Turalba, Angela V. MD*,†,‡; Shah, Ankoor S. MD, PhD*,†,§; Andreoli, Michael T. MD∥; Andreoli, Christopher M. MD*,†,¶; Rhee, Douglas J. MD*,†,§
Open-globe trauma resulting in a full-thickness disruption of the eye wall often causes damage to multiple ocular structures. Elevated intraocular pressure (IOP) and secondary glaucoma can develop after these injuries as a result of trabecular meshwork damage, hyphema, injury to the lens and/or iris, inflammation, peripheral anterior synechiae, vitreous hemorrhage, and topical corticosteroid use.1,2 Globe trauma can also result in ocular blood flow disturbances, which may be independent risk factors for traumatic glaucoma.3 Although the mechanisms of glaucoma after open-globe and closed-globe injuries overlap, the majority of traumatic glaucoma occurs after closed-globe trauma. Studies have reported that 68% to 77% of traumatic glaucoma cases occur after closed-globe injuries and 23% to 32% occur after open-globe injuries.1,4 The overall incidence of glaucoma after ocular contusion ranges from 2% to 43%, depending on the length of follow-up and the criteria used to define traumatic glaucoma.5–7 The functional and structural measures typically used to define glaucoma are often not reliable in traumatized eyes. So previous studies do not specifically define traumatic glaucoma using visual field or optic nerve data, but rely largely on clinician-reported diagnoses or persistently elevated IOP as a marker for traumatic glaucoma.7,8
Only a few studies specifically describe the risk factors and outcomes of elevated IOP and secondary glaucoma after open-globe trauma. A cohort study (N=3627) by Girkin and colleagues found that the risk of developing traumatic glaucoma after penetrating ocular injury was 2.7%. Older age, lens injury, poor visual acuity, and intraocular inflammation were risk factors associated with secondary glaucoma in this cohort.8 Smaller studies have shown that patients with a history of penetrating ocular trauma were more likely to need glaucoma surgery when compared with patients with blunt ocular injury, but that IOP was adequately controlled with medical and surgical treatment in a majority of cases.4,9
The prognosis and clinical management of elevated IOP in the setting of open-globe trauma still remains largely unaddressed in the literature. IOP elevation after these complex injuries can be clinically significant, requiring intervention and long-term follow-up. The aim of our study is to determine the early predictors and long-term outcomes for persistent IOP elevation occurring after open-globe injury. We hypothesize that IOP elevation after open-globe trauma carries a guarded prognosis, and that identifying predictors and clinical outcomes for this entity can be used to guide follow-up care and management of patients at risk.
MATERIALS AND METHODS
A retrospective chart review was conducted on 658 consecutive patients treated for open-globe injuries at the Massachusetts Eye and Ear Infirmary between February 1999 and January 2007. An open-globe injury was defined as a traumatic full-thickness break in the cornea and/or sclera. This cohort represents patients treated by the ocular trauma service either with isolated open-globe injuries or an open-globe injury as part of a multisystem trauma. Initial management of open-globe injuries followed a previously described standardized protocol.10 As part of this protocol, open-globe injuries were surgically repaired within 24 hours of the initial injury, unless limited by delayed presentation or active medical issues. Demographic and clinical data collected included age, sex, information about the time and place of injury, mechanism of injury, details from the initial examination, open-globe repair specifics, follow-up examination findings, and subsequent surgical procedures performed.
Using this initial database, we defined 2 cohorts of patients based on documented maximum IOP at any time after open-globe injury repair. The control group represented patients suffering open-globe injuries who had documented maximum IOP<22 mm Hg at all follow-up visits with a minimum of 2 months of follow-up. The ocular hypertension group consisted of those who had IOP≥22 mm Hg at >1 visit or an isolated IOP≥22 mm Hg that warranted treatment with at least 2 months of follow-up. We excluded patients with <2 months of follow-up, inadequate clinical data on follow-up, and those who had a single, elevated IOP measurement that resolved without treatment.
