Corresponding Author: Chang Ki Yoon, MD, PhD. Department of Ophthalmology, Seoul National University Hospital, Seoul National University College of Medicine, 101 Daehak-ro, Jongno-gu, Seoul 03080, Korea. Retinitis pigmentosa (RP) is one of the most common inherited retinal degenerations and is characterized by progressive rod-cone dystrophy, retinal vessel attenuation, and bone-spicule pigmentary changes, typically beginning in the mid-peripheral retina and progressing toward the macula and fovea. The global prevalence of RP has been estimated to be approximately 1 in 4,000 individuals, affecting more than 1 million people worldwide. Typical symptoms include night blindness followed by progressive peripheral visual field loss, leading to tunnel vision and, in advanced cases, severe visual impairment or legal blindness [1,2].
Cataract formation, particularly posterior subcapsular cataract (PSC), is a well-recognized anterior segment complication of RP. Although PSC is relatively uncommon in the general population, previous studies have reported a substantially higher prevalence of PSC in patients with RP, often occurring at a younger age than age-related cataracts [3,4]. Because patients with RP depend heavily on central vision after progressive peripheral field constriction, even mild PSC involving the visual axis can markedly aggravate visual dysfunction and prompt earlier cataract surgery [5]. Recent studies and systematic reviews have further supported the clinical importance of cataract surgery in RP, while emphasizing that postoperative visual outcomes may be influenced by both retinal status and anterior segment abnormalities [6].
In addition to early cataract formation, RP has been associated with capsular and zonular vulnerability. Zonular insufficiency may be encountered intraoperatively or postoperatively, and RP eyes are prone to anterior capsular contraction, capsulorhexis phimosis, and late in-the-bag intraocular lens (IOL) subluxation or dislocation [7-10]. These abnormalities suggest that instability of the capsular-zonular complex may contribute to postoperative IOL tilt, decentration, and refractive unpredictability. However, most previous reports have focused on clinical observations, surgical outcomes, or case reports of advanced IOL dislocation, and quantitative assessment of subtle postoperative IOL positional instability in RP eyes remains limited.
Anterior-segment optical coherence tomography (AS-OCT) enables non-contact, high-resolution, and reproducible evaluation of anterior segment structures and postoperative IOL position. Previous studies have demonstrated that AS-OCT and three-dimensional reconstruction methods can be used to quantify IOL tilt and decentration objectively [11-13]. Nevertheless, few studies have applied AS-OCT to evaluate postoperative IOL positional stability in patients with RP. Moreover, direct comparative analyses of IOL tilt, IOL decentration, and refractive prediction accuracy, including prediction error (PE) and absolute prediction error (APE), between RP and normal control eyes remain scarce.
Therefore, this study aimed to compare postoperative IOL stability, refractive outcomes, and visual outcomes between RP and normal control eyes after uncomplicated cataract surgery with in-the-bag IOL implantation. Specifically, we evaluated IOL tilt and decentration using AS-OCT, compared PE and APE between groups, and investigated whether RP status was independently associated with IOL instability and refractive PE after adjustment for ocular biometric factors.
Materials and Methods
Ethics statement
This study adhered to the tenets of the Declaration of Helsinki and was approved by the Institutional Review Board of Seoul National University Hospital (No. 2511-050-1690). The requirement for informed consent was waived due to the use of deidentified data and the retrospective nature of the study.
Study design and participants
This retrospective, single-center study included 93 patients, corresponding to 138 eyes, who underwent uncomplicated phacoemulsification with in-the-bag posterior chamber IOL implantation and postoperative AS-OCT. The RP group comprised 36 patients, corresponding to 55 eyes, who underwent cataract surgery between September 2016 and November 2024 and were examined using postoperative AS-OCT between October 2023 and March 2025. The normal control group included 57 patients, corresponding to 83 eyes, who underwent cataract surgery between February 2023 and August 2023 and were examined using postoperative AS-OCT between March 2023 and September 2023. Control eyes were selected from patients who had undergone uncomplicated cataract surgery and routinely received postoperative AS-OCT imaging during a period when this imaging protocol was consistently performed in the clinic.
