Article Text
PDF
Abstract
Purpose To review the published literature on outcomes of keratolimbal allograft (KLAL) for the surgical treatment of limbal stem cell deficiency (LSCD) and corneal blindness after severe corneal chemical injury.
Methods Literature searches were conducted in the following electronic databases: MEDLINE, EMBASE, Science Citation Index, CINAHL, LILACS and the Cochrane Library. Standard systematic review methodology was applied. The main outcome measure was the proportion of eyes with best-corrected visual acuity (BCVA) ≥20/200 at last follow-up. Other measures of allograft success were also collected.
Results We identified six reports in which KLAL outcomes in the eyes after chemical injury could be distinguished. There were no randomised controlled studies. The outcomes of KLAL in 36 eyes of 33 patients were analysed. One study with seven eyes did not specify KLAL follow-up specific to chemical injury. Median postoperative follow-up for the other 29 eyes in 26 patients was 42 months (range 6.2–114 months). In the same 29 eyes, 69% (20/29) had BCVA ≥20/200 at the last follow-up examination. Eighty-nine per cent of all eyes (32/36) underwent penetrating keratoplasty simultaneous or subsequent to KLAL.
Conclusions The number of studies where outcomes of KLAL in eyes with severe corneal chemical injury could be discerned was limited, and variability was observed in outcome reporting. The quality of evidence to support the use of KLAL in LSCD in severe chemical corneal burns was low. Standardisation and longer follow-up are needed to better define evidence-based best practice when contemplating surgical intervention for blindness after corneal chemical injury.
PROSPERO registration number CRD42017054733.
- trauma
- cornea
- stem cells
- treatment surgery
Statistics from Altmetric.com
Introduction
Corneal limbal stem cells continually replenish the corneal epithelium to maintain normal corneal clarity and homeostasis.1 2 When limbal stem cells are damaged by trauma or disease, limbal stem cell deficiency (LSCD) can ensue.3 This condition manifests as a spectrum depending on the degree of limbal stem cell damage, and ranges from punctate corneal epitheliopathy, to persistent corneal epithelial defect, corneal neovascularisation and frank conjunctivalisation of the cornea. The ocular surface in such cases is unstable, leading to decreased vision, corneal stromal scarring and sometimes stromal ulceration. Thus, restoration of limbal function in LSCD is necessary for recovery of useful vision.
Severe LSCD after chemical corneal injury typically requires surgical management. However, standard surgical management for corneal opacity, that is, central penetrating keratoplasty (PK), in cases of LSCD after chemical injury typically leads to grim outcomes.4 Because central keratoplasty does not restore corneal limbal stem cell function in these conditions, the donor graft, like the native cornea it replaces, remains susceptible to subsequent ulceration, infection and failure. To address the problem of LSCD, various cell-based therapies have evolved. Numerous surgical procedures, including limbal autograft, keratolimbal allograft (KLAL), living-related conjunctival allograft (lr-CLAL) and allogeneic cultivated limbal epithelial stem cell transplant (CLET), are based on the premise of stem cell replacement.5–8 In unilateral LSCD, stem cells from the contralateral normal eye can be harvested to treat the diseased eye in a relatively uncomplicated technique known as simple limbal epithelial transplantation.9 10 In bilateral LSCD, because the contralateral eye is also damaged, stem cell restoration currently requires use of limbal tissue from cadaveric donor corneas, or from living relatives, in conjunction with immunosuppressive therapy to prevent tissue rejection and maintain the integrity and function of the allogeneic graft.8 Keratoprosthesis is another surgical option in severe bilateral LSCD. It does not require functional corneal stem cells and obviates systemic immunosuppression. However, complications with this procedure in the setting of chemical ocular injury, in particular severe alkali corneal burns,11 have slowed its adoption.
Ocular chemical injuries are frequently bilateral and often severe, and affect a relatively young population,12–15 with the US incidence rate for chemical ocular burns recently estimated from emergency room discharge coding data to be 51.10 new cases per million persons per year.14 It was previously estimated that eyes with severe chemical ocular burns undergo eight surgical interventions on average, with significant long-term consequences for the overall quality of life in affected patients.15 An expert consensus is lacking whether eyes with severe, bilateral chemical injuries experience benefit from stem cell-based surgical therapies. To address this gap in evidence-based management of chemical ocular injury with secondary corneal blindness, we performed a systematic review of the reported outcomes of KLAL in eyes with chemical injury.
