Article Text

Cost-effectiveness analysis of preloaded versus non-preloaded Descemet membrane endothelial keratoplasty for the treatment of Fuchs endothelial corneal dystrophy in an academic centre
1. Myriam Böhm1,2,
2. Pia Leon1,
4. Stephan Ong Tone1,
5. Tracy Condron1,
6. Ula Jurkunas1
1. 1 Department of Ophthalmology, Massachusetts Eye and Ear Infirmary, Harvard Medical School, Boston, Massachusetts, USA
2. 2 Department of Ophthalmology, Goethe University Frankfurt, Frankfurt am Main, Germany
1. Correspondence to Dr Ula Jurkunas, Ophthalmology, Massachusetts Eye and Ear Infirmary, Boston, MA 02114, USA; Myriam.Boehm{at}gmail.com

## Abstract

Aims To determine the cost-effectiveness of preloaded Descemet membrane endothelial keratoplasty (pDMEK) versus non-preloaded DMEK (n-pDMEK) for the treatment of Fuchs endothelial corneal dystrophy (FECD).

Methods From a societal and healthcare perspective, this retrospective cost-effectiveness analysis analysed a cohort of 58 patients with FECD receiving pDMEK (n=38) or n-pDMEK (n=30) from 2016 to 2018 in the Department of Ophthalmology, Massachusetts Eye and Ear Infirmary, Harvard Medical School, Boston, USA. Exclusion criteria were previous ocular surgeries (other than uncomplicated cataract surgery), including other keratoplasty procedures, ocular pathological conditions as glaucoma, amblyopia, laser treatments, or any retinal or corneal disease. The main outcome parameters were the incremental cost-utility ratio (ICUR) and net monetary benefit (NMB).

Results pDMEK was less costly compared with n-pDMEK (healthcare: $13 886 vs$15 329; societal: $20 805 vs$22 262), with a slighter greater utility (QALY 0.6682 vs QALY 0.6640) over a time horizon of 15 years. pDMEK offered a slightly higher clinical effectiveness (+0.0042 QALY/patient) at a lower cost (healthcare: –$1444 per patient; societal: –$1457 per patient) in improving visual acuity in this cohort of patients with FECD. pDMEK achieved a favourable ICUR and NMB compared with n-pDMEK. Based on sensitivity analyses performed, the economic model was robust.

Conclusions From the societal and healthcare perspective, pDMEK was less costly and generated comparable utility values relative to n-pDMEK. Therefore, pDMEK appears to be cost-effective and cost saving with respect to n-pDMEK. Further long-term follow-up data are needed to confirm these findings.

• cornea
• dystrophy
• eye (tissue) banking
• treatment surgery
• vision

## Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information. The authors confirm that the data supporting the findings of this study are available within the article or its supplementary materials.

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

## Introduction

Fuchs endothelial corneal dystrophy (FECD) is the most common indication for corneal transplantation, followed by corneal edema after cataract surgery in the USA.1 Without surgical treatment, many patients with FECD would have visual impairment and thus impose substantial costs on society.2 In the last two decades, keratoplasty trends have shifted from penetrating keratoplasty (PK) to more customised lamellar techniques, particularly endothelial keratoplasty (EK). In 2019, the Eye Bank Association of America (EBAA) reported a decrease of 5.4% in PKs to 17 409, while EK numbers increased by 4.6% to 30 650, mainly attributable to an increase of 41.2% in Descemet membrane endothelial keratoplasty (DMEK) procedures.3

DMEK is performed by the selective transplantation of Descemet membrane (DM) with the corneal endothelium and has become the preferred method of corneal transplantation for patients with FECD since its introduction in 1998 and 2006.4 5 Besides the steep surgical learning curve and perceived complexity of the procedure, donor graft preparation has been one of the biggest technical challenges due to tissue loss during preparation.6 7 At first, this has considerably slowed down the rate of DMEK surgery adaption by surgeons.8 9 However, since 2006, eye banks have started providing precut and prestripped donor corneal tissue for DMEK, and have demonstrated similar clinical outcomes compared with freshly stripped tissues.10–12 Recently, an increasing number of eye banks in the USA have started to offer preloaded DMEK (pDMEK) donor tissues to further address technical challenges in tissue preparation by surgeons.13 14 pDMEK tissues might lead to standardisation and validation of tissue preparation, reduced tissue wastage, simplify surgery with reproducibility, provide better quality control and reduce surgical time and associated costs.13–15