The inclusion criteria were phacoemulsification with inthe-bag IOL implantation and availability of postoperative AS-OCT imaging suitable for quantitative IOL position analysis. The exclusion criteria were a history of ocular trauma, previous intraocular surgery other than cataract surgery in the study eye, intraoperative complications such as posterior capsular rupture or zonular dialysis requiring non-standard IOL fixation, pseudoexfoliation syndrome with marked phacodonesis requiring capsular tension ring insertion or scleral fixation, corneal opacity or poor image quality precluding reliable AS-OCT analysis, and retinal pathology unrelated to RP that could affect visual outcomes. All included eyes, including those with documented zonular weakness, underwent standard in-the-bag IOL implantation without major intraoperative complications or the need for non-standard IOL fixation.
Outcome measures and imaging
Demographic characteristics, systemic comorbidities, operative details, and postoperative outcomes were obtained from electronic medical records. IOL models were identified from operative records and classified according to haptic design. For statistical analysis, IOL design was categorized as single-piece/open-loop, three-piece, or plate-haptic/four-haptic. TECNIS ZCB00 and TECNIS Eyhance ICB00 (Johnson & Johnson Surgical Vision Inc.), and AcrySof SA60AT and AcrySof IQ SN60WF (Alcon Laboratories Inc.) were classified as single-piece/open-loop IOLs. Sensar AR40e (Johnson & Johnson Surgical Vision Inc.) was classified as a three-piece IOL. ARTIS PL E (Cristalens Industrie) and Akreos AO (Bausch & Lomb) were classified as plate-haptic/four-haptic IOLs. Zonular weakness was assessed based on intraoperative surgical findings documented in the operative records, including zonular laxity, phacodonesis, or capsular bag instability noted during cataract surgery. Ocular biometric parameters, including axial length, anterior chamber depth (ACD), and lens thickness (LT), were measured using the IOL-Master 700 (Carl Zeiss Meditec). Corneal astigmatism was assessed using corneal topography or tomography devices, including Galilei G4 (Ziemer), Orbscan II (Bausch & Lomb), and Atlas 500 (Carl Zeiss Meditec), according to clinical availability.
IOL tilt and decentration were automatically calculated by CASIA2 software ver.50.6A.02 (Tomey Corp.) using three-dimensional analysis of eight meridional AS-OCT images referenced to the corneal topographic axis. A single examiner reviewed all scans for image quality and segmentation adequacy. Interobserver and intraobserver reproducibility were not assessed. Fig. 1A and 1B illustrate representative CASIA2 measurements of IOL tilt and decentration relative to the corneal topographic axis. The postoperative interval was defined as the number of days between cataract surgery and postoperative AS-OCT imaging.
Visual acuity (VA) was converted to the logMAR for statistical analysis. VA improvement was calculated as preoperative logMAR VA minus postoperative logMAR VA, with larger positive values indicating greater postoperative improvement.
Postoperative VA and manifest refraction were assessed at the same visit as AS-OCT imaging. Refractive outcomes included PE, APE, and postoperative residual refractive cylinder magnitude. PE was calculated as the difference between postoperative spherical equivalent and the predicted refractive target, and APE was defined as the absolute value of PE. The predicted refractive target for each eye was calculated using the SRK/T formula with manufacturer-recommended lens constants. The same formula was applied consistently to both RP and normal control eyes for the calculation of PE and APE. Postoperative residual refractive cylinder magnitude was defined as the absolute value of postoperative refractive cylinder.
Goldmann visual field quantification
In exploratory analyses limited to eyes with RP, visual field severity was quantified using Goldmann perimetry. Scanned Goldmann visual field charts were analyzed using Fiji distribution of ImageJ software ver. 1.54p (National Institutes of Health). After calibration using the degree scale of the Goldmann chart, the area enclosed by each isopter was manually traced and measured in square degrees. When internal scotomas were present within the isopter, these areas were separately delineated and subtracted from the gross isopter area to obtain the net seeing visual field area. When multiple residual visual field islands were present, the areas of all seeing regions were summed [14,15].
