Differences in Axial Length and IOLs Power Basing on Alternative A-Scan or Fellow-Eye Biometry in Macular-Off Rhegmatogenous Retinal Detachment Eyes

We observe the potential refractive error basing on alternative A-scan ultrasound and fellow-eye biometry for phacovitrectomy in phakic macula-off rhegmatogenous retinal detachment (RRD) eyes, when IOLMaster fails to obtain data. Vitrectomy without lens extraction was performed for RRD repair. Preoperative axial length was measured using alternative A-scan ultrasound (AL-US). Postoperative axial length in eyes with silicone oil tamponade (AL-SO) and fellow-eye biometry (AL-FE) were obtained using IOLMaster. AL-US, AL-FE and AL-SO were 25.39 ± 2.14 mm, 25.85 ± 2.16 mm and 26.08 ± 2.53 mm, respectively. The Bland-Altman agreements among AL-US, AL-FE and AL-SO were well (95.5%, 21/22 of cases were in LoA). The mean IOLs power calculated using AL-US (Power-US) was 16.81 ± 7.19 D, using AL-FE (Power-FE) was 14.74 ± 6.95 D, and using AL-SO (Power-SO) was 13.54 ± 8.32 D. The difference between AL-US and AL-SO was signi�cant (P < 0.05), while that between AL-FE and AL-SO was not (P > 0.05). The difference between Power-US and Power-SO was signi�cant (P < 0.05), while that between Power-FE and Power-SO was not (P > 0.05). It showed that the alternative A-scan ultrasound leads to signi�cant differences in axial length and IOLs power prediction, while fellow-eye biometry provided similar measurements compared with those of silicone oil-lled eyes after RRD repair.


Introduction
Retinal detachment is the term used to describe the separation between the neurosensory retina and the underlying retinal pigment epithelium.Due to the postoperative complications of the crystalline lens, more than half of phakic patients who underwent vitrectomy for rhegmatogenous retinal detachment (RRD) repair need subsequent cataract surgery within 1 year 1 .Thus, a combined phacovitrectomy was encouraged for patients aged 50 or older for the lower cost of surgery, reduced complications and better visual outcome [2][3][4][5] .However, in some cases (10-17%) of RRD with macular involvement, the performance of the optic biometry (IOLMaster) in measuring the axial length (AL) became impossible because of the limitations of the machine, such as dense media opacity (corneal or lens opacity), poor xation by the patients, or no machine availability 6-8 .Rahman et al. suggested that the AL measured by acoustic biometry (A-scan ultrasound, US) could be an alternative for intraocular lens (IOL) power calculation 8 , and El-Khayat et al. reported that fellow-eye biometry is also recommended, although there has been no published report on outcomes 9 .On the other hand, studies including our previous work showed that both underestimation of the AL and anterior displacement of IOL are potential mechanisms of postoperative myopic shift after phacovitrectomy 10,11 .However, to our knowledge, most researches focusing on myopic shift after phacovitrectomy excluded cases in which the preoperative IOLMaster measurement is absent.Therefore, we conducted the present study to evaluate the agreement of the AL and the predicted IOL power in macula-off RRD eyes among preoperative US, postoperative IOLMaster and fellow-eyes.And our purpose is to evaluate the potential refractive errors after phacovitrectomy according to alternative US or contralateral measurements in eyes which were lack of preoperative IOLMaster data.respectively.Independent samples t-test revealed that there was no signi cant difference in AL-US, AL-FE or AL-SO between genders (all P > 0.05).The mean IOL power calculated using AL-US (Power-US) was 16.81 ± 7.19 diopter (D), the mean IOL power calculated using AL-FE (Power-FE) was 14.74 ± 6.95 D, and the mean IOL power calculated using AL-SO (Power-SO) was 13.54 ± 8.32 D, respectively.The Bland-Altman plots demonstrated that the agreement between AL-US and AL-SO is well since 95.5% (21/22) of samples were included in the 95% limits of agreement (LoA), and the agreement between AL-FE and AL-SO (95.5%, 21/22 in LoA) was similar, as shown in Fig. 1A and B.
However, Repeated Measures ANOVA showed that the difference between AL-US and AL-SO was statistically signi cant (P < 0.05), while that between AL-FE and AL-SO was not (P > 0.05, Fig. 1C).The mean absolute error between AL-US and AL-SO (MAE AL−US−SO ) was 0.69 ± 0.78 mm, and MAE AL−FE−SO was 0.61 ± 0.79 mm.Correspondingly, the difference between Power-US and Power-SO was statistically signi cant (P < 0.05), while that between Power-FE and Power-SO was not (P > 0.05, Fig. 1D).