We analyzed the 2 cohorts using binary logistic regression modeling to define risk factors associated with developing ocular hypertension. We evaluated baseline characteristics including age, sex, preoperative visual acuity, presence of perioperative hyphema, presence of intraocular foreign body (IOFB), presence of lens injury, presence of vitreous hemorrhage, and zone of injury based on the Open-Globe Classification System.11 Zones of injury are defined by the anatomic location of the laceration on the eye—zone I is limited to the cornea, zone II is the region between the limbus and 5 mm posterior to the limbus, and zone III involves the region posterior to zone II. The baseline characteristics were chosen a priori based on the clinical notion that these factors might contribute to traumatic glaucoma. We generated 3 models using binary logistic regression algorithms in SPSS (IBM Corporation, Somers, NY) and MATLAB (The MathWorks Inc., Natick, MA). In model 1, we entered all baseline characteristics listed above in the regression analysis to generate odds ratio (OR) for each of these risk factors in contributing to ocular hypertension after open-globe injury. We considered P values of ≤0.05 as statistically significant.
Since model 1 has the risk of overfitting the data given the many variables entered into the analysis, we created refined models 2 and 3 to confirm the results. Model 2 used a 2-block method where we used the a priori knowledge from Girkin et al8 to build the model in 2 blocks. Their study showed that risk factors for developing ocular hypertension included age, preoperative visual acuity, and lens injury, and so we entered these variables into block 1 of the logistic regression analysis. The analysis computed OR for each of these factors and added or subtracted them from the block analysis as appropriate. Once each of these factors was tested, we performed a stepwise, forward addition of the remaining characteristics as a block 2 analysis in building this model. This addition of parameters was performed by identifying the largest score statistic at each iteration and adding that variable to the model if the P value was <0.05. The final model, model 3, started with all parameters in the analysis and removed individual ones through a stepwise fashion. In each iteration of refinement, the parameter with the least contribution to the fit was removed until all remaining variables were significant. Each of the 3 models were then compared using the Nagelkerke R2 statistic; the larger the value of this statistic, the better the fit of the model. Moreover, by generating a logistic regression model utilizing different building blocks, we were able to evaluate whether the identified risk factors of ocular hypertension in open-globe injury patients were robustly identified across multiple statistical tests.
Follow-up clinical data for the ocular hypertension cohort were then evaluated using information from the initial trauma database and cross-referencing clinical charts from the various subspecialty services at Massachusetts Eye and Ear Infirmary. Clinical outcome data from predetermined follow-up time points were collected for these patients and entered into an additional database. Data collected included visual acuity, IOP, and type of glaucoma intervention needed (ie, medications, anterior chamber (AC) washout, trabeculectomy, and/or implantation of a glaucoma drainage device).
A review of 658 consecutive patients treated for open-globe injuries over the defined 8-year period between February 1999 and January 2007 identified 65 patients who met our inclusion criteria for traumatic ocular hypertension and had a minimum of 2 months of follow-up. We identified 317 patients with IOP<22 mm Hg who had at least 2 months of follow-up. Figure 1 summarizes the groups that met our inclusion criteria. The overall cumulative prevalence of ocular hypertension in our cohort was 17% (65/382). Only 2 patients with a maximum IOP≥22 mm Hg required enucleation for a blind painful eye. In both cases, enucleation was performed within 2 months of the injury, and these patients did not meet the follow-up criteria to be included in the data analysis.
We compared the predetermined baseline characteristics of the ocular hypertension group and the group of excluded patients because of insufficient clinical data. The excluded patients had a higher percentage with IOFBs (32% vs. 8%, P<0.05), but otherwise the groups were statistically similar with respect to the other baseline characteristics analyzed (data not shown).
Table 1 summarizes the multivariate, binary logistic regression analyses we used to identify risk factors associated with the development of ocular hypertension after open-globe injury. Model 1 compared all perioperative parameters of the ocular hypertension cohort with controls to generate OR for each of these characteristics. This analysis showed that increased age (OR 1.03, P=0.001), hyphema (OR 2.35, P=0.025), lens injury (OR 5.19, P≤0.0001), and zone II injury (OR 2.20, P=0.025) all increased the risk of ocular hypertension after open-globe injury. Model 2 utilized the a priori knowledge to guide the analysis, and this showed that the same parameters except hyphema increased risk of ocular hypertension. Finally, model 3 utilized all parameters and removed those not likely to contribute to the model fit, and this showed the same parameters to increase the odds of ocular hypertension as model 1. Overall, each model fit was comparable based on the Nagelkerke R2 values, and they collectively show that lens injury was the strongest predictive factor for ocular hypertension.
