Traumatic ocular injury is the leading cause of vision loss in young adults.1 In contrast to the civilian population, ocular trauma experienced by U.S. service members is usually of increased severity.2 In Operations Iraqi and Enduring Freedom (OIF/OEF, Iraq/Afghanistan), 13% of U.S. service members injured in combat had a traumatic ocular injury requiring evacuation from theater.3 A traumatic ocular injury affecting the posterior segment generally has a poorer prognosis compared to an anterior segment injury.3,4
Proliferative vitreoretinopathy (PVR) is a common cause of decreased vision after retinal detachment (RD) and ocular trauma. Classically, PVR was noted to be secondary to RD and its repair.5–7 However, further research has shown that the etiopathogenesis of PVR is significantly more complicated particularly in patients with trauma.8–10 PVR has been reported after laser treatment of retinoblastoma and treatment of syphilitic uveitis.11,12 In patients with a primary RD, the rate of PVR is 8% to 10%, whereas in those with ocular trauma, it can be as high as between 40% to 60%.10 This high rate in ocular trauma is likely due to the inflammatory cascade, a breakdown of the blood-retinal barrier, and direct introduction of cells from outside the eye.10,13 The development of PVR in association with RD poses a threat to the adequate repair and prognosis of the reattached retina.14,15 The importance of elucidating the risk factors associated with developing PVR specifically in injuries sustained from combat trauma and the connection between PVR and RD is of significant clinical importance. As patients who develop PVR after RD repair have a worse prognosis than those without PVR,16,17 understanding risk factors for PVR development might lead to better identification of high-risk patients and consideration of different more aggressive therapy such as silicone oil versus gas placement.
The goals of this study were to assess the risk factors associated with the development of PVR in all patients and in those with a RD. An evaluation of the rate of poor final visual acuity (VA) in patients with PVR was completed.
Patients and Methods
The Walter Reed Ocular Trauma Database (WROTD) contains data on ocular injuries to U.S. service members evacuated from combat support hospitals (CSHs) in OIF and OEF. Upon evacuation from the CSH, most patients were managed at Landstuhl Regional Medical Center (LRMC) in Landstuhl, Germany, before then being transferred to Walter Reed Army Medical Center (WRAMC) in Bethesda, Maryland, for further treatment. These data were collected over 10 years from 2001 to 2011 and consisted of patient records from initial injury through arrival at WRAMC. Data were collected throughout their stay at WRAMC until they were released back to service or were entered into the Veterans Affairs medical system using Microsoft Excel (Microsoft Corporation, Redmond, WA) and stored on SPSS (IBM, Armonk, NY). Statistical analysis was completed using R statistical software (R Foundation for Statistical Computing, Vienna, Austria).
All cases of ocular trauma from 2001 to 2011 involving U.S. service members or Department of Defense civilians injured in OIF and OEF evacuated to WRAMC were included. All vitreoretinal surgeries occurred at WRAMC, as there was no vitreoretinal surgery capability at a CSH or LRMC. Of these, all cases of documented PVR were studied. Primary outcome measures were the incidence of PVR and risk factors for the development of PVR in all patients and those with RD. Secondary measures were final visual outcome and associated injuries.
The Birmingham Eye Trauma terminology was used to describe injuries.18–20 As previously described the Snellen VA were converted to vision grades.21 Grade 1 was defined as 20/40 or better, Grade 2 was 20/50 to 20/200, Grade 3 was 19/200 to 1/200 (count fingers), Grade 4 was hand motion or light perception, and Grade 5 was no light perception. The ocular trauma scores (OTS) were computed as previously described.22
For statistical analysis, patients with PVR were separated into two different categories. The first group were any patients who had documented PVR, and the second group were those who had a documented PVR and RD. Ocular trauma has a high rate of development of PVR, likely due to the subsequent inflammatory cascade, breakdown of the blood-retinal barrier, and direct introduction of cells from outside the eye.10,13 In Cardillo et al., a study similar to the current study evaluating the rate of PVR in patients with ocular trauma, PVR was noted in 71 injured eyes, but only 54 had a documented retinal detachment.9 For this reason, the authors felt that it was appropriate to evaluate risk factors for the development of PVR in two separate analyses, those with PVR, and those with PVR and RD.
Fisher's exact test and Wilcoxon rank-sum test with continuity correction were used for univariate analysis; Fisher's exact test was also used to compute odds ratios and 95% confidence intervals (CIs). Multivariate analysis was completed using binomial logistic regression. Type 3 likelihood ratio test as implemented in the “car” package was used to evaluate significance.23 Alpha was set at .05 for all analyses.