Discussion
The AL of an eye, which would be measured as the distance from the cornea to the inner limiting membrane using A-scan US 12 , or from the cornea to retinal pigment epithelium using IOLMaster 12 , is the most crucial parameter in the IOL power calculation.A 1-mm change in the AL corresponds to around 2.7-D refractive error in the IOL power 13 .Nepp et al. reported that the mean difference between the US-and IOLMaster-measured AL in SO-lled eyes was 0.4 mm, and only 46% of cases had a value less than 0.3 mm while in 26% the value was greater than 1 mm 14 .Thus, we measured the AL of these SO-lled eyes and normal fellow-eyes in the present study by means of IOLMaster for more accuracy and less deviation 15 .Our results showed that the AL of SO-lled eyes after RRD repair is comparable to that of fellow-eyes, so it was not surprising that IOL power calculated using AL-SO is similarly comparable to that using AL-FE.Nevertheless, there were signi cant differences in both AL and IOL power between US-measured preoperative data of macula-off RRD eyes and IOLMaster-measured postoperative data of SO-lled eyes.The mean AL-US was about 0.68-mm shorter than AL-SO, and the mean Power-US was accordingly 3.27-D greater than Power-SO.Namely, there would be a clinically signi cant myopic shift after SO removal if we performed phacovitrectomy for RRD repair according to the alternative A-scan-based IOL power calculation.
Phacovitrectomy is a safe and effective procedure to treat RRD.Kang et al. 16 and Sakamoto et al. 17 con rmed that there is a tolerable biometric error after the treatment of RRD with phacovitrectomy, which is comparable between acoustic and optic biometry in macula-sparing cases.Although the anatomical and functional results were comparable with those obtained with PPV and delayed cataract surgery, however, the refractive outcomes were less favorable and shifted toward myopia, especially in macula-off cases 18 .When the macula is involved, there would be a mean prediction error of -1.22 ± 2.32-D for the RRD eye and a mean prediction error of -0.01 ± 1.09-D for fellow-eye 9 .Pongsachareonnont et al.
attributed it to the underestimation of 0.59 ± 0.90-mm in AL measurement by IOLMaster 19 .Rahman et al. found that in less than a quarter (13/54, 24.1%) cases IOLMaster could provide available AL measurement 6 .Kim et al. found that the underestimation is associated with macular retinal detachment height 7 .Unlike these above studies, there was no preoperative IOLMaster data as reference, so we could not calculate a user-adjusted AL by combining acoustic and optic biometry as described 8 .Patients with worse vision, greater central macular thickness, and shallow anterior chambers require more caution since they are prone to inaccurate preoperative biometry 20 .In this situation, a delayed cataract surgery or a phacovitrectomy using contralateral AL might be more recommended.
Our results of this study should be interpreted with its limitations in mind.All the participants we recruited underwent SO tamponade, which is the current routine procedure for RRD cases 21 .However, the interference in biometry was not entirely quanti ed, such as under ll/over ll, or the emulsion of SO.Some characteristics of the retinal detachment which might be associated with the accuracy of AL measurement, for example, macular detachment height, were neither controlled.Besides, we have not been obtained the data of nal refractive error from all cases after SO removal, phacoemulsi cation and IOL implantation due to the various degrees of complicated cataract, even though that should be the nal answer.Summarily, according to the present study, if the performance of IOLMaster is failed in eyes with macula-off RRD, ultrasound might lead to a refractive prediction error in IOL power.Phacovitrectomy using fellow-eye biometry as well as secondary IOLs implantation should be the better choice in this situation, in the absence of an anisometropia.

Methods
Design This retrospective, self-control study was conducted in Eye and Ear, Nose and Throat Hospital, Fudan University between October 2017 and June 2019.All study procedures were performed in accordance with the tenets of the Declaration of Helsinki and its amendment.The Ethics Committee of Fudan University approved this study.Written informed consent was obtained from all the participants for publication of their clinical data.
Participants Phakic, macula-off RRD eyes in which the ocular biometric measurements could not be obtained using IOLMaster were included, and then underwent 23-gauge pars plana vitrectomy (PPV) with silicone oil (SO) tamponade but not lens extraction.History of anisometropia, and eyes with scleral buckling surgery, recurrent retinal detachment or other ocular problems, which may affect the biometric measurements, such as corneal scar and lens dislocation, were excluded 15 .
Surgical procedure A standard 23-gauge PPV with the CONSTELLATION® Vision 106 System (Alcon Laboratories, Inc.) was performed in each case.During PPV, the RESIGHT™ Fundus Viewing System (Carl Zeiss Meditec Inc.) was used.Core vitrectomy, mid-peripheral vitrectomy, and vitreous base shaving under scleral depression were performed to remove the vitreous.Per uorocarbon liquid (Per uoron; Alcon Laboratories, Inc.) might be used in some cases depending on the retinal detachment extent.Endolaser photocoagulation was performed around the area of retinal breaks, and uid-air exchange was performed before silicone oil injection (Oxane 5700 centistokes; Bausch & Lomb Inc., Waterford, Ireland).
Measurement techniques Preoperative AL of RRD eyes was measured using A-scan ultrasound biometry (UD-6000 Ultrasonic A/B scanner biometer; Tomey Corporation, Nagoya, Japan) as formerly described 22 instead of IOLMaster.Before SO removal, an optical coherence tomography scan was performed to ensure the macula-on condition in both eyes, and then the AL of postoperative RRD eyes and fellow-eyes were obtained with IOLMaster (Carl Zeiss Meditec Ltd, Jena, Germany) using phakic and SO-lled phakic eyes program, respectively.Calculation of IOLs power was based on the A constant of AcrySof® IQ ReSTOR® SN6AD1 IOL (Alcon Laboratories, Inc., Fort Worth, USA) with SRK/T formulas.
Statistical analysis variables were expressed as the mean ± standard deviation.Statistical analyses were performed using the software package, SPSS Statistics 26.0 for Windows (SPSS Inc., Chicago, IL, USA).Bland-Altman plots were drawn using the MedCalc 15.2.2 (MedCalc Software bvba, Ostend, Belgium).A statistical signi cance was set at P < 0.05.

Declarations Figures
The mean absolute error between Power-US and Power-SO (MAE Power−US−SO ) was 3.27 ± 3.79 D, and MAE Power−FE−SO was 2.36 ± 3.38 D.

Figure 1 The
Figure 1