purpose. To assess the repeatability and reliability of IOLMaster (Carl Zeiss Meditec, Inc., Dublin, CA) axial length and keratometry measurements (K readings) with a soft contact lens on normal eyes. The method is designed for eyes with corneal irregularities or after endothelial keratoplasty.

methods. Biometry was performed on 20 healthy right eyes of volunteer subjects with mean age, 27.3 ± 4.9 years; axial length, 24.77 ± 1.04 mm; and K reading, 43.48 ± 1.69 D. Axial length and keratometry were measured and repeated with −0.5 D So*f*Lens38 (Bausch & Lomb, Rochester, NY) and Acuvue2 (Johnson & Johnson, New Brunswick, NJ) soft contact lenses. Repeatability and reliability were evaluated. Contact lens thickness was measured directly by corneal optical coherence tomography (OCT).

results. Axial lengths increased 59 ± 10 μm with So*f*Lens38 and 134 ± 13 μm with Acuvue2, and these changes correlated with the OCT contact lens thicknesses (*P* = 0.995). The axial length variability remained constant (*P* = 0.18), measuring 24 ± 10 μm for So*f*Lens38 and 23 ± 8 μm for Acuvue2 compared with 20 ± 7 μm with no lens. K readings of 43.08 ± 1.66 D with So*f*Lens38, 42.79 ± 1.57 D with Acuvue2, and 43.48 ± 1.69 D with no lens corresponded to differences of −0.40 ± 0.12 D with So*f*Lens38 and −0.69 ± 0.19 D with Acuvue2. The K-reading variability increased slightly from 0.04 to 0.09 D with either lens.

conclusions. Low-power soft contact lenses enable reliable and repeatable IOLMaster axial length and K-reading measurements. Correcting for the measurable lens thickness and lens effects, a <0.5-D error in the Sanders-Retzlaff-Kraff (SRK) II power formula is predicted.

^{ 1 }Its success relies on accurate preoperative biometry of curvature and intraocular distances, particularly axial length,

^{ 2 }to calculate the appropriate intraocular lens (IOL) power with the appropriate formula, (e.g., Holladay II, Sanders-Retzlaff-Kraff [SRK] II, SRK/T, and Hoffer Q).

^{ 3 }

^{ 4 }

^{ 5 }

^{ 6 }

^{ 7 }

^{ 8 }

^{ 9 }

^{ 10 }

^{ 11 }

^{ 12 }

^{ 13 }

^{ 14 }Optical biometry is fast and efficient compared with ultrasonic immersion techniques

^{ 15 }

^{ 16 }

^{ 17 }

^{ 18 }; however, the patient must have adequate fixation and no advanced cataracts or significant corneal irregularities.

^{ 15 }

^{ 19 }

^{ 20 }Significant corneal irregularities often require transplantation, and new techniques that minimally disrupt the corneal surface are in development. In particular, Descemet’s stripping with endothelial keratoplasty (DSEK) is becoming a promising alternative to traditional penetrating keratoplasty in appropriate cases, such as patients with Fuchs’ corneal dystrophy.

^{ 21 }Preoperative keratometry and axial length measurements can be difficult to obtain in edematous corneas before endothelial keratoplasty combined with implant procedures that require power calculations. Consequently, choosing the appropriate IOL power for the patient undergoing cataract extraction and DSEK can be challenging, as corneal irregularities may limit the surgeon’s ability to measure axial length and corneal curvature accurately by optical partial coherence interferometry (PCI).

*soft contact lens method*is proposed. A soft contact lens provides a smooth interface that permits ocular measurements with PCI. A contact lens of known thickness and refractive power should hypothetically alter axial length and keratometry results (K readings) in a predictable manner. This prospective study involving healthy human eyes was designed to test the accuracy and repeatability of optical biometry in the presence of two different −0.5-D soft contact lenses. These thin, soft contact lenses are expected to disrupt the optical performance of the device minimally and to enhance the optical response of the cornea. The soft lenses have an index of refraction (

*n*) of 1.40 to 1.43 that more closely resembles the cornea (

*n*≅ 1.376) than a hard PMMA lens (

*n*≅ 1.50). This technique was designed to expand the use of PCI biometry to include eyes with corneal irregularities, and clinical practice will help to determine how this potentially useful technique translates in such irregular eyes.

