|Year : 2019 | Volume
| Issue : 1 | Page : 3
Don't let me down: New intraocular lens formula effectiveness in extreme-length eyes in day-to-day practice
Tiago Morais-Sarmento, Ricardo Figueiredo, João Garrido, Ana Luísa Rebelo, Olga Berens, Augusto Candeias
Department of Ophthalmology, Hospital of Espírito Santo of Évora, Évora, Portugal
|Date of Submission||08-Jul-2019|
|Date of Acceptance||10-Jul-2019|
|Date of Web Publication||08-Aug-2019|
Dr. Tiago Morais-Sarmento
Largo Senhor da Pobreza, 7000-811 Évora
Source of Support: None, Conflict of Interest: None
Purpose: This study aimed to assess five intraocular lens (IOL) formula refractive outcomes in eyes with extreme axial length (AL) in the day-to-day practice of a secondary care center.
Design: This is a retrospective consecutive case series.
Methods: From all uneventful phacoemulsification cataract surgeries performed during 2018 (n = 1528), eyes with AL ≤22 mm and ≥25 mm were included and, after applying exclusion criteria, were validated (n = 114). Five IOL power formula predictions were compared to registered postsurgical refractions using IOL Master 500©. Two subgroups were created: eyes with AL ≤22 mm (n = 52) and AL ≥25 mm (n = 62). The formula performance in each subgroup was assessed by values of mean error, mean absolute error (MAE), median absolute error, standard deviation, and frequencies of eyes within lesser values of prediction error (PE).
Results: The formulae presented different PE values in both subgroups. In AL ≤22 mm subgroup, the MAE was 0.622 (±0.120) for Barrett II, 0.684 (±0.153) for Haigis, 0.625 (±0.131) for Hoffer Q, 0.593 (±0.130) for Holladay I, and 0.593 (±0.119) for SRK/T, without statistical significance (P > 0.05). In AL ≥25 mm subgroup, the MAE was 0.409 (±0.110) for Barrett II, 0.739 (±0.159) for Haigis, 1.143 (±0.224) for Hoffer Q, 1.058 (±0.212) for Holladay I, and 0.928 (±0.190) for SRK/T, with statistical significance compared to Barrett II and Haigis (P < 0.001). In this subgroup, the application of Wang–Koch AL adjustment seemed an advantage in Hoffer Q and Holladay I, whereas a disadvantage in Haigis and SRK/T.
Conclusions: Despite performing better than any other formulae, the 4th generation formulae performed worse than initially expected. Formula constants might explain this difference, and different IOL models can be used. However, these limitations are faced in day-to-day practice. Thus, these results hold their value as real-life refractive outcomes in eyes with extreme AL.
Keywords: Cataract extraction, eye axial length, intraocular lens, phacoemulsification, refractive errors
|How to cite this article:|
Morais-Sarmento T, Figueiredo R, Garrido J, Rebelo AL, Berens O, Candeias A. Don't let me down: New intraocular lens formula effectiveness in extreme-length eyes in day-to-day practice. Pan Am J Ophthalmol 2019;1:3
|How to cite this URL:|
Morais-Sarmento T, Figueiredo R, Garrido J, Rebelo AL, Berens O, Candeias A. Don't let me down: New intraocular lens formula effectiveness in extreme-length eyes in day-to-day practice. Pan Am J Ophthalmol [serial online] 2019 [cited 2021 Jan 21];1:3. Available from: https://www.thepajo.org/text.asp?2019/1/1/3/264046
| Introduction|| |
Today, patients not only expect improvement in best-corrected visual acuity solely, but also the most considerable improvement in uncorrected visual acuity as the outcomes of routine and straightforward cataract surgery. Recently, the formulae used to calculate intraocular lenses (IOLs) have come a long way through the crossing from the 3rd generation (Hoffer Q, Holladay I, and SRK/T) to the 4th generation (Barrett II, Haigis, Holladay II, and Olsen) formulae. The significant difference in the latter consists in considering more than the two traditional measurement parameters, from the axial length (AL) and keratometry to including anterior-chamber depth, lens thickness, and horizontal white to white.
This study intends to make a retrospective analysis of five formulae (Barrett II, Haigis, Hoffer Q, Holladay I, and SRK/T) refractive outcomes in our department's phacoemulsification procedures during 1 year in extreme AL eyes.
