Article Text
Abstract
Purpose To assess long-term functional results after vitrectomy for stages 4 and 5 retinopathy of prematurity (ROP).
Methods Retrospective analysis of data from eyes that underwent lens-sparing vitrectomy (LSV) or combined lensectomy and vitrectomy (LV) and had a follow-up of at least 5 years.
Results Eighty-eight eyes of 65 infants who underwent LSV or LV for stage 4 or 5 ROP were included in the study. The mean follow-up was 6.9 years. The anatomic success rate was 89% (17/19) for stage 4A, 63% (24/38) for stage 4B, and 42% (13/31) for stage 5. Forty-five eyes (51%) had measurable visual acuity (VA). The approximate Snellen VA equivalent was 20/550 for stage 4A, 20/1600 for stage 4B, and 20/4000 for stage 5. The remaining 39% (34/88) had light perception or no light perception. Owing to neurological conditions, VA in nine eyes (10%) could not be measured. Anatomic and visual outcomes were not associated with surgical technique.
Conclusions Surgery for stage 4A ROP has better visual success. Acceptable vision for stage 4B and poor for stage 5 ROP, despite retinal reattachment, is thought to be secondary to irreversible injury to either the retina or visual nervous pathways, or both.
- Retina
- Vision
- Treatment Surgery
Statistics from Altmetric.com
Introduction
Retinopathy of prematurity (ROP) is a complex condition characterised by onset of vascular abnormalities in the developing retina, a major cause of visual impairment and blindness in premature neonates.
Screening and treatment guidelines established by the Cryotherapy for ROP and Early Treatment for Retinopathy (ETROP) studies have reduced the rates of stage 4 and 5 ROP and its resultant visual loss.1 ,2 Despite early treatment of high-risk cases, unfortunately 12% of the eyes progress to stage 4 or 5 ROP and need surgical repair.2 Open-sky vitrectomy,3 scleral buckling,4 ,5 lensectomy and vitrectomy (LV) with or without scleral buckling,6 ,7 and lens-sparing vitrectomy (LSV)8 are the surgical techniques that have been used for the management of retinal detachments associated with ROP.
Some studies have reported anatomic outcomes after vitreoretinal surgery for stage 4 and stage 5 ROP.3–8 However, anatomic success may not equate to functional success, and few studies have focused on assessing functional outcomes.9–17 Most of these publications are limited by small sample size, surgery performed by multiple surgeons, a short follow-up period, and reporting bias.
This study describes a retrospectively analysed series of patients with stages 4 and 5 ROP treated by single surgeon in Turkey. All the children were treated in a consistent manner, and were followed-up closely postoperatively to enable early intervention. We aimed to perform a meticulous assessment of visual function by using a standard protocol and to investigate whether such an organised attempt at appropriate diagnosis, a current surgical approach, and postoperative management could yield visual results in the long-term.
Methods
We retrospectively reviewed the records of infants with stage 4 and 5 ROP who underwent vitrectomy between January 1996 and December 2010 performed by a single surgeon (MK) at a tertiary referral centre. Detailed informed consent was taken from the parents before each surgical procedure. The study protocol was approved by the Institutional Review Board of Sisli Memorial Hospital, Istanbul. The study was in accordance with the principles of the Declaration of Helsinki.
The clinical charts were retrospectively reviewed to collect the following data: birth weight, gestational age, sex, clinical characteristics of the ROP, timing of surgery, intraoperative procedures, postoperative complications, follow-up duration, anatomic status, and visual acuity (VA) at the final visit. Patients who had been lost to follow-up and those with insufficient medical records were excluded.
All the infants were examined preoperatively with indirect ophthalmoscopy, and B-mode orbital ultrasonography was performed by an experienced radiologist if indicated.