*f*Lens38 (polymacon 62%, water content, 38%; refractive index, 1.43; approximate center thickness, 0.035 mm; Bausch & Lomb, Rochester, NY) and the Acuvue2 (Etafilcon A, water content, 58%; refractive index, 1.40; approximate center thickness, 0.119 mm; Johnson & Johnson, New Brunswick, NJ).

*f*Lens38 (CL1) and then the Acuvue2 (CL2) lens. One experienced investigator performed all ocular examinations except OCT for each condition. A second skilled investigator performed the OCT measurements. Three central corneal OCT line scans composed of multiple A-scans were used to determine the in situ contact lens thickness with precorneal tear film as described by Muscat et al.

^{ 22 }and Wang et al.

^{ 23 }(Fig. 1) . The IOLMaster was then used to measure axial length and keratometry, including the corneal radius of the two principle meridians, the corneal power, the axes and the astigmatic difference. Each keratometry measurement is the mean value of five individual measurements acquired within 0.5 seconds. The radius of curvature was converted to diopters by using a 1.3375 refractive index. Each K reading was calculated as (K1 + K2)/2, corresponding to the principal meridians and recorded in diopters K.

^{ 24 }Each session proceeded at a similar rate and lasted 73 ± 14 minutes. On average, 12.5 ± 2.4 minutes elapsed between CL1 insertion and the first OCT scan, and 19.2 ± 3.6 minutes elapsed between the first OCT scan and the first IOLMaster scan. CL2 followed at 11.9 ± 2.4 and 18.8 ± 3.4 minutes, respectively.

^{ 25 }

^{ 26 }Reproducibilities from the axial length and corneal radius SD collected at different sessions are ±0.03 and ± 0.02 mm, respectively.

*t*-tests and two-way, repeated-measures ANOVA with the Levene test and the Welch ANOVA, to test the homogeneity of variance (

*P*< 0.05). The reproducibility and reliability of the PCI difference technique used to measure accurate axial lengths was evaluated for each lens. The repeatability and agreement between the lens and no-lens mean K-reading measurements were evaluated by two-way, repeated-measures ANOVA (

*P*< 0.05). Regression analysis was used to assess potential systematic errors correlating to the base biometry.

*P*< 0.0001). The distribution across subjects is shown in Figure 2 . The group mean PCI axial length measured was 24.77 ± 1.04 mm (range, 22.00–26.48) for the no-lens condition, 24.83 ± 1.04 mm (range, 22.07–26.54) with So

*f*Lens38, and 24.90 ± 1.04 mm (range, 22.14–26.60) with Acuvue2. The mean change in axial length associated with each lens was 59 ± 10 μm (range, 41–74) with So

*f*Lens38 and 134 ± 13 μm (range, 111–155) with Acuvue2.

*f*Lens38 and 23 ± 8 μm (range, 13–47) with Acuvue2. Although the mean SD in axial length measurement increased slightly, the variances between the no-lens and both lens conditions were not significantly different (

*P*= 0.18). The two lenses also performed similarly, in that there was no statistically significant difference between the variances of each mean axial length measurement (

*P*= 0.41).

*P*< 0.0001). The mean OCT differences corresponding to the soft lens and precorneal tear film thicknesses (Fig. 2)were 61 ± 6 μm (range, 47–72) for So

*f*Lens38 and 136 ± 3 μm (range, 129–143) for Acuvue2. The axial length differences from PCI were comparable to the corneal thickness differences from OCT for each lens (Fig. 3) . For the So

*f*Lens38 lens condition, the measured PCI axial length difference was 59 ± 10 μm compared with the OCT corneal thickness difference of 61 ± 6 μm. For the Acuvue2 lens condition, the PCI axial length difference was 134 ± 13 μm, compared with the OCT corneal thickness difference of 136 ± 3 μm. The variance introduced by use of either lens was similar (

*P*= 0.995), when the axial length and corneal thickness difference measurements were compared (

*P*= 0.50 for So

*f*Lens38 and

*P*= 0.56 for Acuvue2).

*P*= 0.50 for So

*f*Lens38 and

*P*= 0.56 for Acuvue2, with the Levene test for homogeneity of variance

*P*= 0.02 for So

*f*Lens38 and

*P*= 0.001 for Acuvue2). These results demonstrate the accuracy of using either the IOLMaster or OCT contact lens thickness to determine the actual axial length when a soft lens is applied.