Because Goldmann isopter area depends on stimulus size, areas obtained using different Goldmann targets were not pooled. The III4e Goldmann isopter area was used as the primary visual field severity measure, and the II4e Goldmann isopter area was analyzed separately as a sensitivity measure [16]. Because isopter area showed a skewed distribution, the natural logarithm of the isopter area was used in regression models.
Statistical analysis
Statistical analyses were performed using R software ver. 4.4.1 (R Foundation for Statistical Computing). Continuous variables were compared using Welch t-test, and categorical variables were compared using the chi-square test or Fisher exact test, as appropriate. A p-value < 0.05 was considered statistically significant.
For the primary analyses of IOL positional outcomes, multivariable linear regression analyses were performed with IOL tilt and IOL decentration as dependent variables and RP status as the main independent variable. The models were adjusted for sex, age, axial length, LT, ACD, and preoperative astigmatism.
For postoperative VA analyses, multivariable linear regression analyses were performed with postoperative log-MAR VA as the dependent variable. The models included preoperative logMAR VA, IOL tilt, IOL decentration, and RP status. As a planned secondary analysis, interaction terms between RP status and each IOL positional parameter were included to evaluate whether the associations of IOL tilt and decentration with postoperative VA differed between the RP and normal control groups. IOL tilt and decentration were mean-centered before constructing the interaction terms, and IOL decentration was scaled per 0.1 mm for clinical interpretability. Model A was adjusted for age, axial length, LT, and ACD. Model B additionally included preoperative astigmatism.
For refractive outcome analyses, separate multivariable linear regression models were performed with PE, APE, and postoperative residual cylinder magnitude as dependent variables. These models included RP status, IOL tilt, IOL decentration, sex, age, axial length, LT, ACD, and preoperative astigmatism.
Sensitivity analyses included patient-clustered generalized estimating equation analyses with robust standard errors, separate models additionally adjusted for IOL design or zonular weakness, and analyses restricted to eyes implanted with single-piece/open-loop IOLs. In exploratory analyses limited to RP eyes, postoperative interval was evaluated in relation to IOL tilt and decentration. Visual field severity was assessed using log-transformed III4e Goldmann isopter area, with log-transformed II4e isopter area analyzed separately as a sensitivity measure. Complete-case analysis was used for multivariable models.
Results
A total of 138 eyes of 93 patients were included in this study, comprising 55 eyes of 36 patients with RP and 83 eyes of 57 normal control patients. The analyses were performed on an eye basis. The RP group was significantly younger than the normal control group (61.02 ± 9.39 years vs. 72.06 ± 7.51 years, p < 0.001). Sex distribution (male, 21/55 [38.2%] in the RP group vs. 26/83 [31.3%] in the normal control group; p = 0.465) and laterality (right eye, 26/55 [47.3%] in the RP group vs. 41/83 [49.4%] in the normal control group; p = 0.863) did not differ significantly between the groups. Diabetes mellitus was less frequent in the RP group than in the normal control group (14.5% vs. 30.1%, p = 0.042), whereas hypertension and cerebrovascular disease showed no significant between-group differences.
Regarding ocular biometric parameters, axial length was comparable between the RP and normal control groups (23.66 ± 1.25 mm vs. 23.80 ± 1.21 mm, p = 0.519). However, RP eyes had significantly shallower ACD (2.87 ± 0.56 mm vs. 3.07 ± 0.48 mm, p = 0.028) and greater LT (4.74 ± 0.41 mm vs. 4.49 ± 0.43 mm, p = 0.001). Zonular weakness was observed in 17 RP eyes (30.9%), whereas no normal control eyes showed zonular weakness (p < 0.001). No zonular dialysis, including mild focal zonular dialysis, was documented in any included eye.
