Post-operative Refractive Prediction Error After Phacovitrectomy: A Retrospective Study

Vitreoretinal disorders and cataract are often operated on simultaneously in elderly patients. Pars plana vitrectomy combined with phacoemulsification and intraocular lens (IOL) implantation (phacovitrectomy) reduces recovery time and costs and avoids the need for a second operation [1, 2].

In phacovitrectomy, as in simple phacoemulsification and IOL implantation, achieving the best uncorrected visual acuity, has become more important. However, many authors have reported on a refractive prediction error of approximately − 0.50 dpt after phacovitrectomy [3].

This myopic shift has been attributed to several factors [3]. Simply replacing vitreous with aqueous theoretically causes a myopic shift of approximately − 0.13 dpt because of a slight difference in the refractive index [4‐7]. The actual lens position of the IOL could also differ from the expected lens position because of differences in IOL type [8] and the use of gas tamponade [9‐14]. Moreover, the type of vitreoretinal pathology could influence the IOL power calculation due to errors in axial length measurements [15‐17]. To date, however, there is no consensus on the exact role these factors play in the development of a myopic prediction error.

In the study reported here we evaluated the amount of post-operative refractive prediction error after phacovitrectomy in our facility and attempted to elucidate the various factors involved.

In accordance with guidelines from the ethics committee, we retrospectively included patients who underwent phacovitrectomy with IOL implantation from a single surgeon between October 2012 and February 2016. The indications for performing vitrectomy were either macular hole, macular pucker, or floaters. Patients were excluded if the eye had been operated on previously, if the IOL could not be placed in the capsular bag, or if a suture was left on the tunnel. Based on these criteria, 140 patients were included in the study.

Biometry was performed pre-operatively by ultrasonography using an Ocuscan immersive A-scan ultrasound (US) (Alcon, Fort Worth, TX) with keratometry (Retinomax K-plus 3; Righton, Tokyo, Japan) or optically using partial coherence laser interferometry (IOLMaster 500; Carl Zeiss Meditec AG, Jena, Germany). Pre-operative refractive measurements were obtained using an automatic KR-8100P refractometer (Topcon, Tokyo, Japan). Either a HOYA FY-60AD with an A-constant of 118.4 (HOYA Surgical Optics, Singapore) or Zeiss CT Asphina 409MP with an A-constant of 118.0 (Carl Zeiss Meditec AG) foldable IOL was used. The Holladay-II [18] and SRK/T [19] formulas were available to calculate IOL power, .

Surgery was performed under local or general anesthesia. For cataract removal, a 2.4-mm sclerocorneal incision superiorly and two sideports at the 10 and 2 o’clock positions were made. A continuous curvilinear capsulorhexis was created following which phacoemulsification and aspiration of the crystalline lens were performed. The IOL was implanted in the capsular bag. Vitrectomy was done using a 25 gauge three-port technique. Either air or gas [sulfurhexafluoride (SF₆) or perfluoropropane C₃F₈)] tamponade was used.

Post-operative data included spherical equivalent measurements using the same automated refractometer. The post-operative refractive prediction error was calculated by subtracting the actual refraction from the predicted refraction, using data collected at 1 month post-operatively and, if available, at 3, 6, and 12 months post-operatively. Statistical analysis was performed using appropriate software. Both univariate and multivariate analyses were performed to identify the predictors for post-operative refractive error.

The study was performed in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1964, as revised in 2013.

In our patient population we found an overall myopic refractive error. For the different diagnoses independently, however, only the macular holes showed a myopic error with both the SRK/T and Holladay-II formulas. As such, the existence of a macular hole was found to be associated with the development of a myopic error. Other factors involved in the error were the use of gas tamponade, pre-operative SEQ, axial length, and the use of optical axial length measurement.

The overall amount of post-operative myopic refractive error seen in our study correlates highly with results of other groups, as outlined in the review by Hamoudi et al. [3].

Our results in terms of there being a myopic error only for the patients treated for macular hole, whereas the group of puckers and floaters showed no significant error are in contrast to those reported by Falkner-Radler et al. [14] who describe a larger error for puckers compared to macular holes. These authors hypothesize that gas tamponade, which is routinely used for patients with a macular hole, could lead to zonular elasticity, resulting in a more posterior localization of the IOL once the gas has resolved [14]. On the other hand, many other researchers have postulated that because of the gas tamponade, the IOL is pushed more anteriorly, thereby causing the myopic shift [9‐13]. The latter hypothesis is more in line with our results as we observed a greater myopic error in those patients with macular hole and gas tamponade in the univariate analysis, although neither type of diagnosis nor use of gas tamponade was a significant predictor in the multiple regression analysis.

