Corneal Fine Needle Diathermy with Adjuvant Bevacizumab to Treat Corneal Neovascularization in Children

CLINICAL SCIENCE

Corneal Fine Needle Diathermy With Adjuvant Bevacizumab to Treat Corneal Neovascularization in Children

Uri Elbaz, MD, Kamiar Mireskandari, MBChB, FRCSEd, FRCOphth, PhD, Carl Shen, BMSc, and Asim Ali, MD, FRCSC

orneal neovascularization (CN) in the pediatric popula- Purpose: fi To report the outcomes of corneal ne needle Ction has various etiologies including blepharokeratocon- diathermy (FND) with adjuvant intrastromal and subconjunctival junctivitis (BKC),1 bacterial ulcer, herpes keratitis, anesthetic bevacizumab injection for corneal neovascularization (CN) in cornea, and exposure keratopathy. The cornea in its native children. state is completely devoid of vascular components thereby 2,3 Methods: Medical records of all children who had undergone retaining immunoprivilege. In the event of infectious FND with adjuvant bevacizumab injection were reviewed retro- keratitis, the resulting immune response and new vessel spectively. Treatment efficacy was evaluated by changes in visual formation can help to eradicate the invading organism but can acuity, regression of CN, and clearing of lipid deposits with the cause stromal necrosis, thinning, and scarring that impair aid of slit-lamp color images that were taken before surgical vision. Additionally, irregular astigmatism, edema, haze, and intervention and at last follow-up visit. Postoperative complica- lipid deposition can also occur. Despite elimination of the tions were recorded and served to assess the safety of the inciting cause, CN may often persist. In noninfectious procedure. conditions such as BKC, corneal anesthesia, or exposure keratopathy, the mechanism of CN is poorly understood but is Results: Nine eyes of 9 patients were included in the study. The most likely related to weakness in the integrity and function of mean age of the patients was 8.4 6 4.2 years (4–15 years) and the the blood vessel barrier provided by the limbal stem cells.4 mean follow-up time was 18.7 6 12.2 months (5–35 months). Three In adults, CN has been reported to respond to numerous eyes had a history of herpes simplex keratitis (HSK), 3 eyes had treatment modalities including topical corticosteroids and complete corneal anesthesia, 2 eyes had CN following suture tract nonsteroidal anti-inflammatory drops,5 corneal argon laser infection after corneal transplant for HSK scar and limbal dermoid photocoagulation,6 photodynamic therapy,7 fine needle dia- excision, and 1 eye had blepharokeratoconjunctivitis. After treat- thermy (FND),8,9 electrolysis-needle cauterization,10 repeated ment, 8 eyes had complete CN resolution, and 1 eye with corneal bevacizumab injections (Avastin; Genentech, San Francisco, anesthesia following brain tumor resection had partial regression in CA),11 and ligation of blood vessel origin.12 No treatment vessel distribution and size. Lipid deposition clearance lagged modality has been reported to be completely effective in behind CN resolution. Mean duration of CN before treatment was eradicating blood vessels from the cornea and often, repeated 15.3 6 14.0 months (1–37 months). Mean corrected distance visual treatments are required. Recently, encouraging results of FND acuity before and after surgery was 0.66 6 0.31 and 0.50 6 0.37 and topical and/or subconjunctival bevacizumab administra- logMAR, respectively (P = 0.02). tion before planned keratoplasty were reported in adults.13 In Conclusions: contrast, there are very few reports on CN treatment in Corneal FND with adjuvant bevacizumab injection 8,14 is effective at treating sectorial corneal vessels in children. children. Bevacizumab and FND each have their own disadvan- Key Words: corneal neovascularization, bevacizumab, fine needle tages when applied separately. Bevacizumab is postulated to diathermy be ineffective in more mature chronic vessels that have already acquired a pericyte covering15 and are hence less dependent on (Cornea 2015;34:773–777) vascular endothelial growth factor. FND is thought to induce release of proangiogenic factors,16 collagen shrinkage with potential damage to surroundings of the treated vessel, and Received for publication October 1, 2014; revision received January 27, insufficient response when collateral vessels are still patent.8 It 2015; accepted February 1, 2015. Published online ahead of print March 26, 2015. also does not prevent further new vessel growth. By using From the Department of Ophthalmology and Vision Sciences, Hospital for Sick a combination therapy, we hypothesized that each modality Children, University of Toronto, Toronto, ON, Canada. would compensate for the limitation of the other. The purpose The authors have no funding or conflicts of interest to disclose. of our study was to report the outcomes of a novel combined Reprints: Uri Elbaz, MD, Department of Ophthalmology and Vision Sciences, Hospital for Sick Children, 555 University Ave, Toronto, ON approach of corneal FND with adjuvant intrastromal and M5G 1X8, Canada (e-mail: [email protected]). subconjunctival bevacizumab injection (off-label use) in Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved. a series of pediatric patients with CN. To the best of our

