Corneal transplantation in aniridia‐related keratopathy with a two‐year follow‐up period, an uncommon disease with precarious course

The purpose of this study is to study the frequency, surgical transplantation technique and outcome in patients with aniridia‐related keratopathy (ARK) with two‐year follow‐up period.

transplantation in aniridia. Endothelial failure can be secondary to intraocular surgery due to cataract removal and implantation of a sulcus sutured aniridia IOL in an otherwise fragile aniridic eye.
Penetrating keratoplasty (PK) in ARK has poor results with risk of graft failure and recurrence of ARK corneal changes (de la Paz et al., 2008;Mayer et al., 2003). The stem cell deficiency, present in ARK, is not addressed with PK and it has been suggested that it should be combined with surgical treatment of the limbal stem cell deficiency (Gomes et al., 1996;Kremer et al., 1993). Keratolimbal allograft (KLAL) is an option for pathologies that primarily involve limbal stem cells, such as in ARK, and is therefore usually suggested as the best first-line surgical treatment for ARK (Bakhtiari & Djalilian, 2010;Holland et al., 2003;Kim et al., 2003;Lee et al., 2008). Unlike PK, KLAL addresses the ARK pathogenesis of the limbal stem cells and a previous study showed a median graft survival time of 2 years in PK, compared with 4 years in KLAL (mainly allogenic transplants without HLA matching) (de la Paz et al., 2008). The group with KLAL had a better visual acuity the first 2 years, but the difference was no longer statistically significant after 5 years. Corneal transplantation, regardless of method, did not improve the long-term visual prognosis (after 5 years). However, a stable ocular surface was obtained during longer periods of time when a limbal stem cell transplantation was performed (de la Paz et al., 2008). This indicates a decrease in transplanted limbal stem cells over time, and emphasizes the importance of the limbal milieu for the whole cornea Solomon et al., 2002;Vicente, Bystrom, Lindstrom, et al., 2018). Immunosuppression is one of the most important factors in determining long term ocular surface stability after corneal transplantation with limbal stem cell transplantation, and the success rate of KLAL is reduced in patients with a previous failed penetrating keratoplasty or KLAL Kim et al., 2003). Boston keratoprosthesis is an alternative surgical method, especially in patients with previously failed corneal transplantations, but it also has limited long-term outcomes (Akpek et al., 2007;Holland et al., 2012;Kim et al., 2003). Some studies advocate that the Boston keratoprosthesis should only be used as a last surgical option in the treatment of aniridia, (Shah et al., 2018) while others show comparable results with other techniques when Boston KPro is used as first treatment option (Akpek et al., 2007).
Studying the results after corneal transplantation in ARK is important, as it is a challenging disease, not least when surgery is involved. The disease is uncommon and only a few studies are available on the surgery outcomes in this patient group. For the clinician managing, the ARK-patient with poor and further decreasing visual acuity is thus not easy. Aniridia has an estimated prevalence of 1:70000 in Sweden and estimated point prevalence at birth in Denmark of 1:40000 (Eden et al., 2008;Grindley et al., 1995). This multicentre registry-based study covering two countries, is in this context, a valuable source for more insights and knowledge. The current study period of 16 years includes the time when the experience of KLAL evolved. It was thus interesting to study if this induced a shift in the surgical procedures and outcomes . This study aims to assess the frequency, surgical techniques and outcomes in corneal transplantation for ARK-patients in Sweden and Denmark between 2001 and 2016.

| M ET HOD S
The data for this registry-based study were obtained from the Swedish Cornea Transplant Registry and include all ARK cases with corneal transplantation registered during the years 2001-2016. The study followed the tenets of the Declaration of Helsinki and was approved by the Regional Ethical review Board in Umeå (Dnr 2017/298-31). The Swedish Cornea Transplant Registry includes pre-, peri-and postoperative data about the corneal transplantation procedures in Sweden since 1996, and in Denmark since 2010. The registration is performed by surgeons preoperatively and at the follow-up visit 2 years after the procedure.
All patients with the indication ARK at the time of the first keratoplasty or primary indication ARK in case of retransplantation were included. The preoperative variables were: age at the time of surgery, gender, diagnosis, aim for the surgery (improve vision, reduce pain, tectonic or "other"), lens status, risk factors for transplantation, preoperative corrected distance visual acuity (CDVA) in the operated eye and fellow eye, type of transplantation technique, other procedures performed at the time of keratoplasty, occurrence and type of surgical complications. The postoperative variables included were: follow-up availability (2 years after the procedure), survival and function of the graft, performed retransplantation, ocular comorbidity, rejection, suture problems, glaucoma, infection and other complications, cataract surgery during the two-year follow-up period and postoperative CDVA in the operated and fellow eye. The pre-, peri and postoperative data of each participant is presented in Table S1. Graft survival is defined by whether the graft provides useful vision for the patient. The graft survival is presented as the number and percentage of functioning grafts at the two-year follow-up, and not by graft survival time due to limited documentation of actual failure date.