Eight hundred ninety eyes of 651 U.S. service members were evaluated from the WROTD. A total of 76 eyes (8.5%) of 66 patients that had developed PVR were analyzed. Nineteen patients had bilateral eye injuries and five patients had bilateral PVR. The average age of the patients was 27.6 years (standard deviation: 7.9; range: 19–50 years). Patients were predominantly male (97.4%). Sixty-six (86.8%) injuries occurred in Iraq, and 10 (13.2%) occurred in Afghanistan.
Zones of Injury
Zones of injury were recorded for patients that developed PVR. Zones of injury are defined by whether the injury is an open globe or closed globe injury and based on the anatomic location of the eye injury.24 Closed globe zone 3 injuries were noted in eight eyes (10.5%). Eight eyes (10.5%) had open globe zone 1, 12 (15.8%) had open globe zone 2, and 48 (63.2%) had open globe zone 3 injuries. Of the open globe injuries, six were penetrating injuries, 36 were from intraocular foreign bodies (IOFBs), 12 were from perforating injuries, and 14 were from globe ruptures.
Visual Acuity and Ocular Trauma Score
OTS and initial and final VA were recorded, as shown in Table 1. In the database, four patients (5.3%) were missing final data, and five (6.6%) patients were initially intubated. Twenty-three eyes (30.3%) had no change in VA grade, 31 (40.8%) improved, and 13 (17.1%) declined.
Ocular Trauma Score
Initial Visual Acuity and Final Visual Acuity Grades
To evaluate the importance of PVR relative to other risk factors for final VA less than 20/200, a multivariate analysis was completed on all non-enucleated eyes in the WROTD (Table 2). This analysis studied whether patients had a final VA of grades 1 to 2 (20/200 or better vision) or grades 3 to 5 (19/200 or worse vision) from six features: PVR, RD, perforating injury, globe rupture, optic nerve injury, and phthisis. The OTS was used as the basis for variable selection with RD, globe rupture and perforating injury included as they are part of the OTS. Since afferent pupillary defect (APD; component of OTS) was frequently recorded by a non-ophthalmologist and likely unreliable, optic nerve injury was selected instead. Endophthalmitis, another variable in the OTS, was not included as it was very rare in the WROTD. However, phthisis was included as it was a cause of blindness in the WROTD. PVR was found to be a significant risk factor for a poor final VA (P < .001).
Multivariate Analysis of Risk Factors for Final Visual Acuity Less Than 20/200
Associated Injuries and Surgeries
The following associated injuries were assessed in eyes that developed PVR. Twenty-three (30.3%), 22 (28.9%), and 33 (43.4%) eyes had either one or more corneal, corneoscleral, or scleral lacerations, respectively. A vitreous hemorrhage was seen in 59 (77.6%) eyes. Commotio was noted in 16 eyes (21.1%), chorioretinal rupture was noted in 18 eyes (23.7%), subretinal hemorrhage was noted in 37 eyes (48.7%), intraretinal hemorrhage was noted in 31 eyes (40.8%), and macular hemorrhage was also found in 26 eyes (34.2%). Retinal tears were seen in 45 eyes (59.2%) of patients. RDs were documented in 52 eyes (68.4%). Of these eyes, 45 had a known macula status, with macula-on detachments in nine and macula-off detachments in 36. Choroidal hemorrhage was noted in 37 eyes (48.7%). Further, 31 eyes (40.8%) of patients developed phthisis.
Many patients suffered severe neurologic and orthopedic injuries from improvised explosive device blasts. Systemic neurologic injuries such as cerebral vascular accidents, anoxic brain injury, or spinal cord injury were noted in 19 patients (25.0%), traumatic brain injury in 31 (40.8%), and extremity injury in 36 (47.3%).
In the eyes that had PVR but did not have a documented RD (n = 24), 22 were open globe injuries. Nineteen eyes had a documented IOFB, 17 had a vitreous hemorrhage, and six had a retinal tear. Of the two eyes with closed globe injuries, one had a vitreous hemorrhage with chorioretinal rupture and underwent pars plana vitrectomy (PPV) and pars plana lensectomy; the other eye had noted intraretinal hemorrhage and cataract formation and underwent pars plana lensectomy.