Cornea Volume 34, Number 7, July 2015 www.corneajrnl.com | 773

knowledge, this is the only series reporting this technique in a history of corneal anesthesia had undergone permanent children with CN and among the first reports to present an tarsorrhaphy before the combined treatment. effective treatment modality for this disorder in children. RESULTS PATIENTS AND METHODS Nine eyes of 9 patients were included in this study. The The study was approved by the Research Ethics Board mean patient age was 8.4 6 4.2 years (4–15 years) and the of The Hospital for Sick Children Toronto, Ontario, Canada. mean follow-up time was 18.7 6 12.2 months (5–35 months). We retrospectively reviewed the charts of all patients less The etiology of CN was as follows: 3 eyes had a history of than 18 years who had undergone combined FND with herpes simplex keratitis (HSK), 3 eyes had complete corneal intrastromal and subconjunctival bevacizumab injection. Data anesthesia, 2 eyes had CN following suture tract infection were collected on patient demographics, causes and indica- after lamellar corneal graft for HSK scar and for limbal tions for CN treatment, corrected distance visual acuity dermoid excision, and 1 eye had blepharokeratoconjunctivitis (CDVA), procedural details, and complications. Slit-lamp (BKC). Two of the 3 patients with corneal anesthesia had photographs before surgical intervention and at last follow-up resection of brain tumors affecting the trigeminal nerve in one visit were used to evaluate treatment efficacy. The main patient and trigeminal and facial nerves in the other. The third outcomes measures were regression of abnormal corneal patient had isolated complete corneal anesthesia with a small blood vessels, resolution of corneal lipid deposits, and trigeminal ganglion found on brain imaging and presented changes in CDVA over time. Response to treatment as with recurrent corneal ulcers with subsequent CN. measured by vessel regression and resolution of lipid deposits Patient 4 with HSK had deep anterior lamellar kerato- was graded as complete, partial, no change, or worse than plasty (DALK) performed at the same time as FND and baseline. For statistical analysis, Snellen visual acuity was adjuvant bevacizumab injection; otherwise, all procedures converted to logarithm of the minimum angle of resolution were performed in isolation. All patients presented with (logMAR) units and Student t test was used for comparison of a single (4 eyes) or double (5 eyes) vessel main trunk CDVA before surgery and at last follow-up visit. P , 0.05 involving a single corneal quadrant only. Three patients had was considered significant. multiple previous FND treatments without bevacizumab injections with recurrence of their CN. Mean duration of CN before treatment was 15.3 6 14.0 months (range, 1–37 Surgical Technique months). Lipid deposition was observed in 5 eyes before Corneal FND was performed as previously described treatment. by Pillai et al.8 Briefly, all surgeries were performed in Following treatment, 8 of 9 eyes treated (88.9%) had children under general anesthesia. The needle tip from complete CN resolution with subsequent resolution of corneal a stainless steel needle detached from a 10-0 monofilament edema within 1 month of treatment. They all maintained nylon suture was inserted intrastromally approximately 1 mm corneas devoid of vessels throughout a mean follow-up from the limbus to the level of corneal new vessels. Unipolar period of 17.0 6 11.9 months (5–35 months) after surgery. diathermy was then lightly tapped against the needle and Complete cessation of blood flow was noted on postoperative served to heat the needle in its corneal track to a degree day 1 in all these eyes. Only 1 eye (patient 3) with corneal where blanching of the vessels was noticeable with minimal anesthesia and exposure keratopathy following brain tumor stromal collagen shrinkage. An attempt was made to resection had partial regression in vessel distribution and size. cannulate the vessels or at least disrupt their integrity to Complete lipid deposition clearance was noted in 3 of 5 eyes further enhance the thermal effect. This was repeated as and this lagged behind CN resolution. In 1 patient (patient 4), needed to disrupt all the blood vessels. Once cauterization lipid deposits were removed with removal of the old DALK was complete, bevacizumab 25 mg/mL was injected into the graft. A temporary increase in lipid deposits was noted in 2 deep stroma (up to 2.5 mg in 0.1 mL per each corneal patients, patient 2 (Fig. 1B) and patient 9 (Fig. 2C) at 3 and quadrant involved) in proximity to the vessels until whiten- 3.5 months after surgery, respectively. In patient 2, lipid ing around the vessels was visible and the remainder into the deposition had resolved completely by the 13-month visit subconjunctival space adjacent to areas of CN. (Fig. 1C) and still persisted in patient 9 five months after Postoperatively, all patients were treated with topical the treatment. moxifloxacin 0.5% drops (Vigamox; Alcon Laboratories, Inc, Mean CDVA before and after surgery was 0.66 6 0.31 Mississauga, ON, Canada) for 1 week and prednisolone 1% and 0.50 6 0.37 logMAR, respectively (P = 0.02). Patient 4 drops (Pred Forte; Alcon Laboratories, Inc) with a slow taper who had concomitant FND, bevacizumab injections, and over a couple of months. In steroid responders, prednisolone DALK was excluded from CDVA analysis to isolate the effect was replaced with fluorometholone 1 mg/mL (FML; Allergan, of the investigated treatment on visual acuity. All the remaining Markham, ON, Canada). Patients with a history of herpes 8 eyes treated either maintained or improved their CDVA. stromal keratitis were treated with a weight-adjusted dose of Intracorneal hemorrhage (Fig. 3A) was observed after oral acyclovir before the procedure and indefinitely after the treatment in 2 patients and cleared completely after 1 month. combined treatment. For patients with BKC, all efforts were Figure 3B shows complete resolution of CN with long follow- made to control the underlying disease with conservative up period of more than 2 years in a patient with HSK history measures and systemic antibiotics as needed. All patients with (patient 5) and previous CN recurrence following 2 FND

