**Purpose.**:
To characterize the refractive, keratometric, and corneal aberrometric effect of a specific type of intracorneal ring segment (ICRS) as a function of its thickness and the preoperative conditions of the cornea.

**Methods.**:
A total of 72 consecutive keratoconic eyes of 57 patients ranging in age from 15 to 68 years were retrospectively analyzed and included in the study. All cases had a diagnosis of keratoconus and had undergone implantation of a 160° arc-length KeraRing segment (Mediphacos, Belo Horizonte, Brazil), by femtosecond laser technology. Correlations between ring segment thickness and several clinical parameters were investigated. In addition, a multiple regression analysis was performed to characterize all factors that influence the ring segment effect.

**Results.**:
Significant reductions in central curvature, corneal astigmatism, and comalike aberrations were found after surgery (*P* ≤ 0.03). Moderate and limited correlations were found between ring segment thicknesses and changes in mean keratometry and higher order aberrations (*r* ≤ 0.50, *P* < 0.01). A consistent linear relationship of the superior ring segment thickness to the induced corneal changes, the preoperative cylinder, and the difference in thickness between inferior and superior segments was found (*P* < 0.01, *R* ^{2} = 0.91). An almost identical model was obtained for the inferior ring segment thickness with the only distinction in the factor being the thickness difference between segments (*P* < 0.01, *R* ^{2} = 0.64).

**Conclusions.**:
The selection of the ring segment to implant in keratoconus should be based, not only on refraction and subjective appearance of the corneal topographic pattern but also on corneal aberrometry. This highly customized selection would allow a more predictable outcome.

^{ 1 –22 }These segments act as spacer elements between the bundles of corneal lamellae, producing a shortening of the central arc length (arc-shortening effect) that is proportional to the thickness of the implant (Silvestrini T, et al.

*IOVS*1994;35:ARVO Abstract 2023). As a consequence of this effect, the central portion of the anterior corneal surface tends to flatten, and the peripheral area adjacent to the ring insertion is displaced forward.

^{ 23,24 }In nonpathologic corneas, there is a nearly linear relationship between the degree of central corneal flattening and the thickness of the implanted ring segments.

^{ 25,26 }However, this mechanism of action is not reproduced exactly in the keratoconic cornea. It should be considered that the well-organized lamellar structure of the cornea is lost when the corneal tissue degenerates, as happens in keratoconus.

^{ 27 }The regular orthogonal arrangement of the collagen fibrils is destroyed within the apical scar of the keratoconus.

^{ 27 }Therefore, the effect induced by the ICRS in keratoconus may be different from the effect induced in normal corneas, because the structural properties of the corneal collagen framework are also different.

^{ 3,5,7 –22,28 }A nomogram based on objective data or on an accurate mathematical model characterizing the ICRS effect has not yet been developed or reported. Different limited approaches have been proposed as nomograms for ICRS implantation in keratoconus, some of them based on spherical equivalent refraction or on the subjective appearance of the corneal topographic profile (decentered or not decentered cones). Good visual and refractive outcomes have been reported with all of them.

^{ 3,5,7 –22,28 }However, there are still anecdotal cases of ICRS implantation with minimal keratometric reductions or with no keratometric effect, despite the indications provided by these nomograms. It has been demonstrated that the preoperative manifest refraction or best corrected visual acuity are factors with a limited ability to predict the postoperative visual outcome.

^{ 29 }In contrast, corneal aberrometry has been found to have a great potential for predicting postoperative visual outcome.

^{ 2 }Therefore, there is a need for readjusting the nomograms by using objective clinical data or more complex mathematical corneal models to obtain more predictable results.

^{ 30 }The Alió-Shabayek

^{ 31 }classification was used for grading keratoconus in those cases in which corneal aberrations were evaluated. In all cases ICRS implantation was indicated because of the existence of reduced BSCVA and/or contact lens intolerance.