Also, one could suspect that the accuracy of the axial length measurement is involved in the myopic error for macular holes. An accurate measurement of axial length is of great importance for a smaller post-operative refractive error. An error of 0.1 mm in axial length causes a post-operative refractive error of 0.27 dpt [3, 15, 20]. In some cases in our study, US was the preferred diagnostic tool due to dense media opacities; in others, US was chosen merely for logistical reasons. Theoretically, the use of an optical device, such as the IOL-Master optical biometer, to measure axial length would be more reliable as it measures up to the retinal pigment epithelium, whereas US measures up to the internal limiting membrane which can be affected by the specific disease entity [15‐17]. Also, even though at our facility we use immersion US biometry, US measurement can be affected by unintentional indentation of the cornea during acquisition. However, our comparison of patients measured with the IOL-Master and those measured with US revealed no difference in axial length, not for the overall population, nor for the specific disease entities independently. Moreover, even if an error in US axial length measurement for macular holes would be at play here, one would expect it to be an overestimation of axial length, resulting in a post-operative hyperopic refractive error, not a myopic one.

A interesting observation is that overall, and for the group of puckers in particular, we did see a larger post-operative myopic prediction error for those patients measured with the IOL-Master device compared with those measured with US. Rationally, we cannot explain such a difference for the puckers as we expect an underestimation of axial length by US to result in more myopic error.

There has been discussion as to what extent replacing vitreous with aqueous influences post-operative refraction. Some researchers state that, based on the schematic eye, the refractive index of vitreous is virtually identical to that of aqueous and therefore does not explain the myopic error [21, 22]. In reality, the refractive index of vitreous (1.3346) is slightly higher than that of aqueous (1.3336), which results in a myopic shift of approximately − 0.13 dpt [4‐7]. In phacovitrectomy for floaters no error in axial length measurement is expected, and no alterations in retinal anatomy is involved in the surgery. Hence, in these patients one would expect an error to the degree explained solely by vitreous replacement. With the Holladay-II formula we indeed observed such an error (− 0.14 dpt). However, this was not statistically significant, possibly due to very low numbers in this particular group.

The stability of the IOL position could in theory be greatly affected by IOL features, such as edge design, haptics design, and materials. It has been postulated that for phacovitrectomy the axial movement (and therefore changes in refraction) is greater in one-piece IOLs than in three-piece IOLs [8]. In our center, however, we routinely place either a HOYA one-piece with a Zeiss one-piece IOL in the bag. Therefore, no comparison between one- and three-piece IOLs could be made in our study. No difference in ultimate refractive error was noted between the HOYA IOL (C-loop haptics) and Zeiss IOL (plate haptics) IOL in univariate or multivariate analysis.

We observed a greater myopic refractive error with the Holladay-II formula than with the SRK/T formula. This difference can possibly be explained by the fact that the Holladay-II formula is often used in highly myopic eyes as it generally tends to select an IOL of greater power than would other formulas, hence preventing an inadvertent post-operative hyperopia [23].

Jee et al. described a larger post-operative myopic error in highly myopic eyes that was related to a change in axial length in these eyes [24]. Although this correlation was not significant in our population, we did find a correlation between the pre-operative refractive state of the eye and the post-operative refractive prediction error. It is possible that highly myopic eyes are more susceptible to axial elongation as a result of the phacovitrectomy because of a thinner sclera and lower ocular rigidity. In the multivariate model, both pre-operative spherical equivalent and pre-operative axial length were strong predictors of the post-operative refractive error.

Our study is limited by its retrospective nature and the heterogeneity in pre-operative refractive state, diagnosis, IOL type, technique for axial length measurement, and usage of tamponade. To address the matter of population heterogeneity, we obtained refractive outcomes within the different subpopulations. Also, this allowed us to elucidate to which extent these factors are involved in the myopic error. For this reason, we chose not to select only one group of each parameter for our analyses.

Furthermore, as outlined by Sheard [25], the A-constant is a value that adjusts the IOL power calculation for systematic errors resulting from various factors, such as the diversity in biometry measurement devices, patient population, and surgical technique. In our facility, when planning the operation it is yet uncertain who will perform the operation. Aristodemou et al. [26] also demonstrated that with current small-incision phacoemulsification, no significant difference is found when using surgeon-specific constants. Therefore, we have chosen not to alter the constants for each surgeon but rather to use the standard A-constant. The myopic error in our population of phacovitrectomy is highlighted when our results are compared to those of studies using non-optimized formulas in patients undergoing phacoemulsification only, as these studies show a hyperopic error of over + 0.50 dpt [26‐29].

As evidenced by our study, not every factor seems to be as important in each individual patient, suggesting one cannot generalize all patients into one group, but instead a more tailored approach is probably warranted. Based on our results, for example, a greater myopic error is to be expected in patients with a pre-operative myopic refractive state and who undergo phacovitrectomy for a macular hole with the usage of gas tamponade. As for the future, we endeavor to overcome the phenomenon of a refractive prediction error for phacovitrectomies. To further elucidate the role of gas tamponade in this matter, we have set up a prospective study in which changes in lens position and axial length are further investigated in a more homogenous population with respect to the vitreoretinal pathology and (pre)operative factors, such as lens type and axial length measurements.