Prior cost-effectiveness studies have shown that deep anterior lamellar keratoplasty and Descemet stripping (automated) endothelial keratoplasty (DSEK or DSAEK) are cost-effective compared with traditional full-thickness penetrating keratoplasty.16–19 A recent study revealed that DMEK generated greater utility and was less costly compared with DSAEK, thus supporting DMEK as a superior and cost-saving procedure.20 To date, no study has quantified and compared the benefits and costs of pDMEK relative to n-pDMEK.

The purpose of this study was to determine the cost-effectiveness of pDMEK compared with n-pDMEK for the treatment of FECD in a cohort of patients undergoing EK at the Massachusetts Eye and Ear Infirmary (MEEI).

## Material and methods

In this study, a trial-based cost-effectiveness analysis comparing pDMEK and n-pDMEK was performed from a societal and healthcare perspective. The study used single study-based estimates based on retrospective 1 year visual acuity (VA) and complication data from patients who were treated for FECD by corneal fellows under the supervision of one experienced corneal surgeon at MEEI, Harvard Medical School, between 2016 and 2018. All subjects provided written informed consent for the surgical procedure.

### Study population

The population of the study included consecutive patients with FECD who underwent either pDMEK or n-pDMEK, with or without combined cataract extraction (Triple DMEK). Exclusion criteria were previous ocular surgeries (other than uncomplicated cataract surgery), including other keratoplasty procedures and other ocular pathological conditions as glaucoma, amblyopia, previous laser treatment, or any documented evidence of retinal or corneal disease that would limit recovery of VA after surgery.

### Study procedures

Cornea fellows performed all surgeries with one experienced surgeon. The experienced surgeon started performing n-pDMEK in 2014 and incorporated pDMEK in the practice in 2017. This study included patients who received n-pDMEK or pDMEK from 2016 to 2018 after at least 70 DMEK cases performed by the experienced surgeon. Some cornea fellows started their learning curve with n-pDMEK and then switched to pDMEK procedure, whereas others directly started with pDMEK.

Various US eye banks provided the preloaded and non-preloaded donor corneas (CorneaGen, SightLife (including formerly Northeast Pennsylvania Lions), Lions VisionGift eye banks). At these eye banks, each technician has successfully completed a rigid formalised training programme, as previously published.21 Trained eye bank technicians prestripped n-pDMEK and pDMEK tissues donor corneas using a modiﬁed submerged cornea using backgrounds away technique.22 23 Next, grafts were marked with a S-stamp and punched using a Moria guarded punch (Moria, Doylestown, Pennsylvania, USA). For the preparation of the n-pDMEK graft, the final step performed by the eye bank technician was the quality evaluation of the graft by specular microscopy and slit lamp evaluation and placing of the cornea containing the prestripped tissue in an Optisol-filled Krolman viewing chamber (Krolman, Boston, Massachusetts, USA). In the operating room, the surgeon completed the tissue preparation process by punching, completing the stripping, staining and finally loading the tissue in a Viscojet injector 2.2.

In contrast, for the preloaded graft preparation, the peripheral endothelium–Descemet membrane complex surrounding the punched graft is removed by the eye bank technician and the graft is lifted and allowed to spontaneously scroll into the endothelium-out conformation. The graft is then stained for 4 min with trypan blue (0.06%, C-blue; Stephens Instruments, Lexington, Kentucky, USA), washed with balanced salt solution (BSS; Alcon, Fort Worth, Texas, USA) and submerged in Optisol-GS (Bausch & Lomb, Irvine, California, USA). The DMEK scroll is subsequently drawn into a Straiko-modiﬁed Jones tube, and after quality evaluation placed in an Optisol-ﬁlled Krolman viewing chamber (Krolman, Boston, Massachusetts, USA). The pDMEK tissue is shipped to the operating room, where the surgeon removes the modified Jones tube from the viewing chamber and attaches it to a precut 14 French catheter tubing attached to a BSS-filled 6 cc syringe.

