Intraocular foreign bodies (IOFB) are a. Intraocular foreign bodies. Factors influencing final visual outcome. Patients and Methods

Transcription

1 Intraocular foreign bodies. Factors influencing final visual outcome Christophe Chiquet 1, Jean-Christophe Zech 1, Philippe Denis 1, Patrice Adeleine 2 and Christiane Trepsat 1 Department of Ophthalmology 1, Edouard Herriot Hospital, Lyon, France Department of Biostatistics 2, Hospices Civils de Lyon, Lyon, France ABSTRACT. Purpose: To identify the prognostic factors of poor visual outcome (visual acuityæ6/240) in eyes with intraocular foreign bodies. Methods: The records of 95 consecutive patients were retrospectively reviewed for 6 years ( ). All eyes underwent a primary surgical repair and foreign-body removal (electromagnet or vitrectomy). The mean follow-up period was 25 months (6 72 months). Single analysis and multiple logistic stepwise regression analysis were performed to determine predictors of poor vision. Results: Thirty patients (31.6%) showed 6/240 or worse vision at the end of their follow-up period. Three significant predictive factors had independent and combined effects on post-operative visual outcome: a corneo-scleral entry wound (odds ratio (OR)Ω14.5, pω0.001), largest diameter of IOFB (ORΩ1.21, pω 0.01) and the presence of secondary retinal detachment (ORΩ9.48, pω0.0002). Post-operative complications included traumatic cataracts (51%), retinal detachments (28%) and phthisis bulbi (8%). Conclusion: Using multivariate analysis, corneo-scleral entry wound, largest diameter of foreign body and secondary retinal detachment were found to be predictors of poor visual outcome after intraocular foreign body removal. Our results suggest that patients with high-risk intraocular foreign body trauma should be candidates for pars plana vitrectomy rather than electromagnet procedure. Key words: intraocular foreign body perforating trauma ocular injury prognostic factors retinal detachment vitrectomy. Acta Ophthalmol. Scand. 1999: 77: Copyright c Acta Ophthalmol Scand ISSN Intraocular foreign bodies (IOFB) are a major cause of ocular trauma and of legal blindness. Young adults, especially men, are the most likely victims, as a result of industrial, agricultural or firearm injuries. Our goal was to analyse the visual outcome after IOFB extraction in order to identify the main prognostic factors of final vision. Previous studies have described various aspects of IOFB injuries or penetrating ocular trauma, including demographic (Tielsch et al. 1988, Klopfer et al. 1992, Baker et al. 1996), histopathological characteristics (Clearly & Ryan 1979a, 1979b, Winthrop et al. 1980, Punnomen 1990) or clinical findings and visual outcome (De Juan et al. 1983; Shock & Adams 1985; Williams et al. 1988; Punnomen & Laatikainen 1989; Behrens-Baumann & Praetorius 1989; Slusher 1990; Ahmadieh et al. 1994). Predictive factors of visual outcome could aid the clinician in choosing appropriate surgical and medical management of IOFB trauma and in counseling the patients. To establish predictive clinical variables for final vision, we correlated final visual acuity (VA) with standard clinical parameters using univariate analysis. A multiple logistic model was also applied to estimate the odds ratio and to predict final vision from the clinical factors. Patients and Methods A chart review was performed on the records of 95 patients with IOFBs admitted at Edouard Herriot Hospital (Lyon, France) between January 1990 and December Patient selection Patients were considered eligible for the retrospective study under the following criteria: the foreign body was retained in the eye and caused entrance laceration, the IOFB extraction was not performed elsewhere and the follow-up period was 6 months or more. Patients were excluded for any of the following reasons: perforating injuries (with or without IOFB), corneal foreign body, follow-up period less than 6 months after the injury. Examination Data related to the baseline examination systematically included: age and sex of the patient; cause and mechanism of the injury; type, size (largest diameter) and location of the IOFB; number of foreign bodies; pre-operative VA; prolapsed intraocular contents; hyphaema; iris and lens damage; vitreous haemorrhage; any retinal damage including retinal incarceration or detachment and types of retinal tears; presence of endophthalmitis. The details of the initial and subsequent surgeries were recorded: delay of therapy; method of foreign body removal; pars plana vitrectomy (PPV); pars plana lensectomy; scleral buckle procedure; use of internal tamponade; cryopexy; cataract surgery and enucleation. Initial surgical procedure All entry wounds were cleared of prolapsed material and repaired. For magnetic IOFB, a pars plana scleral incision 321