In this study, using the IOL Master 500© IOL power calculator, it was possible to study and compare the performance of five different formulae when using two different IOLs, with the final goal of obtaining real-life results from day-to-day practice, which are usually different from the results of theoretical studies.
| Methods|| |
This study abided to the ethics code based on the tenets of the Declaration of Helsinki and was approved by Hospital Espírito Santo de Évora's Ethics Committee. The acquisition of measurements was carried out using IOL Master 500 software (v7.3.0.0048, Carl Zeiss Meditec, Jena, Germany), which is a partial coherence interferometry-based biometer. According to Cooke and Cooke, in extreme short eyes (AL ≤22 mm), the best formulae should be Barrett II and Holladay I, whereas in large eyes (AL ≥25 mm), the best formulae should be Olsen and Haigis. These results contradicted the previous results of the superiority of Hoffer Q in short eyes and of Holladay I in long eyes.
Our group evaluated the electronic medical records of all patients who were submitted to uneventful phacoemulsification cataract surgery from January 1, to December 31, 2018. We selected all eyes with AL ≤22 mm or ≥25 mm (as considered by the IOL Master 500© software, Carl Zeiss Meditec, Jena, Germany, and by Cooke and Cooke). At our department, the patients were implanted with AKREOS AO MI60© IOL (Bausch and Lomb©, Laval, Quebec, Canada) or with AMO Tecnis 1 ZCB00© IOL (Johnson and Johnson©, New Brunswick, New Jersey, United States of America).
All eyes with ocular pathology, which abandoned follow-up, which did not get subjective manifest refraction before 90 days postsurgery, or which were the second eyes (as suggested by the Protocols for Studies of Intraocular Lens Formula Accuracy) were excluded from this study. An overview of our selection and validating criteria is depicted in [Figure 1].
The postoperative assessment comprised subjective manifest refraction acquired between 30 days and 90 days postoperatively.
The Haigis, Hoffer Q, Holladay I, and SRK/T were available in IOL Master 500© IOL power calculation software (v7.3.0.0048). The Wang–Koch (WK) AL adjustment published in 2011 was applied to the Haigis and Hoffer Q formulae, whereas the WK adjustment published in 2018 was applied to the Holladay I and SRK/T formulae. The Barrett II Universal formula is freely available on the Asian-Pacific Association of Cataract and Refractive Surgeons (https://www.apacrs.org/barrett_universal2/).
Regrettably, the Holladay II and Olsen formulae are not freely accessible and, as such, were not available for comparison in this study.
Preoperative measurements (biometry measurements)
All eyes had three parameters measured before surgery, namely AL, keratometry, and anterior-chamber depth. While the Hoffer Q, Holladay I, and SRK/T only consider AL and keratometry, the Barrett II Universal considers AL, keratometry, and anterior-chamber depth, with optional consideration of horizontal white to white and lens thickness. The Haigis formula considers AL and anterior-chamber depth.
The prediction error (PE) is defined as the difference between the subjective postsurgical spherical equivalent (SE) and the SE predicted by each formula using the IOL power implanted. As such, a positive PE points to a more hyperopic refractive outcome, and a negative PE points to a more myopic refractive outcome.
Formula constant for each intraocular lens
According to the Protocols for Studies of Intraocular Lens Formula Accuracy, whenever in the absence of enough available clinical datasets, the constants from the User Group for Laser Interference Biometry (ULIB) should be applied. As Barrett II Universal constant optimization was not possible and as Haigis a 1 and a 2 constant optimization was not possible due to sample size (n < 200), we considered the ULIB optimized constants for the two IOLs used, as described in [Figure 2].
|Figure 2: User Group for Laser Interference Biometry optimized constants for each intraocular lens modelB|
Click here to view
Computations and calculations
While PE was calculated using Microsoft Office Excel 2016©(Microsoft©, Redmond, Washington, United States of America), the mean error (ME), the mean absolute error (MAE), the median absolute error, the standard deviation, and the statistical significance were calculated with IBM© SPSS© 25, IBM©, Armonk, New York, United States of America). As our samples present a nonnormal distribution according to Shapiro–Wilk and Kolmogorov–Smirnov tests, statistical significance was calculated by Wilcoxon and Friedman tests, using the 4th generation formulae available as reference (Barrett II Universal and Haigis). The results were considered statistically significant whenever P < 0.05.
| Results|| |
From the 1528 eyes submitted to routine phacoemulsification procedures, the inclusion criteria selected 170 eyes, whereas the exclusion criteria removed 36 eyes. Thus, the entrance criteria validated 134 eyes in 114 patients, comprising 62 eyes with AL ≤22 mm (in 62 patients) and 52 eyes with AL ≥25 mm (in 52 patients), as shown in [Figure 2].