Surgical procedures
All patients underwent 3-port vitrectomy through a pars plicata approach, with or without lensectomy.8 ,15 After lateral canthotomy and a standard conjunctival opening, a 2.5 mm, 20-gauge infusion cannula was placed 1 mm posterior to the limbus (ie, through the pars plicata) in the inferotemporal quadrant. Additional sclerotomies were made at the 10 and 2 o'clock positions, either limbal or 1 mm posterior to the limbus. A lensectomy was performed if the lens was found to impede adequate management of anterior-posterior traction, or if there was limiting intraoperative visualisation. Vitreous and membranes were removed, and anterior-posterior and circumferential tractions were released using the Associate 2500 vitrectomy system (DORC, Zuidland, Netherlands). For visualisation of the residual vitreous and posterior hyaloid membrane, after removal of the anterior and core vitreous, 0.2–0.3 mL (40 mg/mL) triamcinolone acetonide aqueous suspension was injected into the mid-vitreous cavity, followed by partial or complete removal of the posterior hyaloid. Other surgical procedures included retrolental membrane dissection, epiretinal membrane delamination, forceps membrane peeling, fluid-air exchange, and fluid-gas exchange. No intravitreal anti-vascular endothelial growth factor (anti-VEGF) was given preoperatively or intraoperatively. The parents were advised to use a topical antibiotic and corticosteroid drops four times a day for 1 month during the postoperative period.
Follow-up examination
All the infants were routinely examined on day 1 after surgery, then 1 week, 1 month, and every 3 months for a year, and then yearly. All the patients were closely followed up in our institution and were referred to a paediatric ophthalmologist for comprehensive postoperative visual rehabilitation. Ametropic and anisometropic amblyopia were managed by correction with contact lenses or spectacles and subsequent eye patching.
VA measurements were performed using the Tumbling E or Early Treatment Diabetic Retinopathy Study (ETDRS) charts. Following a standardised protocol, the ETDRS charts were used with a maximum distance of 4 m, but the distance was reduced to 1 m or to 0.5 m if needed to permit letter identification in an eye with poor vision.18 In some patients, VA testing was not performed because of developmental limitations (the chart-measured vision as ‘fixes and follows’ or ‘central-steady-maintained’) or if both eyes had vision of light perception (LP) or worse.
Anatomic success was defined as complete retinal attachment for stage 4A ROP, partial retinal detachment for stage 4B, and at least posterior pole attachment for stage 5.
Data were analysed using the Kruskal-Wallis test, Pearson χ2 test, Fisher’s exact test, Mann-Whitney test, and Spearman's rank correlation depending on the variables. A value of p<0.05 was considered statistically significant. Statistical analyses used SPSS V.20.0 (SPSS, Inc, Chicago, Illinois, USA).
Results
Of the 136 patients who underwent vitrectomy for stage 4A, 4B or 5 ROP whose charts were reviewed, 88 eyes of 65 patients who had been followed up for at least 5 years were included (table 1).
None of these patients had previously undergone intravitreal anti-VEGF injection or a scleral buckle procedure. Of the 65 infants included, 34 were male (52%). Overall, the mean±SD gestational age at birth was 28.4±2.3 weeks, mean±SD birth weight was 1212±355 g, mean±SD postconceptional age at the time of surgery was 43.1±6.2 weeks, and mean±SD interval between initial laser and surgery was 6.5±4.8 weeks. The mean follow-up period was 6.9±1.9 years. Of the 88 eyes, 51 (58%) had received prior retinal ablative treatment.
LSV was performed on 39 eyes (44%) and combined LV was performed on 49 eyes (56%) (table 2). Six eyes after LSV (4 stage 4A and 2 stage 4B) and one after LV (stage 5) required reoperation (mean 9 weeks later) due to vitreous haemorrhage or reproliferation. Favourable anatomic and visual outcomes were achieved in only two (stage 4A) of these seven eyes despite the second surgical procedure. There were no significant differences in anatomic success between LSV and LV for stage 4A (93% vs 75%, p=0.39) or for stage 4B (67% vs 57%, p=0.35).
Forty-five eyes (51%) had measurable VA at the final visit and 34 (39%) had vision of LP or no LP (NLP). Owing to various neurological conditions, VA could not be measured in nine eyes (10%) that had the ability to fix and follow. Table 3 summarises the long-term visual results after vitrectomy for stages 4 and 5 ROP.
Of the 11 eyes (58% of all stage 4A eyes) for which a VA could be measured, 42% (8/19) had VA of 20/200 or better, and 16% (3/19) had ambulatory VA of 20/2000 or better. Of the 22 eyes (58% of all stage 4B eyes) for which a VA could be measured, seven (32%) had VA of 20/400 or better. The mean approximate Snellen VA equivalent of the 11 eyes that underwent surgery for stage 5 ROP and had measurable VA (35% of all stage 5 eyes) was 20/4000 (range 20/2000–20/20 000).