2 using an extra-ocular giant magnet was primarily proposed if the IOFB appeared magnetic, was located preretinally or in the vitreous, without major intraocular lesion. If the IOFB was non-magnetic, intraretinal, associated with dense vitreous haemorrhage or encapsulated, a vitrectomy was performed for IOFB removal (delay : 1 15 days). The delay between injury and initial surgical procedure also depended on the time of the patient s presentation at the hospital. All of these patients systematically received intravenous antibiotic therapy involving fluoro-quinolone (ofloxacine) associated to fosfomycin or piperillin, lasting at least 5 days. Subsequent surgery When appropriate, a second surgical procedure was performed such as cataract surgery, retinal detachment (RD) surgery or enucleation. Posterior chamber implantation was proposed in some cases. Argon laser photocoagulation around a retinal tear was performed when necessary, in eyes with relatively clear media and attached retina. Retinal tears in RD were localised and then treated with cryotherapy. If necessary external buckling procedures and C2F6 gas were then carried out. Outcome evaluation The outcome and the follow-up data reviewed were: anterior and posterior complications after surgical procedures; postoperative best-corrected VA and followup period. Follow-up data were obtained from the outpatient charts and from the referring ophthalmologist when the patient was postoperatively reviewed elsewhere. which factors were associated with either a good or a poor final vision outcome. To determine which of these variables were combined prognostic factors for visual outcome, logistic regression models were fitted using a backward stepping procedure. The odds ratio (OR) which provides an estimate of the relative risk of poor visual outcome was calculated. Multivariate logistic regression analysis was performed using factors that were found to be significant predictors of final visual outcome in univariate analysis (with p 0.25). In the estimated multiple logistic model, p is the probability of having a poor visual outcome and X is a clinical variable, which can have a value of 0, 1 or 2 (according to the coding). The logistic coefficient b i is a measure of the importance of each X i in the prediction of the probability of poor visual outcome after controlling for all other variables. For increased clarity, percentages were rounded to the nearest digit. Statistical significance was defined as a two-tailed P value of less than Results Our study included 95 patients (95 eyes) with a 6-month minimum follow-up period (range: 6 72 months, mean: 25 months). There were 88 men (92%) and 7 women. Their age varied from 10 to 79 years (mean, 31.6 years). There were 47 right eyes (49%). The hospitalisation delay after the injury ranged from a few hours to three days in 92 % of the cases. Fifty-two (54%) of the injuries were work-related and twenty-three per cent of the trauma were due to firearms. The IOFBs were sharp (35%), angular or round (36%), or irregular (29%). All IOFBs were metallic and were proved to be magnetic in 61 % of the cases. The size of IOFB was defined by its largest diameter (mean: 3.5 mm, range: 0.5 to 25 mm) and 73 % of IOFBs were smaller than or equal to 3 mm. A single IOFB was present in 92 % of the eyes. The commonest ocular findings upon admission were: corneal wound (66%), prolapse or damage of the iris (48%), hyphaema (42%), lens damage (42%), vitreous haemorrhage (46%) and retinal detachment (RD) (3%). The sites of perforation were frequently the cornea (52%), followed by the sclera (34%) and the association cornea sclera (15%). On admission, one patient presented endophthalmitis (septic retinopathy) three days after the injury. Pre-operative IOFB localisation was achieved most commonly by orbital x-rays (36 patients, 38 %) or by orbital x-rays associated with a computerised