SUPPLEMENTAL DEXTROSE IN THE INFUSION FLUID DURING DIABETIC VITRECTOMY

Abstract

Purpose:

Historically, supplemental dextrose to infusion fluid has been used to reduce the need for intraoperative lensectomies to maintain visualization during diabetic vitrectomy. Valved, small-gauge vitrectomy has reduced surgical time and decreased intraoperative fluid flow. Assessment of supplemental dextrose in modern vitrectomy is presented in this study.

Methods:

A retrospective cohort study of diabetic patients undergoing vitrectomy was conducted. The dextrose group received supplemental dextrose in the infusion fluid, while the nondextrose group used a standard balanced salt solution (BSS Plus). Group assignment was per surgeons’ typical practice patterns. Eyes with tractional retinal detachments were also evaluated as a subgroup.

Results:

Three hundred thirty phakic eyes were included. Supplemental dextrose was used in 199 eyes (60.3%). One unplanned lensectomy was performed in this series, in the nondextrose group, not statistically different from the dextrose group, with zero lensectomies (P = 0.4). Cataract survival curves overlapped for all eyes and for the tractional retinal detachment subgroup.

Conclusion:

In modern vitrectomy, unplanned lensectomy is rare. No difference was observed in the rate of intraoperative lensectomies or overall postoperative cataract course with or without dextrose supplementation to the infusion fluid. Standard solutions appear to be adequate for infusion, even for diabetics.

During pars plana vitrectomy, infusion fluids maintain globe stability and intraocular pressure through the infusion of fluids as vitreous is extracted. These infusion fluids irrigate the retina and posterior lens capsule and, thus, must be safe for all intraocular structures for the duration of surgery. Designed by several pioneers in the field, the balanced salt solution (BSS) plus contains a mix of saline, glutathione, calcium, bicarbonate, and glucose.¹ During vitrectomy, avoidance of lens and corneal opacification is essential to allow visualization to the fundus and successful completion of surgical maneuvers. Unplanned lensectomy during vitrectomy renders the patient aphakic and creates a unicameral eye, possibly increasing the risk for neovascular glaucoma.²

Previous animal and human research has shown a marked increase in glucose, sorbitol, and fructose concentrations in the lenses of diabetic eyes.^3,4 The excess sugar accumulation suggests that standard BSS Plus, while approximating the aqueous of a normoglycemic patient, is relatively hypotonic to the diabetic lens. If the diabetic lens is “hypertonic” to the infusion solution, it may hydrate and become opacified.

Haimann et al³ found, in a rabbit model, that supplementing dextrose to BSS Plus avoided this lens opacification in diabetic eyes. Standard BSS Plus has an osmolarity of 305 mOsm, while fortification with 3 mL of 50% dextrose results in an osmolarity of 320 mOsm.⁵ The proposed mechanism for lens protection is that the higher osmolarity of the fortified infusion fluid may more closely parallel that of the lens, which would not hydrate and thus retain its clarity. The animal model was followed by a randomized controlled trial, published in 1984, showing that the addition of 3 mL of 50% dextrose to BSS Plus reduced the need for intraoperative lensectomy during vitrectomy for diabetic tractional retinal detachments from 30% to 0%. Of note, the surgeries in Haimann and Abrams⁵’s study were 20-gauge vitrectomy, and the average surgical time was two or more hours. Subsequent to the publication of Haimann and Abrams’s research, supplementation of dextrose in the infusion fluid became standard protocol during diabetic vitrectomy in some centers, with Cherfan et al⁶ even studying this approach, “as was customary for patients with diabetes” in nondiabetics.

Over the past 30+ years, surgical techniques have improved, along with advances in instrumentation and visualization. In modern, small-gauge, valved, sutureless vitrectomy, surgical times are often shorter, and control of fluidics has improved, with less time for irrigation contact and less volume of fluid circulating into the eye. The hyperglycemic irrigating solution has also been evaluated in the nondiabetic population, with no difference in postoperative lens changes compared with the normoglycemic infusion.⁶ The role of supplemental dextrose in modern diabetic vitrectomy is evaluated in this study.