^{ 32 }In this study, the following topographic data were evaluated and recorded with all corneal topographic devices: corneal dioptric power in the flattest meridian for the 3-mm central zone (K1), corneal dioptric power in the steepest meridian for the 3-mm central zone (K2), and mean corneal power in the 3-mm zone (KM).

_{3}

^{±1}), coma-like RMS (computed for third-, fifth-, and seventh-order Zernike terms), spherical-like RMS (computed for fourth- and sixth-order Zernike terms), and higher order residual RMS (computed considering all Zernike terms except those corresponding with primary coma and spherical aberration). The corresponding Zernike coefficient for primary spherical aberration (Z

_{4}

^{0}) was also reported with its sign.

^{ 2,4,6 –8,13 }

^{ 2,8 }In 24 (33.3%) eyes, only one ring segment was implanted, whereas in the remaining 48 (66.7%) eyes, two segments were necessary.

^{ 2,33 }No explantations or reposition of the ring segments were needed during these first 3 months after surgery.

*t*-test for paired data was performed for all parameter comparisons between preoperative and postoperative examinations or consecutive postoperative visits. When parametric analysis was not possible, the Wilcoxon rank sum test was applied, to assess the significance of differences between preoperative and postoperative data, with the same level of significance used in all cases (

*P*< 0.05; SPSS ver. 15.0 for Windows; SPSS, Chicago, IL).

^{ 31 }14 (26.9%) eyes had a grade I cone, 16 (30.8%) a grade II cone, 8 (15.4%) a grade III cone, and 14 (26.9%) a grade IV cone.

*P*< 0.01, Wilcoxon test). A statistically significant mean improvement in UCVA of 3 lines was noted (

*P*< 0.01, Wilcoxon test). BSCVA showed a statistically significant improvement of ∼1 line (

*P*< 0.01, Wilcoxon test).

Parameter (Range) | Preop | Postop | P * |
---|---|---|---|

UCVA, logMAR | 0.99 ± 0.67 | 0.70 ± 0.35 | <0.01 |

(0.10 to 2.78) | (0.13 to 1.30) | ||

Sphere, D | −3.70 ± 4.82 | −2.51 ± 4.46 | <0.01 |

(−21.00 to +4.00) | (−17.50 to +3.25) | ||

Cylinder, D | −4.07 ± 2.40 | −2.86 ± 1.91 | <0.01 |

(−9.50 to 0.00) | (−9.00 to 0.00) | ||

SE, D | −5.64 ± 5.00 | −3.99 ± 4.50 | <0.01 |

(−22.25 to +1.00) | (−19.00 to +2.00) | ||

BSCVA, logMAR | 0.36 ± 0.27 | 0.27 ± 0.23 | <0.01 |

(0.00 to 1.30) | (0.00 to 1.30) |

*P*< 0.01, Wilcoxon tests; Fig. 2). Specifically, a mean central flattening effect of 2.45 ± 2.45 D was obtained. Regarding corneal aberrations, a significant decrease was observed in the RMS values for coma-like aberrations (

*P*= 0.03, Wilcoxon test) and corneal astigmatism (

*P*= 0.01, Wilcoxon test) at 3 months after surgery (Table 2). The primary spherical aberration term became, on average, more positive after surgery, although the change did not reach statistical significance (Table 2).

