Comparison of measurements obtained by an optical low coherence interferometry biometer and a swept source optical coherence tomography biometer in two datasets

INTRODUCTION

Accurate biometry measurements are essential for optimal intraocular lens (IOL) power calculation in cataract surgery.^[1] Contemporary optical biometers assess key ocular parameters, including axial length (AL), flat keratometry (K1) and steep keratometry (K2), anterior chamber depth (ACD), lens thickness (LT) and white-to-white (WTW) distance. The IOLMaster^® (Carl Zeiss Meditec, Jena, Germany), introduced in 1999, pioneered the use of partial coherence interferometry for non-contact biometry.^[2,3] Despite its efficiency, this device necessitated visual axis realignment between parameter measurements due to disparate measurement principles.

Technological advancements have yielded new biometers utilising novel methodologies. This study aims to compare biometric measurements obtained from two such instruments: The Lenstar^® (Haag-Streit AG, Switzerland), which employs optical low-coherence reflectometry (OLCR), and the Anterion^® (Heidelberg Engineering, Heidelberg, Germany), which utilises swept-source optical coherence tomography (SS-OCT). The comparison will be conducted within a National Health Service (NHS) cataract clinic setting, analysing two distinct datasets:

1. Scans acquired on different days by multiple technicians, simulating real-world clinical variability

2. Same-day scans performed by a single technician, minimising operator-dependent variables.

Through analysis of these datasets, we seek to evaluate the agreement between these advanced biometers and assess their potential impact on clinical practice. This study’s findings may inform the selection and utilisation of biometric devices in cataract surgery planning, potentially influencing surgical outcomes and patient care.

MATERIALS AND METHODS

This study adhered to the tenets of the Declaration of Helsinki. Informed consent was obtained from all patients. The study was registered as an audit (number OPH_099).

RESULTS

Descriptive analysis

Table 1 summarises the descriptive analysis of the Lenstar and Anterion values along with ICC values, P-values and Pearson correlations.

Table 1: Comparison between Lenstar and Anterion biometry values and parameters.

	Lenstar1	Anterion1	P2	r3	ICC (95% CI)	Mean Difference (Lenstar-Anterion)
Dataset 1
AL (mm)	23.78±1.37	23.78±1.39	0.66	0.99	0.998 (0.997–0.998)	0.005±0.128
K1 (D)	43.22±1.47	43.18±1.45	0.18	0.94	0.971 (0.960–0.979)	0.054±0.490
K2 (D)	44.25±1.60	44.14±1.39	<0.01	0.96	0.981 (0.973–0.987)	0.101±0.418
ACD (mm)	3.17±0.44	3.21±0.43	<0.01	0.96	0.974 (0.961–0.983)	−0.044±0.132
LT (mm)	4.45±0.46	4.60±0.46	<0.01	0.92	0.933 (0.712–0.972)	−0.153±0.181
WTW (mm)	11.87±0.77	11.75±0.48	0.02	0.54	0.650 (0.518–0.745	0.121±0.646
Dataset 2
AL (mm)	23.27±±1.44	23.26±1.45	0.35	0.99	0.999 (0.999–1.000)	0.010±0.067
K1 (D)	43.14±2.02	43.10±2.16	0.56	0.98	0.988 (0.978–0.994)	0.041±0.454
K2 (D)	44.47±1.76	44.37±1.84	0.11	0.98	0.987 (0.976–0.993)	0.100±0.393
ACD (mm)	3.09±0.38	3.07±0.36	0.51	0.82	0.901 (0.816–0.947)	0.023±0.225
LT (mm)	4.35±0.78	4.80±0.59	<0.01	0.59	0.639 (0.185–0.826)	−0.452±0.643
WTW (mm)	11.89±0.68	11.71±0.63	<0.01	0.82	0.882 (0.753–0.940)	0.180±0.340

¹Mean±standard deviation, ²Paired t-test, ³Pearson’s correlation coefficient. AL: Axial length, ACD: Anterior chamber depth, WTW: White-to-white, LT: Lens thickness, ICC: Intraclass correlation coefficient, CI: Confidence interval