*f*Lens38, and 42.79 ± 1.57 D (range, 40.80–46.08) with Acuvue2 (Table 2) . These mean K readings are significantly different (

*P*< 0.0001). The distribution of K-reading differences across subjects is shown in Figure 4 . The effect of each lens was assessed by comparing the lens versus no-lens mean K-reading differences. So

*f*Lens38 corresponded to a mean K-reading difference of −0.40 ± 0.12 D and Acuvue2 to a mean K-reading difference of −0.69 ± 0.19 D. These values include the curvature and lens effects associated with each −0.5 D contact lens. The lens effect should be minimal in a low-power lens. If one approximates the lens effect using the lens power, the magnitude of the effective differences adjusts to 0.10 ± 0.12 D for So

*f*Lens38 and −0.19 ± 0.19 D for Acuvue2. The adjusted mean K readings were distinct from those with no lens (

*P*= 0.0011 for So

*f*Lens38 and

*P*= 0.0002 for Acuvue2); however, these differences may be clinically acceptable, considering that the variances were within 0.5 D and within the minimum power tolerance allowed for pseudophakic IOLs.

^{ 27 }Accuracy improved when the K readings were adjusted by the experimentally determined, lens-specified mean K differences (Fig. 5) .

*f*Lens38, and 0.09 ± 0.06 D (range, 0.02–0.28) with Acuvue2 (Table 2) . The mean K-reading variance increased slightly with each lens; however, the increase was small (0.05 D) and similar for each lens (

*P*= 0.84).

*P*< 0.01). To test whether the subject’s base corneal thickness, axial length, and K reading influenced the technique, regression analyses were performed (Figs. 7 8 9) . A statistically significant error is observed only in the CL2-difference K readings, compared with both base axial length and corneal curvature.

^{ 5 }

^{ 7 }

^{ 8 }

^{ 28 }

^{ 29 }

^{ 30 }However, to provide a basic example of the influence of uncorrected errors in axial length and K reading with either lens, we applied the SRK II equation,

^{ 4 }

^{ 31 }where the dioptric lens power equals A

_{Constant}− 2.5 × axial length − 0.9 × K reading. According to the formula, the axial length error has a proportionally greater influence on the predicted postsurgical refractive power than the does the K-reading error (ratio, 2.5:0.9 or 2.78:1).

^{ 32 }The change in axial length due to the added soft contact lens is expected to be a small percentage, or 50 to 150 μm of ∼25 mm. For reference, an error in axial length of 100 μm would result in a refractive error of 0.28 D.

^{ 2 }For our emmetropic to moderately myopic eyes, the SD in IOLMaster-measured K reading increased by 0.05 D, and axial length did not significantly increase in the presence of a soft contact lens. As shown in this study, one can determine the change in axial length to within 0.013 mm with the IOLMaster and 0.006 mm with the corneal OCT. Assuming the maximum uncertainties in axial length and K reading attributable to the technique, one can expect an error of ±0.13 D for So

*f*Lens38 and ±0.20 D for Acuvue2 when applying this formula. These errors contribute an uncertainty of <0.5 D to the power calculation. IOLs are currently calibrated to correct eyes within ±0.5 D; therefore, the error is considered to be clinically acceptable.

^{ 33 }

^{ 34 }

^{ 35 }

^{ 36 }

^{ 37 }

^{ 38 }

^{ 39 }The soft contact lens method is not intended for these cases; instead, it is intended for cases with irregular surfaces and irregular astigmatism (i.e., allowing measurement in situations in which optical biometry cannot be accurately measured otherwise). With a rigid lens, the corneal curvature will be measured as the front curvature of the lens, irrespective of the underlying corneal curvature. Our hypothesis was that the curvature of the cornea with a low-power soft contact lens would be predictably related to the underlying corneal curvature and would subclinically distort the IOLMaster biometry.

*f*Lens38 and Acuvue2. The variability and accuracy of axial length measurements did not significantly change with either lens. The variability in K reading increased slightly (0.05 D) in the presence of either lens and could be accurately adjusted to subtract either lens contribution. The Acuvue2 K-reading difference was sensitive to base axial length and K reading, possibly reflecting a change in fit across the sample. These K-reading differences were greater for shorter base axial lengths and steeper base K readings. Rates of change of −0.11 D/mm between 26.5- and 22-mm base axial length and 0.08 D per diopter base K reading between 41 and 47 D were measured. The Acuvue2 K-reading differences could be compensated accordingly; otherwise, a maximum K-reading difference of −0.31 D can be expected, as observed in the extreme cases encountered. The So

*f*Lens38 did not show this dependence and may yield more accurate measurements when applied, irrespective of the base biometry.