Parameter | Preop | Postop | P |
---|---|---|---|

Higher order RMS, μm | 3.73 ± 1.97 | 3.24 ± 1.44 | 0.09* |

(0.78 to 10.21) | (1.08 to 8.21) | ||

RMS for corneal astigmatism, μm | 3.21 ± 2.16 | 2.50 ± 1.73 | 0.01* |

(0.33 to 10.86) | (0.20 to 8.69) | ||

Primary coma RMS, μm | 3.11 ± 1.75 | 2.66 ± 1.46 | 0.13† |

(0.33 to 8.38) | (0.65 to 7.65) | ||

Z40, μm | −0.24 ± 0.94 | −0.01 ± 0.73 | 0.09† |

(−2.06 to 2.69) | (−1.58 to 1.67) | ||

Residual RMS, μm | 1.56 ± 1.28 | 1.53 ± 0.68 | 0.85† |

(0.47 to 8.10) | (0.46 to 3.18) | ||

Spherical-like RMS, μm | 1.25 ± 0.89 | 1.21 ± 0.58 | 0.78* |

(0.24 to 6.38) | (0.33 to 2.73) | ||

Comalike RMS, μm | 3.46 ± 1.86 | 2.94 ± 1.45 | 0.03* |

(0.40 to 9.96) | (0.84 to 7.94) |

_{3}

^{±1}; primary spherical aberration, Zernike term Z

_{4}

^{0}; residual aberrations, all Zernike terms, except Z

_{3}

^{±1}and Z

_{4}

^{0}; spherical-like aberrations, fourth- and sixth-order Zernike terms; comalike aberrations, third- and fifth-order Zernike terms.

Postoperative Outcome or Change | Correlation | Correlation Coefficient | P |
---|---|---|---|

Change in sphere, D | IST | 0.279 | 0.02 |

Preop sphere | −0.62 | <0.01 | |

Change in mean keratometry, D | IST | −0.435 | <0.01 |

SST | −0.484 | <0.01 | |

IST-SST | 0.300 | 0.02 | |

Preop LogMAR UCVA | −0.432 | <0.01 | |

Preop sphere | 0.537 | <0.01 | |

Preop LogMAR BSCVA | −0.365 | 0.01 | |

Preop mean keratometry | −0.480 | <0.01 | |

Change in the RMS value for corneal residual HOA, μm | IST | 0.501 | <0.01 |

SST | 0.437 | <0.01 | |

Change in manifest cylinder, D | Preop cylinder | −0.64 | <0.01 |

Change in spherical-like RMS, μm | Preop primary coma RMS | −0.382 | 0.01 |

Preop coma-like RMS | −0.375 | 0.01 |

_{p}is the preoperative cylinder, DifKM is the change in mean keratometry after surgery, and DifIST is the difference between the thickness of the inferior and superior ring segments.

Models | R ^{2} | Adjusted R ^{2} | P | Cook's Distance | Residuals Statistics | Durbin-Watson Test | Multicolinearity Tolerance |
---|---|---|---|---|---|---|---|

Model 1 | 0.84 | 0.83 | <0.01 | 0.09 ± 0.26 | ≤100 μm 91.23% | 1.75 | 0.84–0.88 |

≤50 μm 63.16% | |||||||

Model 2 | 0.62 | 0.59 | <0.01 | 0.09 ± 0.26 | ≤100 μm 87.72% | 1.75 | 0.84–0.88 |

≤50 μm 63.16% | |||||||

Model 3 | 0.91 | 0.90 | <0.01 | 0.12 ± 0.31 | ≤100 μm 92.68% | 1.96 | 0.57–0.77 |

≤50 μm 75.61% | |||||||

Model 4 | 0.64 | 0.59 | <0.01 | 0.12 ± 0.31 | ≤100 μm 90.24% | 1.96 | 0.57–0.77 |

≤50 μm 68.29% |

_{p}, DifKM, and DifIST are as defined for models 1 and 2; and DifRMSHOA is the change in the RMS value for corneal higher order aberrations.

*P*≥ 0.09) and the absence of influential points or outliers. The lack of multicollinearity and the independence of the residuals were also confirmed (Table 4).