Dataset 1 demonstrated excellent reproducibility between instruments for AL, K1, K2, ACD and LT (ICC > 0.9). However, WTW showed poor reproducibility (ICC = 0.65). Dataset 2 showed excellent reproducibility for AL, K1 and K2 (ICC > 0.9), good reliability for ACD and WTW, but poor reliability for LT. The main difference between datasets was that dataset 1 showed poor reliability for LT, whereas dataset 2 showed poor reliability for WTW measurements. Bland–Altman plots for agreement analysis are presented in Figures 1-12. The thick black line represents the mean difference between devices, and the dotted lines represent the upper and lower LoA.

Close

Figure 1:

Bland–Altman plot of axial length dataset 1. The mean difference is represented by the solid line and the 95% limits of agreement by the dotted line. AL: Axial length.

Export to PPT

Close

Figure 2:

Bland–Altman Plot of axial length dataset 2. The mean difference is represented by the solid line and the 95% limits of agreement by the dotted line. AL: Axial length.

Export to PPT

Close

Figure 3:

Bland–Altman plot of K1 dataset 1. The mean difference is represented by the solid line and the 95% limits of agreement by the dotted lines.

Export to PPT

Close

Figure 4:

Bland–Altman plot of K1 dataset 2. The mean difference is represented by the solid line and the 95% limits of agreement by the dotted lines.

Export to PPT

Close

Figure 5:

Bland–Altman plot K2 dataset 1. The mean difference is represented by the solid line and the 95% limits of agreement by the dotted lines.

Export to PPT

Close

Figure 6:

Bland–Altman plot of K2 dataset 2. The mean difference is represented by the solid line and the 95% limits of agreement by the dotted lines.

Export to PPT

Close

Figure 7:

Bland–Altman plot of anterior chamber depth (ACD) of dataset 1. The mean difference is represented by the solid line and the 95% limits of agreement by the dotted lines.

Export to PPT

Close

Figure 8:

Bland–Altman plot of anterior chamber depth (ACD) dataset 2. The mean difference is represented by the solid line and the 95% limits of agreement by the dotted lines.

Export to PPT

Close

Figure 9:

Bland–Altman plot of lens thickness (LT) of dataset 1. The mean difference is represented by the solid line and the 95% limits of agreement by the dotted lines.

Export to PPT

Close

Figure 10:

Bland–Altman plot of lens thickness (LT) of dataset 2. The mean difference is represented by the solid line and the 95% limits of agreement by the dotted lines.

Export to PPT

Close

Figure 11:

Bland–Altman plot of white-to-white (WTW) for dataset 1. The mean difference is represented by the solid line and the 95% limits of agreement by the dotted lines.

Export to PPT

Close

Figure 12:

Bland–Altman plot of white-to-white (WTW) for dataset 2. The mean difference is represented by the solid line and the 95% limits of agreement by the dotted lines.

Export to PPT

Axial length (AL)

AL showed no statistically significant difference between biometers for both datasets (P = 0.66 and P = 0.35), with high reproducibility (ICC > 0.9) and mean differences of 0.01 ± 0.13 mm and 0.01 ± 0.07 mm. Bland–Altman plots demonstrated low differences between measurements and narrow LoA [Figures 1 and 2].

Keratometry

While there was no statistically significant difference in K1 for both datasets (P = 0.18 and P = 0.56), Bland–Altman plots showed moderate variability within ± 1 dioptre for both datasets [Figures 3 and 4]. K2 showed statistically significant differences for dataset 1 (P < 0.01) but not for dataset 2 (P = 0.11). Bland–Altman plots for K2 demonstrated greater variability within ± 1 dioptre for both datasets [Figures 5 and 6]. Mean differences between Lenstar and Anterion for K1 and K2 were 0.05 ± 0.49 D and 0.10 ± 0.42 D for dataset 1, and 0.04 ± 0.45 D and 0.10 ± 0.39 D for dataset 2, with high reproducibility in both groups and datasets (ICC > 0.9).