Methods

Institutional Review Board approval was obtained from the University of Wisconsin. This study adhered to the tenets of the Declaration of Helsinki. This retrospective cohort study evaluated diabetic patients undergoing vitrectomy for any indication from October 2013 to October 2021; eyes with a tractional retinal detachment were identified as a subset. The dextrose group was defined as having supplementation of 3 mL of 50% dextrose added to the infusion, while the non-dextrose group used standard BSS Plus. Group assignment was per surgeons’ typical practice patterns, whereas some surgeons routinely requested the addition of dextrose in phakic, diabetic patients. Administration of the supplemental dextrose was confirmed by chart review. Inclusion criteria included postoperative follow-up of at least 90 days, as well as a phakic lens status at the time of surgery.

If both eyes met the inclusion criteria, each was included for analysis. As bilateral surgery was not temporally related (i.e., mean one year between eyes) and many had divergence in dextrose supplementation, bilateral cases were analyzed independently and not clustered.

Means are presented with SD and compared with the student’s t-test, with statistical significance set at 0.05. Interquartile range (IQR) is also reported in some instances for further context of data variability. Fisher’s exact test was used to compare categorical variables.

Duration to cataract surgery after vitrectomy between the dextrose group assignment was assessed by Kaplan–Meyer curves. A two-sided nonparametric weighted log-rank test was used to assess for intersecting hazards between groups,⁷ and a one-sided test was used to assess for time-dependent superiority between groups.⁸ With evidence of nonproportional hazards from these tests, a parametric accelerated failure time regression was fit to model time to cataract extraction given the perioperative covariates of age, duration of vitrectomy, fasting glucose, and hemoglobin A1c. The log-logistic parametric distribution was chosen on the basis of optimal model fit (Akaike information criterion). Interaction effects between perioperative covariates and the dextrose group indicator were sequentially tested and chosen in accordance with optimal model fit criteria. The interaction between fasting glucose and dextrose produced the best candidate model and was selected as the final model. Last, an adaptive index procedure⁹ was fit using the final parametric model to better understand the interaction between fasting glucose and dextrose. This is a data-driven approach for estimating a binary decision rule to stratify population risk.

Results

The medical record was reviewed of 786 eyes of diabetic patients undergoing vitrectomy. Pseudophakia was noted in 221 eyes, aphakia in 17, and anterior segment pathology obscuring view to the lens in two eyes; these eyes were excluded. Overall, 157 cases were missing pre- or postoperative data and were excluded. The remaining were 389 preoperatively phakic eyes; 59 eyes had planned cataract extraction at the time of vitrectomy and were removed for further analysis. The resulting 330 eyes (50 bilateral) were evaluated.

Supplemental dextrose was used in the infusion fluid in 199 eyes (60.3%). Baseline characteristics are listed in Table 1. Median follow-up was 2.6 years (interquartile range 1.3–4.5 years). Annual sample sizes in this study were 330 eyes at baseline, and 180, 89, 46, 27, 20, and <10 eyes for years 1 to 6+, respectively.

Table 1.

Baseline Characteristics of Preoperative Variables of Dextrose- and Non-dextrose-Supplemented Groups

	Dextrose n (%)	No Dextrose n (%)	P
N	199 eyes	131 eyes
	167 patients	113 patients
Sex (male)	95 (56.9%)	61 (54.0%)	0.63
Diabetes management
Insulin	106 (63.5%)	74 (65.5%)	0.73
Oral medication	101 (60.5%)	59 (52.2%)	0.17
	Mean (SD)	Mean (SD)
Age	53.9 (13.4)	55.1 (13.1)	0.40
Weight (kilograms)	93.4 (24.9)	98.7 (30.0)	0.13
Duration of diabetes (years)	19.5 (11.0)	20.5 (11.3)	0.58
Hemoglobin A1c	8.1 (2.0)	8.2 (1.9)	0.71
Fasting glucose	157.2 (66.9)	169.3 (62.1)	0.13
Preoperative visual acuity
logMAR	1.3 (0.8)	1.2 (0.8)
Snellen	20/400	20/315	0.30

logMAR, logarithm of minimum angle of resolution.