FIGURE 1. Corneal neovascularization resolution phases. Slit-lamp photography (upper row) with time-matched Visante OCT (Carl Zeiss, Oberkochen, Germany) images (lower row). A, Preoperative images showing neovascularization pattern, sectorial corneal haze (A) and localized stromal edema (A9). B, Clinical phase II: 3 months after surgery, a complete resolution of CN with more prominent lipid deposition. OCT (B9) shows significant decrease in corneal edema with more stromal hyperreflectivity corresponding to the increase in lipid deposition. C, Clinical phase III: 13 months after surgery, no recurrence of CN, mild localized scar, and completely resolved lipid deposits. OCT image (C9) shows reorganization of stroma with thickened posterior lamella (slowly resuming normal stromal thickness), less anterior hyperreflectivity, and epithelial thickening compensating for the mild stromal loss.

treatment attempts. Similarly, Figure 2 (patient 9) shows 10–84 years) with a complete resolution rate of 57.1% after presurgery (Fig. 2A) and postsurgery (Figs. 2D, E) images a follow-up period of 6 to 24 months. A more recent study21 demonstrating complete resolution of CN and corneal edema demonstrated similar long-term efficacy results after FND following the combined treatment. solely with 68.1% regression rate in CN after 5 years. Koenig No infections or corneal melting were noted. The et al13 recently reported the outcomes of a combined treatment procedure was very well tolerated with no reports of of FND with topical and/or subconjunctival bevacizumab significant pain after surgery. administration before planned keratoplasty in 16 eyes. A significantly greater reduction in corneal NV was found in patients who had subconjunctival bevacizumab injection at DISCUSSION the time of FND together with bevacizumab eye drops Corneal neovascularization has detrimental effects on compared with FND only and postoperative bevacizumab vision and poses a major challenge to the ophthalmologist eye drops. The reported complete CN resolution rate was especially in the young population where visual development 68.8% (11 of 15 eyes); however, they did not evaluate the is immature. Various treatment modalities have been sug- effect of intrastromal injection. The bevacizumab intrastromal gested in the management of this complex problem. However, injection maintains the highest stromal concentration over none are completely effective and high recurrence rates time and prevents regrowth of CN as was shown recently in necessitate repeated treatment in many cases. Very few an animal model by Dastjerdi et al.22 studies were reported in children with variable success.8,14 We selected the longer lasting stromal injection as our Here, we report on the combination of corneal FND with favored technique in these children aiming for a one-time adjuvant intrastromal and subconjunctival bevacizumab in- treatment under general anesthesia. This was done immedi- jections in a pediatric population with satisfactory results after ately after FND treatment to saturate the corneal stroma and a mean follow-up period of more than 18 months. enhance the duration of effect of bevacizumab. In children, in Recently, bevacizumab has been gaining popularity in the contrast to adults, subconjunctival injections cannot be management of CN. In a previous meta-analysis,17 it was found administered repeatedly as an office procedure but only that both topical and subconjunctival bevacizumab achieve under general anesthesia, and our goal was to limit the significant reduction in the area of CN. Other small cohort number of treatments required. Of note, 2 of our patients with studies18–20 investigating the efficacy of intrastromal bevacizu- complete resolution of CN following combined treatment mab injection demonstrated further encouraging results. previously had multiple CN recurrences when FND alone Another study11 focusing on combined subconjunctival and was used. In these patients, once bevacizumab was given by peripheral intrastromal bevacizumab injections in 12 eyes also intrastromal and subconjunctival injection following FND, showed marked CN regression following treatment; however, we did not observe CN recurrence even after extended none of the treated eyes showed complete resolution of CN. follow-up beyond 2 years. This demonstrates the additive Fine needle diathermy was first introduced by Pillai effect of the intrastromal bevacizumab injection specifically et al8 in a mixed adult and pediatric population (age range, in children where CN likely behaves more aggressively than