Anterior chamber depth (ACD)

The Bland–Altman plot for ACD showed narrow variability in dataset 1 [Figure 7] but wider variability for dataset 2 [Figure 8]. ACD showed statistically significant differences between samples (P < 0.01) for dataset 1. Dataset 2 did not show statistically significant differences (P = 0.51). ACD mean difference for dataset 1 showed Anterion with greater measurements (−0.04 ± 0.13 mm), contradicting dataset 2 which had Lenstar with greater measurements (0.02 ± 0.26 mm). ACD demonstrated high reproducibility between devices in both datasets (ICC > 0.9).

Lens thickness (LT)

LT showed statistically significant differences between samples (P < 0.01) with mean differences of −0.15 ± 0.18 mm for dataset 1 and −0.45 ± 0.64 mm for dataset 2. LT had high reproducibility between devices for dataset 1 (ICC > 0.9) but poor to moderate reproducibility for dataset 2 (ICC = 0.64). Bland–Altman plots showed a spread of differences for dataset 1 [Figure 9] and very wide LoA with a spread of differences for dataset 2 [Figure 10].

White-to-white (WTW)

The mean difference in WTW measurements between Lenstar and Anterion was 0.12 ± 0.65 mm for dataset 1 and 0.12 ± 0.34 mm for dataset 2, both statistically significant (P = 0.02 and P < 0.01, respectively). Reproducibility was poor to moderate for dataset 1 (ICC = 0.54) and good for dataset 2 (ICC = 0.82). The Bland–Altman plot for dataset 1 [Figure 11] showed similar values for measurements above 11 mm but outliers and larger differences for measurements below 11 mm. Dataset 2 [Figure 12] showed a spread between the LoA.

DISCUSSION

In this study, we evaluated the similarity of biometry measurements between Lenstar and Anterion for AL, flat (K1) and steep (K2) keratometry, ACD, LT and WTW distance in two datasets. The first dataset comprised a larger cohort where biometrics were not necessarily performed on the same day or by the same technician. The second dataset consisted of biometrics conducted on the same day by the same technician.

Our results demonstrated high similarity between the measurements for AL, K1, K2 and ACD in both datasets. The greatest similarity was observed in AL, with an ICC of 0.99 and excellent inter-device agreement of values. Both Lenstar and Anterion have previously shown good repeatability of AL measurement in other studies.^[10-12] Our findings align with recent literature.^[13]

Keratometry values were flatter in Anterion measurements compared to Lenstar in both datasets for both K1 and K2, but only statistically significant in dataset 1 for K2. These flatter Anterion measurements are consistent with previous publications.^[6,14,15] The difference in keratometry values can be attributed to the distinct measurement methods employed by each device. Anterion performs a 65-radial scan at a 3 mm zone, whereas Lenstar takes K readings from 32 points at 1.65 mm and 2.3 mm optical zones. The mean difference in both K1 and K2 in both datasets was small (<0.10 dioptres) and thus not clinically relevant for IOL calculation.

In dataset 1, Anterion measured ACD deeper than Lenstar, and this difference was statistically significant. Our values correlate with other recent studies that have found Anterion to measure deeper ACD compared to Lenstar and other biometers.^[6,14,15] Conversely, dataset 2 showed Lenstar with deeper ACD measurements compared to Anterion. However, the presence of outliers in the smaller sample size of dataset 2 may have skewed the mean difference. ACD measurement is crucial, as a change of 0.10 mm accounts for a 0.14 D refractive error change.^[16] These findings have implications for short eyes with AL <22 mm, where inclusion of ACD, LT and WTW in the Barrett Universal II lens calculation formula can result in a clinically significant difference in calculated IOL power.^[17]