Regarding surgical intervention, 70 eyes (92.1%) underwent a PPV and 15 (19.7%) eyes of patients had a scleral buckle. Moreover, five eyes (6.6%) underwent a subsequent enucleation. Tamponade use was recorded: 38 eyes were filled with silicone oil, 17 with C3F8, five with SF6, and three with air. Seven had undocumented tamponade selection. Time to initial retina surgery was available in only 41 patients, and the time to surgery on average was 22.56 days (range: 3 to 87 days). Thus, time to vitreoretinal surgery was not included in statistical analysis. In addition of those that underwent retinal surgery for RD, 21 of 41 patients (51.2%) had a successful reattachment.
Risk Factors for PVR Development
A further statistical analysis using univariate analysis was completed to review risk factors for the development of PVR in all patients and those with RD as described in the methods section. In all patients with PVR, there were multiple different significant risk factors (Table 3A). In patients with RD, IOFB (P < .001), unsuccessful repair (P = .002), and macular hemorrhage (P ≤ .04) were identified as significant risk factors (Table 3B)
Univariate Analysis to Assess for Risk Factors the Development of PVR in all Patients (n = 76)
Univariate Analysis for Risk Factors the Development of PVR in Patients With Retinal Detachment (n = 52)
This study sought to examine the risk factors associated with the development of PVR in globe injury after combat ocular trauma in U.S. service members. This is a unique patient populace due to severity of injury as evidenced by more than 70% of patients having an OTS less than 65 portending a poor visual prognosis.22 The more severe the trauma experienced by the eye, the more likely the development of PVR is to occur.10 Ocular trauma has a high rate of development of PVR, secondary to the subsequent inflammatory cascade, breakdown of the blood-ocular barrier, and direct introduction of cells from outside the eye.10,13 In patients with ocular trauma and/or RD, there is a risk of retinal hypoxia and apoptosis of photoreceptor cells.5,25–27 The combination of inflammation, ischemia, and blood further activates inflammatory cells resulting in the release of cytokines and growth factors.10 When these proteins interact with the retinal pigment epithelium (RPE), glial and Müller cells, it results in their proliferation. Particularly in RPE cells, their interaction with vitreous cytokines results in an epithelial to mesenchymal transformation allowing migration into the vitreous.28,29 RPE and/or Müller cells that transitioned into intravitreal fibroblasts along with extraocular fibroblasts introduced via trauma transdifferentiate into myofibroblasts.30 These myofibroblasts result in the contracture of cellular membranes pulling on the vitreous and in patients with RD this can prevent the retina from healing and reattaching.10,31
PVR results in a worse prognosis.14–17,33–33 In the WROTD, approximately 30% of ocular injuries had a final VA of less than 20/200, but in patients with PVR, approximately 80% had a final VA less than 20/200.21 As such, in our study, PVR development was found to be positively associated with a final VA less than 20/200 (P < .001) in multivariate analysis.
High myopia,34 open globe injuries,9,35 and vitreous hemorrhage9,36 have all been found in previous studies to be risk factors for the development of PVR. In our study, there were multiple risk factors characteristic of both closed and open globe injuries. However, critically associated injuries (P = .49) were not risk factors. Patients with more severe systemic injuries may not be at an increased risk for PVR, although this will require further study.
In patients with a RD, suffering an injury from an IOFB resulted in an increased risk for developing PVR (P < .001). Trauma related to IOFB onset remains a significant complication for future PVR development.37–41 In a previous study from the WROTD41 the risk factors for the development of PVR in patients with an IOFB injury were studied. In this study of all IOFBs in the WROTD,41 not just IOFBs in patients with RD as in the current study, IOFBs were not associated with a poor final visual outcome less than 20/200. An increased risk for development of PVR was noted in OTS 0 to 65 injury (P < .001), RD (P < .001), subretinal hemorrhage (P = .001), macular hemorrhage (P = .001), and posterior zone of injury (P = .002). For total number of IOFB (P = .97), volume of the largest IOFB per eye (P = .02) and length of time to removal (P = .11), there was not a significant associated risk with PVR development.
