Discussion
The aim was to compare refractive predictability and endothelial cell loss in FLACS compared with CPS with a 6-month follow-up.
The timeframe for endothelial cell recovery is 1 to 180 days.15 To evaluate the long-term effect of FLACS on the endothelium, we therefore measured endothelial cell loss at day 180. We detected a significant difference in ECL between FLACS and CPS at both day 40 and 180 with a mean difference of 152 cells/mm2 (30% cell loss reduction in FLACS compared with CPS) at day 40 and 103 cells/mm2 at day 180 (21% cell loss reduction in FLACS compared with CPS). A patient with a healthy endothelium can endure a cell loss of 100–160 cells/mm2 without suffering from corneal oedema; however, for patients suffering from pre-existing endothelial cell loss, this amount of cell loss might cause corneal oedema.
In CPS, there are ECL rapports between 1.4% and 23%.8 16–19 Comparative studies between FLACS and CPS describes the ECL percentages as 5.8%–13.7% in the CPS group and 4.3%–17.06% in the FLACS group.9–12 20 21 Our findings with a difference in ECL percentages of 12.89%–18.19% are in the high end of these findings.9–12 20 21 A problem when comparing ECL results in comparative studies is the heterogeneity in the cell counting technique. Validation studies report that the most reliable cell counts are obtained by choosing the clearest image, which is used for automated cell count performed by Image-Net and then manually correcting any incorrectly drawn cell borders.13 14 We used this technique. However, most studies examining ECL and FLACS do not mention how ECD are counted, which might explain the different ECL results found in other reports. Another problem with previous studies is that the majority of these are non-randomised studies with a short follow-up of about 3 months making them inadequate to evaluate the long-term effect. To our knowledge, none of the comparative studies between CPS and FLACs have examined endothelial hexagonality. We saw an initial change of hexagonality at day 40 which was more profound in the FLACS group compared with CPS (5% vs 1.8% p>0.05). At day 180, the hexagonality change was normalised in FLACS while CPS reported a 2% hexagonality change (p>0.05). Our findings are in agreement with previous findings examining hexagonality after cataract surgery in patients with a healthy endothelium.8 16 19
When plotting the effect of different CDE levels on ECL (figure 2), we found that the relationship between ECL and CDE is non-linear, and CDE use above 10 U/S causes a doubling or more in ECL compared with CDE use below 10 U/S. In figure 3, ECL and the use of CDE in the two groups are plotted, showing that the initial ECL is higher in FLACS until a CDE level of approximately 10 U/S, after which the curve flattens. This suggests that one can use up to 10 U/S of CDE in CPS without causing more cell loss than with FLACS. The higher amount of ECL at <10 U/S in FLACS compared with CPS suggests that factors other than CDE are involved in ECL. These other factors might be manual handling, turbulence, knife time and fluid use—in our study, we found that FLACS was associated with increased fluid use compared with CPS. This could be due to lens fragments that are not completely divided by FLACS compared with the cracking method used in CPS.
Figure 3Correlation between cumulative dissipated energy (CDE) and endothelial cell loss (ECL) in femtosecond laser-assisted cataract surgery (FLACS) and conventional phaco surgery (CPS). At day 180, the mean ECL was 326 cells/mm2 in FLACS and 465 cells/mm2 in CPS. FLACS had greater ECL than CPS up to CDE energy levels of around 10 U/S.
Our univariate analysis found that both fluid use and operation method were significant for ECL. Both variables became non-significant in the multivariate model (table 6), leaving only CDE and preoperatively ECD as significant for ECL. This indicates that the main contributor to ECL is CDE and operation method becomes non-significant as it is correlated to energy; thus, as FLACS uses less CDE than CPS, it has a lesser impact on the ECL.
In our study, we found a 33% decrease in CDE by FLACS. We used the operation method ‘divide and conquer’ and not chopping. The energy difference between CPS and FLACS would probably have been smaller if a chopping method had been used.8 However, we chose divide and conquer as this method was the preferred method by our experienced surgeon. Chopping technique can be more surgeon dependent and might lead to more manipulation in the eye with more impact on the endothelium.8
We report five cases of complications: two cases in CPS eyes and three cases in FLACS eyes. The types of complications were different between the two groups but of equal severity. Four of the patients with complications cancelled their second eye surgery and, thus, were not included in the data analysis. Because the complications were equally distributed between the two groups, there was no skewness in the data.
