Background/aims To include intraoperative measurements of the anterior lens capsule of the aphakic eye into the intraocular lens power calculation (IPC) process and to compare the refractive outcome with conventional IPC formulae.
Methods In this prospective study, a prototype operating microscope with an integrated continuous optical coherence tomography (OCT) device (Visante attached to OPMI VISU 200, Carl Zeiss Meditec AG, Germany) was used to measure the anterior lens capsule position after implanting a capsular tension ring (CTR). Optical biometry (intraocular lens (IOL) Master 500) and ACMaster measurements (Carl Zeiss Meditec AG, Germany) were performed before surgery. Autorefraction and subjective refraction were performed 3 months after surgery. Conventional IPC formulae were compared with a new intraoperatively measured anterior chamber depth (ACD) (ACDIntraOP) partial least squares regression (PLSR) model for prediction of the postoperative refractive outcome.
Results In total, 70 eyes of 70 patients were included. Mean axial eye length (AL) was 23.3 mm (range: 20.6–29.5 mm). Predictive power of the intraoperative measurements was found to be slightly better compared to conventional IOL power calculations. Refractive error dependency on AL for Holladay I, HofferQ, SRK/T, Haigis and ACDintraOP PLSR was r2=−0.42 (p<0.0001), r2=−0.5 (p<0.0001), r2=−0.34 (p=0.010), r2=−0.28 (p=0.049) and r2<0.001 (p=0.866), respectively,
Conclusions ACDIntraOP measurements help to better predict the refractive outcome and could be useful, if implemented in fourth-generation IPC formulae.
- Intraocular pressure
- Lens and zonules
- Optics and Refraction
- Treatment Surgery
Statistics from Altmetric.com
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.
Although intraocular lens (IOL) power calculation (IPC) improved during the last decades, approximately 5%1–3 of all patients need a refractive correction of more than ±1 dioptre after cataract surgery. As shown by Norrby,4 the main source (about 35%) of this postoperative deviation from the aimed refraction is the incorrect prediction of the final IOL position. This error increases to 50% if the postoperative subjective refraction is not taken into account as a source of error. To calculate IOL power, the so-called effective lens position (ELP) is used. ELP is a virtual position that is assumed by using preoperative measurements, such as corneal radii, axial eye length (AL) and, in some formulae, the anterior chamber depth (ACD).5–10 Another option is to use non-paraxial ray tracing, or thick lens formulae.11–13 One problem that all these methods have in common is the uncertainty of the final IOL position. It should be mentioned that ray tracing methods use a real (anatomical) postoperative ACD that can directly be compared with the measured postoperative ACD, whereas, this is not the case for conventional IOL power calculation formulae that use the fictitious ELP.
Recently, it was shown that intraoperative optical coherence tomography (OCT) measurements of the anterior lens capsule (ACDIntraOP) after implanting a capsular tension ring (CTR) was a better prediction factor for the postoperative IOL position compared to other factors, such as preoperatively measured ACD, lens thickness, or corneal radii.14
The aim of this study was to evaluate if prediction of the postoperative refraction was more accurate if ACDIntraOP measurements are used for IPC formulae.
Materials and methods
This prospective study included patients, who were scheduled for cataract surgery. Thirty hypermetropic eyes (emmetropic IOL power >23.0D), thirty normal eyes (emmetropic IOL power 18.0D to 23.0D) and 10 myopic eyes (emmetropic IOL power <18.0D) were included. Exclusion criteria were preceding ocular surgery or trauma, pseudoexfoliation syndrome, intraoperative floppy iris syndrome and a preoperative visual acuity of less than 0.05 Snellen. All the research and measurements followed the tenets of the Declaration of Helsinki and were approved by the local ethics committee. Written informed consent was obtained from all patients in the study.
Preoperatively, the eye to be operated was examined at the slit-lamp, and optical biometry was performed using the IOLMaster 500 (Carl Zeiss Meditec AG, Germany). Furthermore, partial coherence interferometry (PCI) measurements (ACMaster, Carl Zeiss Meditec AG, Germany) of the ACD were performed preoperatively. IOL power calculation was performed using the SRK/T formula in all cases, except in those cases with an AL of below 22.0 mm. In these eyes the Hoffer Q formula was used.