For the 65 patients that developed traumatic ocular hypertension, the average length of follow-up was 20.6 months (range, 2 to 102 mo). Forty-eight (74%) patients had >6 months of follow-up, whereas 37 (57%) patients had >12 months of follow-up. The average follow-up for controls was 11.3 months (range, 2 to 89 mo). The average maximum IOP for patients with ocular hypertension was 33.4 mm Hg, which occurred at the median follow-up of 21 days (25th percentile, 2 d; 75th percentile, 74 d; range, 1 d to 77 mo). The average IOP in the overall cohort normalized over time (Fig. 2). At all follow-up time points, the majority of patients with ocular hypertension maintained an IOP between 10 and 21 mm Hg (Table 2).
Table 3 classifies the patients based on the interventions used to manage elevated IOP, and when analyzed separately, each group followed a similar trend in IOP lowering over time. The majority (74%) of patients were treated medically. Patients treated with medications alone had an average maximum IOP=31.6 mm Hg occurring at a median follow-up of 20 days. Average IOP was 17.9 mm Hg at 1 year, 14.5 mm Hg at 2 years, and 17.2 mm Hg at 5 years. Of the patients treated medically, 35 (73%) were not using IOP-lowering medications at the last documented visit. For patients who were observed, average maximum IOP was 24.2 mm Hg occurring at a median follow-up of 33 days. Average IOP for this group was 16.8 mm Hg at 6 months and 16.3 mm Hg at 1 year. The patients who required an AC washout but no additional glaucoma surgery had an average maximum IOP of 36.5 mm Hg occurring at a median follow-up of 2 days. Average IOP was 14 mm Hg at 3 months and 13.3 mm Hg at 6 months. Four of the 5 patients were not on any medications at the last follow-up, while 1 patient was on maximal medical therapy and an IOP=8 mm Hg at the last follow-up visit.
Of the 8 patients who underwent glaucoma surgery, 5 had Ahmed valve placement, 2 had trabeculectomy with mitomycin-C, and 1 had Baerveldt tube shunt placement. Four patients had an AC washout before their tube shunt surgery. Five patients had their initial glaucoma surgery within 6 months after open-globe repair, whereas the remaining 3 patients had surgery between 15 and 30 months after the initial injury. Two patients required repeat glaucoma surgery, including a revision of an Ahmed valve and placement of an Ahmed valve after a failed trabeculectomy. Average maximum IOP for these patients who underwent surgery was 48.5 mm Hg with an average of 3.4 medications occurring at a median follow-up of 2.1 months. Average IOP was 18 mm Hg at 1 year, 25.3 mm Hg at 2 years, 17.8 mm Hg at 3 years, and 12.3 mm Hg at 5 years. Patients were on an average of 0.7 medications on the last documented visit with 4 patients on no medications on the last documented follow-up visit.
Median visual acuity improved over time for all groups. Median preoperative visual acuity of hand motions improved to a median visual acuity of 20/60 at both 12 and 36 months for the overall cohort of patients with ocular hypertension (Fig. 3). Moreover, most of the improvement in visual acuity occurred within the first 3 months after the injury and then remained stable through 12 and 36 months.
Persistent IOP elevation is a significant complication after open-globe injury that often requires treatment. The clinical course, response to treatment, and long-term outcomes of persistently elevated IOP after open-globe trauma are largely unknown. Our study addresses this gap in the literature by confirming the risk factors previously described, as well as describing outcomes in patients with ocular hypertension after open-globe injury. Contrary to our initial hypothesis, the cohort with ocular hypertension after open-globe repair had favorable outcomes, on average, with improved visual acuity and normalization of IOP over time. Increased age is a risk factor for traumatic ocular hypertension, and open-globe injuries in the geriatric population are associated with poorer outcomes.12 However, our patients with elevated IOP still represent a younger population with a median age of 47 years, which may explain their better outcomes. The group with traumatic ocular hypertension had improvement in their visual acuity from presentation through the first 3 months after injury. Visual acuity improved in our cohort likely from the treatment and resolution of comorbid eye conditions that result from anterior segment injuries, such as corneal laceration, hyphema, and traumatic cataract. Visual outcomes after open-globe injuries are more significantly related to factors other than elevated IOP, that have been implicated in previously described prognostic models.13–15
The majority of our patients with ocular hypertension maintained IOP between 10 and 21 mm Hg across all treatment groups over time. Bai et al4 also reported successful IOP control in their patients with secondary glaucoma after open-globe and closed-globe injuries. This may be reflective of multiple mechanisms underlying IOP elevation after trauma, which can be amenable to tailored therapy. For example, steroid-related IOP elevation may resolve after cessation of steroids, and in cases of lens-related glaucoma, surgical removal of the lens may be sufficient in lowering IOP to a safe level. Good IOP control, however, does not signify the course of glaucomatous disease. In assessing the development and progression of glaucoma, our study is limited by the lack of optic nerve head imaging and visual field data during follow-up visits. These tests for glaucomatous optic neuropathy are limited and easily confounded by other ocular conditions; media opacities, traumatic optic neuropathy, and retinal pathology in traumatized eyes often make visual field and optic nerve testing unreliable in these cases.