### Decision model

A cost-effective analytical model (figure 1) was created using Tree-Age Pro Suite V.2009 (Tree Age Software, Williamstown, Massachusetts, USA) for patients who underwent pDMEK and n-pDMEK transplantation. Both costs and utilities were discounted at 3% per year.

Figure 1

Decision analytical model. The branches from the preloaded Descemet membrane endothelial keratoplasty (pDMEK) model were the same as the non-preloaded Descement membrane endothelial keratoplasty (n-pDMEK) model. Squares represent decision nodes, circles represent chance nodes and triangles represent terminal nodes. Under every chance node the probability of an event was placed. A decision tree was used to compare pDMEK versus n-pDMEK for the average patient with Fuchs endothelial corneal dystrophy (FECD). pDMEK and n-pDMEK have expected visual acuities of 0.12 and 0.13 logMAR, respectively, in the first year after surgery.

### Visual acuity data

VA was measured using Snellen VA charts for all patients before and at all follow-ups (from 1 to 12 months) in the clinic after EK.

Both preoperative and postoperative VA results refer to corrected distance VA (CDVA) and were converted to logarithm of the minimal angle of resolution (logMAR) equivalent units for analysis in this study.

### Complications

The 1-year incidences of postoperative complications of the patient cohort (pDMEK vs n-pDMEK) were obtained from a retrospective chart review of MEEI patients. These complications include intraocular pressure (IOP) elevation, macular edema, partial graft detachments (rebubbling rates) and secondary keratoplasty.

There is no long-term data on corneal graft failures or endophthalmitis rates available for pDMEK. Potential complications like endophthalmitis occur rarely after EK, and none has been reported in previous case reports, thus, the incidence is hard to assume. For the purpose of this study, the graft failure rates and endophthalmitis rates of both surgeries were assumed to be equal since no graft failures or endophthalmitis were observed in this cohort of patients, thus not influencing the comparison of utility gains of both groups. In line with Gibbons et al 20 and other studies on DMEK procedure, which reported the uneventful resolution of most postoperative complications with appropriate medical and surgical management, postoperative complications, such as cystoid macular edema (CME), were assumed to resolve completely.

### Sensitivity analysis

A one-way deterministic sensitivity analysis was conducted to test the model results to the following key parameters: (1) surgery time, (2) rebubbling rate, (3) graft tissue costs. Because surgery time between for pDMEK and n-pDMEK was responsible for most of the difference in initial costs between the two procedures, sensitivity analyses were performed to investigate its impact on the cost-effectiveness of pDMEK relative to n-pDMEK. Moreover, the rebubbling rate and graft tissue costs were also varied to evaluate the magnitude of its contribution to the relative cost-effectiveness of pDMEK. Incorporating the results from one-way sensitivity analyses, a two-way sensitivity analysis on surgery time and rebubbling rate was performed.

### Statistical analysis

The statistical analysis was performed using SPSS Statistics for Mac (V.24.0, International Business Machines). The Kolmogorov-Smirnov test was used to test for normal distribution of the data. Differences between treatment groups were statistically tested with independent samples t-test for data fitting a normal distribution. If normal distribution was not confirmed, the Mann-Whitney U test was used. A χ2 test was used if appropriate. Between eyes correlation was measured for 10 patients with bilateral FECD using the intraclass correlation coefficient (ICC). Calculating the ICC of the preoperative utility for these eyes showed a poor correlation of 0.207, indicating that bilateral eyes of these patients should be included in the study in order not to reject valid data and reduce the potential power of the study. A p value of less than 0.05 was considered statistically significant. The data are presented as mean±SD throughout this paper.

## Results

The cases were analysed from a healthcare perspective. Sixty-eight eyes from a cohort of 58 patients that underwent EK were included in the study. Baseline demographic data are shown in table 1. Thirty-eight eyes underwent pDMEK tissues and 30 eyes underwent n-pDMEK, with a mean age at the time of surgery of 67.7 and 68.3 years old, respectively. There were no significant differences in mean age (p=0.950), sex (p=0.861), clinical FECD stage (p=0.957) or type of surgery (p=0.853) distributions between the two groups, making them comparable for cost-effectiveness analysis. There was no significant difference in mean preoperative CDVA between eyes that underwent pDMEK (0.45±0.25 logMAR) compared with n-pDMEK (0.46±0.27 logMAR, p=0.807). However, the surgery time was significantly shorter in the single pDMEK group (35 min) than in the single n-pDMEK group (55 min) (p<0.001). The shorter surgery time was also detected when performing subgroup analyses that included comparison of only triple DMEK cases and both single and triple DMEK cases analysed together (p<0.001, table 1).