*f*Lens38 and Acuvue2 are manufactured by different processes: The So

*f*Lens38 is cast-molded and the Acuvue2 is stabilized soft-molded. Also, different cast-molded lenses can perform differently.

^{ 40 }Other differences include the water content and thickness, which could influence the rigidity of the lens.

*f*Lens38 and Acuvue2. The subtraction technique has been tested and shown to be a statistically reliable technique in PCI-based optical biometry. Consequently, this method has the potential to improve the accuracy of IOL power calculations in patients needing combined cataract extraction and DSEK, by enabling accurate PCI measurements in eyes with irregular corneal surfaces. To assess this hypothesis, a prospective study evaluating the accuracy of the soft lens method in eyes with coexisting cataract and corneal disease is warranted.

**Figure 1.**

**Figure 1.**

**Figure 2.**

**Figure 2.**

Condition | Mean AL | Mean SD | Mean AL Difference | Mean CL Thickness, OCT Diff. | AL Diff.- OCT Diff. |
---|---|---|---|---|---|

No CL | 24.77 ± 1.04^{*} | 20 ± 7^{, †} ^{, ‡} | — | — | — |

Range | 22.00–26.48^{*} | 10–38^{, †} ^{, ‡} | — | — | — |

+CL1 | 24.86 ± 1.04 | 24 ± 10 | 59 ± 10^{, †} | 61 ± 6^{, †} | −2 ± 10^{, †} |

Range | 22.07–26.54 | 14–52 | 41–74^{, †} | 47–72^{, †} | −19–13^{, †} |

+CL2 | 24.90 ± 1.04 | 23 ± 8 | 134 ± 13 | 136 ± 3 | −2 ± 13 |

Range | 22.14–26.60 | 13–47 | 111–155 | 129–143 | −25–22 |

*N*= 10 consecutive measurements), and AL differences are presented for 20 subjects for the conditions of no contact lens (No CL), with So

*f*Lens38 (+CL1), and with Acuvue2 (+CL2). The mean contact lens thicknesses for each lens condition corresponding to the corneal OCT difference measurements are also presented. The mean OCT measurements for CL1 and CL2 were 2 μm larger than the corresponding IOLMaster AL measurements. Although the variability was slightly greater with CL2, the two techniques detected statistically similar changes in thickness associated with either lens (

*P*= 0.995).

**Figure 3.**

**Figure 3.**

Condition | Mean K-reading | Mean SD | Mean K-reading Difference^{, ‡} | Adjusted Mean K Difference^{, §} | Adjusted Mean K Difference^{, ∥} |
---|---|---|---|---|---|

No CL | 43.48 ± 1.69^{*} | 0.04 ± 0.02^{*} ^{, †} | — | — | — |

Range | 41.21–47.12 | 0.00–0.11^{*} ^{, †} | — | — | — |

+CL1 | 43.08 ± 1.66 | 0.09 ± 0.09 | −0.40 ± 0.12^{*} | 0.10 ± 0.12^{*} | 0.00 ± 0.12^{*} |

Range | 41.00–46.72 | 0.02–0.41 | −0.57–−0.20^{*} | −0.07–0.30^{*} | −0.17–0.20^{*} |

+CL2 | 42.79 ± 1.57 | 0.09 ± 0.06 | −0.69 ± 0.19 | −0.19 ± 0.19 | 0.00 ± 0.19 |

Range | 40.80–46.08 | 0.02–0.28 | −1.04–−0.30 | −0.54–0.20 | −0.35–0.39 |

*f*Lens38 (+CL1), and with Acuvue2 (+CL2). The mean K-reading differences correspond to each low-power contact lens effect and include curvature, fit, and tear film irregularities. The estimated lens contribution, subtracting the thin, soft contact lens power (−0.5 D) of each lens, shifts the K-readings to within 0.5 D of the actual keratometry. The change in K reading is lens specific (Fig. 6) , and so a more accurate adjustment is obtained by subtracting the mean K-reading differences associated with each lens, as shown.

**Figure 4.**

**Figure 4.**

**Figure 5.**

**Figure 5.**

**Figure 6.**

**Figure 6.**

**Figure 7.**

**Figure 7.**

**Figure 8.**

**Figure 8.**

**Figure 9.**

**Figure 9.**