tomography scan (59 patients, 62%). The foreign body was embedded in the surface of the retina (23%), located in the vitreous (48%), in the sclera (16%) or in the anterior segment (13%). The pre-operative VA in the 95 admitted patients ranged from 6/6 to «no light perception» (Fig. 1). Surgical management Foreign body removal was attempted in each patient. Magnet extraction was per- Statistical methods The following prognostic variables were studied: age of patients, cause of injury (firearms, other mechanisms), size (millimetres), shape (sharp, angular or round, irregular) and final location of the IOFB (anterior segment, vitreous, retina or choroid), initial VA (Æ6/240, ±6/240), type of entry wound (corneal, corneo-scleral, scleral), initial hyphaema, vitreous haemorrhage, intraocular tissue prolapse, scleral wound, initial or secondary RD, post-traumatic cataract and the surgical procedure for IOFB removal (PPV). The statistical analysis was performed using the Statistical Package for the Social Science program (SPSS). Each factor was analysed by chi-square analysis (incorporating Yates correction) to determine Fig. 1. Visual acuity before and after surgery plotted against frequency. 322

3 formed initially in 56 eyes and was successful in 72% (40 eyes) of these cases. For IOFB removal, PPV was performed in 30 eyes with a median delay of 7 days after the injury, including the 15 cases of failure of magnet extraction. Twentyeight foreign bodies (93%) were successfully removed using this technique. Twenty-five IOFBs could be removed using forceps, without PPV or electromagnet, because of their location within the sclera or the anterior segment. Management of complicated IOFBs required secondary vitreous surgery in 14 cases. Secondary PPV was performed for the following reasons: post-operative RD (10 eyes, mean delay after traumaω2.5 months), vitreous haemorrhage (3 eyes, mean delayω1 month), and epiretinal membranes associated with vitreous haemorrhage (1 eye, delayω4 months). Twenty-seven eyes received scleral buckling and/or an encircling procedure. C2F6 gas was used for 23 patients. Traumatic cataract required pars plana lensectomy (23 cases, mean delay after trauma: 17 days), extracapsular cataract extraction with posterior IOL (4 cases, mean delay: 14 months) or phakoemulsification (2 cases, mean delay: 23 days). Two secondary enucleations were performed in our series. Post-operative Complications Traumatic cataract developed in 48 out of 95 patients (51%). Twenty of these 48 patients (42%) achieved a final VA better than or equal to 6/12. Twenty-seven patients (28%) developed delayed RD. This complication was the largest single cause of blindness: final VA was worse or equal to 6/30 in 68% of the cases. Two patients with total RD and no light perception could not be operated. Clinical factors listed in Table 1 were studied using chi-square analysis to determine risk factors of secondary RD. Firearm injury (pω0.01) and vitreous haemorrhage (pω0.012) were found to be significantly correlated with secondary RD. One patient presented a septic retinopathy on admission. No organisms were cultured. Six patients developed a traumatic glaucoma. Phthisis bulbi occurred in 8 patients. No patient developed sympathetic ophthalmia, siderosis or chalcosis bulbi. Visual outcome and predictive factors of final vision In case of cataract occurring as a late complication, visual functions were Table 1. Univariate analysis for 95 eyes with IOFB. Number Patients with poor visual of outcome (% or mean for Factor patients qualitative variable) p value Age (years) Firearms 0.02 O Yes (54.5%) O No (24.7%) Shape of IOFB O Sharp 33 6 (18.2%) O Irregular 28 8 (28.6%) O Angular or round (47.1%) Largest diameter of IOFB (mm) Number of IOFB IOFB location 0.88 O Anterior segment 12 3 (25%) O Vitreous body (32.6%) O Retina 22 8 (36.4%) O Sclera 15 4 (26.7%) Entry wound 0.03 O Corneal (32.7%) O Corneo-scleral 14 8 (57.1%) O Scleral 32 6 (18.6%) Initial visual acuity O Æ6/ (43.3%) O ±6/ (10.3%) Initial central corneal wound 0.01 O Yes (58.8%) O No (25.6%) Initial hyphaema O Yes (47.5%) O No (20%) Initial prolapse of intraocular tissue O Yes (71.4%) O No (20.3%) Initial intravitreous haemorrhage 0.17 O Yes (38.6%) O No (25.5%) Secondary cataract 0.03 O Yes (41.7%) O No (21.3%) Secondary retinal detachment O Yes (57.1%) O