The ICC values differed between the datasets for LT and WTW. LT was found to be greater in Anterion compared to Lenstar in both our datasets. This contradicts other literature that found SS-OCT biometers to have lower values compared to OLCR biometers.^[5,18,19] However, it agrees with a study by Vasavada et al., who found higher LT in an SS-OCT biometer compared to an OLCR biometer, although only for dense cataract.^[20] Our data included dense cataracts, which could account for the differences, as SS-OCT has demonstrated improved measurements in dense cataracts.^[21]

Clinically, LT impacts fourth-generation formulae, which use LT to predict the final IOL position, but it has no significance on predicted IOL power in older formulae.^[22] Haigis and Hill-RBF V2.0 have been shown to be significantly influenced by LT, independently of ACD, with a myopic shift with thin lenses and a hyperopic shift with thick lenses.^[23] Srivannaboon et al. found that when using the Holladay II formula, measurements from a partial coherence biometer with and without LT values showed a difference of only 0.02D on predicted postoperative spherical equivalent refraction.^[17] However, Lam found that using LT rather than age-based LT estimation for the Holladay II formula resulted in a statistically significantly lower mean absolute refractive error, and in 30% of cases, measured LT led to a different IOL choice.^[24] A recent study found that excluding LT led to statistically significant differences in absolute prediction error in several formulae.^[25]

WTW ICC measurements were also contradictory between the datasets. Other similar studies have shown poor ICC values in WTW compared to other parameters, but not as low as our value in dataset 1.^[14] WTW was statistically smaller in Anterion compared to Lenstar, but the Bland–Altman plot for dataset 1 revealed several outlier values where the Lenstar measurements were significantly smaller, with a difference of >2 mm. This discrepancy was particularly noticeable in WTW measurements of <11 mm, where differences between the biometers widened in both datasets. Clinically, this is relevant for implantable Collamer lens and anterior chamber IOL surgery, where sizing is partly based on WTW values. The significant WTW measurement difference could be due to the different measurement techniques employed by the devices. Anterion measures the endpoint of the posterior corneal surface (Schwalbe’s line), whereas Lenstar uses the image of the iris and the eye radii obtained from keratometry.

Our study had several limitations. Only the best scan was taken for analysis rather than an average of scans. Patients with prior corneal surgery, refractive laser treatment, keratoconus or pterygium were excluded from the study. However, patients with dense cataracts were not excluded, nor were other challenging cases such as high myopes, previous ophthalmic surgeries or ocular comorbidities. Hence, a wide range of short and long ALs were included in the study.

CONCLUSION

This study conducted within the NHS, comparing Lenstar and Anterion biometers across two distinct datasets, yields several important findings. First, the consistency of measurements taken on different days by various operators suggests that these biometers are robust to typical clinical variability, a valuable attribute in busy hospital settings where same-day measurements are not always feasible.

Both instruments demonstrated excellent agreement and intradevice consistency for AL, K1 and K2 and ACD. However, statistically significant differences were observed in LT and WTW measurements, irrespective of operator or timing of scans. Notably, WTW measurements below 11 mm showed greater discrepancies between devices, warranting caution in clinical scenarios where precise WTW values are critical, such as in implantable Collamer lens and anterior chamber IOL sizing.

These findings have important clinical implications, particularly for the calculation of IOL power in cataract surgery. The discrepancies in LT measurements may impact newer biometry formulae that incorporate this parameter, potentially affecting IOL power calculations. Furthermore, the variations in ACD, LT and WTW measurements could be particularly significant for short eyes and when using fourth-generation IOL calculation formulae.

While our study demonstrates the overall reliability of these advanced biometers, it also highlights the need for careful interpretation of certain parameters. Clinicians should be aware of these potential discrepancies when using these devices interchangeably or when applying their measurements to various IOL calculation formulae. Future research should focus on the clinical impact of these measurement differences on post-operative refractive outcomes.

In summary, while Lenstar and Anterion provide largely comparable measurements for key biometric parameters, the observed differences in LT and WTW emphasise the importance of consistent device selection in longitudinal patient assessments and careful consideration when applying these measurements to surgical planning.