FIGURE 2. Pre and postoperative images of patient 9. A, Preoperative image. Nasal corneal neovascularization in the patient with a history of herpes simplex keratitis showing 2 vessel stems crossing the corneal center with lipid deposits (white arrow) and mild temporal edema (slit view). B, Ten days after combined treatment. Cessation of blood ﬂow and discontinuation of blood vessels tract is demonstrated. C, D, and E, Three and a half months after combined treatment. Accentuation of lipid deposition (C), resolved temporal edema with minimal thinning (D, slit view), and complete CN resolution, quiet conjunctiva, and moderate residual scarring (E).

in adults and therefore presents with higher recurrence rates complete resolution rate with the FND-alone treatment shown after FND alone. in adult population (88.9% vs. 57.18–68.1%,21 respectively). In a small case series,14 the efficacy of yellow dye laser The only patient in our series who had partial CN resolution in treating CN in awake pediatric patients was demonstrated, following treatment (patient 3) had a history of complete although with a short follow-up period (3 and 14 months in 2 corneal anesthesia following neurosurgery with subsequent cases and unspecified follow-up time for the third case). damage to the trigeminal and facial nerves. The microenviron- Interestingly, the only patient in this case series with a partial ment of the persistently insensate cornea in combination with response to yellow dye laser treatment had a history of HSK. In poor lid function and impaired blinking is detrimental to the our cohort, all patients with HSK experienced complete ocular surface and may have contributed to reduced success in resolution of CN following treatment with significant improve- this case.23 In addition, the only patient who had partial lipid ment in vision. Moreover, our overall complete resolution rate deposition clearance (patient 9) has not had enough follow-up shows clinically significant superiority over the reported time to enable complete lipid clearance and we anticipate that

FIGURE 3. Immediate and late postoperative images of patient 5. A, Intracorneal hemorrhage 1 day after surgery. B, Twenty-six months after combined treatment. Complete CN resolution with moderate residual scarring.