Our study shows that IOP elevation in these patients occurs at a median duration of 3 weeks, but primarily between 2 days and 2.5 months. This knowledge can help guide the timing of IOP surveillance and treatment in high-risk eyes after an open-globe injury. Our study indicates that most of these patients can be treated with IOP-lowering medications alone, although some may need surgical intervention to control IOP. Twelve percent of patients required glaucoma surgery for IOP control, with most patients undergoing surgery within 6 months of the initial trauma and 3 patients requiring glaucoma surgery >2 years after the injury. In cases where glaucoma surgery was performed within the first 2 months of the injury, hyphema and vitreous hemorrhage were important factors contributing to IOP elevation, whereas chronic angle closure glaucoma was often the indication for glaucoma surgery performed more than a year after an open-globe injury. This suggests that clinicians counsel patients with open-globe injuries that routine, long-term (maybe life-long) follow-up is necessary. With longer follow-up, we may find more patients in our cohort who require surgical intervention for IOP control. Previously published rates of glaucoma surgery after ocular trauma vary significantly. In a study of combat ocular trauma, the rate of glaucoma-implant surgery was 1.3% after both open-globe and closed-globe trauma.6 In contrast, Ozer et al9 showed that 58% of patients with traumatic glaucoma after penetrating injury needed IOP-lowering surgery, typically within 6 months of the injury. Differences in the rates and timing of surgical intervention could be partially explained by the nature of the injuries, patient demographics, extent of follow-up, and regional differences in practice patterns.
Not unexpectedly, the rate of traumatic ocular hypertension in our cohort (17%) was higher than the rate of traumatic glaucoma previously reported as 2.7% after penetrating ocular trauma.8 Presumably, not all patients with traumatic ocular hypertension develop glaucomatous optic neuropathy, and the criteria used to define traumatic glaucoma can affect reported rates. Our study investigates a hypertensive ocular disorder resulting from multiple mechanisms and uses IOP elevation as a marker for traumatic glaucoma. In the prior series, the diagnosis of traumatic glaucoma was based on a survey completed by an ophthalmologist at a 6-month interval follow-up visit.8 Sihota and colleagues had defined traumatic glaucoma as a chronically elevated pressure in a cohort of patients with ocular contusion. In that series of 92 patients, 43% developed traumatic glaucoma after closed-globe trauma.7 This rate is significantly higher than prior documented rates of glaucoma after ocular contusion from studies that used different criteria for traumatic glaucoma. Using ocular hypertension as a surrogate to study traumatic glaucoma has a higher sensitivity and is potentially more relevant in clinical situations where persistently elevated IOP will prompt treatment.
Our study confirms prior risk factors associated with ocular hypertension after open-globe and closed-globe injuries, including increased age, lens injury, and hyphema.5,8,9 We also found that zone II injury (defined as a full-thickness laceration involving the region between the limbus and up to 5 mm posterior to the limbus) is also a predictive factor for developing ocular hypertension, presumably because of direct injury to the trabecular meshwork and alterations to aqueous outflow. However, we cannot propose a direct causal mechanism of IOP elevation in this cohort since open-globe injuries are complex and IOP elevation can result from multiple mechanisms depending on the nature of the trauma and the subsequent interventions used to treat the eye. Many of these patients undergo extensive surgical rehabilitation for their open-globe injury, and IOP elevation can also be a direct result of having multiple intraocular surgeries.16 All the patients who underwent glaucoma surgery after open-globe repair had between 1 and 3 intraocular operations from the initial injury to the time of glaucoma surgery. These operations were usually complex and often included pars plana vitrectomy, AC washout, cataract extraction, and/or penetrating keratoplasty. All the patients needing glaucoma surgery also had at least partial synechial closure of the angle, hyphema and/or vitreous hemorrhage that contributed to their IOP elevation. Unlike other reports, vitreous hemorrhage and preoperative visual acuity were not significant risk factors in the development of traumatic ocular hypertension in our analysis. Prior studies found intraocular inflammation as a risk factor for developing glaucoma, but our database did not adequately quantify inflammation as a baseline characteristic and could not be assessed in our analysis.