Preoperative and postoperative VAs were both worse in the RP group than in the normal control group. Preoperative VA was 1.11 ± 0.76 logMAR in the RP group and 0.59 ± 0.54 logMAR in the normal control group (p < 0.001). Postoperative VA was also worse in the RP group (0.69 ± 0.86 logMAR vs. −0.04 ± 0.11 logMAR, p < 0.001). Although both groups showed postoperative visual improvement, the magnitude of improvement was smaller in the RP group than in the normal control group (0.42 ± 0.62 logMAR vs. 0.62 ± 0.53 logMAR, p = 0.046).
The RP group showed significantly worse refractive outcomes and greater IOL positional instability. PE was more myopic in the RP group than in the normal control group (−0.25 ± 0.65 D vs. 0.08 ± 0.47 D, p = 0.002), and APE was greater in the RP group (0.53 ± 0.45 D vs. 0.36 ± 0.31 D, p = 0.018). Postoperative residual cylinder magnitude was also greater in the RP group than in the normal control group (1.11 ± 0.64 D vs. 0.81 ± 0.53 D, p = 0.005) (Table 1). IOL tilt was significantly greater in the RP group than in the normal control group (5.90° ± 2.52° vs. 4.87° ± 1.38°, p = 0.007), as was IOL decentration (0.29 ± 0.19 mm vs. 0.20 ± 0.11 mm, p = 0.002) (Table 1 and Fig. 2A-2D). The distribution of IOL design differed significantly between the groups. All normal control eyes were implanted with single-piece/open-loop IOLs, whereas the RP group included 28 eyes with single-piece/open-loop IOLs, 17 eyes with three-piece IOLs, and 10 eyes with plate-haptic/four-haptic IOLs (p < 0.001). The interval from cataract surgery to AS-OCT imaging was substantially longer in the RP group than in the normal control group (1,107.73 ± 618.84 days vs. 35.58 ± 25.45 days, p < 0.001) (Table 1). In exploratory analyses limited to RP eyes, postoperative interval was not significantly associated with IOL tilt or IOL decentration. However, because postoperative interval was strongly associated with group status, this analysis could not exclude a time-dependent contribution to the observed between-group differences in IOL position.
Fig. 3A and 3B show polar coordinate plots of IOL tilt and decentration according to laterality. These plots demonstrated a broader distribution of IOL positional parameters in RP eyes, supporting greater postoperative IOL positional variability in the RP group.
Multivariable linear regression models for IOL tilt and decentration
In multivariable linear regression analysis adjusted for sex, age, axial length, LT, ACD and preoperative astigmatism, RP status was independently associated with greater IOL tilt and decentration. RP eyes showed a 1.040° greater IOL tilt than normal control eyes after adjustment (95% confidence interval [CI], 0.216-1.864; p = 0.014). RP status was also independently associated with greater IOL decentration, with an adjusted β of 0.095 mm (95% CI, 0.029-0.161; p = 0.005).
Among ocular biometric parameters, shallower ACD was associated with both greater IOL tilt and greater IOL decentration. For each 1-mm increase in ACD, IOL tilt decreased by 1.198° (95% CI, −2.196 to −0.200; p = 0.019), and IOL decentration decreased by 0.123 mm (95% CI, −0.202 to −0.043; p = 0.003). In the IOL decentration model, older age, longer axial length, and greater preoperative astigmatism were also significantly associated with greater IOL decentration (Table 2).
Multivariable linear regression models for refractive outcomes
In multivariable analyses for refractive outcomes, RP status was independently associated with a more myopic PE after adjustment for IOL tilt, IOL decentration, sex, age, axial length, LT, ACD, and preoperative astigmatism. The adjusted β for RP status was −0.463 D (95% CI, −0.731 to −0.195; p < 0.001). RP status was also independently associated with greater APE, with an adjusted β of 0.294 D (95% CI, 0.115-0.472; p = 0.001).