### Comparative effectiveness

pDMEK resulted in similar 1-year VA outcomes compared with n-pDMEK, with an expected CDVA of 0.12 logMAR and 0.13 logMAR, respectively (p=0.827). The first postoperative year utility gains for pDMEK and n-pDMEK compared with no surgical treatment were 0.0543 and 0.0540, respectively. Summing the discounted expected utility gains over 15 years resulted in slightly different but comparable total expected utilities with 0.6682 and 0.6640 QALYs for pDMEK and n-pDMEK, respectively.

### Medical costs

Total costs for each of the procedures are detailed in table 2. From the societal and healthcare perspective, the average cost was $19 872/$13 348 in the pDMEK group and $21 354/$14 829 in the n-pDMEK group. Complication rates and the associated costs are detailed in table 3. The incidence-weighted costs of treating complications from the societal and healthcare perspective associated with pDMEK were $933.55 and$538.18, whereas the cost associated with n-pDMEK were $906.01 and$497.35, respectively. The total cost associated with each procedure was determined by adding the cost of the primary surgery to the complication cost. The total estimated cost associated with pDMEK was $20 805 from the societal and$13 886 from the healthcare perspective and with n-pDMEK $22 262 and$15 329, respectively, with a difference in costs of $1457 and$1444 per patient.

Table 3

Medical costs of postoperative complications associated with endothelial keratoplasty from a healthcare perspective: preloaded Descemet membrane endothelial keratoplasty (pDMEK) vs non-preloaded Descemet membrane endothelial keratoplasty (n-pDMEK)

### Cost-utility analysis

Cost-utility estimates are presented in table 4. From a societal perspective, the average 15-year cost of pDMEK surgery was $20 805 and$22 262 for n-pDMEK. From a healthcare perspective, the cost of pDMEK was $13 886 and$15 329 for n-pDMEK. Compared with n-pDMEK, pDMEK had a slightly higher incremental QALY of 0.0042 and a lower incremental cost of –$1457 from the societal perspective and –$1444 from the healthcare perspective and thus the incremental cost-utility ratio favoured pDMEK over n-pDMEK. Over 15 years, pDMEK was the dominant procedure, since it was slightly more effective and less expensive than n-pDMEK. Consequently, pDMEK would be plotted in the southeast quadrant of a cost-effectiveness plane.29 Since the incremental utility was negative in the dominant intervention, the incremental cost-utility ratio can be reported as ‘cost-saving’ as suggested by the cost-effectiveness analyses reported recommendation. The NMB was positive, implying that the value of the incremental benefit exceeded the incremental costs. pDMEK was cost-effective over a range of cost-effectiveness ceilings calculated from 0$(if society were only willing to consider interventions that were cost-saving) to$100 000 per QALY gained (table 4).

Table 4

Cost-effectiveness and cost-utility analyses of preloaded Descemet membrane endothelial keratoplasty (pDMEK) vs non-preloaded Descemet membrane endothelial keratoplasty (n-pDMEK)

### Sensitivity analysis

The deterministic analyses were expressed in terms of NMB rather than the ICER because pDMEK procedure was dominant compared with n-pDMEK, and hence had a negative ICER. A negative ICER cannot be directly interpreted because it can correspond to either the dominant (positive QALY and negative cost) or dominated (negative QALY and positive cost) quadrants of the cost-effectiveness plane. A positive NMB shows that the pDMEK procedure is cost-effective at a threshold of $100 000 per QALY gained versus n-pDMEK in the given scenarios. A negative NMB indicates that pDMEK may not be a cost-saving option at this threshold. Figure 2 shows results of one-way sensitivity analysis by calculating the NMB of the key model parameters. The outcomes are represented in a tornado diagram expressed in terms of the NMB at threshold$100.000 per QALY.