As shown in [Table 1], the short eyes subgroup presented a mean AL value of 21.633 (±0.087) mm and was implanted with AKREOS AO MI60 in 59.68% procedures and with AMO Tecnis 1 ZCB00 in 40.32% procedures. The long eyes subgroup presented a mean AL value of 26.732 (±0.526) mm and was implanted with AKREOS AO MI60 in 53.85% procedures and with AMO Tecnis 1 ZCB00 in 46.15% procedures.
|Table 1: Axial length descriptive statistics and intraocular lenses implanted in each group|
Click here to view
As shown in [Table 2], in eyes with AL ≤22 mm, the SRK/T formula registered the lowest MAE value and overall, the highest rates of lesser PE. On the other hand, the Haigis formula registered the highest MAE value and the Hoffer Q formula registered, overall, the lowest rates of lesser PE. However, these results did not achieve a statistically significant difference, as shown in [Table 3].
|Table 2: Formulae performance for short (AL=22 mm) and long eyes (AL=25mm)|
Click here to view
As shown in [Table 2], in eyes with AL ≥ 25 mm, the Barrett II Universal formula registered the lowest MAE value and the highest rates of lesser PE. On the other hand, the Hoffer Q formula registered the highest MAE value and the lowest rates of lesser PE. The Barrett II Universal Formula results achieved a statistically significant difference in the reduction of MAE and the reduction of the rates of lesser PE. Regarding Haigis, the results showed statistically significant differences in MAE value when compared to all formulae, but only showed statistically significant improvements in rates when compared to the Hoffer Q formula, as shown in [Table 3].
Analyzing the results of the AL WK adjustment in long eyes (AL ≥ 25 mm), the results showed lower MAE values and higher rates of lesser PE in the Hoffer Q and Holladay I formulae, whereas the results showed higher MAE values and smaller rates of lesser PE in the Haigis and SRK/T formulae [Table 4].
|Table 4: Formulae performance for long eyes (AL=25 mm) with Wang-Koch adjustment|
Click here to view
| Discussion|| |
Regarding the short-eyes subgroup (AL ≤22 mm), all the formulae showed a hyperopic ME except for Haigis. Furthermore, when considering the P values in ME [Table 3], the Haigis always shows a statistically significant difference (which can be explained by the myopic shift), whereas the Barrett II only shows a statistically significant difference against Hoffer Q. However, when considering the P values of MAE, there were no statistically significant differences. When assessing the rates of patients with lower PE values [Table 2] and [Table 3], there was a trend for higher percentages with the Holladay I formula (until ±0.75 D cohort) and with the Haigis and SRK/T (in +0.75 D and +1.00 D cohorts), while not statistically significant (P > 0.05). The Barrett II Universal formula seemed to perform worse than the Haigis, Holladay I, and SRK/T, which is a surprising result when comparing to Cooke and Cooke's results that described the Barrett II Universal formula as being superior to the Haigis and SRK/T [Table 5].
|Table 5: Comparison of lower prediction error (PE) frequencies with PE frequencies values from Cooke & Cooke|
Click here to view
Regarding the long-eyes subgroup (AL ≥25 mm), all formulae showed a hyperopic ME except for the Barrett II Universal, which showed the ME value most approximate to null (−0.337, ±0.125). Moreover, when considering the P values in ME [Table 3], the Barrett II Universal and Haigis formulae showed statistically significant differences from the 3rd-generation formulae (P < 0.001, Wilcoxon test). However, when considering the P values of MAE, the Barrett II Universal and Haigis formulae showed statistically significant differences from every other formula. Regarding the percentage of patients with lower PE [Table 3], there was a statistically significant advantage of the Barrett II Universal overall formulae (P < 0.01, Friedman test) and of the Haigis formula over Hoffer Q (P < 0.05, Friedman test).