IOL tilt and IOL decentration were not independently associated with PE or APE in these models. Greater preoperative astigmatism was associated with greater APE (β = 0.106; 95% CI, 0.001-0.211; p = 0.048), whereas LT showed a borderline association with PE (β = 0.335; 95% CI, 0.000-0.670; p = 0.050).
Although RP eyes showed greater postoperative residual cylinder magnitude in the unadjusted comparison, RP status was not significantly associated with postoperative residual cylinder magnitude after multivariable adjustment (β = 0.138 D; 95% CI, −0.098 to 0.374; p = 0.248). In this model, preoperative astigmatism was the only factor significantly associated with postoperative residual cylinder magnitude (β = 0.462; 95% CI, 0.324-0.601; p < 0.001) (Table 3).
Multivariable linear regression models for postoperative VA
In multivariable linear regression models for postoperative logMAR VA, preoperative logMAR VA was the strongest factor associated with postoperative VA in both models. In model A, preoperative logMAR VA was significantly associated with postoperative logMAR VA (β = 0.426; 95% CI, 0.295-0.558; p < 0.001). RP status was also independently associated with worse postoperative log-MAR VA at the mean levels of IOL tilt and decentration (β = 0.530; 95% CI, 0.319-0.742; p < 0.001).
The interaction between RP status and IOL decentration showed a borderline association with postoperative log-MAR VA in model A (β = 0.105/0.1 mm; 95% CI, −0.001 to 0.211; p = 0.053). This finding remained similar after additional adjustment for preoperative astigmatism in model B (β = 0.105; 95% CI, −0.002 to 0.211; p = 0.055). In contrast, the interaction between RP status and IOL tilt was not statistically significant in either model. Although the direction of the estimates suggested a potentially greater adverse association of IOL decentration with postoperative VA in RP eyes, the interaction did not reach statistical significance (Table 4).
Exploratory analyses incorporating visual field severity
Among RP eyes with available Goldmann perimetry, exploratory analyses showed no significant association between log-transformed III4e Goldmann isopter area and postoperative VA, refractive outcomes, IOL tilt, or IOL decentration. Similar findings were observed when log-transformed II4e Goldmann isopter area was used as a sensitivity measure.
Sensitivity analyses
Patient-clustered generalized estimating equation analyses were performed to account for the inclusion of both eyes from some patients. These analyses showed results consistent with the primary analyses. RP status remained significantly associated with greater IOL tilt (β = 1.040; 95% CI, 0.165-1.914; p = 0.020) and greater IOL decentration (β = 0.095; 95% CI, 0.040-0.150; p < 0.001). RP status also remained significantly associated with PE (β = −0.463; 95% CI, −0.812 to −0.114; p = 0.009) and APE (β = 0.294; 95% CI, 0.076-0.511; p = 0.008), whereas the association with postoperative residual cylinder magnitude remained non-significant (β = 0.138; 95% CI, −0.089 to 0.366; p = 0.233) (Supplementary Table 1).
In IOL design-adjusted sensitivity analyses, the associations between RP status and IOL tilt or decentration were attenuated and no longer statistically significant. However, RP status remained significantly associated with more myopic PE and greater APE after additional adjustment for IOL design (Supplementary Table 2). In analyses restricted to eyes with single-piece/open-loop IOLs, RP status remained significantly associated with PE and APE, whereas its associations with IOL tilt and decentration were not statistically significant (Supplementary Table 2).
In zonular weakness-adjusted sensitivity analyses, RP status remained significantly associated with greater IOL tilt, greater IOL decentration, more myopic PE, and greater APE. In the same models, clinically documented zonular weakness was not statistically significant. Neither RP status nor zonular weakness was significantly associated with postoperative residual cylinder magnitude (Supplementary Table 3). In exploratory analyses limited to RP eyes, postoperative interval was not significantly associated with IOL tilt or IOL decentration.
