A leading challenge in optical biometry is the axial measurement of eyes with dense cataracts. PSC and mature cataracts commonly cause measurement acquisition failure. It has been reported that failure rates vary significantly depending on the optical technology used to measure axial length, with SS-OCT-based biometers producing the lowest failure rates [12]. As mentioned earlier, few studies have specifically analyzed the technology’s degree of success in eyes with dense cataracts [13,14,15,16,17,18]. The main purpose of this study was to evaluate acquisition success rates using different SS-OCT-based biometers in eyes with dense cataracts. To the best of our knowledge, there have been no studies that evaluated these optical biometers using objective metrics for describing cataract density.

Our results revealed statistically significant differences in the axial length acquisition success rates between biometers (P = 0.014). The acquisition rates of the SS-OCT biometers [Argos® (100%), IOLMaster® 700 (98.04%) and Anterior® (94.12%)] were significantly higher than the PCI biometer [Pentacam® AXL (60.78%)]. Additional file 6: Table S1 shows the axial length acquisition success rates obtained in eyes with dense cataracts from different clinical studies using optical biometers based on PCI, OLCR and SS-OCT technologies. It shows the cataract type for each study sample as well as the type of cataract that could not be measured. Hirnschall et al. [13] evaluated the use of the IOLMaster® 700 SS-OCT for eyes in which the axial length could not be measured using the IOLMaster® 500 PCI biometer. The authors looked at 23 subjects with failed acquisition using the IOLMaster® 500 and only two could not be measured with the IOLMaster® 700 SS-OCT. The 2 patients whose axial length could not be measured with the IOLMaster® 700 had nuclear cataracts. The study reported an acquisition success rate of 91.3%, and our value was 98.04%. The authors concluded that SS-OCT technology significantly improves the rate of attainable axial eye length measurements, especially in eyes with PSC, but also in eyes with dense nuclear cataracts, except for white cataracts. The main reason for SS-OCT’s greater success rate is the use of a longer wavelength (1,055 nm) compared to PCI technology (780 nm), as shorter wavelengths mean shallower penetration due to scattering [29]. Henriquez et al. [14] studied 45 eyes with cataracts of grade ≥ 3 (using LOCS III) for nuclear color (NC) and nuclear opalescence (NO), PSC and/or cortical cataract. They used the IOLMaster® 700 SS-OCT, Galilei G6 OLCR and Pentacam® AXL PCI biometers and assessed lens density by means of the LOCS III and PNS metric. The mean and SD for the PNS score in their sample was 3.21 ± 1.33. After three attempts, the Pentacam® AXL, Galilei G6 and IOLMaster® 700 could measure axial length in 37.7%, 42.2%, and 84.4% of the eyes, respectively. These authors reported the mean cataract density based on LOCS III for those eyes that could not be measured using the three devices (see Additional file 1: Table S1 for details). Note that these three devices use different wavelengths—1055 nm (IOLMaster® 700), 880 nm (Galilei G6) and 474 nm (Pentacam® AXL)—and higher wavelengths produce a better signal-to-noise ratio hence improving tissue penetration (but higher signal-to-noise ratio is also achieved due to different detection schemes used in those devices and different types of detector technologies). Additionally, wavelength correlates with the acquisition success rate. Our acquisition rates were higher for the IOLMaster® 700 and Pentacam® AXL. Note that the mean LOCS III and PNS scores were higher than those for our sample when axial length could not be measured (specifically, for the Pentacam® AXL biometer, the values were 4.00 ± 1.08 and 2.76 ± 1.52, respectively). Vasavada et al. [15] compared the Lenstar LS900 OLCR and the Tomey OA-2000 SS-OCT and found that the LS900 was unable to measure axial length in 22.58% eyes, whereas the Tomey OA-2000 failed in only in 1.6% eyes. Note that the study sample included eyes with low-grade cataracts. The authors concluded that axial length measurement on OCLR failed in one-fifth of the subjects with dense cataracts, which consist of white mature, dense PSC with a posterior capsule plaque or posterior polar cataracts.