Figure 2

Tornado diagram of the range of net monetary benefits at willingness-to-pay threshold $10 000 per quality-adjusted life years (QALY) for the key model parameters from a societal and healthcare perspective. Ranges: surgery time: 35–60 min; re-bubbling rate: 2.4%–82%, Graft tissue costs:$3800–$4800. The light blue segments represent the effect of decreasing the variable value, and the dark blue segment represents the effect of increasing the variable. The results of one-way sensitivity analyses on the key parameters show that the parameters examined within the reasonable sensitivity ranges produced the results supporting the main conclusion that pDMEK dominates n-pDMEK procedure. The higher total expected costs of n-pDMEK compared with pDMEK were relatively robust to changes in surgery time. n-pDMEK is still dominated by pDMEK at a surgery time of 40 min with NMB of$321.01 and $334.30 for the healthcare and societal perspective, respectively. The partial graft detachment rates of the pDMEK and n-pDMEK groups were varied from 2.4% up to 82%. At a graft detachment rate of 82%, pDMEK was still the dominant procedure with a NBM of$1099.64 from the healthcare perspective and $432.37 from the societal perspective. Similarly, varying the graft tissue costs of pDMEK and n-pDMEK from$3800 to $4800 yielded positive NBMs in all scenarios, leaving pDMEK as the dominate strategy (figure 2). Because surgery time and rebubbling rate are central to the model, a two-way sensitivity analysis on these parameters was conducted. pDMEK was the dominant strategy, when the surgery time of n-pDMEK varied between 36 and 55 min and the rebubbling rate was between 2% and 30% with NMB values lying between$1388 and $21.69. In two-way sensitivity analysis, n-pDMEK became only cost-effective over pDMEK at equal surgery time, which is not expected, combined with a rebubbling rate of 28% (NMB: –$19.44) or higher.

## Discussion

In this study, a cost-effectiveness analysis was performed to quantify the relative costs of pDMEK and n-pDMEK for treatment of FECD from a healthcare and societal perspective.

The study is highly useful helping surgeons to transition from DSAEK to DMEK, with DMEK being more efficient but more complicated to perform. In 2019, the EBAA reported 17 428 DS(A)EK and 13 215 DMEK cases, representing around $340 million per year for endothelial keratoplasties, which highlights the importance of this study. The CDVA results and complication rates of this cohort of DMEK patients were comparable to those reported in prior literature with surgeon-stripped, eye-bank prestripped and preloaded tissue.6 30–33 Both pDMEK and n-pDMEK significantly improved VA in patients with FECD with no significant difference between both groups after 12 months. pDMEK was associated with lower costs (healthcare:$13 886 vs $15 329, societal:$20 805 vs $22 262) and a slightly higher utility (QALY 0.6682 vs QALY 0.6640) compared n-pDMEK over a time horizon of 15 years. pDMEK also offered a comparable clinical effectiveness (+0.0042 QALY/patient) at a lower cost (healthcare: –$1,444/patient, societal: –$1457/patient) in improving VA in this cohort of patients with FECD. Moreover, pDMEK achieved a favourable ICUR and NMB (healthcare:$1864, societal: $1878). In this economic model, pDMEK was the dominant intervention from a healthcare and societal perspective, since it was more cost-effective and cost saving relative to n-pDMEK. The cost-effectiveness analyses demonstrated that n-pDMEK had increased opportunity costs due the longer surgery time and variations in rebubbling rates between both groups. However, sensitivity analyses on both factors demonstrated the robustness of the model. It is important to highlight that pDMEK remains the less costly procedure compared with n-pDMEK even though pDMEK eye bank tissue cost was about$500 more expensive in the base case than the n-pDMEK tissue costs. Sensitivity analysis on graft tissue costs demonstrated cost savings of pDMEK over n-pDMEK in all cost scenarios from $3800 to$4600. Based on sensitivity analyses performed, the economic model of this study was robust.

Prior studies reported partial graft detachments after DMEK surgery that required air injection ranging from 2.4% to 82%.20 A study on prestripped tissues prepared by a single eye bank noted a rebubbling rate of 27.5% (11 of 40 eyes), which is comparable to the rebubbling rate in the n-pDMEK group of the current study.33 Moreover, two recently published studies of clinical outcomes on pDMEK also reported similar rebubbling rates as the current study of 19.6% (9 of 46 eyes) and 16.6% (16 of 111 eyes), respectively.6 13 Thus, the numbers used in the base case seem to be representative of the procedures. However, a study with larger sample size is needed to investigate whether there is a significant difference in rebubbling rates between pDMEK versus n-pDMEK surgeries.