The timing of vitreoretinal surgical intervention after globe trauma is highly controversial. A recent systematic review of trauma in the civilian population looked at the timing of outcomes of vitreoretinal surgery after traumatic RD and found mixed results in regard to early (less than 7 days) intervention versus greater than 7 days surgical intervention.1 The rate of PVR in the review varied with those who had repair in less than 7 days ranging from 39% to 44%, whereas in those who had repair more than 7 days later was 5% to 56%. Another study in a civilian population found that if the interval of injury and vitrectomy was greater than 28 days there was a large risk for PVR development (odds ratio: 139.25; 95% CI, 1.13–6.52).35 A recent randomized, controlled study looking at vitrectomy in patients with open globe injuries, found that patients that underwent early vitrectomy (less than 4 days) versus delayed vitrectomy (10 to 14 days) had greater visual improvement (gain of 3 lines; 38 vs. 12%; P = .041) and lower rates of PVR (14 vs. 40%; P = .054).42 This study found no significant difference in the rate of RD between the two groups (86 vs. 88%; P = 1.0).42 In our study as discussed above, unfortunately only two-thirds of patients had a documented time to surgery making statistical analysis of this factor challenging. Our military populace is very different than the standard RD populace in that all are secondary to trauma, whereas in the civilian populace, only approximately 1% of RD is related to trauma.43 The patient population in our study is also highly unique as it is a young male population with serious injuries as evidenced by approximately 70% of eyes having an OTS score less than 65. These differences make comparisons between military and civilian trauma inappropriate, therefore making assumptions about timing for surgery based off military experience difficult.
A previous study found that macula-off RD is over four times more likely to lead to the development of PVR;44 however, in our study, macular status was not shown to be a significant risk factor for PVR development (P = .5). However, macular hemorrhage was found to be a significant risk factor for the development of PVR (P = .04). It is unclear to the authors why macular hemorrhage versus subretinal or intraretinal hemorrhage elsewhere in the retina was a risk factor for the development of PVR, and further study is needed. It is likely that the subretinal or choroidal hemorrhage was very large in these patients and included the macula but was not isolated to the macula like a subretinal neovascular membrane.
Thus, from a military perspective, if consideration for vitreoretinal surgery in a war zone becomes necessary, it is critical that the best supplies and equipment are available to manage these patients to ensure the best possibility of surgical reattachment. In our clinical practice, eyes that are at greatest risk for PVR formation — notable retinal detachment, presence of IOFB, and hemorrhage inclusive of the macula justify more urgent evacuation to mitigate the likelihood of PVR formation. Of note, these scenarios — specifically the presence of sub-macular hemorrhage — reduce visual prognosis, as well. That said, earliest possible vitreoretinal intervention must be considered in all cases of posterior segment trauma to minimize the likelihood of PVR formation. In future wars, if longer evacuations are anticipated have a vitreoretinal surgeon at a CSH or closer to the combat theater, such as at LRMC, may be necessary to ensure best patient outcomes.
Some vitreoretinal surgeons have proposed ophthalmic or systemic immunosuppressive or anti-inflammatory medications as an adjunct therapy to manage proliferative vitreoretinopathy.45–48 Although these treatments show promise, during the study period, they were generally not considered given the complex injury patterns suffered by the patients. Injuries typically included both anterior and posterior segment trauma, as well as significant systemic injuries. Anecdotally, we rarely used medications such as intravenous or oral steroids as a surgical adjunct in isolated cases. However, these cases were typically the most severe injuries, and any data analysis would be skewed by the severity of the injuries employed. Moreover, we did not collect use of adjunct medications such as periocular or systemic steroid usage or triamcinolone use during vitrectomy as variables in our dataset.
There were multiple limitations to this study. Early in OIF and OEF, there were no electronic health records making patient tracking from site of injury to WRAMC difficult. At times, the records were incomplete making data analysis complicated. In addition, initial VA were first taken when the patients were extubated, and this could have been multiple days after injury and/or surgery; thus, making OTS and initial VA outcomes difficult to evaluate. In addition, approximately 40% of patients had a traumatic brain injury making the initial visual acuity data potentially unreliable. Another critical limitation is that only 41 patients had a documented time to surgery, thus making it difficult to complete statistical analysis on the data related to timing of surgery. Lastly, it is possible that PVR developed in some patients prior to first intervention, however this was not recorded in the database.
To conclude, this paper represents a unique series of U.S. service members sustaining eye injuries with delayed initial RD repair (time to surgery was 22.56 days; range: 3 to 87 days). PVR is a common complication in combat ocular trauma occurring in nearly 10% of injured patients evacuated to WRAMC. Risk factors for PVR development varied and were characteristic of both open and closed injuries. In patients with RD, it is critical to ensure a successful reattachment. IOFB is a significant risk factor for subsequent development of PVR. Finally, nearly 80% of these patients had a final VA less than 20/200, and PVR is a significant cause of blindness in deployed U.S. service members with ocular trauma.