During surgery, continuous OCT imaging was performed and screenshots of the video were taken in a second step after surgery. Surgery was performed under topical anaesthesia in all cases by one experienced surgeon (OF). A self-sealing clear corneal incision with a 2.8 mm single-bevelled steel blade, and one paracenteses, were performed. Incision was followed by the injection of an ophthalmic viscoelastic device (OVD), capsulorhexis, phacoemulsification and irrigation/aspiration of cortical material. After injection of a cohesive OVD (Healon, Abbott Medical Optics, USA) a capsular tension ring with 11.0 mm diameter (CTR, Morcher, Germany) was implanted. Then the OVD was removed rigorously and the infusion bottle of the phacomachine was adjusted to 27.2 cm above eye level to ensure a relatively normal intraocular pressure of 20 mm Hg using a handpiece of a bimanual irrigation/aspiration set held through the paracenteses. At this time point, a screenshot from the continuous recording of the OCT was taken. Then, injection of a cohesive OVD into the capsular bag was performed as standard procedure. The IOL (Tecnis ZCB00) was implanted into the capsule bag using the Emerald AR injector (both Abbott Medical Optics, USA). Following the implantation of the IOL, the viscoelastic substance was aspirated thoroughly using a bimanual I/A set. Care was taken to completely remove OVD from behind the IOL by slightly displacing and tilting it and reaching behind the optic with the aspiration cannula. Postoperative therapy within the first month was ketorolac gtt three times daily for 1 month.
Additional examinations at 3 months, postoperatively, were autorefraction in IOL mode (Topcon RM 8800, Topcon, Japan) and subjective refraction using the cross-cylinder method. For this paper, only subjective refraction was used for further analysis.
Intraoperative OCT set-up and analysis
This method is explained elsewhere in detail.14 In short, a time-domain OCT with a wavelength of 1310 nm (Visante, Carl Zeiss Meditec AG, Germany) was used. This anterior segment OCT was shown to be highly reproducible for measurements of the ACD15 and of the IOL/crystalline lens.16 In this project, a prototype of a combination of this anterior segment OCT and an operating microscope (OPMI 200, Carl Zeiss Meditec AG, Jena, Germany) was used to allow continuous measurements of the cornea and the anterior lens capsule after implanting a CTR. The distance between the centre of the anterior surface of the cornea and the centre of the anterior lens capsule was used for further analysis (=ACDIntraOP).
For statistical analysis, Microsoft Excel 14.0 for Mac (Microsoft, USA) with a Statplus:mac V.18.104.22.168 plug-in (AnalystSoft, USA), an Xlstat 2012 plug-in (Addinsoft, USA) and a Solver plug-in in Excel (Microsoft, USA) were used. Furthermore, SPSS V.21.0 for Mac (IBM, USA) was used for histograms and scatter plots. Missing data observations were excluded from analysis. For statistical modelling Partial Least Squares Regression (PLSR) was performed with Xlstat 2012.
To optimise all conventional IPC formulae constants, the Solver software was used to back-calculate the postoperative refractive outcomes that produce zero mean error (Excel file developed by SN).
In total, 70 eyes of 70 patients (60% women and 40% men) were included in this study. Descriptive data are depicted in table 1. ACDIntraOP measurements were not possible in four cases. Ten patients were lost to follow-up, six due to general health problems, and four due to incompliance.
Mean intraoperative ACD (measured from the endothelium of the cornea to the anterior lens capsule) of the aphakic eye was 5.11 mm (SD: 0.66; range: 3.63–6.84). Mean postoperative subjective refraction (SE) at 3 months was −0.71 D (SD: 0.73 D; range: −3.0–0.6). In four cases postoperative refractive outcome was aimed for –2.5D. All constants of all IPC formulae were optimised to allow comparison. The newly derived constants for Holladay I, HofferQ, SRK/T, Haigis formulae were sf=1.88, ACD=5.63, A=119.1 and a0=−1.36, respectively. For the Haigis formula, only a0 was optimised and a1 and a2 were taken from the ULIB data base.17
In the next step, a PLSR model for the ACDIntraOP measurement was generated to predict the postoperative manifest refraction (SE). AL, ACDIntraOP, and the used IOL power were defined as explanatory variables. Additionally, the aimed refraction was added to the equation. The resulting prediction equation is shown below (PIOL=IOL power, AL=axial eye length in mm, as given by the IOLMaster 500 and ACDintraOP=intraoperatively measured distance between the endothelium of the cornea and the anterior lens capsule after implanting a capsular tension ring plus the preoperatively measured corneal thickness in mm). Corneal radii were not found to have a predictive power and were excluded from the model.
Tables 2 and 3 show the distribution of the refractive error with different IPC formulae using optimised constants. None of the conventional formulae was found to be significantly superior to another, when compared using multiple comparison testing and a Bonferroni correction of 0.005. Intraoperative ACDintraOP measurements were more powerful predictors compared to conventional IOL power calculations, but these differences were not found to be significant (intraoperative measurement vs Holladay: p=0.957; vs HofferQ: p=0.786; vs SRK/T: p=0.992; vs Haigis: p=0.996).
Using Pearson correlation, refractive error dependency on AL for Holladay I, HofferQ, SRK/T, Haigis and ACDintraOP PLSR was r2=−0.42 (p<0.0001), r2=−0.5 (p<0.0001), r2=−0.34 (p=0.010), r2=−0.28 (p=0.049) and r2<0.001 (p=0.866), respectively. Table 4 shows the same results using a one-way analysis of variance instead of Pearson correlation.