Although risk factors for developing traumatic glaucoma after open-globe and closed-globe injuries overlap, the frequency of secondary glaucoma in ocular contusion led us to conjecture that the mechanism of globe rupture has an influence on the development of traumatic ocular hypertension. We hypothesized that blunt ruptures were more likely to result in the development of ocular hypertension compared with penetrating injuries. Using the OTS classification of mechanism,13 we found that blunt rupture was not a predictive factor in the development of elevated IOP (data not shown). This probably reflects the multifactorial nature of open-globe injuries and the numerous mechanisms that are thought to be involved in the development of ocular hypertension and secondary glaucoma in these cases. We are limited in our ability to comment on IOFB as a risk factor for glaucoma as a significant proportion of patients excluded from our study had IOFB injuries and were lost to follow up elsewhere or had inadequate documentation of their clinical course.
To analyze which baseline characteristics were predictive in the development posttraumatic ocular hypertension, we performed multivariate binary logistic regression analyses that can be applied to a predictive model for IOP elevation after open-globe trauma. Similar predictive models have been developed to determine visual outcomes after open-globe injury.13,14 These models have been shown to have high predictive accuracy and potentially can be used clinically in the management and counseling of these trauma patients.15 However, these prognostic models do not specifically address traumatic glaucoma as an outcome. A follow-up study developing a predictive model for IOP elevation will be helpful in identifying patients at risk for traumatic glaucoma. This current study demonstrates that IOP control is attainable in those who develop ocular hypertension after open-globe injuries, so having a prognostic tool would be helpful in identifying and treating patients at risk.
1. De Leon-Ortega JE, Girkin CA .Ocular trauma-related glaucoma.Ophthalmol Clin N Am. 2002; 15:215–223.
2. Milder E, Davis K .Ocular trauma and glaucoma.Int Ophthalmol Clin. 2008; 48:47–64.
3. Martini E, Guiducci M, Campi L, et al .Ocular blood flow evaluation in injured and healthy fellow eyes.Eur J Ophthalmol. 2005; 15:48–55.
4. Bai H, Yao L, Wang D, et al .Causes and treatments of traumatic secondary glaucoma.Eur J Ophthalmol. 2009; 19:201–206.
5. Girkin CA, McGwin G Jr, Long C, et al .Glaucoma after ocular contusion.J Glaucoma. 2005; 14:470–473.
6. Weichel ED, Colyer MH, Ludlow SE, et al .Combat ocular trauma visual outcomes during operations Iraqi and Enduring Freedom.Ophthalmology. 2008; 115:2235–2245.
7. Sihota R, Kumar S, Gupta V, et al .Early predictors of traumatic glaucoma after closed globe injury.Arch Ophthalmol. 2008; 126:921–926.
8. Girkin CA, McGwin G Jr, Morris R, et al .Glaucoma following penetrating ocular trauma: a cohort study of the United States Eye Injury Registry.Am J Ophthalmol. 2005; 139:100–105.
9. Ozer PA, Yalvac IS, Santana B, et al .Incidence and risk factors in secondary glaucomas after blunt and penetrating trauma.J Glaucoma. 2007; 16:685–690.
10. Andreoli CM, Andreoli MT, Kloek CE, et al .Low rate of endophthalmitis in a large series of open globe injuries.Am J Ophthalmol. 2009; 147:601–608.
11. Pieramici DJ, Sternberg P Jr, Aaberg TM Sr, et al .A system for classifying mechanical injuries of the eye (globe).Am J Ophthalmol. 1997; 123:820–831.
12. Andreoli MT, Andreoli CA .Geriatric traumatic open globe injuries.Ophthalmology. 2011; 118:156–159.
13. Kuhn F, Maisiak R, Mann L, et al .The Ocular Trauma Score (OTS).Ophthalmol Clin North Am. 2002; 15:163–165.
14. Schmidt GW, Broman AT, Hindman HB, et al .Vision survival after open globe injury predicted by classification and regression tree analysis.Ophthalmology. 2008; 115:202–209.
15. Man CY, Steel D .Visual outcome after open globe injury: a comparison of two prognostic models-the Ocular Trauma Score and the Classification and Regression Tree.Eye. 2009; 24:84–89.
16. Andreoli MT, Andreoli CM .Surgical rehabilitation of the open globe injury patient.Am J Ophthalmol. 2012; 153:856–860.