No (20.9%) IOFB extraction using pars plana vitrectomy 0.2 O Yes (40%) O No 40 9 (22.5%) evaluated after lens extraction. Thirtytwo patients (31%) had 6/240 or worse vision at the end of their follow-up period. Table 1 presents significant predictors of poor visual outcome (Æ6/240) identified by univariate analysis: firearm injury, shape and size of IOFB, initial VA (Æ6/240), corneo-scleral wound, initial hyphaema, prolapse of intraocular tissue, secondary cataract and secondary RD. With regard to good visual outcome (6/ 12 or better), the univariate analysis determined the same factors influencing the final visual outcome. The results of the multiple logistic regression analysis (Table 2) showed that three variables were significant predictive factors which had a combined effect on post-operative poor visual outcome: a corneo-scleral entry wound (odds ratioω 14.5), the largest diameter of IOFB (ORΩ1.21) and the presence of secondary retinal detachment (ORΩ9.48). Multiple logistic regression analysis indicated that similar independent factors predicted a good visual outcome (6/12 or better), i.e. absence of corneo-scleral entry wound (pω0.0013), size of IOFB (pω 0.01) and absence of secondary RD (pω ). 323

4 Table 2. Significant predictors of poor visual outcome for 95 eyes (Multivariate logistic regression analysis). Logistic coefficient Standard Exp (logistic Factor (j) Numerical variable X i (b i ) error p value coefficient) 1. Secondary RD 0ΩAbsence 1 1ΩPresence Entry wound 0ΩCorneal 1 1ΩScleral ΩCorneo-scleral IOFB Size Largest diameter (mm) Constant term bωª Estimated multiple logistic model: log p/(1ªp)ωª3.05π2.25 X 1 π2.67 X 2 π0.19 X 3 The odds ratio (Exp (logistic coefficient) provides a practical estimate of the relative risk of poor visual outcome. For instance, the risk of poor visual outcome is multiplied by a factor of 1.21 with IOFB size increases of 1 mm and multiplied by a factor of 14.5 when the patient had a corneo-scleral entry wound (when compared to a patient with corneal entry wound). IOFB: intraocular foreign body. RD: retinal detachment. Discussion Using univariate analysis, we showed that initial VA, type and size of IOFB, initial corneo-scleral wound, prolapse of ocular tissue or hyphaema were significantly associated with poor visual outcome. Consistent with previously reported clinical series, initial VA is an important prognostic indicator of visual outcome after penetrating trauma (Baker et al. 1996; De Juan et al. 1983; Sternberg et al. 1984; Hutton & Fuller 1984; Esmaeli et al. 1995) and after IOFB injury (Williams et al. 1988; Punnomen & Laatikainen 1989). Our analysis of predictive factors emphasizes and confirms that the visual outcome is strongly influenced by the nature of penetrating trauma and the extent of the initial wounds. As shown in our study, initial hyphaema was common (42%) and was found to be associated with poor final visual results. In other respects, our results and several reports (Pieramici et al. 1996; Williams et al. 1988; Punnomen & Laatikainen 1989; Hutton & Fuller 1984) demonstrated that the visual outcome is affected unfavourably by blunt injury, particularly after firearm accidents. In our series, a final VA better than 6/ 19 was obtained in 59% of our patients. Whichever method of extraction used, the percentage of patients with final VA better than or equal to 6/19 ranged from 39% to 66% in several studies (Percival 1972; Coleman et al. 1987; Williams et al. 1988; Pieramici et al. 1996). In the present study, we also used multivariate statistical analysis to identify independent predictive factors. The logistic model is suitable for analysing each risk factor independently and the possible statistical interactions between each variable. Using multiple logistic regression analysis, we found that the presence of an initial corneo-scleral entry wound, the IOFB size and post-operative RD were significant predictors of poor visual outcome. For example, our results may assist the ophthalmologist as they are based on the OR associated with the size of IOFB. Indeed, this OR provides a practical estimate of the relative risk of poor visual outcome and is multiplied by a factor of 1.21 with IOFB size increase of 1 mm. Several authors showed that the size of IOFBs is the most important factor in the prognosis and that the high enucleation