A prior cost-effectiveness study on DMEK over DSAEK in the USA demonstrated that DMEK generated greater utilities and was less costly than DSAEK. Therefore, DMEK was the dominant procedure and was cost saving with respect to DSAEK.20 Similarly, previous studies have shown the cost-effectiveness of performing DSAEK over PK.16 17 19 The estimation of DMEK costs from a healthcare and societal perspective and QALY in two recent studies by Gibbons et al presents comparable values as this cohort of patients with FECD undergoing DMEK surgery, further supporting the economic model of the current study.20 34

### Limitations of the study

The general limitations of the current study are that it was based on retrospective data of a relatively small cohort of patients and the theoretical nature of the model with assumptions made for analyses purposes. However, since pDMEK is a rather new procedure, there are no randomised clinical trials published yet. Moreover, there are several specific limitations of this study. First, using the complication data of a brief timeline precluded the observation of corneal graft failures in this cohort of patients, which might affect the utility gains associated with each procedure. However, a study that investigated pDMEK tissues found an endothelial cell loss of 30.9% (24.5%–80.0%) at 6 months postoperatively.35 This is comparable to what another study found for non-preloaded eye bank prestripped DMEK tissues with an endothelial cell loss of 30.5% (3.8%–67.4%) after 6 months, supporting the assumption that there is no significant difference of post-surgery ECC between pDMEK and n-pDMEK tissues.33 Second, a single study-based estimates approach in an academic centre was chosen. It is important to note that depending on the surgeon’s experience surgery time of DMEK procedure may vary. Third, the model was applied to corneal transplants that were supplied and performed in the USA only, where medical and surgical costs may be significantly different from the other countries. Fourth, the current study uses utilities based on time trade-off; this does not allow for direct comparison with studies using utilities derived with other methods. Moreover, 5 out of 10 patients receiving bilateral surgery underwent pDMEK in one eye and n-pDMEK procedure in the fellow eye. Finally, this trial-based analysis at hand followed a deterministic model approach. It lacks a probabilistic sensitivity analysis, which would be recommended for future analysis. In conclusion, this cost-effectiveness analysis reveals that pDMEK and n-pDMEK procedures both are highly cost-effective treatments for patients with FECD. However, in this economic model, pDMEK appears more cost-effective and cost saving for treating FECD from a healthcare and societal perspective. The outcomes of this cost-effectiveness study will be of interest to corneal surgeons who consider adopting pDMEK relative to n-pDMEK. In particular, the role of surgery time, eye-bank tissue costs and intraoperative complication rates, as rebubbling are key drivers of the decision to adopt pDMEK over n-pDMEK. The model presented allows informed decision-making and to adequately allocate finite healthcare resources. As we gain more information about long-term outcomes of pDMEK relative to n-pDMEK, further research is needed to continually reassess outcomes of these procedures and their relative cost-effectiveness.

## Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information. The authors confirm that the data supporting the findings of this study are available within the article or its supplementary materials.

## Ethics statements

### Ethics approval

Institutional review board approval was obtained from the Human Studies Committee of the MEEI. The study adhered to the tenets of the Declaration of Helsinki and was conducted in compliance with the rules and regulations of the Health Insurance Portability and Accountability Act.

## Footnotes

• Contributors MB: design of the work, data collection, data analysis and interpretation, drafting the article, critical revision of the article, final approval of the version to be published. PL: data collection, critical revision of the article, final approval of the version to be published. AW: data collection, critical revision of the article, final approval of the version to be published. SOT: data collection, critical revision of the article, final approval of the version to be published. TC: data collection, critical revision of the article, final approval of the version to be published. UJ: design of the work, data analysis and interpretation, drafting the article, critical revision of the article, final approval of the version to be published.

• Funding This work was supported by European Society of Cataract and Refractive Surgeons (ESCRS) Peter Barry Fellowship.

• Competing interests None declared.

• Provenance and peer review Not commissioned; externally peer reviewed.