Methods
This systematic review was conducted as per the Meta-analysis of Observational Studies in Epidemiology guidelines.16 The systematic review protocol was registered at the Prospective Register for Systematic Reviews (PROSPERO; http://www.crd.york.ac.uk/PROSPERO/display_record.php?ID=CRD42017054733).
Search strategy
The search strategy for this review is reported in online supplementary appendix A. The search was conducted in April 2017 in the following electronic databases: MEDLINE, EMBASE, Science Citation Index, CINAHL, LILACS and the Cochrane Library. A search was also conducted on guidelines.gov for any guidelines published on this topic. PROSPERO was searched for ongoing or recently completed systematic reviews. All reports published before 1 April 2017 were screened. No specific dates or years were used as time limits. No language restrictions were used. Reference lists were scanned to identify additional applicable studies.
Supplementary file 1
Inclusion and exclusion criteria for considering a study for review
Reports in English Language were considered for analysis. If after review of the full text we were unable to discern the outcomes of KLAL in LSCD specific to ocular chemical injury as opposed to other aetiologies, the report was excluded from the systematic review. Reports where it was not possible to differentiate outcomes between eyes with chemical and thermal burns were also excluded. Letters, conference abstracts, case reports, review articles, editorials and animal studies were excluded. Reports where eyes with chemical injuries were a part of a larger series but only one eye with chemical injury was suitable for analysis were excluded.
Inclusion criteria for an individual case for analysis
Eligibility criteria necessary for including a case in the systematic review were (1) documentation of severe ocular chemical injury, (2) age ≥18 years, (3) preoperative best-corrected visual acuity (BCVA) <20/200, (4) intervention with KLAL and (5) follow-up of ≥6 months after KLAL surgery.
Article selection
Of all the studies that were suitable for inclusion, the titles and abstracts were screened by two authors (SSS and HNS). Disagreements were resolved by consensus of all four authors. Selected full-text articles were retrieved and assessed for eligibility.
Data extraction
From the full-text studies that were suitable for inclusion, information was recorded on a data extraction form. Qualitative data recorded included study design, study population, criteria by which LSCD was defined and diagnosed, clinical outcome, type and duration of immunosuppression, postsurgical intervention, and complications. Individual patient data were analysed from each study. Quantitative data extracted included the number of eyes with ocular chemical injury with clearly defined outcomes, BCVA, length of follow-up and number of ocular surgeries after KLAL. The primary outcome measure was the proportion of eyes with BCVA ≥20/200 at the last postoperative visit. Data were aggregated at the level of each individual patient who fulfilled the inclusion criteria. A descriptive synthesis was undertaken.
Quality assessment
To rate individual studies based on study design, a scale based on the Scottish Intercollegiate Guidelines Network (www.sign.ac.uk/assets/qrg50.pdf, accessed 1 April 2017) was used. Two masked reviewers (SSS, HNS) independently assessed the quality of all the included studies using a validated checklist consisting of 19 questions. Any disagreements were resolved by consensus or arbitration by the senior author (JC). This checklist was developed by using a combination of two quality appraisal checklists specifically devised for interventional case series.17 18 Items that were applicable to this study were used.
Results
A total of 1448 articles of potential interest were identified by the original literature search (figure 1). No ongoing or recently completed systematic reviews were found on PROSPERO. No guidelines on KLAL after chemical injury could be found on www.guidelines.gov. After removal of the duplicated records, 1092 records were screened. After reviewing the title and abstract, 1025 were excluded. Sixty-seven full-text articles were then assessed for eligibility. A total of six reports met our criteria for inclusion in this systematic review and were further analysed.8 19–23 Individual studies were rated (figure 2, table 1). There were no randomised controlled studies; all were non-comparative case series. Post-KLAL outcomes for 36 eyes with chemical injuries could be evaluated in these six studies over a median follow-up of 42 months (range 6.2–114 months). Of the 33 patients, 31 were male. Other descriptive information for each study is presented in table 2. One study that reported outcomes for KLAL in 23 eyes with varied indications reported a mean post-KLAL follow-up for all 23 eyes, but did not specify post-KLAL follow-up periods for the seven eyes operated on for LSCD after chemical burn (seven patients).