Mean surgical time was 59.8 (SD 36; interquartile range 35.3–76) minutes with no difference between dextrose groups (P = 0.3). Only one unplanned lensectomy was performed during vitrectomy in this series, in the non-dextrose group (0.7%), not statistically different from the dextrose group, with zero unplanned lensectomies (0%; P = 0.4). Between groups, a similar ratio remained phakic at the last follow-up (37.7% for the dextrose group; 34.4% without dextrose; P = 0.6). Use of long-term tamponade (gas or silicone oil) was not significantly different between groups (63.3% with dextrose, 55.7% without; P = 0.2).

The probability of cataract extraction with and without dextrose supplementation is shown in Figure 1. The two-sided weighted log-rank test suggested evidence for intersecting hazards between dextrose groups (P = 0.04), which justifies our choice of a parametric formulation without the proportional hazards assumption. Furthermore, the one-sided test lacked statistical evidence for a difference between groups but revealed a trend toward the benefit of dextrose supplementation postoperatively after the first year (P = 0.09). This trend may not be clinically significant; at three years, approximately three-quarters of both groups had undergone cataract extraction.

Fig. 1.

Probability of cataract extraction surgery for eyes with and without dextrose supplementation.

Multivariable modeling assessed the interaction of dextrose and pre- and perioperative variables of age, duration of vitrectomy, fasting glucose, and hemoglobin A1c to duration of cataract extraction. Age and surgical duration were the only variables to exhibit a significant effect, both with a negative trend (i.e., higher age and surgical duration are expected to shorten the time to cataract extraction). However, these variables did not interact with dextrose supplementation. Interestingly, the main effects of fasting glucose and dextrose presented insignificant results, but the interaction between fasting glucose and dextrose was significant (Table 2).

Table 2.

Multivariable Model Assessing the Interaction of Dextrose Supplementation With Pre- and Intraoperative Variables and Time to Cataract Extraction

Covariate	β	AF (95% CI)	P
Age	−0.041	0.96 (0.95, 0.97)	< 0.001
Surgical duration	−0.007	0.99 (0.99, 1.00)	< 0.001
Preoperative visual acuity	0.026	1.03 (0.60, 1.76)	0.93
Hemoglobin A1c	0.025	1.02 (0.94, 1.11)	0.56
Fasting glucose	−0.009	0.99 (0.74, 1.32)	0.95
Dextrose	0.002	1.00 (0.99, 1.00)	0.31
Dextrose × fasting glucose	0.005	1.00 (1.00, 1.01)	0.03

AF, acceleration factor (e^β).

Considering the opposing coefficient sign between fasting glucose (−0.009) and the interaction effect (0.005) suggests that without dextrose, fasting glucose exhibited a negative trend (higher glucose accelerates time to cataracts), but upon dextrose administration fasting glucose exhibited a positive trend (higher glucose decelerates time to cataracts). We used the results of the adaptive index procedure to better understand this interaction relationship, which estimated a fasting glucose threshold of <120 mg/dL as a high-risk subgroup for dextrose administration. The acceleration factor is 0.63, indicating that, on average, those with fasting glucose <120 mg/dL with dextrose will experience a 37% decrease in time to cataract extraction compared with no dextrose administration, suggesting a clinically meaningful effect. However, the 95% confidence intervals (0.37–1.09) indicate uncertainty in this estimate.

In the subset of patients who underwent vitrectomy for diabetic tractional retinal detachments (86 eyes), the mean surgical time was 86.8 (SD 37.2; interquartile range 60.3–105.8) minutes. Kaplan–Meyer analysis showed overlapping cataract survival plots between dextrose- and non-dextrose-supplemented groups (Figure 2).

Fig. 2.

Probability of cataract extraction surgery for eyes undergoing vitrectomy for diabetic tractional retinal detachment, with and without dextrose supplementation.

Discussion

In modern, small-gauge vitrectomy, intraoperative unplanned lensectomy is rare. No difference in the intraoperative lensectomy rate or overall postoperative cataract course was observed with or without dextrose supplementation to the infusion fluid.

Haimann and Abrams⁵ reported that intraoperative lensectomy was required for 30% of eyes without dextrose supplementation during vitrectomy, with an average surgical duration of two hours. However, our large series had only unplanned lensectomy in eyes without dextrose supplementation, with a rate of <1%. While the volume of infusion fluid used in the current series is not available, improvements in surgical techniques, such as valved trocars and reduction in average surgical duration, likely have allowed a decreased volume of infusion fluid and resulted in sufficient visibility to complete surgeries with lens preservation.