| Statistics
The visual acuity was recorded in decimal scale to the registry. A large portion of the participants had visual levels of counting fingers (CF), hand motion (HM), or light perception (P) which have an ordinal character, more than a parametric one. Thus, the median pre-and postoperative visual acuity was compared using sign test instead of e.g. Wilcoxon signed-rank test. The difference in follow-up time depending on functioning graft was compared using Wilcoxon rank-sum test and the cases lost to follow-up depending on preoperative inflammation were tested with Chi-Squared test. The statistical significance was determined at p < 0.05. The data were analysed using RStudio (2016).

| R E SU LT S
In total, 36 cases (eyes with ARK from 26 patients) who underwent corneal transplantation during 2001-2016 in Sweden and Denmark, were included in this study. PK was the most common procedure, chosen in 58.3% (n = 21) of the eyes, followed by a combination of PK and limbal stem cell transplantation in 13.9% (n = 5) and KLAL in 13.9% (n = 5), Boston keratoprosthesis in 8.3% (n = 3) and anterior lamellar transplantation in 5.6% (n = 2). All patients that underwent surgery with a Boston keratoprosthesis were between 60 and 75 years old and without registered previous ocular surgery. Retransplantation was performed in 13 cases (36.1%), with one previous graft in median (range 1-3). The chosen surgical technique for retransplantation was PK in 9 cases, a combination of PK with limbal stem cells in 3 other cases and anterior lamellar transplantation in one case, Figure 1. The development over time regarded surgical method is shown in Figure 2. The primary reason for the transplantation was improvement of vision. Only one eye had surgery for tectonic reason, and no surgery was registered due to reduction of pain or other indication. Concomitant surgical procedures performed were amniotic membrane transplantation (11.1%; n = 4), tarsorrhaphy (8.3%; n = 3), cataract extraction with IOL implantation (8.3%; n = 3) and scleral IOL fixation (2.8%; n = 1). In addition, at the time of the surgery, 44.4% (n = 16) of the eyes were pseudophakic, 25.0% (n = 9) were phakic and 30.6% (n = 11) were aphakic. The most common risk factors other than the preoperative diagnosis were vascularization (61.1%; n = 22), followed by glaucoma (41.7%; n = 15), and inflammation (19.4%; n = 7), Table 1. No perioperative complications were registered in any of the 36 procedures.
At the time of transplantation, the median age of the included cases was 42 years and 63.9% (n = 23) were female, Table 1. Follow-up data were available in 26 cases (eyes), of which 22 had both pre-and postoperative CDVA registered (21 of 22 from unique patients). The time of follow-up was 2.03 years (IQR 0.20) in median. The postoperative median CDVA in the operated eye showed a trend towards improvement from HM (0.001 Decimal) to CF (0.01 Decimal), but without statistical significance with the use of sign test. The retransplantations showed similar results, Table 1. Each patient's CDVA before and after the corneal surgery is shown in Figure 3, including surgical method. There was no statistically significant difference in median CDVA in the fellow eye at base line compared with 2 years later (CF).
At follow-up, planned 2 years after surgery, grafts were functioning in 16 of the 26 cases available to follow-up (61.5%), Table 1. The groups were too small to compare graft function depending on surgical method. The cause of graft failure was recurrency of ARK in four cases (40%), irreversible graft rejection in two cases (20%), infection in one case (10%) and other or unknown in three cases (30%). The follow-up time was 730 days (IQR 44) in the group with a functioning graft and 866 days (IQR 397) in the group with a failed graft (P = 0.01). Recurrence of ARK during 2 years after the surgery was registered in 19.2% (n = 5) of the 26 cases, infection in the graft in 19.2% (n = 5), rejection of the graft in 11.5% (n = 3) and glaucoma was seen in 34.6% (n = 9). Six cases (23.1%) had two or more of the conditions above. The cases with graft rejection episodes had all non-functioning grafts at follow-up. Three of the four cases with concomitant amniotic membrane transplantation were available for follow-up and all suffered from a non-functioning graft. There was a trend towards an increased proportion of non-functioning graft, graft rejection, glaucoma, infection, and ARK relapse at follow-up among retransplantation compared with primary transplantation, but the difference was not statistically significant, Table 1. There was no definite association between preoperative vascularization, inflammation, or glaucoma and CDVA at 2 years-follow up, but the sample size was too small for proper statistical analysis. There was a tendency for more preoperative vascularization, inflammation and glaucoma in the group with nonfunctioning grafts at follow-up compared with the group with functioning grafts, Table 2. Further, that trend was similar in the group with retransplantation compared with primary surgery, but the differences were without statistical significance. Preoperative inflammation was present in 50.0% (n = 5) of the cases lost to follow-up compared with 7.7% (n = 2) in the group available to follow-up (p = 0.02), and there was also a trend for more preoperative vascularization and glaucoma in the group lost to follow-up.