Flow chart detailing the literature search approach for studies on outcomes of keratolimbal allograft (KLAL) for limbal stem cell deficiency following corneal chemical injury.
Bar graph showing the proportion of studies meeting specific quality measures, for systematic review of keratolimbal allograft outcomes in chemically injured eyes with limbal stem cell deficiency.
Studies on keratolimbal allograft for chemical injuries: quality items per study
Descriptive information of studies included for analysis
Three out of the six studies evaluated eyes with both unilateral and bilateral chemical injuries, while the other three studies did not mention if KLAL was performed on patients with unilateral or bilateral chemical injuries. Four studies failed to mention if the chemical injury was due to acid or alkali. In the other two studies, all the ocular injuries in cases meeting inclusion criteria for this systematic review were secondary to alkali agents. None of the studies mentioned the grade of chemical burn in the acute stage. One study documented the severity of LSCD in eyes in the chronic stage,20 but this was not mentioned in the other studies. The time interval between the chemical burn and KLAL surgery was mentioned in only one study.21
LSCD was assessed clinically in all six studies. Impression cytology was performed in four studies. However, the results of impression cytology were presented in detail in only one study.19
Prior ocular surgical interventions were documented in five out of six studies. Out of the six studies surveyed, four studies described outcomes in eyes that received KLAL as the primary limbal stem cell transplantation procedure, while two studies focused on outcomes of KLAL in eyes that had undergone prior limbal stem cell transplantations. Overall, prior to KLAL, 13 eyes underwent 23 PKs (average 1.77 per eye) (range 1–6), 6 eyes underwent 7 KLALs, 11 eyes underwent cataract extraction, and four eyes underwent glaucoma procedures. Three out of six studies clearly mentioned the number of eyes with pre-existing glaucoma.8 19 22
The surgical approach was described in detail in all six studies, with only minor variation in technique between studies. Briefly, abnormal fibrovascular tissue on the recipient ocular surface is excised. A cadaveric donor corneoscleral rim is obtained after central trephination of a donor cornea. The posterior lamella of the donor is excised, and the donor limbus sutured into the recipient limbal area.19 The medications and dosages given for immunosuppression were described in all studies, but treatment duration varied widely. Out of the six studies, four studies clearly described a uniform immunosuppression plan for all patients, while two studies used different immunosuppression regimens in different patients and did not clearly state which patients received which form of immunosuppression. The most common systemic immunosuppressive agent given was oral ciclosporin, described in five out of six studies, at doses ranging from 2.5 to 10 mg/kg daily, and for various durations ranging from 1 week preoperatively to indefinitely after surgery. Other systemic agents given included oral tacrolimus (1 mg/kg/ day) and oral mycophenolate mofetil (1 g/day). In five studies, these agents were administered along with a systemic corticosteroid immediately after surgery in the form of intravenous dexamethasone (8 mg/day) or intravenous methylprednisolone (1 mg/kg), and continued in the form of oral prednisone (1 mg/kg daily) for a month following surgery. Indicated periodic monitoring for blood levels of immunosuppressive agents was described in three out of six studies. Criteria for discontinuation of systemic immunosuppressive agents were mentioned in only two out of six studies. Topical ocular ciclosporin in addition to systemic immunosuppression was used in all eyes in two out of six studies. In one study, topical ciclosporin was administered in selected eyes, but detailed information on individual patients was not provided.
The main outcome measure of this systematic review was the proportion of eyes with postoperative BCVA ≥20/200 at the last recorded postoperative visit. Postoperative visual acuity specific to chemical ocular injury was clearly described for individual patients in all studies except one. Other than improvement in visual acuity, five out of six studies mentioned additional measures of KLAL success, including a clear and stable corneal epithelium on clinical examination without epithelial defect and reduced stromal vascularisation at the end of the follow-up period. Decreased pain was considered a positive outcome measure in one study,21 but this was assessed retrospectively from the medical records of KLAL recipients. Evaluation of the corneal epithelium postoperatively with impression cytology to verify a corneal phenotype was done in one study,19 but specific results were not discussed.