FLACS produces a more centred circular capsulotomy compared with CPS. A more precise capsulotomy would likely improve the position of the IOL and, thus, the predictability of the IOL power and the refractive results. We found no significant difference in MAE (table 3 and figure 4) even when adjusting for myopia, hyperopia and corneal astigmatism. These results concur with previous findings by other authors.3 22–24 We found that our target refraction trended towards a more hyperopic refraction than intended (figure 5). In contrast to our findings, Conrad-Hengerer et al performed a randomised controlled trial examining refractive and visual outcome in 100 patients with 180 days of follow-up.25 They found significant less MAE in FLACS (92% within 0.5 D) compared with CPS (71% within 0.5 D). This may be due to their exclusion of patients with high myopia, hyperopia or corneal astigmatism of more than 1.5 D.
Figure 4Comparison of postoperative MAE at 180 days after surgery between FLACS and CPS. Ten per cent achieved the attempted refraction in both FLACS and CPS. Fifty-six per cent treated by FLACS ended with an MAE of less than 0.5 D, compared with 64% treated by CPS. We detected no significant difference in CDVA or UDVA outcome between the two groups. CDVA, corrected distance visual acuity; CPS, conventional phaco surgery; FLACS, femtosecond laser-assisted cataract surgery; MAE, mean absolute error; UDVA, uncorrected distance visual acuity.
Figure 5Achieved vs attempted refraction. Most eyes achieved an attempted refraction of zero spherical equivalent (SEQ). The circles below the line indicate that the eyes achieved a more myopic refraction than intended and vice versa. We found a subtle trend towards a more hyperopic refraction. CPS, conventional phaco surgery; FLACS, femtosecond laser-assisted cataract surgery.
We detected a significantly better UDVA in the CPS group at day 40 compared with the FLACS group. However, this difference was non-significant at day 180. We detected no difference in CDVA between the two groups, at day 40 or day 180.
A possible explanation for the non-superior refractive results in FLACS could theoretically be that FLACS causes a more unstable capsulorhexis leading to more IOL decentration and lens tilt. However, Panthier et al recently found that the mean diameter, mean deviation from intended rhexis size and mean deviation error were greater in CPS compared with FLACS a long with their finding with no significant differences in in UDVA, CDVA and MAE between the two groups.26
Mastropasqua et al and Toto et al examined lens decentration in the sagittal plane and both reported significantly less lens decentration in FLACS compared with CPS at day 180.3 24 In monofocal lenses, multiple reports state that decentration becomes clinically relevant with a decentration of >0.4 mm.27 28 Mastropasqua et al and Toto et al found lens decentration in both CPS and FLACS to be less than the clinical relevant 0.4 mm.3 24
Limitations in our study is the lack of an objective or standardised subjective grading of the cataract grade preoperatively. This was not performed as cataract grading was not of primary concern instead we used energy consumption to evaluate ECL. We chose to use the patient as an intraindividual control and therefore randomised one eye leaving the other to get the opposite treatment than the randomised eye. It would have been possible to randomise all eyes instead of only one of the two eyes. We chose the former to reduce interindividual variation and because randomisation of all eyes would require inclusion of twice the number of eyes.
We included two follow-up controls. Follow-up at day 40 was to examine refractive predictability not too long after surgery and follow-up at day 180 was to examine the long-term results of FLACS and CPS. However, it is possible that the interval between these two follow-ups are too long to detect a significance, and it would have been interesting to have a follow-up closer to the surgery date and after 3 months. We used the surgical technique ‘divide and conquer’, and we might have had different energy and ECL outcome if chopping technique had been used. Also, we used the LensAR Laser System (LENSAR) and Infiniti Vision System (Alcon Laboratories), and it would have been interesting to test a fluid-based interface system such as the Centurion Vision System (Alcon Laboratories). Furthermore, we did not measure IOL decentration or lens tilt.
Our study is strengthened by its study design, which includes randomised choice of operation, a large patient cohort, use of a single experienced surgeon, counting of ECD by manually correcting the automated drawn cell borders by a blinded observer (as validation studies suggest), and blinded optometrist refraction performed at day 40 and day 180, as well as the use of the patients as their own control: one eye operated on by FLACS and the other eye by CPS.