Discussion
In this study, RP eyes demonstrated greater postoperative IOL positional instability and refractive unpredictability than normal control eyes after uncomplicated cataract surgery with in-the-bag IOL implantation. RP eyes had significantly greater IOL tilt and decentration, and RP status was associated with both parameters in the primary multivariable models. However, the marked difference in postoperative AS-OCT timing between groups limits attribution of these positional differences specifically to RP. RP eyes also showed a more myopic PE and greater APE than normal control eyes. Patient-clustered sensitivity analyses yielded consistent results.
IOL design-adjusted sensitivity analyses attenuated the associations between RP status and IOL tilt or decentration, which were no longer statistically significant after adjustment for IOL design. This finding suggests that postoperative IOL positional instability may be influenced by both RP-associated capsular-zonular vulnerability and IOL design, and that their independent effects could not be fully separated in this cohort. Thus, IOL design may partially account for the observed positional differences, which should not be interpreted as an effect of RP alone. In contrast, the associations of RP status with PE and APE remained significant after adjustment for IOL design.
RP eyes also showed greater postoperative residual cylinder magnitude in the unadjusted comparison. However, this association was attenuated after multivariable adjustment, and preoperative astigmatism was the only factor significantly associated with residual cylinder magnitude. Therefore, postoperative residual cylinder should be interpreted as a secondary refractive finding rather than a primary RP-associated outcome.
The direct association between IOL instability and postoperative VA was less evident. In the postoperative VA models, the interaction between RP status and IOL decentration showed only a borderline association with postoperative VA, whereas the interaction between RP status and IOL tilt was not significant. Therefore, the visual impact of IOL instability in RP should be interpreted cautiously, whereas the association between RP and refractive unpredictability was more clearly observed.
Anterior segment biometric features in RP
RP eyes underwent cataract surgery at a younger age and showed greater LT and shallower ACD than normal control eyes. These findings may reflect disease-specific anterior segment characteristics in RP rather than age-related cataract alone, consistent with a recent biometric study reporting thicker lenses and shallower anterior chambers in cataract eyes with RP [17]. Increased LT and reduced ACD may be related to cataract maturation, lens hydration, and instability of the lens-capsule complex caused by zonular weakness. These changes may affect surgical difficulty, capsular bag behavior, and effective lens position estimation.
IOL instability and capsular-zonular vulnerability
RP eyes showed significantly greater IOL tilt and decentration than normal control eyes. This finding is consistent with previous reports describing zonular insufficiency, anterior capsular contraction, capsulorhexis phimosis, and late in-the-bag IOL subluxation or dislocation in RP [7-10,18-21]. The higher prevalence of zonular weakness in the RP group supports the concept that RP is associated with capsular-zonular vulnerability as well as retinal degeneration.
Zonular weakness-adjusted sensitivity analyses suggested that the associations between RP status and IOL positional or refractive outcomes were not fully explained by clinically documented zonular weakness alone. However, zonular weakness was observed only in RP eyes and may represent part of the RP-associated capsular-zonular vulnerability rather than an independent confounder; therefore, its independent effect should be interpreted cautiously.
The mean IOL decentration in the normal control group was approximately 0.2 mm, comparable to values reported in normal pseudophakic eyes [22,23]. AS-OCT and three-dimensional reconstruction methods allow quantitative assessment of IOL tilt and decentration [11-13]. Using this approach, the present study demonstrated that RP eyes may have greater postoperative IOL instability even after uncomplicated in-the-bag IOL implantation. Progressive zonular weakness, capsular contraction, and asymmetric capsular bag forces may contribute to IOL tilt, decentration, and, in severe cases, late in-the-bag IOL dislocation.
The substantially longer postoperative imaging interval in the RP group may also have contributed to the observed positional differences. Although postoperative interval was not significantly associated with IOL tilt or decentration in RP-only exploratory analyses, this finding does not exclude time-dependent positional changes because postoperative interval was strongly confounded with group status. Therefore, the between-group differences in IOL tilt and decentration should be interpreted cautiously.