| DI SC US SION
Aniridia with ARK is a rare but severe disease, with no curative treatment, and with several therapeutic challenges. Surgical procedures are usually left as a final F I G U R E 1 The distribution of corneal transplantation technique at primary surgery and retransplantation. Notice that PK is the most common technique chosen, regardless of primary surgery or retransplantation. PK, penetrating keratoplasty; PK SC, penetrating keratoplasty with limbal stem cells; KLAL, keratolimbal allograft; KPro, Boston keratoprosthesis; ALK, anterior lamellar keratoplasty. therapy step in the management of aniridia cases, when visual acuity is severely reduced and the symptomatology grave (Lee et al., 2008). The decision to do surgery is hard for a number of reasons. All kind of ocular surgery, even due to cataract or glaucoma, tends to worsen ARK, and risks pushing the patient's corneal status over the edge (Eden et al., 2010;Landsend et al., 2021;Tsai et al., 2005). The severe vision loss caused by corneal opacities, glaucoma and cataract in ARK must be weighed against the risk and potential benefit of surgery. This study assesses the frequency, surgical techniques and outcomes in corneal transplantation for ARK-patients in Sweden and Denmark between 2001 and 2016.
Over a period of 16 years, 36 corneal transplantation procedures were performed in ARK cases and registered in the Swedish Cornea Transplant Registry. The frequency of surgery was stable over the studied time although some variation in surgical methods reflects the doubtful results despite the development of various techniques. PK was the most common method until 2008 when KLAL and a combination of PK with limbal stem cells became more frequently used. This shift can be attributed to the dissemination of knowledge about the benefits of including limbal stem cells during transplantation which is done in KLAL . At the end of the study period, PK was again the most common method which could be due to a lack of surgical competence with very few cases per each center, unsatisfactory results, the need of a successful stem cell transplantation and long term systemic immunotherapy. However, the choice of method has not affected the total number of transplantations in any substantial way. The Year of surgery Boston keratoprosthesis was used in three patients with a median age of 65 years (IQR 5). Even though corneal transplantation in patients with ARK can be complicated due to altered anterior eye morphology and usually is performed in low-visibility conditions (de la Paz et al., 2008), no peri-operative complications were registered in the present study.
All participants in the current study had severe vision impairment preoperatively, with best preop-CDVA of CF in the operated eye and 0.1 as best in the fellow eye. The primary reason for corneal transplantation was visual improvement in almost all cases, even though visual acuity in aniridia patients is known to be reduced due to other consequences of the disease, such as glaucoma, cataract, foveal hypoplasia and optic nerve hypoplasia (Nishida et al., 1995). This study shows a trend towards better visual acuity, (median CDVA: CF) compared with before the surgery (median CDVA: HM). A fairly large proportion of cases had glaucoma (36.6%) and had undergone an infection (19.2%) at the follow-up visit after 2 years, which could have influenced the poor visual outcome together with other unknown factors. At twoyear follow-up, 61.5% of the grafts available for follow-up were functioning. But, in other words: 38.5% of the grafts failed and did not provide any useful visual function. Furthermore, ARK relapse was not uncommon (19.2%). Other causes of severe vision loss besides the corneal opacities in ARK cases must be considered carefully before corneal transplantation is performed. The risk of a potential vicious circle towards a worse clinical situation after surgery must be weighed against the potential gain.
An improved visual acuity from HM to CF may seem limited, but it can have a substantial functional impact that mitigates the disability for some years. Nevertheless, caution should be taken in interpreting this data in an over optimistic form, as other studies have shown that these improvements tend to disappear in the long term (de la Paz et al., 2008). Ten of the 36 eyes were unfortunately not available to follow-up and can cause bias if they were not randomly lost. If the included cases were more successful than the missing ones, the results of this study could be falsely optimistic. With more preoperative inflammation in the group lost to follow-up than the group available for follow-up, that is a possibility. Efforts to complete the data have been made.