rate is associated with a large IOFB (Roper-Hall 1954; Percival 1972; Johnston 1971). Consistent with previously reported clinical series of ocular trauma, wounds limited to the cornea had a better prognosis than corneo-scleral or scleral wounds (De Juan et al. 1983; Punnomen & Laatikainen 1989; Sternberg et al. 1984; Ahmadieh et al. 1993). Punnomen & Laatikainen (1989) observed that 70% of eyes with a posterior segment entry site became blind as compared with only 20% of those with an anterior segment entry site. In agreement with previous reports (Heimann et al. 1983; Karel & Diblik 1995), our results showed that RD was a crucial factor for poor visual outcome after IOFB injuries. Two multivariate analyses (Hutton & Fuller 1984; Brinton et al. 1982) demonstrated that eyes with RD had a worse prognosis in severe ocular trauma. In our study, retinal impact injury, entry site or tissue prolapse did not appear significantly correlated to secondary RD. Nevertheless, we found that firearm injury and vitreous haemorrhage were indicative of secondary RD. Percival (1972b) noted that RD occurred in 22% of injured eyes with IOFB, particularly if large foreign body, vitreous haemorrhage or vitreous loss were observed at the initial examination. A more recent study on 1564 patients with ocular trauma (Cardillo et al. 1997) showed that a vitreous haemorrhage, a persistent intraocular inflammation or a posteriorly located wound were independent predictive factors for the development of post-traumatic proliferative vitreoretinopathy. Severe vitreous haemorrhage is associated with intraocular fibroblastic proliferation and traction RD in experimental (Clearly & Ryan 1979a, 1979b; Winthrop et al. 1980) and clinical studies (Johnston 1971; Percival 1972b; Faulborn et al. 1977) and leads to a poor prognosis (Brinton et al. 1982; De Juan et al. 1983; Ahmadieh et al. 1993). According to several reports (Ahmadieh et al. 1993; De Juan et al. 1984); clinical management of IOFB injuries using PPV may salvage the most severely injured eyes. Experimental studies have confirmed the effectiveness of PPV in the treatment of posterior segment trauma (Clearly & Ryan 1981; Gregor & Ryan 1983a, 1983b), especially the prevention of traction RD. In our retrospective study, magnet extraction was used for IOFB extraction when possible and initial PPV was used for non magnetic IOFB, retinal or encapsulated IOFB, or media opacities obscuring the view of posterior segment, including cases of dense vitreous haemorrhage. Secondary PPV was performed for post-operative complications, such as persistent vitreous haemorrhage, epiretinal membranes or RD. As observed in other studies (Williams et al. 1988; Punnomen & Laatikainen 1989), we did not find a statistically significant difference in final visual outcome between eyes with IOFB removed using PPV or magnetic extraction. In that respect, final VA after IOFB injury seemed to be preferentially determined by the type and extent of ocular damage rather than the type of surgery. In our study, severe IOFB injuries associated 324

5 with initial vitreous haemorrhage (pω 0.01) or caused by BB pellet (p 0.005), i.e. lead pellet that measures 0.44 cm in diameter, were more common in the PPV group. Since the visual outcome is known to be unfavourably affected by these factors, we think that PPV may be beneficial for eyes severely injured with IOFB. This conclusion was also drawn by De Juan et al. (1984); Esmaeli et al. (1995) and Pieramici et al. (1996) in the management of penetrating ocular injury. Nevertheless, according to the final VA results, electromagnet extraction remains a viable treatment for selected IOFB, particularly in the early period after the ocular trauma (Coleman et al. 1987). References Ahmadieh H, Soheilian M, Sajjadi H, Azarmina M & Abrishami M (1993): Vitrectomy in ocular trauma. Factors influencing final visual outcome. Retina 13: Ahmadieh H, Sajjadi H, Azarmina M, Azarmina M, Soheilian M & Baharivand N (1994): Surgical management of intraretinal foreign bodies. Retina 14: Baker RS, Wilson MR, Flowers CW, Lee DA & Wheeler NC (1996): Demographic factors in a population-based survey of hospitalized, work-related, ocular injury. Am J Ophthalmol 122: Behrens-Baumann W & Praetorius G (1989): Intraocular foreign bodies. 