Six studies comprising 36 eyes with chemical burns were analysed for postoperative visual acuity ≥20/200. One study with seven eyes of seven patients did not describe individual follow-up times or postoperative visual acuities. In the remaining five studies, 20 of 29 (69%) eyes had BCVA ≥20/200, with a median follow-up of 42 months (range 6.2–114 months) (table 2). KLAL was considered ‘successful’ by the authors of the studies reviewed in 19 of 30 (63.33%) eyes. One study comprising six eyes did not assess the outcomes in terms of success or failure.
Among the 36 eyes receiving KLAL, 32 (88.9%) also underwent a central keratoplasty. Of the 36 eyes, 31 (86.11%) received a full-thickness PK (11 simultaneous to KLAL, 15 after KLAL and 5 not specified), while one eye underwent lamellar keratoplasty after KLAL. The time between KLAL and subsequent central keratoplasty ranged from 5 weeks to 14.8 months. Concurrent cataract surgery was performed in 8 of the 11 eyes that also received a PK at the same time as KLAL, while in one eye, cataract surgery was performed at the time of subsequent PK. These data were not available in all studies. A glaucoma drainage device was implanted concurrently with KLAL in one eye and after KLAL in four eyes. Further surgical interventions, such as repeat KLAL, repeat PK and further surgeries to reduce intraocular pressure, were required in 22 of 36 (61.1%) eyes, although one study did not clearly mention postoperative surgical interventions. After the primary KLAL procedure, a KLAL rejection was suspected in seven eyes when focal engorgement of vessels was seen at the limbus. Eight repeat KLAL procedures were performed in four eyes. Rejection of the central corneal graft was noted in 12 of 32 (37.5%) eyes, and a repeat PK was required in eight eyes. Post-KLAL rejection, other alternative stem cell procedures (lr-CLAL, CLET) were performed in two eyes. Glaucoma was mentioned as a postoperative complication in three out of six studies, but was not discussed in detail in any study. From the data that were available, it is evident that at least four eyes required a glaucoma drainage device post-KLAL, while it was medically managed in the rest. Immunosuppression-related complications were mentioned for one patient who had persistent hypertension and hyperbilirubinaemia believed to be related to oral tacrolimus administration.
Our quality assessment checklist used the following answers: ‘yes’, ‘no’ and ‘unclear’ (table 1). The answer ‘yes’ represented a positive measure of quality among all 19 questions (table 1, figure 2). Out of six studies, four (66.67%) clearly stated the objective of the study. Out of six studies, four (66.67%) collected data retrospectively, while one study mentioned that data were collected prospectively and one study failed to comment on the manner in which data were collected. Three studies did not clearly mention the institution from which the patient data were collected, while three studies collected it from a single centre. The recruitment period was mentioned in four (66.67%) out of six studies. All six studies mentioned the characteristics of all the patients included in the study. Criteria for including a patient in each study were not clearly discernible in any of the six studies. In only one out of six studies was it clear from the baseline data presented that participants entering the study had a similar degree of LSCD severity. Prognostic factors, for example, pre-existing glaucoma and eyelid deformities, both were clearly identified in only one (16.67%) out of six studies. Two out of six studies mentioned that KLAL surgery was performed by a single surgeon, while four studies did not specify who performed the surgeries. Out of six studies, five (83.33%) clearly described the additional surgical interventions. Outcome measures were established prior to the intervention and mentioned in the introduction or the methods sections in three (50%) out of six studies.
The persons who assessed outcomes were not blinded to the intervention in any of the six studies. BCVA ≥20/200 at final postoperative visit could be discerned in eyes with chemical injuries in five (83.33%) out of six studies. In one study, we could not determine the final visual acuity because the data were not presented specific to patients with chemical burns. In five out of six studies, BCVA specific to patients with chemical injury was described both before and after KLAL surgery. Three out of six studies (50%) had a mean follow-up period of more than 3 years. None of the six studies reported the number of patients lost to follow-up. Adverse events were reported in all six studies.