When considering the role of WK AL adjustment in the long-eyes subgroup (AL ≥25 mm), there seems to be an advantage to apply it when using the Hoffer Q or Holladay formulae and a disadvantage when using the Haigis or SRK/T formula. However, it should be duly noted that this adjustment is not readily available on the biometer measurement software, forcing an external calculation with posterior manual re-entry of the value into the biometer calculation software. Bearing in mind the day-to-day practice reality, this adjustment seems to bring little improvement for the effort demanded and, as so, does not seem worthy of routine implementation in nonspecialized centers.
Our results are different from the expected values based on the results published by Cooke and Cooke, as shown in [Table 5]. Furthermore, the results are different, whether we analyze MAE values or analyze the percentage of patients with lower PE values.
First, our group was not able, as most secondary care centers (nonspecialized in refractive surgery), to optimize the formula constants to our sample, whereas the literature shows results based on sample-optimized constants. Second, our group implanted two IOL models (AKREOS AO MI60© and AMO Tecnis 1 ZCB00©), different from the mostly used IOL model, Acrysof SN60WF© (Alcon©), in the published literature. Thus, the limitations of this study might account for the central part of the divergence (the small sample size, the nonoptimized sample formula constants, and the IOLs used). Nonetheless, while these limitations exist and are acknowledged, these are the same limitations faced by clinicians in the day-to-day practice and need to be addressed not only when considering real-life results but also when translating them to the routine practice of any nonspecialized center.
| Conclusions|| |
The main take-home message from this study is the remarkable difficulty in calculating the adequate IOL power for eyes with extreme ALs, whether short or long. When comparing the old and new IOL calculating formulae, the results show some superiority of the Haigis and SRK/T formulae in short eyes and a clear advantage of the Barrett II Universal formula in large eyes. Despite these real-life results from day-to-day practice showing differences from the already-published data, these results confirm an improvement when switching from the 3rd-generation formulae to the 4th generation, just not as much as expected or wished for.
Other cited material
A-Barrett GD. Barrett Universal II Formula. Singapore, Asia-Pacific Association of Cataract and Refractive Surgeons. Available at: http://www. apacrs. org/barrett_universal2/.
B-User Group for Laser Interference Biometry (ULIB) optimized constants. Available at: http://ocusoft. de/ulib/c1. htm.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Hoffer KJ. The Hoffer Q formula: A comparison of theoretic and regression formulas. J Cataract Refract Surg 1993;19:700-12. Erratum in: J Cataract Refract Surg 1994;20:677 and 2007;33:2-3.
Holladay JT, Prager TC, Chandler TY, Musgrove KH, Lewis JW, Ruiz RS. A three-part system for refining intraocular lens power calculations. J Cataract Refract Surg 1988;14:17-24.
Retzlaff JA, Sanders DR, Kraff MC. Development of the SRK/T intraocular lens implant power calculation formula. J Cataract Refract Surg 1990;16:333-40.
Fang JP, Hill W, Wang L, Chang V, Koch DD. Advanced intraocular lens power calculations. In: Cataract and Refractive Surgery. Berlin, Heidelberg: Springer Berlin Heidelberg; 2006. p. 31-46.
Haigis W, Lege B, Miller N, Schneider B. Comparison of immersion ultrasound biometry and partial coherence interferometry for intraocular lens calculation according to Haigis. Graefes Arch Clin Exp Ophthalmol 2000;238:765-73.
Cooke DL, Cooke TL. Comparison of 9 intraocular lens power calculation formulas. J Cataract Refract Surg 2016;42:1157-64.
Aristodemou P, Knox Cartwright NE, Sparrow JM, Johnston RL. Formula choice: Hoffer Q, Holladay 1, or SRK/T and refractive outcomes in 8108 eyes after cataract surgery with biometry by partial coherence interferometry. J Cataract Refract Surg 2011;37:63-71.
Hoffer KJ, Aramberri J, Haigis W, Olsen T, Savini G, Shammas HJ, et al.
Protocols for studies of intraocular lens formula accuracy. Am J Ophthalmol 2015;160:403-50.
Wang L, Shirayama M, Ma XJ, Kohnen T, Koch DD. Optimizing intraocular lens power calculations in eyes with axial lengths above 25.0 mm. J Cataract Refract Surg 2011;37:2018-27.
Wang L, Koch DD. Modified axial length adjustment formulas in long eyes. J Cataract Refract Surg 2018;44:1396-7.
[Figure 1], [Figure 2]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5]