Tamaoki et al. [16] assessed the Argos®, IOLMaster® 700 and OA-2000 SS-OCT biometers reporting acquisition rates of 89.9%, 63.6% and 80.8%, respectively. The IOLMaster® 700 had a significantly lower acquisition rate than that of the Argos® (P < 0.0001) and OA-2000 (P = 0.011) biometers, but there was no statistically significant difference between the Argos® and OA-2000 in this regard. The cataract type that resulted in failed acquisitions were mainly mature or white cataracts (see Additional file 1: Table S1). Data could not be acquired for three eyes with grade 4 nuclear sclerotic cataract accompanied by PSC using the IOLMaster® 700. However, acquisition was successful with the other two biometers. As there were only six cases of grade ≥ 4 nuclear sclerotic cataracts accompanied by PSC in their study, the authors suggested that further investigation is needed to elucidate how this condition may affect axial length measurements. The same authors [17], compared the IOLMaster® 700 and Argos® SS-OCT biometers in eyes with a nuclear hardness ≥ 4 [30, 31]. If the Argos® biometer failed to measure the axial length using the standard mode, it was measured using the ERV mode. They did not observe any statistically significant difference between the acquisition success rates for the Argos® in standard mode (69.5%) and the IOLMaster® 700 (61.5%; P = 0.083). However, they found that the overall acquisition success rate of the Argos® standard and ERV modes combined (93.4%) was significantly higher than that of the IOLMaster® 700 (P < 0.001). Among the 82 eyes that could not be measured using the IOLMaster® 700, 25 eyes had a white cataract. Of these 25, the Argos® successfully measured the axial length of 16.0% in standard mode and 60.0% in ERV mode. The percentages obtained in our study were higher than those reported by Tamaoki et al. [16, 17]. In another study, González-Godínez et al. [18] used the IOLMaster® 700 and IOLMaster® 500 and reported a failure rate of 68.57% and 21.43%, respectively (P = 0.007). The analysis revealed that PCI acquisition success rates were 69.23% of NO4, 66.6% of P3 and 15.3% of mixed cataracts, while for SS‑OCT they were 100% of NO4, NO5, P3 and P5 and 76.9% of mixed cataracts. The IOLMaster® 700 biometer had rates of 100% of NO4, NO5 and P3, and 88.8% of P4. They reported failure rates for a mixed cataract group (composed of PSC P > 3 and cortical C ≥ 4 or nuclear opalescence NO ≥ 4) of 100% when employing the PCI biometer and 40% for the SS‑OCT. The authors suggested that the cut‑off for the SS-OCT biometer may well be up to P4 and NO5. As for dense nuclear opacity above NO5 and intumescent cataracts, immersion ultrasound biometers remain the best option.

As highlighted by Tamaoki et al. [17], the difficulty of measuring axial length in white cataracts is due to light scattering even when using SS-OCT biometers operating at long wavelengths (see failures for SS-OCT as a function of LOCS III score in our study, Fig. 1, bottom panel). Tamaoki et al. [17] reported that the axial length of 48.6% of white cataracts could not be measured using the Argos® in standard mode but acquisition was successful in the ERV mode. In our study, axial length could not be measured with the Argos® in standard mode for three eyes with grade 6 LOCS III scores, but all of them were successfully measured with the ERV mode. OCT sensitivity decreases with depth and particularly in the case of dense cataracts. This precludes retinal pigment epithelium layer detection and vitreous length measurement due to light energy attenuation. In ERV mode, optical path length is measured using the same principle as that used in enhanced depth imaging for choroidal imaging [17, 32]. We believe this mode is an excellent option for measuring axial length in complicated cases. Other authors [33] have recently concluded that pharmacologic pupil dilation improved the quality of the IOLMaster® 700’s biometrically measured axial length in patients with low-quality measurement due to dense cataract. High-quality axial length measurements were successfully obtained in 60 of the 79 eyes (75.95%) following pupil dilation in this study. The mean standard deviation of the measurements obtained decreased significantly (P < 0.001) and the mean difference before and after pupil dilation was 0.03 ± 0.07 mm (P < 0.001).

Regarding the agreement between devices for the different parameters, our results revealed statistically significant differences between the five biometers (Table 1, P < 0.05). The post-hoc Tukey test showed that statistical significance depended on which pairwise comparisons were analyzed between biometers and which parameter was being examined. Specifically, significant differences were found between devices for all ACD and LT comparisons, and some for K1, K2, WTW and axial length (Table 2, P < 0.05).