Discussion
Blinding corneal opacity after chemical injury represents a significant therapeutic challenge. To identify the best clinical practice approach to corneal blindness after chemical ocular injury that is evidence-based, we performed a systematic review of published studies of KLAL in eyes with LSCD after chemical injury. The retrospective nature of most studies and the lack of standardisation between studies generated difficulty in the assessment. Most studies we reviewed included patients with LSCD caused by a spectrum of diseases. Outcomes specific to chemical injuries were often difficult to discern, limiting the number of studies that could be evaluated in a systematic fashion. In some studies, it was impossible to differentiate the outcomes of eyes with chemical burns from eyes with thermal burns.24–27 Existing literature suggests that eyes may respond differently to chemical and thermal injuries.28 29 Taking this into consideration, we sought to review cases of chemical injury alone. The pathogenesis of corneal injury in alkali injury is different from that of acid injury,30 which suggests that post-KLAL success rates may differ for these two injuries. In four out of six studies, the pH (acid or alkali) of the chemical ocular injury was not differentiated between cases, and in the other two studies all the patients who met our inclusion criteria had alkali injuries. Therefore, we could not draw conclusions as to whether KLAL produced different outcomes in alkali as compared with acid burns. We also reviewed reports in which both autografts and allografts were performed, in which we could not calculate outcomes specific to KLAL.31–34
The degree of postinjury LSCD very likely influences the likelihood of success of subsequent interventions, with patients with partial LSCD experiencing good outcomes with amniotic membrane transplantation alone.35 While our aim was to focus on KLAL outcomes in severe LSCD after chemical ocular injury, most studies did not specify the severity of LSCD prior to KLAL. Therefore, we only included eyes with preoperative BCVA <20/200 in this review. In some studies, outcomes were mentioned in eyes with less than 6 months of follow-up after KLAL. These eyes were excluded from analysis, as we considered <6 months to be too short a period to determine success.19 36 Inclusion and exclusion criteria were not clearly delineated in most studies. Two studies did not consider the outcome of primary KLAL uniformly in all eyes and reported outcomes of only repeat KLAL in some eyes. Negative outcomes of KLAL in these eyes could have been influenced by prior sensitisation to foreign antigens. Almost two-thirds (63.8%) of the eyes receiving KLAL had undergone prior major surgical interventions including prior PK, which could also have influenced the outcome of KLAL in these eyes. Other unreported variables that may have influenced the success of KLAL in the reports we reviewed included the type of injury and its severity, and the acute care provided to the patient at the time of the burn. In addition, burn-related abnormalities of the eyelids and tear film, including entropion/ectropion, dry eye, symblepharon and keratinisation, can affect surgical outcomes,30 but were not universally described in the six studies. Only 50% of the studies we reviewed reported KLAL outcomes after a mean follow-up of more than 3 years. Shorter follow-up times would be expected to result in better reported outcomes.
Importantly, chemical ocular injury can predispose eyes to glaucoma.37 38 Because glaucoma tends to be worse in alkali injury and those of greater severity, the same eyes most likely to develop LSCD and be considered for KLAL surgery would be expected to have the highest risk of glaucoma.38 Treatment of acute chemical ocular injury typically includes topical corticosteroids; KLAL recipients also typically receive topical and systemic corticosteroids, further contributing to a concern about corticosteroid-induced glaucoma in this patient cohort. The prevalence of preoperative glaucoma in those patients receiving KLAL and the incidence of glaucoma postoperatively were not reported in half of the studies we reviewed. None of the studies mentioned postoperative evaluation of the optic disc. Under these circumstances, it was difficult to determine whether failure to achieve BCVA ≥20/200 in KLAL recipients was due to particularly severe preoperative LSCD, recurrent LSCD due to allograft failure or rejection, severe preoperative and/or postoperative glaucoma, adnexal abnormalities, or related in some way to complications of prior or subsequent interventions.
Improvement in BCVA was used to define KLAL success in all the studies we reviewed. Because KLAL may improve the quality of the corneal epithelium without a reduction in stromal scarring, and subsequent PK may be necessary to improve vision in some eyes after KLAL, visual acuity may not be the optimal primary outcome measure in the assessment of KLAL effectiveness. The timing of PK relative to KLAL may also influence the visual recovery. One report suggested that performing PK with KLAL at the same session was associated with markedly reduced KLAL survival after 2 years compared with when PK was performed afterwards, although this difference was not statistically significant due to insufficient case numbers.39 Another report suggested a tendency for the KLAL combined with PK to fail earlier than the KLAL followed by PK (failure on average at 12.5 months vs 20.5 months, respectively).21 These results were not limited to eyes with chemical injury. In our review, outcomes of KLAL with simultaneous PK versus KLAL with delayed PK could not be compared because of insufficient information.