IOL design may also have contributed to postoperative IOL position. Non-single-piece IOLs were used only in RP eyes, and previous studies have suggested that IOL haptic design and orientation can influence postoperative IOL tilt and decentration, although findings across different designs have been inconsistent [24-27]. In the present study, adjustment for IOL design attenuated the associations between RP status and IOL tilt or decentration. Therefore, the observed positional differences may reflect the combined influence of RP-associated capsular-zonular vulnerability and IOL design rather than an effect of RP alone.
Refractive implications and PE
RP status was clearly associated with refractive unpredictability. RP eyes showed a more myopic PE and greater APE than normal control eyes, and these associations remained significant after adjustment for IOL positional parameters, ocular biometric factors, and preoperative astigmatism. These findings suggest that refractive outcomes are less predictable in RP eyes. The associations also remained significant after adjustment for IOL design and in analyses restricted to single-piece/open-loop IOLs, suggesting that refractive unpredictability in RP eyes was not explained solely by differences in IOL design.
Importantly, IOL tilt and decentration were not independently associated with PE or APE. Thus, refractive unpredictability in RP eyes may reflect variability in effective lens position and broader capsular-biometric factors, including capsular bag behavior and anterior segment biometry, rather than the measured positional parameters alone. Because effective lens position is a major determinant of postoperative refraction, unstable capsular bag behavior may reduce the accuracy of IOL power prediction [28]. Recent evidence also suggests that capsular tension ring implantation may improve IOL power prediction accuracy in cataract eyes with RP, supporting the role of capsular bag stability in refractive predictability [29]. In this context, IOL instability may represent a clinical manifestation of underlying capsular-zonular vulnerability rather than an isolated determinant of refractive error.
Postoperative residual cylinder magnitude showed a different pattern from PE and APE. Although RP eyes had greater residual cylinder magnitude in the unadjusted comparison, RP status was not independently associated with residual cylinder after multivariable adjustment. Instead, residual cylinder was primarily associated with preoperative astigmatism, suggesting that it may be more strongly influenced by preexisting astigmatic status than by RP-associated IOL positional instability.
Visual implications of IOL instability
Although RP eyes had greater IOL tilt and decentration, the direct visual effect of IOL instability was not definitively established in this study. Postoperative VA in RP is strongly influenced by retinal factors, making the effect of subtle IOL misalignment difficult to isolate using postoperative logMAR VA alone [30,31]. The interaction analysis suggested a possible greater adverse association between IOL decentration and postoperative VA in RP eyes, although this finding did not reach conventional statistical significance. The interaction between RP status and IOL tilt was not significant, and high-contrast logMAR VA may not fully capture glare, contrast loss, or higher-order aberrations associated with IOL misalignment [32].
To further explore the role of RP severity, exploratory analyses were performed using Goldmann isopter area in RP eyes. Neither log-transformed III4e nor II4e Goldmann isopter area was significantly associated with postoperative VA, refractive outcomes, IOL tilt, or IOL decentration. These results suggest that visual field severity alone did not clearly explain the worse visual or refractive outcomes in RP eyes, although they should be interpreted cautiously because of the limited number of eyes with available perimetry.
Taken together, these findings suggest that surgeons should carefully assess zonular status and capsular bag stability in RP eyes, even when cataract surgery is uncomplicated. Postoperative monitoring for anterior capsular contraction and IOL positional change may also be important. Although capsular tension rings may help stabilize the capsular bag in selected cases, they do not completely eliminate the risk of capsular contraction or late IOL instability in RP [7,29,33]. Therefore, capsular support devices should be considered individually according to intraoperative zonular status.
From a refractive perspective, RP eyes may be more prone to myopic shift and larger APE, even after uncomplicated cataract surgery. Therefore, refractive outcomes should be interpreted cautiously when counseling patients with RP. Although residual cylinder was greater in RP eyes in unadjusted comparisons, this difference was mainly explained by preoperative astigmatism after adjustment, emphasizing the importance of careful preoperative assessment of corneal or refractive astigmatism. The poorer postoperative VA in RP eyes should be interpreted as a multifactorial outcome influenced by baseline visual function, retinal degeneration, refractive unpredictability, and, potentially, IOL decentration.