*r*= −0.32,

*P*= 0.04; models 3 and 4,

*r*= −0.47,

*P*< 0.01). In addition, statistically significant differences in some corneal aberrometric coefficients were found between cases with residuals >50 μm and those with residuals ≤50 μm for the first two models (model 1 and 2, RMS for astigmatism

*P*= 0.02, RMS for residual higher order aberrations

*P*= 0.01, RMS for coma-like aberrations

*P*= 0.02; Wilcoxon test). In the two linear models that involved corneal higher order aberrations, differences between cases with residuals >50 μm and those with residuals ≤50 μm were near the limit of statistical significance for the RMS values corresponding to corneal astigmatism and spherical-like aberrations (

*P*= 0.06, Wilcoxon tests; Fig. 5). It should be noted that only 13 cases had residuals >50 μm, whereas the remaining cases had lesser residuals. This trend observed for corneal astigmatism and spherical-like aberrations was also observed in the other two linear models.

^{ 1 –15,20,28 }This keratometric reduction was the main reason for the change in refraction and the increase in UCVA. However, this keratometric change was dependent on several preoperative factors as keratometry and BSCVA. As mentioned, the ring segments implanted in the midperiphery have been shown to induce a shortening of the central arc length (arc shortening effect [Silvestrini T, et al.

*IOVS*1994;35:ARVO Abstract 2023]) and then a flattening of the central portion of the anterior corneal surface.

^{ 23,24 }Regarding corneal aberrometric changes, significant changes were found in corneal astigmatism and coma-like aberrations. This aberrometric improvement could be in relation with the significant improvement found on average in logMAR BSCVA. It should be remembered that primary coma has been demonstrated to have a very negative impact on visual acuity due to the optical blur that it induces.

^{ 33 }In a study by our research group, a significant reduction of higher order aberrations was found after KeraRing implantation, using the femtosecond laser technology in keratoconus, but only in those eyes with a magnitude of coma aberration larger than 3 μm.

^{ 8 }In addition, it should be mentioned that a nonsignificant change in the primary spherical aberration toward less negative values was observed. This change in primary spherical aberration and also the reduction in coma-like errors were consistent with the reduction of the localized corneal steepening that was present in the keratoconic eyes.

^{ 25,26 }Besides these moderate relationships, we found that some clinical changes also correlated significantly with some preoperative conditions, as the magnitude of the spherocylindrical error or the corneal curvature. Therefore, it seems clear that some factors influence the visual and refractive outcomes achieved with the KeraRing segments. In other words, this process cannot be represented by means of a simple linear model with two variables. The effect achieved with each KeraRing segment is a multifactorial process depending on the ocular preoperative conditions and on the thickness of the implant. It should be remembered that all segments were implanted according to the same surgical criteria (inner and outer diameters of 4.8 and 5.7 mm, respectively, and ring placement at ∼80% of the depth of the cornea). The diameter and the depth of the implant are also factors in the final effect achieved with the ring segments,

^{ 24 }but these factors have not been modified in the present study. It should be considered that corneal changes induced by the ICRS must be in relation to the structural properties of the collagen framework in the corneal stroma. The stroma accounts for 90% of corneal thickness, and evidently its mechanical properties define, for the most part, the mechanical properties of the whole corneal structure. In the normal cornea, there is a preferred orientation of collagen lamellae along the horizontal and vertical directions, but this trend is maintained to within approximately 1 mm from the limbus, where a circular or tangential disposition of fibrils occurs.

^{ 34 }However, this well-organized lamellar structure is lost when the corneal tissue degenerates, as happens in keratoconus.

^{ 27 }The regular orthogonal arrangement of the collagen fibrils is destroyed within the apical scar of the keratoconus.

^{ 27 }Therefore, the ICRS effect in keratoconus seems to be a more complicated phenomenon that needs a more complex mathematical model. The current investigation was conducted to define an approach to devising such a model through multiple linear regression analysis. A more accurate model should be defined in the future, considering clinical parameters and also accurate measurements of the structural and mechanical properties of the corneal tissue.

