Nowadays optical biometry represents the gold standard technique for axial length measurements in ophthalmology with an accuracy of around 0,05 mm. (J. Holladay, 2009)
This method is based on the principle of partial coherence interferometry (PCI) where the device sends a monochromatic wavefront of infrared light into the eye and captures it when the light reflects back from the retina, allowing an accurate estimation of the optical axial length along the visual axis, from the corneal apex to the foveal pigment epithelial layer.
Differently, because ultrasound immersion biometry does not depend on patient fixation, it measures only the anatomical axial length, from the corneal apex to the posterior pole. For IOL selection most important is to calculate the optical axial length, which as shown by Zaldivar et colleagues, especially in long eyes (> 26mm) is shorter than the anatomical axial length of a mean of 0.8 mm leading to a possible hyperopic refractive error when using an ultrasound biometry for calculations.35
Another inconvenience of the ultrasound biometers is related to the fact that they are highly dependent on the examiner experience. Optical biometry allows a contact free measurement of the optical axial length avoiding miscalculations related to ocular compression.36,37
Moreover, ultrasound waves stop in front of the retina at the level of internal limiting membrane (ILM) instead of at the level of the photoreceptors or retinal pigment epithelium (RPE). On average, retinal thickness is considered to be 200 µm. However, retinal thickness is highly variable especially in macular diseases. Thus 200 µm addition is not accurate for all patients. For this reason, calculation based on this technique are more likely to be affected in case of the presence of epiretinal membrane.
Although axial length measurements using a PCI technology should not be affected by retinal thickening due to macular pathologies, in literature are reported cases of myopic spherical errors after combined phacovitrectomy for iERM related to the presence of a second biometric peak.38,39
The main purpose of our study was to evaluate the effect of changes in retinal thickening on ultrasound immersion and optical biometry measures as well as the consequences of these changes on the postoperative refraction after combined phacovitrectomy for iERM.
According to literature, (Kovacs et al) we found a significant increase in ultrasound axial length (UAL) and a significant decrease in central macular thickness following the combined procedure. (p < 0,05) Our assumption is that the reduction in macular thickness following the ERM-ILM peeling, is responsible of the increase in the immersion ultrasound axial length. Spearman’s rank coefficient shows a negative correlation (rho= -0,3455) between the two variables, but not statistically significant. (p > 0,05)
Differently, optical biometry measurements should not be affected by the retinal thickening. However, we recorded a minimal but statistically significant postoperative reduction in the optical axial length. (p < 0,05)
This relative shortening of the eye is previous described after cataract extraction. In 1975, Binkhorst considered the decrease in axial length (AL) after cataract surgery when developing his formula for intraocular lens (IOL) power calculation. This same principle was applied when incorporating partial coherence interferometry into intraocular lens (IOL) power calculations. It recommended the use of a 0.12-mm correction factor for acrylic lenses and 0.08 mm for polymethylmethacrylate lenses to align preoperative and postoperative measurements. The IOL Master 500 has also incorporated this factor, adding 0.1 mm in cases of pseudophakic measurements.40
Our study found a decrease in the optical AL even with this correcting factor. This is already described in literature. The authors speculate this reduction could be related either due to a real decrease in the AL measurement after the lens removal or an inaccuracy in the suggested correcting factors.40,41
A measurement error of 0,1 mm is estimated to result in a postoperative refractive error of 0,28 D. 24 While some studies have shown no significant differences in the predicted and postoperative refractive error (RE) for combined phacovitrectomy and cataract surgery only groups, it is widely accepted that epiretinal membrane (ERM) may alter the accuracy of biometric estimes.42 Studies have demonstrated that eyes which undergo vitrectomy in addition to phacoemulsification for ERM are characterized by a mild myopic shift in postoperative refraction. In most studies induced myopia ranged from − 0.4 to -0.3 D.15,43,44
Thus, some suggest to add + 0,5 D to the predicted refraction in order to achieve the emmetropia.25
These findings suggest that the surgical procedure may mimic an absolute or relative shortening of axial length, a phenomenon which has been demonstrated when a tamponing agent such as gas is used, as it may push the IOL forward and alter the effective lens position (ELP).45 However, Jeoung et al. found no difference in postoperative myopic shift between eyes with intraocular tamponade and eyes without intraocular tamponade.20
Another possible cause of the myopic shift could be a miscalculation in biometry measurements. Kovacs et al, linked the myopic shift following phacovitrectomy for macular epiretinal membrane to the extent of preoperative macular thickening. They concluded that the observed myopic shift results from an erroneous IOL calculation, which occurs due to an underestimation of axial length (AL) with US immersion biometry because of the thicker macula.25,46 This could be particularly relevant in immersion biometry; however, Kim et al. showed a myopic shift for both A-scan and optic biometry (IOL Master 500).47 Differently, a strong correlation between preoperative predicted and postoperative achieved refraction was found in a study of 13 patients.48
Although previous studies have reported an increase in refraction error with a relative myopic shift following combined phacovitrectomy due to a preoperatively thicker retina (Kovacs et al), our study did not show a statistically significant dioptric shift either myopic or hyperopic in both subgroups (Optical and Immersion Biometry).
What is interesting is that the comparison between the optical dioptric shift and the immersion dioptric shift showed poor agreement between the two techniques.
A correlation was found between the preoperative CMT and the optical dioptric shift (rho = 0,6242 p = 0,0557). Differently the preoperative CMT did not correlate with the immersion dioptric shift (rho= -0,0303 p = 0,9326).
We interpreted these findings as a result of the presence of the epiretinal membrane which led to unpredictable refractive outcomes compared to the optical technique. Because US immersion biometry calculate the anatomical axial length, this could lead to a hyperopic spherical error as well as compensate a myopic shift induced by the postoperative flattening of the retina. Another source of error could be the irregularity in thickness of the diseased retina which affects US measurements, as well as oblique angle of incidence between the US waves and the ILM.
A correlation was found also between preoperative predicted refraction and actual spherical equivalent (SE) both for US immersion (rho = 0,9; p < 0,05) and optical biometry (0,511; p > 0,05), and also between preoperative predicted refraction and actual spherical equivalent (SE) both for US immersion (rho = 0,9; p < 0,05) and optical biometry (0,511; p > 0,05). However, analyzing Bland-Altman diagrams of the difference between predicted refractive target and refractive outcome both for optical and US biometry, we found a clinically significant dispersion of the data.
There are several limitations of this study: the limited number of patients included, the absence of a control group, the heterogeneity of the population (wide range of axial lengths, 21,88–27,35 mm) and the different biometric formulas used, and also the limited follow up time (up to 3 months).