Results for K showed that the minimum mean differences were obtained for the comparison between the Anterion® vs. IOLMaster® 700 (− 0.038 D) for K1 and between the Anterion® vs. Pentacam® AXL (0.040 D) for K2, but neither case was statistically significant (P > 0.05). The maximum mean differences were between the Argos® vs. Pentacam®, with both K1 (0.156 D) and K2 (0.206 D) returning a statistically significant difference (P < 0.05). It has been suggested that a difference of 1.00 D in the K value would cause a difference of about 1.40 D in the IOL power calculation [35]. If we used the K values to calculate the IOL power and took 0.206 D as the maximum mean difference, then it would lead to a difference of about 0.28 D in the IOL power. The small differences in agreement reported for all K measurement comparisons led to clinically insignificant changes in the IOL power calculation because of the 0.50 D step in IOL manufacturing. Nevertheless, it is worth noting that the range of LoA varied as a function of the comparison and all comparisons were > 1.00 D, which is broad enough to produce a significant change in IOL power (see Table 2 and Additional file 1: Fig. S1). Others have reported that K values cannot be interchanged between the Anterion®, IOLMaster® 700 and Pentacam® AXL biometers [10]. The mean differences reported for WTW measurements varied from − 0.01 mm for the Argos® vs. the IOLMaster® 700 to − 0.20 mm between the Anterion and IOLMaster® 700 (Table 2). The LoA range was ≥ 0.60 mm, which may be clinically significant. For IOL power, we did observe differences when using the Holladay 2 and Barrett formulas with WTW as a variable. WTW measurements between these devices cannot be considered interchangeable. Tañá-Rivero et al. concluded that WTW data may be considered interchangeable between the Anterion® and IOLMaster® 700, and the Anterion® and Pentacam® HR but not between the IOLMaster® 700 and Pentacam® HR [36], and that the IOLMaster® 700 measured the largest WTW distances and the Pentacam® AXL the shortest [10]. As for ACD, we found statistically significant differences between all pairwise comparisons, although they were very small (about 0.1 mm). A previous study comparing the Anterion® and Pentacam® HR biometers reported similar mean differences [39] and the Anterion®, IOLMaster® 700 and Pentacam® AXL can be used interchangeably [10]. On average, it has been reported that a 1 mm deviation in ACD could lead to a refractive error of 1.50 D in IOL power [34] (maximum LoA was about 0.30 D, resulting in a change in IOL power of less than 0.50 D). Therefore, we can conclude that the ACD differences reported between biometers will not affect IOL power calculation. Mean differences for LT were less than 0.05 mm with a maximum LoA range of 0.378 mm. Note that a 0.2 mm increase in LT would change the IOL power by 0.20 D. However, taking into account our mean differences, this may not have a clinically significant impact on the IOL power calculation when using the Olsen or Holladay 2 formulas [37, 38]. We believe the three devices can be used interchangeably for LT measurements, as has been reported previously for comparisons of SS-OCT biometers [10, 40]. Our mean differences obtained for axial lengths between optical biometers were all less than 0.1 mm, except when compared to the OcuScan® RxP which returned higher values (up to 0.234 mm for the comparison between Pentacam® AXL and OcuScan® RxP; see Table 2). If we consider that a 0.1 mm error in axial length would yield a refraction error of about 0.27 D [34], then the differences between optical biometers would not affect the IOL power calculation but would affect any comparisons with ultrasound. Hence, we believe that only optical biometers can be used interchangeably (as previously reported with SS-OCT and PCI [10] in non-dense cataracts and healthy eyes [9]). However, the LoA ranges reach high values, surpassing the limits considered clinically negligible, and this must be taken into account in the IOL power calculation.

Only Henriquez et al. [14] has compared the IOLMaster® 700 and Pentacam® AXL in 45 eyes with mature cataracts. They reported significant inter-device differences for axial length and K2 (P = 0.012 and 0.034, respectively), but not for K1 and ACD (P > 0.1). The absolute mean difference was 0.05 mm for axial length and 0.33 D for K2. Our results are distinct because we have found statistically significant inter-device differences for all parameters (K1, K2, WTW and ACD), but not for axial length (P > 0.99, with a mean difference of − 0.023 mm and a LoA range of 0.784 mm; see Table 2). As discussed previously, the mean LOCS III and PNS scores were higher than those observed for our sample where the measurement was not possible, which could be a source of differences between the studies. Tamaoki et al. [17] compared pre- and postoperative axial length measurements and found mean absolute errors of 0.05, 0.10, 0.08 and 0.12 mm for the Argos®, IOLMaster® 700, Argos® in ERV mode and ultrasound biometers, respectively. The absolute difference for the Argos® was significantly lower than for the IOLMaster® 700, Argos® in ERV mode and ultrasound biometers (P < 0.001, P = 0.008 and P < 0.001, respectively). Comparing the Argos® with and without ERV, they indicated that despite being higher than the difference obtained with the standard mode, this value can be considered clinically negligible. They also found an absolute error of 0.12 mm using the ultrasound biometer, almost the same as that observed with the ERV mode. They suggested that measurements with the ultrasound biometer are also important since 22% of cases showed an axial length error ≥ 0.2 mm. It should be noted that for pre- and post-surgical measurements, it helps to consider the LoA when performing the Bland–Altman analysis [11] and, secondly, the Argos® measures axial length using segmental refractive indices and the components of the eye are different in this comparison (with more variability if different IOLs are used, as indicated by Tamaoki et al. [17]). Gonzalez-Godinez et al. [18] compared the IOLMaster® 700 and an ultrasound biometer and found a poor level of agreement: the IOLMaster® 700 mean axial length was 0.15 mm shorter than the ultrasound biometer (P = 0.005) with an LoA of 1.56 mm. Our values, when compared to another ultrasound biometer, were slightly higher: 0.180 mm and 2.049 mm for the mean difference and LoA, respectively. Differences between techniques (including which retinal layer is measured, velocity and refractive index changes in dense cataracts when using ultrasound biometry) may explain the poor agreement. A recent network-based big data analysis demonstrated that when considering the measurement of axial length, contact ultrasound biometry obtains lower values compared with optical biometers [41].