^{ 25,26 }However, this effect was limited by the preoperative manifest cylinder of the eye, which seems to be in relation with the instability of the keratoconic cornea. The third implicated factor, the difference in thickness between the inferior and superior segments, represents the interaction between both ring segments, and it correlated positively with the thickness of the inferior ring segment, but inversely with the thickness of the superior ring segment. This result means that the most positive combination for KeraRing consists of a superior ring segment that is thinner than the inferior implant. The goodness of fit of these models was confirmed by testing the homoscedasticity of the models, the correlation between residuals and the multicollinearity. The predictability of the superior ring segment thickness model was good, with 91.23% of unstandardized residuals <100 μm and 63.16% ≤50 μm. However, the predictability of the inferior ring segment thickness model was moderate, with 87.72% of unstandardized residuals <100 μm and 63.16% ≤50 μm.

^{ 2,35 }It should be considered that larger amounts of corneal higher order aberrations are present in the more advanced keratoconic corneas.

^{ 31,36 }In such cases, the biomechanical alteration seems to be more pronounced. Indeed, in previous work, our research group found a significant correlation between the corneal resistance factor (CRF) parameter measured with the ocular response analyzer (ORA; Reichert) and the magnitude of corneal spherical-like aberrations.

^{ 36 }All the topographic and aberrometric alterations in keratoconic eyes are the consequence of the biomechanical changes that occur in the corneal structure. Therefore, the improvement in the predictability of the models for the ring segment thicknesses, when the corneal higher order aberrations are included, could be the consequence of introducing an additional factor in relation to the corneal biomechanical status. In other words, the introduction of the aberrometric factor could be an indirect manner of considering part of the corneal biomechanical factor. In any case, this indirect contribution of aberrometry to corneal biomechanics is limited, and it does not account for the total biomechanical effect. The predictability of the superior ring segment thickness model including corneal aberrations was quite good, with 92.68% of unstandardized residuals <100 μm and 75.61% ≤50 μm. The predictability of the model for the thickness of the inferior segment was also good but a little bit more limited, with 90.24% of unstandardized residuals <100 μm and 68.29% ≤50 μm.

^{ 36 }Therefore, it seems clear that the corneal biomechanical status is a limiting factor for the developed nomogram and would have significant relevance in more advanced keratoconus. Currently, there is no nomogram for ICRS implantation that takes into account the specific biomechanical properties of the cornea. One reason for this is that the analysis of the corneal biomechanical properties of the cornea in vivo is not an easy task in clinical practice. To this date, only one device has been developed for the clinical evaluation of corneal biomechanics: the ORA (Reichert).

^{ 37 }This device is an adaptation of a noncontact tonometer, which allows the measurement of the intraocular pressure as well as two new metrics referred to as corneal hysteresis (CH) and CRF. The exact differences between these two biomechanical parameters, as well as the exact contributions of the elastic and viscous components to the magnitude of these parameters are not yet completely understood. However, although we do not know the exact physical meaning of these parameters, CH and CRF have been shown to be very useful for characterizing the biomechanical properties of the cornea in the clinical practice.

^{ 38 }Indeed, as previously mentioned, the keratometry and the magnitude of corneal higher order aberrations have been shown to correlate inversely with the CRF in keratoconus.

^{ 36 }

^{ 39 }To this date, there are no published studies in which the visual, refractive, and keratometric outcomes were compared after ICRS implantation via some of these incision locations. Theoretically, the ideal location would be the steepest corneal meridian, as most surgeons do currently, because this kind of incision would reduce the corneal power of the steepest meridian, and it would increase the flattest keratometric reading. This solution would minimize the corneal and manifest astigmatism. However, significant reductions in manifest cylinder have also been achieved in eyes with the incision located elsewhere.

^{ 39 }In our study, an incision on the steepest corneal meridian was used in most of the cases, with very few cases with the incision distant from the steepest corneal meridian. We think that this factor caused very little variability in the outcome. Indeed, we have obtained predictable models for the selection of the ring segments according to the intended corneal change and the preoperative conditions of the eye. Furthermore, the most important issue was the position of the ring segments, and that was always parallel to the flattest corneal meridian.