This study has several limitations. First, its retrospective design may have introduced selection bias and residual confounding. The sample size was determined by the availability of eligible patients with suitable postoperative AS-OCT imaging, which may have limited the power to detect smaller differences, subgroup effects, or interaction effects. In addition, the normal control group was not individually matched to the RP group. Although multivariable analyses adjusted for age and ocular biometric parameters, unmeasured baseline differences may remain.
Second, the postoperative imaging interval differed substantially between groups, with AS-OCT performed much later in the RP group than in the normal control group. This difference may have introduced systematic bias because IOL tilt and decentration may change over time. Although postoperative interval was not significantly associated with IOL tilt or decentration in RP-only exploratory analyses, time-dependent positional changes could not be fully disentangled from disease-related effects because postoperative interval was strongly confounded with group status. Therefore, the observed between-group differences in IOL position cannot be attributed specifically to RP alone. In addition, study-specific interobserver and intraobserver reproducibility of AS-OCT measurements was not assessed, although previous studies have reported good repeatability of CASIA2 measurements of IOL tilt and decentration [34].
Third, some surgical and IOL-related factors that may influence postoperative IOL stability were not uniformly available, including IOL material, capsulorhexis size, detailed intraoperative capsular or zonular findings, and surgeon-related differences in surgical technique. In addition, IOL design was not balanced between groups, and nonsingle-piece IOLs were used only in RP eyes. Although IOL design-adjusted and single-piece/open-loop restricted sensitivity analyses were performed, the independent effects of RP status and IOL design could not be fully separated.
Fourth, analyses were performed on an eye basis, and some patients contributed both eyes. Although patient-clustered sensitivity analyses showed results consistent with the primary analyses, residual effects of within-patient correlation cannot be completely excluded. Zonular weakness-adjusted sensitivity analyses were also performed; however, zonular weakness was identified based on operative records and was observed only in RP eyes. Therefore, the independent effects of RP status and clinically documented zonular weakness could not be fully separated.
Finally, detailed retinal structural status, including ellipsoid zone integrity, macular thickness, cystoid macular edema, and foveal involvement, was not incorporated into the primary models. Although Goldmann isopter area was evaluated in exploratory analyses limited to a subset of RP eyes, the limited availability of these data precluded its inclusion in the primary between-group models. Therefore, residual confounding by retinal disease severity may have limited our ability to isolate the visual impact of IOL instability.
Future prospective studies with standardized longitudinal AS-OCT imaging, balanced IOL design selection, detailed surgical information, retinal structural assessment, and standardized visual field testing are needed to clarify the progression of IOL positional changes and improve refractive prediction strategies in RP eyes.
In conclusion, eyes with RP showed greater postoperative IOL tilt and decentration and less predictable refractive outcomes than normal control eyes after uncomplicated cataract surgery with in-the-bag IOL implantation. However, these positional differences should not be attributed to RP alone because postoperative AS-OCT timing differed markedly between groups, non-single-piece IOLs were used only in RP eyes, and the associations with IOL tilt and decentration were no longer statistically significant after adjustment for IOL design. In contrast, RP status remained associated with a more myopic PE and greater APE, suggesting persistent refractive unpredictability in RP eyes.
The visual impact of IOL instability was suggestive but not definitive. Although IOL decentration showed a borderline tendency toward a greater adverse association with postoperative VA in RP eyes, this association did not reach statistical significance. In exploratory analyses, visual field severity, as assessed using Goldmann isopter area, was not significantly associated with postoperative VA or IOL positional/refractive outcomes in the available RP subset; however, this finding should be interpreted cautiously because of limited data. Quantitative AS-OCT assessment and careful long-term monitoring may be useful for detecting subtle IOL misalignment, anticipating refractive unpredictability, and optimizing postoperative management in patients with RP.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement1
Print
