We have previously shown the utility of the ZAP software in providing reproducible measurements of posterior cornea arc lengths from ASOCT images . In this study, ZAP was used to measure the graft parameters in ASOCT images of post-DSAEK eyes. We performed blinded intra-observer repeated measurements with the ZAP software on ASOCT images of endothelial keratoplasty in order to ascertain the repeatability of graft measurements which found that ZAP was able to provide reproducible measurements of DSAEK graft parameters from ASOCT images.
Although useful in providing anatomical detail of anterior segment structures, ASOCT does have its limitations. Images need to be dewarped with software algorithms to correct for index transitions , and calculations of anterior segment dimensions are also dependent on assumptions of cornea refractive index. The time-domain ASOCT scans at 2000 axial lines per second and patient movement will affect the quality of the image and the accuracy of the dimensions measured from ASOCT scan. Therefore quantitative imaging on ASOCT may not always correlate with the actual tissue measurements. Our study compared an ex-vivo tissue measurement (the trephine diameter of the DSAEK graft), with the arc length of the graft measured post operatively on ASOCT. We found that among the in-vivo ASOCT graft measurements taken, the graft posterior arc length correlated best with graft trephine diameter. This may be expected since it would be less affected by the recipient cornea posterior curvature or the thickness of the graft, and the donor trephinations were all performed from the endothelial surface downwards.
Most of the grafts in this series were performed in eyes with pseudophakic bullous keratopathy where larger grafts with greater replacement of functional cornea endothelial cells were desirable. In our series, the mean ratio of posterior graft arc length to recipient posterior cornea arc length was 0.712 +/− 0.056 (Range 0.504 to 0.852). There was a small negative trend between the ratio of posterior graft arc length to posterior cornea arc length and the posterior cornea arc length, which suggested that we could have used larger grafts in eyes with larger posterior corneal arc lengths (Figure 5). For eyes in our series where the posterior cornea arc length was larger than the 75th percentile (13.52mm), the ratio of graft posterior arc length to posterior cornea arc length was less than the mean (0.712) in 70.4% of these eyes. This suggests that 70.4% of grafts in patients with PCAL> 13.5mm were undersized.
We feel that sizing of the DSAEK graft is not a “one size fits all” procedure. It is known that cornea dimensions can vary with ethnicity and adult stature . Intra-operative graft sizing based on visual assessment of the horizontal white to white diameter does not take into account the cornea curvature, and may also be an underestimate in eyes with significant arcus or pannus. The main advantage of using the ZAP software is that it allows the surgeon to quantitatively assess the size of the recipient posterior corneal surface as opposed to the anterior surface. Using the anterior surface of the cornea may be appropriate for sizing for penetrating keratoplasty but it is inappropriate for EK surgery not only due to the position of the EK graft but also because the EK surgeon can potentially transplant a larger graft since there is no concern regarding the limbal vasculature. There is no published information as to the optimal graft size for endothelial keratoplasty and hence also the optimal ratio of graft posterior arc length to posterior cornea arc length, although it has been suggested that a larger graft with the same endothelial cell density would provide a greater total number of functioning endothelial cells to the recipient and may support greater long term endothelial cell survival [26, 27]. Although one retrospective study demonstrated no significant difference in the loss of endothelial cell density comparing 8mm grafts and grafts larger than 8mm , this study only included patients with Fuch’s endothelial dystrophy where the endothelial dysfunction is mainly in the central cornea and larger grafts may provide less of an advantage. Comparative studies between DSAEK and PK have shown that although DSAEK has a greater initial endothelial cell loss [29, 30], the subsequent cell loss is at a slower rate in comparison to PK . This may be related to the larger grafts with greater number of cells inserted during DSAEK. A large graft is not always an advantage, as an oversized graft may be more difficult to insert, and is also less tolerant to decentration, with a higher risk of crowding of the anterior chamber angle, contact with the peripheral iris leading to the development of peripheral anterior synechiae, and hence the potential for a higher risk of both angle-associated secondary glaucoma, and allograft rejection. The importance of DSAEK graft sizing may be of greater significance in our population of Asian eyes where the incidence of narrow angles and angle closure glaucoma is much higher than in Western and European populations . Studies have also found that DSAEK grafts induce an initial hyperopic shift that decreases somewhat over time . This hyperopic shift is reported to be correlated with central graft thickness, graft trephine diameter as well as the thickness gradient between the centre and periphery of the graft [6, 7, 33]. Examining the relationship between the posterior and anterior graft arc length, in addition to the graft thickness and diameter, with the final refractive outcome, will allow us to better predict this hyperopic shift with the ZAP software.
In future, we aim to perform further studies to prospectively analyze the effect of graft size and the ratio of graft diameter to posterior cornea arc length, on postoperative outcomes including endothelial cell count and refractive outcome. Once we establish an optimal graft size ratio, we can estimate the appropriate graft trephine diameter based on the cornea posterior arc length from the pre-operative ASOCT image. Our multiple regression analysis showed that, a model using posterior cornea arc length and AC width measured on ASOCT would allow us to reasonably estimate the graft arc length based on the trephine diameter chosen (Table 4).
The use of ASOCT imaging and ZAP software in our study does have some limitations. It has previously been shown that there can be difficulty in detecting the sclera spurs on some ASOCT scans . In some images, the internal surface of the sclera formed a smooth continuous line with no inward protrusion or change in curvature which made it impossible to detect the sclera spur. In addition, some ASOCT images were of suboptimal quality, which also affects the ability to accurately identify the sclera spur. Some of these poor quality images could not be processed by the ZAP software even if the sclera spur could be identified; this figure was 9.9% in our study. The study was also limited in that measurements were performed on horizontal (nasal/temporal) ASOCT scans only. We chose to use only horizontal scans since these scans have been shown to be the most consistent with respect to obtaining high quality images for ZAP software analysis, and also recognizing that most surgeons measure the horizontal rather than vertical white-to-white diameter . In addition, ASOCT imaging of vertical scans can be limited by eyelid anatomy especially in our population of Asian eyes with small palpebral apertures. A further limitation to the software is that it can only be used on time domain scans that image the whole length of the anterior segment. The next generation of Fourier domain ASOCT scanners whilst providing better quality images have the disadvantage of not being able to scan the whole anterior segment in one scan.