without the presence of corneal vessels this will be resolved 3. Niederkorn JY. Immune privilege and immune regulation in the eye. Adv completely over time as was demonstrated for patient 2. Immunol. 1990;48:191–226. 4. Chang JH, Gabison EE, Kato T, et al. Corneal neovascularization. Curr Our clinical observations have led us to hypothesize Opin Ophthalmol. 2001;12:242–249. a triphasic clinical response to FND treatment (Fig. 1). In the 5. Harvey PT, Cherry PM. Indomethacin v. dexamethasone in the initial phase, there is fast regression of blood vessels due to suppression of corneal neovascularization. Can J Ophthalmol. 1983;18: the imposed thermal injury. In the second phase, the 293–295. endothelial cell pumping mechanism easily overcomes the 6. Nirankari VS, Baer JC. Corneal argon laser photocoagulation for neovascularization in penetrating keratoplasty. Ophthalmology. 1986; corneal edema in the area of the old leaky vessels. However, 93:1304–1309. once the corneal edema resolves, lipid deposits are more 7. Epstein RJ, Stulting RD, Hendricks RL, et al. Corneal neovasculariza- accentuated, similar to the process occurring after absorption tion. Pathogenesis and inhibition. Cornea. 1987;6:250–257. of macular edema and formation of hard exudates in the 8. Pillai CT, Dua HS, Hossain P. Fine needle diathermy occlusion of corneal vessels. Invest Ophthalmol Vis Sci. 2000;41:2148–2153. retina. Hence, a phase of worsening lipid deposition may be 9. Faraj LA, Elalfy MS, Said DG, et al. Fine needle diathermy occlusion of observed before it is slowly cleared over a period of a few corneal vessels. Br J Ophthalmol. 2014;98:1287–1290. months in the third phase. 10. Wertheim MS, Cook SD, Knox-Cartwright NE, et al. Electrolysis-needle The limitations of our study include its retrospective cauterization of corneal vessels in patients with lipid keratopathy. – nature and lack of a control group. This is mostly due to the Cornea. 2007;26:230 231. fi 11. Yeung SN, Lichtinger A, Kim P, et al. Combined use of subconjunctival rarity of CN in children. In addition, we have not quanti ed and intracorneal bevacizumab injection for corneal neovascularization. the vascularization area before and after treatment as was Cornea. 2011;30:1110–1114. suggested previously24; however, all patients had involvement 12. Chang JH, Garg NK, Lunde E, et al. Corneal neovascularization: an anti- of a single quadrant solely, and only 1 patient with complete VEGF therapy review. Surv Ophthalmol. 2012;57:415–429. 13. Koenig Y, Bock F, Kruse FE, et al. Angioregressive pretreatment of corneal anesthesia had partial response to treatment. More- mature corneal blood vessels before keratoplasty: fine-needle vessel over, the sample size was too small to determine if age, coagulation combined with anti-VEGFs. Cornea. 2012;31:887–892. gender, etiology, previous procedures, and preoperative 14. Lueder GT, Culican S. Yellow dye laser treatment of vascularized characteristics impacted the therapeutic outcome in our corneal stromal scars in pediatric patients. Arch Ophthalmol. 2008;126: – cohort. Finally, we have not assessed corneal sensation in 564 566. fi 15. Cursiefen C, Hofmann-Rummelt C, Kuchle M, et al. Pericyte recruitment all patients and therefore the effect of corneal sensation de cit in human corneal angiogenesis: an ultrastructural study with clinico- on the response to treatment could not be evaluated. pathological correlation. Br J Ophthalmol. 2003;87:101–106. In conclusion, we show that corneal FND with adjuvant 16. Junghans BM, Collin HB. The limbal vascular response to corneal injury. intrastromal and subconjunctival bevacizumab injection is an An autoradiographic study. Cornea. 1989;8:141–149. 17. Papathanassiou M, Theodoropoulou S, Analitis A, et al. Vascular effective method to treat corneal vascularization in the endothelial growth factor inhibitors for treatment of corneal neovascula- pediatric population. We believe this technique simplifies rization: a meta-analysis. Cornea. 2013;32:435–444. the postoperative drop regimen by avoiding the need for 18. Oh JY, Kim MK, Wee WR. Subconjunctival and intracorneal bevacizu- prolonged bevacizumab or steroid drops or repeat anesthesia mab injection for corneal neovascularization in lipid keratopathy. – for repeated subconjunctival injections in children. Cornea. 2009;28:1070 1073. 19. Hashemian MN, Zare MA, Rahimi F, et al. Deep intrastromal bevacizumab injection for management of corneal stromal vascularization after deep anterior lamellar keratoplasty, a novel technique. Cornea. ACKNOWLEDGMENTS 2011;30:215–218. 20. Mohammadpour M. Deep intrastromal injection of bevacizumab for the We thank Cynthia VandenHoven, BAA, CRA and Leslie management of corneal neovascularization. Cornea. 2013;32:109–110. MacKeen, BSc, CRA, Medical Imaging Specialists from the 21. Trikha S, Parikh S, Osmond C, et al. Long-term outcomes of Fine Needle Department of Ophthalmology and Vision Sciences, Hospital Diathermy for established corneal neovascularisation. Br J Ophthalmol. for Sick Children, University of Toronto, Toronto, Ontario, 2014;98:454–458. Canada for all the images presented herein. 22. Dastjerdi MH, Sadrai Z, Saban DR, et al. Corneal penetration of topical and subconjunctival bevacizumab. Invest Ophthalmol Vis Sci. 2011;52: 8718–8723. REFERENCES 23. Ferrari G, Hajrasouliha AR, Sadrai Z, et al. Nerves and neovessels 1. Hamada S, Khan I, Denniston AK, et al. Childhood blepharokeratocon- inhibit each other in the cornea. Invest Ophthalmol Vis Sci. 2013;54: junctivitis: characterising a severe phenotype in white adolescents. Br J 813–820. Ophthalmol. 2012;96:949–955. 24. Kirwan RP, Zheng Y, Tey A, et al. Quantifying changes in corneal 2. Streilein JW. Ocular immune privilege: therapeutic opportunities from an neovascularization using fluorescein and indocyanine green angiography. experiment of nature. Nat Rev Immunol. 2003;3:879–889. Am J Ophthalmol. 2012;154:850–858.