Postoperative cataract formation is a common occurrence after vitrectomy, believed to be secondary to reactive changes of lens proteins intra- and postoperatively as the antioxidant vitreous is removed.^7,10,11 Most post-vitrectomy patients require cataract extraction within two years. Diabetes increases cataract development, likely secondary to glucose fluctuations that cause lens swelling and opacification. Paradoxically, diabetic patients have a lower rate of post-vitrectomy cataract extraction compared with the nondiabetic population, perhaps secondary to underlying ischemia and decreased oxygenation of the posterior segment.¹¹

With the supplementation of 3 mL of 50% dextrose to standard BSS Plus, the resulting glucose concentration is 400 mg/dL, compared with standard BSS Plus of 100 mg/dL.⁵ Haimann and Abrams⁵’s rationale for the dose of supplemental dextrose was that the intraocular glucose correlated to serum glucose, and matching the patient’s elevated glucose and hyperosmolarity with supplementation of dextrose to BSS Plus would reduce lens opacification. Subsequent analysis confirms the correlation between aqueous and serum glucose concentration.⁸ However, separate vitreous analysis shows a modest relationship between serum and vitreous glucose (r = 0.4), and a markedly elevated glucose over 200 mg/dL may correspond to a vitreous glucose of 80 mg/dL.⁹ The importance of the glucose concentrations of the intraocular fluids, aqueous versus vitreous, on cataract formation may be a topic of further investigation.

Our multivariable model of pre- and perioperative variables revealed several key factors influencing time to cataract surgery. Age and vitrectomy duration significantly emerged as the main effects, indicating that these factors independently contribute to the timing of cataract surgery, such that older patients and those undergoing longer vitrectomy procedures exhibit earlier intervention of cataracts. Other literature report similar findings.¹² Conversely, the administration of dextrose, along with levels of hemoglobin A1c and fasting glucose as main effects, did not show a significant impact on the timing of cataract surgery. This suggests that these metabolic factors, in isolation, may not alter the timeline for cataract surgery in a clinically meaningful way.

Subsequent interaction analyses imply that the influence of dextrose administration on time to cataracts was not modified by the patient’s age, vitrectomy duration, or hemoglobin A1c, whereas the influence of dextrose on time to cataracts may be moderated by preoperative fasting glucose levels. Specifically, sooner cataract surgery was observed in the patients with normal fasting glucose (<120 mg/dL) who received supplemental dextrose. Similarly, glucose fluctuations in the nonsurgical diabetic are known to increase lens opacification.¹³ The differences observed in this analysis may not be clinically significant, as most eyes eventually have cataract surgery, and the confidence interval indicates uncertainty in the estimate. Hemoglobin A1c, a helpful marker for glycemic control over several months, did not interact with dextrose. Both of these findings may emphasize the lack of benefit of supplemental dextrose, even in a poorly controlled diabetic. Furthermore, for the tractional retinal detachments subset, the cataract survival curves for these longer, more complex cases overlapped, with no difference with dextrose supplementation.

In this series, approximately half of the diabetic patients had cataract extraction by the second year post-vitrectomy. Literature reports of duration to cataract surgery vary widely, with Gupta et al¹⁴ reporting one half in 1 year, Ostri et al¹⁵ and Glassman et al¹⁶ reporting one third in 2 years, Smiddy and Feuer¹⁷ reporting 15% at 2 years, and one quarter at 3 years in a study by Farawami et al.¹¹ Variations likely reflect practice patterns with different thresholds for cataract surgery.

Limitations of this study include the retrospective nature of this analysis. Cataract surgery was used as a marker for the development of visually significant lens opacification, as done by others^11,17; however, other reasons for cataract extraction or observation may exist, and thresholds for the decision for surgery vary widely among ophthalmologists.

In conclusion, standard BSS Plus appears to be adequate for infusion fluid, even for diabetic patients, including those with tractional retinal detachments. Intraoperative lensectomy rates were not increased when dextrose was not added to the infusion. Postoperative cataract course is also no different with or without dextrose, even when assessing for interactions with markers for surgical or disease complexity, such as age, duration of surgery, and hemoglobin A1c.