Acute limbal allograft rejection after KLAL can be difficult to identify24 and can be missed without follow-up examination at the close intervals typical of a prospective clinical trial. An important secondary outcome measure in all KLAL studies reviewed was the postoperative appearance and stability of the corneal epithelium. However, none of the examiners were masked to the intervention, and preoperative and postoperative impression cytology to demonstrate the preoperative absence of goblet cells and postoperative restoration of a corneal phenotype were performed in only a few of the 36 eyes in the six studies. Immunosuppression regimens after KLAL also differed widely, and it was impossible to determine the impact of a specific immunosuppressive regimen on KLAL success. However, one report cited better outcomes (BCVA of ≥20/200 in 80% of patients with a mean postoperative follow-up of 5 years) in patients who received immunosuppression indefinitely with at least two systemic agents.22 In this study by Liang and coworkers, the authors administered an aggressive immunosuppressive regimen, including oral prednisone, mycophenolate mofetil and tacrolimus, and corrected adnexal risk factors for KLAL failure, including symblephara, trichiasis and lid position abnormalities, prior to KLAL surgery.
We have not systematically compared the outcomes of KLAL in chemical injury with other techniques previously described as treatments, including lr-CLAL, cultivated oral mucosal epithelial transplantation (COMET), Allo-CLET and Boston keratoprosthesis. In the largest published study of CLAL using human leukocyte antigen (HLA)-matched, living-related tissue for bilateral LSCD after chemical burn (12 eyes, with 63.3 months mean postoperative follow-up), of the eight operated eyes with BCVA <20/200 preoperatively, 50% achieved ≥20/200 at the last recorded postoperative visit.40 In the largest published study of Allo-CLET for chemical injury (19 eyes, with 22 months mean postoperative follow-up), 31.6% of the eyes achieved BCVA ≥20/200 postoperatively.41 In the largest study describing use of the Boston keratoprosthesis type I after chemical injury (42 eyes, mean follow-up 40.2 months), 61.9% achieved a BCVA of ≥20/200 at last postoperative visit.42 These results suggest that KLAL and Boston keratoprosthesis might produce similar results in LSCD after chemical injury, but a systematic review for keratoprosthesis implantation in this patient population would be required before drawing this conclusion.
Study limitations
We attempted to reduce potential bias in this review by conducting a systematic search across multiple databases and by having two reviewers screen the titles and the abstracts of all studies to arrive at a final list of articles to be included in the systematic review. However, there were limitations to this study. Only articles in English were included in this review, and there are studies on this topic in other languages. Also, we did not include ‘grey’ literature such as conference abstracts. Therefore, studies that have been presented at conferences but have not yet been published were not reviewed.
Conclusions
Large variability was observed in outcome reporting, and the overall quality of evidence was judged to be low. Stratification of patients by injury type and severity, the degree of secondary LSCD and corneal opacity, comorbidities possibly affecting outcomes, and length of follow-up are needed to better define evidence-based best practices in surgical interventions in corneal blindness secondary to chemical ocular injury. Because of heterogeneity in disease presentation, inadequate reporting, and a gap in appropriate, objective outcome measures, the current evidence available is not sufficient to provide a recommendation for or against KLAL in corneal blindness due to LSCD after chemical ocular injury.
References
Footnotes
Contributors SSS performed literature searches, read potential manuscripts, collated the data and wrote the paper. HNS assisted with deciding on which manuscripts met inclusion criteria, and edited the paper. EIP assisted with interpretation of the data and edited the paper. JC conceived the project and edited the paper. All authors read and approved the final version of the manuscript.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement The authors will freely share, on reasonable request, the search results and lists of excluded articles and reasoning for their exclusion.
Presented at Presented as a Poster at the Association for Research in Vision and Ophthalmology Annual Meeting, Baltimore, Maryland, 2017.
Linked Articles
- At a glance