. Second, the concentration of vitreous fibers in the porcine eye is not identical with that of the human eye. It is therefore possible that different forces influence the gas jet in humans with retinal detachment. In addition, gas does not expand in the cadaveric eye, and conclusions on the long-term status of the bubble should be drawn with caution. 

In conclusion, the factors found to be important in reducing the formation of multiple gas bubbles in a porcine eye model for pneumatic retinopexy were depth of injection and speed of injection. Our results favor the practice of slow and shallow injections. A moderately brisk injection may increase the likelihood for the formation of multiple intraocular gas bubbles. 

 Submitted for publication November 16, 2006; revised January 7, 2007; accepted March 29, 2007.

 Disclosure: D.H. Bourla, None; A. Gupta, None; J.P. Hubschman, None; N. Bourla, None; F. Yu, None; S.D. Schwartz, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Summary of Results

	Shallow Injection (n = 216)			Deep Injection (n = 108)
Depth of injection
Eyes with one bubble (%)	63.4			27.8
Eyes with two bubbles (%)	30.1			36.1
Eyes with more than two bubbles (%)	6.5			36.1

	Shallow			Deep
		Slow (n = 108)	Brisk (n = 108)	Slow (n = 54)	Brisk (n = 54)
Speed of injection
Eyes with one bubble (%)	75.9	50.9	44.4	11.1
Eyes with two bubbles (%)	24.1	36.1	35.2	37.0
Eyes with more than two bubbles (%)	0.0	13.0	20.4	51.9
Number of bubbles (mean ± SD)	1.2 ± 0.4	1.6 ± 0.7	1.9 ± 1.2	2.8 ± 1.3

Table 2.

 

Summary of Results for Various Needle Gauges

	Shallow Injections											Deep Injections
		Slow					Brisk					Slow					Brisk
Needle gauge	25	27	30	25	27	30	25	27	30	25	27	30
Bubbles (mean ± SD)	1.2 ± 0.4	1.2 ± 0.4	1.3 ± 0.5	1.5 ± 0.7	1.6 ± 0.7	1.8 ± 0.8	1.8 ± 1.0	1.7 ± 1.1	2.2 ± 1.4	2.6 ± 1.3	2.7 ± 1.4	3.2 ± 1.3

Figure 1.

 

Capillary instability. (A) High-speed photograph showing the breakup of a fluid column after it was injected through a capillary. (B) Schematic diagram of the capillary force that causes collapse of the fluid. Reprinted, with permission, from Funada T, Joseph DD. Viscous potential flow analysis of capillary instability. Int J Multiphase Flow. 2002;28:1459–1478. © 2002 Elsevier.

Figure 2.

 

Kelvin-Helmholtz instability with Rayleigh’s capillary force. Vortices are formed when two fluids are moving in different speeds due to sheering forces between the fluids. Capillary perturbations induce an array of spiral vortices. Illustration by David Le Beck.

The authors thank Chih-Ming Ho, Nicolas Bremond, and Daniel Joseph for their gracious help in outlining the physical principles behind the results.