The visual outcomes of stage 4A were better than those of stage 5 ROP (p=0.001). Earlier retinal ablation was associated with better visual results for stage 4B ROP (p=0.04). There was no statistically significant difference in the visual results between LSV and LV for stage 4B (p=0.39), but numbers were too small to allow statistical comparison for stage 4A.
Glaucoma occurred most often in aphakic eyes (92%, 11/12), and was controlled with topical ocular hypotensive agents. One patient underwent glaucoma surgery. Table 4 summarises the long-term postoperative complications after vitrectomy for stages 4 and 5 ROP.
Discussion
While 51% of eyes with stage 4A (63%), stage 4B (58%), and stage 5 (35%) had measurable VA, the remaining 49% had no functional vision or exhibited behaviour consistent with form vision, but had developmental comorbidity precluding measurement of VA. Of the eyes with measurable VA following LSV or LV for stage 4A ROP, 42% (8/19) had VA of 20/200 or better and 16% (3/19) had ambulatory VA of 20/2000 or better. The anatomic success rate for stage 4A ROP was 89%.
Stage 4A ROP without surgical intervention leads to poor visual outcomes, and only 20% of eyes develop VA of 20/200 or better.19 Singh et al12 focused on long-term visual outcomes (mean age at last follow-up, 7.4 years) and reported slightly better results for LSV with or without scleral buckle for stage 4A ROP. Of the 23 eyes that underwent surgery for stage 4A ROP, 21 (91%) were successfully attached. The mean VA of the 16 eyes (70%) for which a Snellen VA could be measured was 20/189, and 12 eyes (75%) had VA of 20/400 or better. The remaining seven eyes (30%) had LP or NLP, or fixing and following. It was concluded that the placement of a scleral buckle in conjunction with LSV did not adversely affect VA. Prenner et al9 and Lakhanpal et al10 reported VA outcomes better than most reported results. Although the average age was only 3.5 years, Prenner et al reported that 11 (48%) of the 23 eyes achieved VA of 20/40 or better after undergoing LSV for stage 4A ROP. The overall mean VA was 20/58 and no child had a VA worse than 20/200. Lakhanpal et al10 achieved reattachment in all eyes and reported mean visual VA of 20/62 (range 20/50–20/100) for stage 4A ROP (14 eyes) treated with LSV. The mean age at VA testing was only 3.7 years. These results suggest that eyes with stage 4A ROP have the potential for good visual results after LSV. By contrast, in the only prospectively followed cohort of patients with stage 4A ROP from the ETROP study at 6 years, vitrectomy was associated with macular attachment in only five of 16 eyes (31%).11 Only one of 16 (6%) eyes had VA of 20/200 or better, six had VA worse than 20/200, and eight had LP or NLP. The authors considered that the worse outcomes might be related to the indications for surgery, timing of surgery, use of lensectomy/vitrectomy as well as LSV, and the experience of the vitreoretinal surgeons.
The staging of ROP is based on the clinical findings determined by ophthalmoscopic examination. Although LSV in stage 4A was associated with better functional and anatomical outcomes, it is possible that some of the stage 4A cases in our study population might actually be stage 4B. In addition, 21% of stage 4A eyes were managed with LV as well, suggesting a very progressive stage 4A with dragging at least of the photoreceptors. This could explain why these eyes achieved VA outcomes worse than previously reported.
The anatomic and visual outcomes following surgery for stage 4B ROP have generally been poorer. We found that 22 of the 38 eyes (58%) had measurable VA, with seven of these (32%) having VA of 20/400 or better. The anatomic success rate for stage 4B ROP in our study population was 63%. El Rayes et al13 reported results similar to ours in 24 eyes having LSV and 32 eyes having LV for stage 4B ROP. The structural outcomes after LSV and LV were 75% and 72%, respectively. VA following surgery ranged between 20/300 and hand motions (median follow-up period, 3.3 years). VA of 20/800 or better was noted in 24 eyes (43%). Lakhanpal et al10 reported VA outcomes for stage 4B substantially better than most published results: mean VA 20/200 (range 20/126–20/502) for 16 eyes treated with LSV. VA was tested at an early age (only 3.7 years). Again, by contrast, in the ETROP study, of 10 eyes undergoing vitrectomy with or without scleral buckling for stage 4B ROP, only two (20%) achieved measurable VA, which was worse than 20/200 at 6 years.11 In another study, Singh et al12 reported a mean VA of 20/846 for four of the nine eyes that underwent surgery for stage 4B ROP (all of the four eyes had an LSV with scleral buckling) at an average age of 8.6 years. Three of these four eyes had VA of 20/800 or better.