297 consecutive cases. Ophthalmologica 198: Brinton GS, Aaberg TM, Reeser FH, Topping TM & Abrams GW (1982): Surgical results in ocular trauma involving the posterior segment. Am J Ophthalmol 93: Cardillo JA, Stout JT, LaBree L, Azen SP, Omphroy L, Cui JZ, Kimura H, Hinton DR & Ryan SJ (1997): Post-traumatic proliferative vitreoretinopathy. The epidemiologic profile, onset, risk factors, and visual outcome. Ophthalmology 104: Clearly PE & Ryan SJ (1979a): Experimental posterior penetrating eye injury in the rabbit: II. History of wound, vitreous and retina. Br J Ophthalmol 63: Cleary PE & Ryan SJ (1979b): Method of production and natural history of experimental posterior penetration eye injury in the rhesus monkey. Am J Ophthalmol 88: Clearly PE & Ryan SJ (1981): Vitrectomy in penetrating eye injury. Results of a controlled trial of vitrectomy in an experimental posterior penetrating eye injury in the rhesus monkey. Arch Ophthalmol 99: Coleman DJ, Lucas BC, Rondeau MJ & Chang S (1987): Management of intraocular foreign bodies. Ophthalmology 94: De Juan E Jr, Sternberg P Jr & Michels RG (1983): Penetrating ocular injuries: types of injuries and visual results. Ophthalmology 90: De Juan E, Sternberg P, Michels RG & Auer C (1984): Evaluation of vitrectomy in penetrating ocular trauma : a case-control study. Arch Ophthalmol 102: Esmaeli B, Elner SG, Schork MA & Elner VM (1995): Visual outcome and ocular survival after penetrating trauma. Ophthalmology 102: Faulborn J, Atkinson A & Olivier D (1977): Primary vitrectomy as a preventive surgical procedure in the treatment of severely injured eyes. Br J Ophthalmol 61: Gregor Z& Ryan SJ(1983a): Complete and core vitrectomies in the treatment of experimental posterior penetrating eye injury in the rhesus monkey. I clinical features. Arch Ophthalmol 101: Gregor Z& Ryan SJ(1983b): Complete and core vitrectomies in the treatment of experimental posterior penetrating eye injury in the rhesus monkey. II histologic features. Arch Ophthalmol 101: Heimann K, Paulmann H & Tavakoli U (1983): The intraocular foreign body: principles and problems in the management of complicated cases by pars plana vitrectomy. Int Ophthalmol 6: Hutton WL & Fuller DG (1984): Factors influencing final visual results in severely injured eyes. Am J Ophthalmol 97: Johnston S (1971): Perforating injuries: a five year study. Trans Ophthalmol Soc UK 91: Karel I & Diblik P (1995): Management of posterior segment foreign bodies and longterm results. Eur J Ophthalmol 5: Klopfer J, Tielsch JM, Vitale S, See LC & Canner JK (1992): Ocular trauma in the United States. Eye injuries resulting in hospitalization, 1984 through Arch Ophthalmol 110: Percival SPB (1972a): A decade of intraocular foreign bodies. Br J Ophthalmol 56: Percival SPB (1972b): Late complications from posterior segment intraocular foreign bodies. Br J Ophthalmol 56: Pieramici DJ, MacCumber MW, Humayun MU, Marsh MJ & de Juan E (1996): Openglobe injury. Update on types of injuries and visual results. Ophthalmology 103: Punnomen E & Laatikainen L (1989): Prognosis of perforating eye injuries with intraocular foreign bodies. Acta Ophthalmol (Copenh) 67: Punnomen E (1990): Pathological findings in eyes enucleated because of perforating injury. Acta Ophthalmol (Copenh) 68: Roper-Hall MJ (1954): Review of 555 cases of intra-ocular foreign body with special reference to prognosis. Br J Ophthalmol 38: Shock JP & Adams D (1985): Long term visual acuity results after penetrating and perforating ocular injuries. Am J Ophthalmol 100: Slusher MM (1990): Intraretinal foreign bodies: management and observations. Retina 10 (Suppl): S50 S54. Sternberg P, De Juan E, Michels RG & Auer C (1984) : Multivariate analysis of prognostic factors in penetrating ocular injuries. Am J Ophthalmol 98: Tielsch JM, Parver L & Shankar B (1988): Time trends in the incidence of hospitalized ocular trauma. Arch Ophthalmol 107: Williams DF, Mieler WF, Abrams GW & Lewis H (1988): Results and prognostic factors in penetrating ocular injuries with retained intraocular foreign bodies. Ophthalmology 95: Winthrop SR, Cleary PE, Minckler DS & Ryan SJ (1980): Penetrating eye injuries: a histopathological review. Br J Ophthalmol 64: Received on September 10th, Accepted on January 23rd, Corresponding author: Professor Christiane Trepsat Department of Ophthalmology Edouard Herriot Hospital Place d Arsonval Lyon cedex 03 France Tel: Fax: