The accuracy of intraocular lens (IOL) power calculation and the subsequent refraction of the pseudophakic eye mainly depend on the prediction of the postoperative IOL position and on the accuracy and precision of preoperative measurements of different parameters, such as axial length and, to a lesser extent, keratometry.1–4 The postoperative IOL position, usually defined as the effective lens position, is influenced by several parameters, such as the IOL design and the capsulorhexis features.5–7 Any axial movement of the IOL caused by a fibrotic reaction of the capsule is likely to induce a refractive change of the eye. For this reason, it has been postulated that femtosecond laser-assisted cataract surgery may improve the refractive outcome of IOL implantation. Several studies have shown that femtosecond laser-assisted cataract surgery capsulotomies are more symmetric, allow a complete overlying of the capsulorhexis border on the IOL optic, lead to a better IOL positioning in the early postoperative period, and lead to a symmetrical and uniform long-term capsule fibrosis.8–13 Although Filkorn et al. reported better outcomes after femtosecond laser-assisted cataract surgery than standard phacoemulsification,14 there has been no final demonstration that femtosecond laser-assisted cataract surgery really improves the refractive accuracy of cataract surgery with IOL implantation.
This study compares the IOL axial movements and the related refractive changes during 6 months of follow-up after femtosecond laser-assisted cataract surgery and conventional cataract surgery, evaluates the influence of capsulotomy features on IOL positioning, and assesses IOL centration. In addition, we aim to investigate whether femtosecond laser-assisted cataract surgery improves the refraction prediction error by third-generation IOL power formulas with respect to conventional cataract surgery.
Patients and Methods
The prospective study (continuous quality control) was performed at the Eye Clinic, Department of Medicine and Science of Aging, University of “G. d’Annunzio,” Chieti-Pescara, Italy. The study was approved by the local ethics board and adhered to the tenets of the Declaration of Helsinki. All patients provided written informed consent to participate in the study.
The inclusion criteria were age between 65 and 75 years, nuclear cataract of grade 3 (nuclear opalescence of 3; Lens Opacities Classification System III), and corneal endothelial cell count greater than 1,200/mm2. The exclusion criteria were poor pupil dilation, pathologies that can present alterations of the anterior segment (eg, corneal opacities, keratoconus, chronic uveitis, zonular dialysis, pseudoexfoliation syndrome, glaucoma, and diabetes), other ocular pathologies that can impair visual function, previous anterior or posterior segment surgery, and intraoperative or postoperative complications.
Eighty eyes of 80 patients (age range: 65 to 75 years) who were candidates for cataract surgery were included in the study. Consecutive patients were enrolled into two groups: the femtosecond laser group (40 eyes) and manual group (40 eyes). The enrollment was sequential. The manual group consisted of patients who underwent conventional surgery prior to the availability of the LensX laser system (Alcon LensX Inc., Fort Worth, TX). The femtosecond laser group included consecutive patients who underwent surgery after an adequate period of training (2 weeks) on the new system.
Before cataract surgery, patients underwent a complete ophthalmologic examination including slit-lamp evaluation, applanation tonometry, and ophthalmoscopy through dilated pupils.
Preoperatively, keratometry, anterior chamber depth (ACD), and axial length were measured by means of an IOLMaster (software version 5.4.3.0002; Carl Zeiss Meditec AG, Jena, Germany). In both groups, a monofocal aspheric IOL (AcrySof SN60WF; Alcon Laboratories, Inc., Fort Worth, TX) was implanted. The IOL power was calculated using the Hoffer Q, Holladay 1, and SRK/T formulas with a targeted refraction of emmetropia (0.00 diopters [D]).15–17 The IOL power choice was based on the Hoffer Q formula for short (< 22 mm) eyes, the Holladay 1 formula for average (range: 22 to 24.49 mm) and medium-long (range: 24.50 to 25.99 mm) eyes, and on the SRK/T formula for long (> 26.00 mm) eyes.18
A single experienced surgeon (LM) performed all femtosecond laser-assisted and manual procedures, which included corneal incisions, anterior capsulotomy, phacoemulsification, and IOL implantation. The eyes were dilated and topical anesthesia was administered repeatedly before starting the procedure. All surgical procedures were performed using standard surgical equipment.
LensX Femtosecond Procedure
In the femtosecond laser group, the capsulotomy, lens fragmentation, and corneal incisions were performed using the LensX platform. The capsulotomy diameter was 4.9 mm. The upper and lower deltas of the capsule were set to 330 µm. The energy was 6 µJ with a spot separation of 3 µm and a layer separation of 3 µm. Capsulotomy was manually centered on the pupil center by the surgeon via video microscopic imaging at the beginning of the case with the pupil dilated.
Lens fragmentation was performed after capsulotomy. A temporal 2.7-mm three-plane primary clear corneal incision and a secondary 1-mm one-plane corneal incision were performed at the end of the femtosecond procedure.
In the manual group, a temporal 2.75-mm three-plane primary clear corneal incision and a secondary one-plane corneal incision were made using disposable keratome knives. The manual capsulorhexis was performed using the continuous curvilinear capsulorhexis technique using cystotome and rhexis forceps with an intended diameter of 4.9 to 5.0 mm. In all cases, a divide-and-conquer technique was performed for lens fragmentation.
Standard Surgical Procedures for Both Groups
In both groups, standard phacoemulsification was used to complete the surgery with combined longitudinal/torsional ultrasound mode using the Alcon Constellation System (Alcon Laboratories, Inc.). IOLs (AcrySof SN60WF; Alcon Laboratories, Inc.) were implanted in the capsular bag with a Monarch III injector and Monarch D Cartridge (Alcon Laboratories, Inc.). The incisions were neither hydrated nor sutured.
Postoperative therapy consisted of ofloxacin 0.3% and dexamethasone 0.2% eye drops four times daily for 3 weeks.
At 7, 30, and 180 days after surgery, each patient underwent an evaluation of uncorrected and corrected distance visual acuity using the logMAR scale.
Subjective refractive sphere and cylinder were measured for all patients. In all cases, the postoperative ACD was measured at each follow-up visit with the Visante anterior segment optical coherence tomography (model 1000; Carl Zeiss Meditec, Inc., Dublin, CA). Cross-sectional scans were obtained for each patient and scans with the best quality, in terms of visibility of anterior segment, were chosen to measure the ACD with the anterior segment dual-line scan mode. The patients were seated and directed to maintain fixation on the internal target. An image chamber tool was used to measure the ACD, defined as the distance from the central corneal epithelium to the anterior IOL surface. All scans were taken by a single examiner (LT). All data were calculated from the best image obtained in a series of three images.
In addition, capsulorhexis/capsulotomy assessments were performed using retroillumination photographs obtained with maximum pupil dilation at 7, 30, and 180 days postoperatively. The images were blinded for patient information and randomized prior to analysis. All digital images were imported and analyzed with an in-house closed-source software developed in MatLab 2009b (MathWorks, Natick, MA), as previously described.10 Capsulorhexis/capsulotomy size and IOL centration were evaluated for each image as previously described.10
Assessment of Prediction Error in Refraction
To calculate the prediction error in refraction with each formula, the postoperatively measured refraction was subtracted from the predicted refraction (based on the IOL actually implanted) for all eyes of both groups.15–17 The manifest refraction was measured 1 month after surgery, which is when refractive stability can be expected with small-incision clear corneal surgery and the type of IOL implanted, and after 6 months.19–21
For each group, the predictions made by the Hoffer Q, Holladay 1, and SRK/T formulas were optimized in retrospect by adjusting the personalized ACD, surgeon factor, and A constants to give an arithmetic prediction error of zero in the average case, according to the method described by Hoffer15,18 and Olsen.2 As a result, it was possible to evaluate the statistical error because representing the optimum prediction error rather than offset errors related to incorrect lens constants or systematic errors in the measuring environment.
The median absolute error in refraction prediction with each formula was then compared between the femtosecond laser and manual groups.
Sample Size and Statistical Analysis
An estimation of the number of eyes that needed to be enrolled was based on the main endpoint, which was the difference in distance between IOL centroid and pupil centroid at 180 days after surgery between the femtosecond laser and manual group. Assuming a difference in variation of at least 0.08 mm in distance between the two groups, using a t test for unpaired data at a level of 0.05 with 80% power, and the common standard deviation of 0.10 mm, approximately 26 patients were needed in each group. This value was based on previous studies comparing femtosecond laser-assisted capsulotomy and manual capsulorhexis.13 The calculation was performed using PASS 2005 (NCSS, LLC, Kaysville, UT).
All quantitative variables were summarized as mean and standard deviation. Qualitative variables were summarized as frequency and percentage. The Shapiro–Wilk test was used to detect departures from normality distribution.
Student’s t test for unpaired data was applied for assessing the quantitative variables between groups, whereas the chi-square test was applied for qualitative variables. Linear mixed-effects models for repeated measurements were used to analyze the effect of surgery on continuous outcome variables (visual parameters, refractive parameters, capsulorhexis area, and IOL centroid–pupil centroid distance). Models were used to regress measures with patients as a random effect on the fixed-effect factor (group) assuming unstructured covariance matrix. The crossover effect of time and group was entered as an interaction term for each outcome variable. When normality of data was not verified, data transformation was applied before the model.
Contrast analysis, priori specified, was also used to evaluate the difference between groups at each time period analyzed. All statistical analyses were conducted using SPSS software version 11.0 (SPSS, Inc., Chicago, IL).
Patient demographics are listed in Table 1. All surgical procedures were completed and were uneventful. No patient was lost to follow-up.
In both groups, 100% of patients showed a corrected distance visual acuity 0.3 logMAR or greater (> 20/40 Snellen) at each follow-up visit. The differences in corrected distance visual acuity between the two groups were not statistically significant.
The median absolute error was not significantly different between the two groups with all formulas at each follow-up control ranging between 0.29 (Hoffer Q) and 0.64 D (Hoffer Q) in the femtosecond laser group and between 0.24 (SRK-T) and 0.55 D (Hoffer Q) in the manual group (Table A, available in the online version of this article).
Median Absolute Error In Refraction Prediction
The mean arithmetic error was zero for all formulas as a result of constant optimization. The optomized constants for the Hoffer Q, Holladay 1, and SRK/T formulas are shown in Table B (available in the online version of this article).
Optimized Constants for the Hoffer Q, Holladay 1, and SRK/T Formulas
Mixed-model analysis indicated the presence of statistically significant differences in postoperative subjective and objective spherical equivalent (P < .001) between groups. The effect of time after surgery was also statistically significant (P = .043) (Table 2). At 7 days, the mean spherical equivalent was statistically significantly different between groups with a slightly more myopic spherical equivalent refraction in the femtosecond laser group compared to the manual group (P < .05, contrast analysis). At 30 and 180 days, the mean spherical equivalent showed a hyperopic shift in the femtosecond laser group and a myopic shift in the manual group from 7 days postoperatively; the shift was 0.17 ± 0.23 D in the femtosecond laser group and −0.23 ± 0.10 D in the manual group (P < .001).
Postoperative Outcomes by Group
At 7 days postoperatively, the mean ACD was 4.71 ± 0.29 mm (range: 4.12 to 5.33 mm) in the femtosecond laser group and 4.75 ± 0.30 mm (range: 4.30 to 5.25 mm) in the manual group and increased to 4.77 ± 0.10 mm at 180 days in the femtosecond laser group and decreased to 4.64 ± 0.29 mm at 180 days in the manual group (Figure 1). Figure 1 shows the value of ACD for the two groups at different postoperative controls. The test measures demonstrated no significant change of ACD for a period of time (P = .073) and showed that the change was significantly different depending on the group (P = .048). The group × time interaction was not statistically significant (P = .769). The overall ACD change was higher in the manual group (−0.06 ± 0.03) compared to the femtosecond laser group (0.03 ± 0.02) (P < .001) (Figure 2).
Mean postoperative anterior chamber depth (ACD) change over time for each group. The bars represent the standard error of the mean. ACD was defined as the distance from the central corneal epithelium to the anterior intraocular lens surface.
Mean postoperative change of anterior chamber depth (ACD). The bars represent the standard error of the mean. Negative values indicate a decrease and positive values an increase in ACD. ACD was defined as the distance from the central corneal epithelium to the anterior intraocular lens surface. ***P < .001 for the manual group.
Mixed-model analysis showed a statistically significant difference in the area of the capsulotomy or capsulorhexis between the groups (P < .001), whereas the effect of time after surgery was not statistically significant (P = .899). The group × time interaction was not statistically significant (P = .245). Particularly, the area of the capsulorhexis that was produced manually was significantly lower than the area of the capsulotomy obtained with LensX femtosecond laser-assisted cataract surgery at all time periods (P < .05, contrast analysis). The capsulotomy area increased slightly from 7 to 180 days in the femtosecond laser group (0.6 ± 1.5 mm2) and decreased slightly in the manual group (−0.5 ± 1.8 mm2) (Table 2, Figure 3). There was a deviation of the obtained capsulotomy/capsulorhexis area (1.6 ± 0.7 mm2) compared to the expected intended area at 180 days after surgery in the femtosecond laser group and a deviation of 3.4 ± 1.3 mm2 (P < .001) in the manual group.
Mean postoperative rhexis area for each group. The bars represent the standard error of the mean. *P < .05, contrast analysis.
The distance between the pupil centroid (geometric center) and IOL centroid was significantly lower in the femtosecond laser group compared to manual group at all time periods (P < .001). At 30 and 180 days, the distance between the two centroids was greater in the manual group than in the femtosecond laser group (P < .05, contrast analysis). IOL decentration increased for a period of time in both groups to 0.24 ± 0.06 mm in the manual group compared to 0.18 ± 0.03 mm in the femtosecond laser group (P < .001) (Table 2).
To the best of our knowledge, this is the first study evaluating IOL axial movements for a period of time using anterior segment optical coherence tomography after femtosecond laser-assisted cataract surgery and conventional cataract surgery, evaluating the influence of capsulotomy features on IOL axial movements, and assessing the prediction error in both techniques.
In the current study, the ACD showed a greater change in the manual group (−2.4%) compared to the femtosecond laser group (0.6%) from 7 to 180 days postoperatively. The mean capsulorhexis area produced manually was significantly lower than the area of the capsulotomy obtained with femtosecond laser at all time periods. The capsulotomy area increased slightly from 7 to 180 days in the femtosecond laser group and decreased slightly in the manual group.
The values of IOL decentration were significantly higher in the manual group compared to the femtosecond laser group at all time periods. IOL decentration increased for a period of time in both groups. Considering that in both groups the same IOL was implanted and that there were no statistically significant differences between the two groups in axial length, target refraction, simulated keratometry, preoperative ACD, or implanted IOL power, we believe that the reason for IOL axial changes was related to the different features of the capsulotomies/capsulorhexes. The capsulorhexis area was smaller in the manual group compared to the femtosecond laser group and a greater deviation between achieved area and intended area was detected in the manual group compared to the femtosecond laser group. The smaller rhexis size in the manual group was due to human error in estimating the true measurement of the rhexis diameter by the surgeon when performing continuous curvilinear capsulorhexis. In the femtosecond laser group, the rhexis size was set by the surgeon preoperatively in the femtosecond laser system and performed by the laser source, thus eliminating the estimation error.
It is known that the forward IOL shift occurring during the first days postoperatively is caused by the haptic compression force decay against the capsule contraction. This phenomenon is exacerbated for a period of time due to capsule fibrosis and shrinkage. It is possible to hypothesize that smaller rhexis, such as in the manual group, counteracts with greater resistance the haptic compression force and this phenomenon could increase for a period of time due to both fibrosis of the capsule and the natural decay of haptic compression force, whereas larger rhexis, such as in the femtosecond laser group, could exert a weaker resistance to haptic compression force.
In our study, the mean spherical equivalent showed a myopic shift in the manual group and a slight hyperopic shift in the femtosecond laser group from 7 to 180 days after surgery, which could be related in part to the greater anterior-posterior shift of the IOL observed in the manual group compared to the femtosecond laser group. In addition, IOL decentration that was significantly higher in the manual group compared to the femtosecond laser group soon after surgery and during the follow-up could have contributed to changes in postoperative refraction. Nevertheless, the two techniques did not show significant differences in the refraction prediction error at 1 and 6 months postoperatively because the median absolute error with all formulas was not significantly different between the two groups. These results are in contrast to the findings of a previous study,14 and further investigation on larger samples is required to understand whether femtosecond laser-assisted cataract surgery can really improve the refractive outcome of our surgery.
Several studies have examined IOL positioning after cataract surgery and the related refractive results focusing on factors determining IOL axial changes. Nejima et al.7 reported a reduction of ACD measured by Scheimpflug camera, and an associated myopic shift in eyes implanted with three-piece angulated IOLs compared to a single-piece flat haptics IOL (Acrysof IOL) during a 12-month follow-up period. Other studies reported higher stability of single-piece flat haptics IOLs (Acrysof IOLs) compared to three-piece IOLs with better refraction.22,23 Better haptic memory and lack of angulation were advocated for the better results of the single-piece IOLs.
On the other hand, Savini et al.5 observed higher mean absolute error after implantation of a one-piece Acrysof IOL (mean absolute error range: 0.15 to 0.19 D) compared to a three-piece Acrysof IOL (mean absolute error range: 0.23 to 0.30 D). They believed that one-piece IOLs had a greater decay of haptic compression force against the capsular bag compared to three-piece IOLs. The edge design and particularly the sharp edge have been related to less posterior capsule opacification and less IOL displacement compared to the round edge.24,25
Eom et al.6 analyzed IOL axial movements and predicted refractive error of three IOL models during a 6-month follow-up period. ACD was measured by means of anterior segment optical coherence tomography and defined as the distance from the central corneal endothelium to the anterior IOL surface. They observed smaller axial movement of the non-angulated C-loop longer overall length IOL (Acrysof IQ SN60WF IOL) compared to the other two angulated plate haptic shorter overall length IOLs, emphasizing the concept that a shorter angulated haptic cannot fully support the capsular bag during the postoperative period. The predicted refractive error was not significantly different between the three IOLs.6
The difference in outcomes of different studies may be the result of the different IOL features evaluated and the limited reproducibility of the measurements methods.
Our study showed that femtosecond laser-assisted cataract surgery induced a lower overall variability of ACD for a period of time compared to conventional cataract surgery with more stable postoperative refraction. Nevertheless, no differences were observed between the two techniques in the refractive predictability of IOL power calculation. Although they are not statistical, it might be mentioned that in the femtosecond laser group, the Hoffer Q formula (0.29) showed the lowest median absolute error compared to the Holladay 1 (0.38) and the SRK/T (0.43) formulas.