Anatomic and functional outcomes following vitreoretinal surgery for stage 5 ROP are worse than those for stage 4A or 4B. In our study, only 11 (35%) of 31 eyes had measurable vision, an approximate Snellen VA equivalent of 20/4000 (range 20/20 000–20/2000). The remaining 20 (65%) had VA of LP or NLP, or fixing and following only. The anatomic success rate for stage 5 ROP was 42%. Cusick et al15 has reported the largest series for infants with stage 5 ROP having vitreoretinal surgery. With regard to anatomic outcomes for the 608 eyes, data were classified as follows: 28% success, 5% partial success, 55% failure, and 11% lost eye. From 14% of the eyes achieving form vision, only eight (4%) of 183 eyes achieved VA better than 5/200, and 85% of the eyes had LP or NLP. In the ETROP study, 10 eyes had vitrectomy with or without scleral buckling for stage 5 ROP. No eye achieved measurable VA. Of the 10 eyes, one had LP and eight had NLP at the last visit at 6 years.11 Kono et al16 reported a case series of 51 eyes with stage 5 ROP, managed with LV as the initial procedure. Of the 33 eyes with unsuccessful initial surgery, 15 underwent a second surgical procedure, including an encircling scleral buckle and vitrectomy with gas or silicone oil tamponade with endophotocoagulation. The mean follow-up period was 64.4 months. Retinal reattachment was achieved in 47% of cases and VA was better than 20/600 in three eyes. Choi and Yu17 reviewed the VA and anatomic outcomes of 87 eyes that had undergone vitrectomy for advanced ROP, 78 of which had stage 5 ROP. Vitrectomy with lensectomy was performed in 80 eyes, and LSV in seven eyes. After a mean follow-up period of 35.6 months, anatomic success was achieved in 57% of cases. With regard to VA outcomes, LP was achieved in 48%, fixing and following was noted in 17%, and 7% of eyes were able to identify form.
Our study has several limitations, such as the single-centre retrospective design with a single surgeon performing both preoperative and postoperative evaluation. There is no description of features of ROP, including circumferential clock hours of fibrovascular tissue, and aggressive posterior ROP, preoperatively. We had a significant number of patients for whom follow-up data (>5 years) could not be obtained. The patients who were lost to follow-up may have had poor functional and anatomic outcomes. We also had several patients with additional neurological abnormalities that prevented reporting of VA.
Snellen letters have been regarded as the gold standard for determining resolvable VA.20 VA has been determined in previous studies by a variety of means, Teller or Allen optotypes being preferred.9 ,10 ,13 However, it has been suggested that Allen optotypes overestimate VA in children.20 The lack of standardisation in evaluation makes it difficult to compare the outcomes of different studies.
In conclusion, LSV and LV for stage 4A may have favourable anatomic results leading to functional vision in the long-term. Anatomic and visual outcomes are acceptable for stage 4B and poor for stage 5 ROP. The treatment of ROP should be weighted toward avoiding advanced ROP. Appropriate screening, timely peripheral ablation and vitreoretinal surgery remain the best alternatives for the management of ROP. Large prospective randomised trials and standardisation of VA testing would be helpful to investigate the indications for each surgical technique, optimal timing of surgery, and postoperative management at each stage of ROP.
Acknowledgments
We thank O Uysal for assistance with statistical analysis.
References
Footnotes
Contributors MK, MH, ISM and AS were involved in the conception and design of the study. MK, MH and AS were involved in the analysis, interpretation and critical revision of the article. MK, MH, and ISM were involved in the drafting of the manuscript. MK, MH, ISM and SA were involved in the final approval of the article. MH and ISM were involved in the data collection and literature research.
Competing interests None declared.
Parental consent Obtained.
Ethics approval Institutional Review Board of Sisli Memorial Hospital, Istanbul.
Provenance and peer review Not commissioned; externally peer reviewed.