To compare the predictive power of preoperative ACD measurements with ACDintraOP measurements without a CTR and with a CTR, a PLSR model was created using the following parameters: AL, mean keratometry, used IOL power and all three ACD measurements. Therefore, the effect was tested separately for short eyes (below 22.0 mm) and all other eyes (more than 23.0 mm; there was no eye with exactly 23.0 mm). For both groups, ACDintraOP measurements with a CTR were superior to other ACD measurements, followed by ACDintraOP measurements before implanting a CTR. In both groups, preoperative measurements showed the least predictive power of the postoperative refraction (spherical equivalent).
The aim of this study was to include intraoperative ACDintraOP measurements of the aphakic eye into the IPC process. To our knowledge, this is the first study of its kind and showed that the ACDintraOP with a continuous intraoperative OCT device is a useful parameter for IPC. Other attempts, to calculate the IOL power using intraoperative measurements showed little success: intraoperative autorefraction has problems, such as misalignment, or insufficient patient fixation,18 and intraoperative wavefront aberrometry was found to be too inaccurate for IPC.19 Additionally, the main source of error, namely predicting the postoperative IOL position4 remains. In this study, the intraoperative position of the anterior lens capsule in aphakic eyes was used to predict the postoperative IOL position. It was assumed that this position is very similar to the anterior aspect of the haptics of the IOL, immediately after surgery. This assumption was derived from the idea of the lens haptic plane concept.20 Furthermore, it follows the intentions by Holladay et al8 to use anatomical ACD measurements. In a first approach, it was tried to directly implement the ACDintraOP measurement into conventional IPC formulae. However, due to their empirically developed nature this method was not found to be successful. Therefore, a simple PLSR model was created to use the ACDintraOP together with AL, Kmean and aimed postoperative refraction. PLSR has some advantages compared to simple multiple linear regression modelling that were explained elsewhere.14 In short, multiple linear regression modelling suffers from ‘overfitting’. This means that increasing the number of explanatory variables will always result in a higher predictive power (an increased r2), but this is an erroneous improvement, because it only counts for the investigated sample, but not for patients outside the investigated sample. Another problem is the colinearity, as multiple linear regression should only be used if all explanatory variables are independent from each other, and this is not the case for most of the factors in IOL power calculation, such as AL and ACD. These and other problems can be significantly reduced with PLSR. Furthermore, PLSR uses a matrix-based regression that allows a weighting of all explanatory variables as well as a more powerful analysis with a significantly lower sample size.
Although the sample size was large enough to show that ACDintraOP PLSR modelling was superior to conventional IPC formulae concerning AL dependency, this formula was only created to compare conventional optimised formulae with a formula including ACDintraOP, but it should not be seen as a new IPC formula. In this study, we avoided first-generation modelling techniques, such as multiple linear regression models that were used in the past to predict the postoperative ACD for several reasons. One advantage of PLSR is the weighting of explanatory variables, which could also significantly lower AL dependency compared to conventional IPC formulae.
In four cases, intraoperative measurements were not successful. In these cases, the resolution of the OCT scans was not good enough to clearly detect the anterior lens capsule properly, or the anterior lens capsule was not taut at the measurement time point (2 cases). The first problem could be solved in the future by using an OCT with a higher resolution and/or signal to noise ratio. The second problem was partly taken care of by using a CTR. Although intraoperative measurements without a CTR were shown to be a better predictor of the postoperative IOL position compared to preoperative ACD measurements, however, the superiority of intraoperative measurements with a CTR was even higher. One aim for the future will be to improve the intraoperative measurement technique without a CTR to avoid the additional costs of a CTR.
Another problem that should be mentioned is the possibility of vitreous hydration during cataract surgery. This would result in flawed intraoperative measurements. However, it is currently not possible to detect vitreous hydration with the intraoperative OCT set-up used in this study.
This study showed that intraoperative measurements of the position of the anterior lens capsule after having removed the lens content are useful to better predict the postoperative refractive outcome compared to conventional IPC formulae. The next steps could be to automate the process of intraoperative measurements and to implement these capsule position measurements into fourth-generation formulae or ray tracing using the real anatomical IOL position instead of a fictitious ELP in the future.
Contributors N Hirnschall: contribution to concept and design, analysis of the data, writing the manuscript. S Norrby: contribution to analysing the data, reviewing the manuscript. M Weber: collection of data, reviewing the manuscript. S Maedel: collection of data, reviewing the manuscript. S Amir-Asgari: collection of data, reviewing the manuscript. O Findl: concept and design, analysing the data, reviewing the manuscript.
Competing interests None.
Ethics approval Local ethics committee of the City of Vienna.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement The authors have no proprietary or financial interest in any of the materials or equipment mentioned in this study. The trial was sponsored by the ‘Jubilaeumsfond’ of the Austrian National Bank (project number 14052).