Additionally, the group with non-functioning grafts had longer follow-up time, indicating that the risk of graft failure increases with longer follow-up time. Previous studies, with a follow-up time of 3 years in average, have reported failure rates of 64% up to 100% with centered PK (Kremer et al., 1993;Lim et al., 2017). Other studies show that PK is ineffective in the long-term with high risk of rejection and high graft failure rates, as the underlying stem cell deficiency in aniridia is not treated. It has even been postulated that PK as a single procedure should not be performed, instead it should be combined with a concomitant or a previous treatment addressing the limbal stem cell deficiency (Gomes et al., 1996;Kremer et al., 1993). An improved success rate has been reported with KLAL in the management of ARK. (Kim et al., 2003) Even though limbal transplantation has better results in the short term, the long term results are poor regardless the type of transplantation. (de la Paz et al., 2008) The sample size in this current study was too small to compare graft survival depending on surgical method. Concomitant risk factors were present in a majority of the cases and highlight both clinical characteristics of ARK, with vascularization and inflammation but also glaucoma. These circumstances are not optimal in corneal transplantations and considered high risk surgery but is inevitable and chronic in many ARK cases.
This study covers nearly all corneal transplantations performed in two nations over 16 years in a very rare and severe disease, aniridia. It shows real-world data, reduces the bias caused by local and regional variations and shows some changes over time. A limitation is a low number of cases, which makes the study under powered to compare the different surgical methods, risk factors and their effect on graft survival and vision. At the same time, the study shows that the Scandinavian corneal surgeons comply with the evidence-based advice to be cautious in the choice of ARK treatment, and only plan a corneal transplantation as a last option.
The already collected data is a great advantage of registry-based studies compared with other study designs, where the longitudinal data enable analyses of diseases over time, particularly important in rare diseases, unusual factors, and complications. The effectiveness and safety of treatments and possible risk factors can be studied through data from real-life patients in real-life clinical settings. Still, the registry-based studies carry their own limitations. Above all, the variables are fixed. For example, more detailed preoperative data, valid for the ARK group of cases, would have been desirable but the Swedish Cornea Transplant Registry is not designed specifically to follow only ARK, instead it includes all diagnosis that undergo a corneal transplantation and thereby has limitations regarding ARK. Neither is the use of immunosuppressive treatment available even if it is the most important factor for long term ocular surface stability after corneal transplantation with limbal stem cell transplantation in these patients.  Another limitation is the registry's follow-up time of 2 years, leading to a lack of more long-term data. This appears to be specifically relevant in interpreting the results in patients with ARK.
In conclusion, more than half of the ARK-patients had a functioning graft 2 years after the corneal transplantation, even though the majority of them were affected by complications as recurrence of ARK in the transplant, rejection and glaucoma. The visual acuity gain was modest at best, but might enable a few years of improved visual function in a severely affected patient group. Limbal stem cell transplantations were introduced and performed during the study period, but PK without limbal stem cell transplantation was still the most common surgical method at the end of the study period.
The decision to perform corneal transplantation in ARK patients is difficult and must be considered individually in each case due to its specific factors, still this study can be helpful in that process, not least by pointing out precautions. Hopefully future medical development will lead to more successful outcomes for this difficult disease.

CON F L IC T OF I N T E R E ST
The authors declare no conflict of interest.