Predicting contact lens base curve using corneal topography in keratoconus

Steven Mathews, O.D., Ph.D., F.A.A.O.* K. Tom Xu, Ph.D.+ Jay Bradley, M.D.* Joel George, M.D.* Texas Tech University HSC *Department of Ophthalmology & Visual Science +Department of Family & Community Medicine 3601 4th Street Lubbock, Texas 79430

Abstract Purpose: Rigid contact lenses are required to improve visual acuity in all but the mildest of keratoconic eyes. The current literature offers the novice keratoconic rigid lens fitter a confusing collection of recommendations on how to initiate the fitting of a keratoconic cornea. The aim of our study is to offer the keratoconic contact lens fitter improved guidance on picking the first trial lens base curve. Methods: Our retrospective study gathered data from 200 previous keratoconic rigid lens fits. We used multivariate linear regression to investigate the relationship between final base curve data and prefit corneal topography parameters to determine which variables would have best predicted the final base curve. In addition, we also used univariate regression to investigate the recommendations proposed by previous studies using our own data for comparison purposes. Results: The best predictors (R2=0.786) on our keratoconic eyes that had been measured by a Keratron Scout were a linear combination of Steep Sim-K and Steep Normalized Curvature. The best predictors (R2=0.682) on our keratoconic eyes that had been measured by a Tomey TMS-1 were a linear combination of Flat Sim-K, Asymmetry Index, and Irregularity Index. The R2 values from our univariate linear regressions ranged from a low of 0.286 from the Tomey Flat K data to a high of 0.667 from the Scout Average K data. None of the univariate regressions had a slope near 1 or a Y intercept near zero. Discussion: The regression formulas from this study should help a novice keratoconic rigid lens fitter get a better start at learning the difficult task of fitting keratoconic corneas. In addition, experienced keratoconic rigid contact lens fitters might be able to use these formulas to arrive at a final best fit base curve more quickly than previously.

Introduction Keratoconus is a progressive non-inflammatory thinning of the cornea1, 2. Because the eye is essentially a collagen sac filled with a pressurized fluid, this thinning produces a corneal ectasia that disrupts the normal regular optics of the cornea. While spectacles and soft contact lenses can often provide adequate visual acuity in mild keratoconus, rigid contact lenses become the optical correction of choice once the disease becomes more severe3-5. Achieving a stable rigid contact lens fit on a keratoconic cornea that also provides good vision and maintains corneal health can be arduous. This point is perhaps best illustrated by three papers from the late 1980s6-8. Smiddy and colleagues8 reviewed the charts of 115 consecutive keratoconus patients seen over a five year period at the Wilmer Eye Institute. Most of these patients had been referred by physicians for keratoplasty because of previous unsuccessful contact lens wear. Smiddy et al. found that 87% of these contact lens failures could be fit successfully at the Wilmer Cornea Service. In addition, 69% of patients successfully fit did not require keratoplasty over an average follow-up period of 63 months. Similarly, Belin and colleagues6 did a retrospective case analysis of patients seen over a two year period. All 33 of the keratoconus patients in this study had been referred for surgery due to contact lens intolerance. 47 of 61 eyes (77%) from these patients were successfully refitted with rigid contact lenses. Later that same year, the same group7 extended their previous study to two and a half years and found essentially the same result. Again, all keratoconus patients had been referred for surgery due to contact lens intolerance. In this second study, 80% of 71 eyes were successfully refitted with rigid contact lenses. Taken together, these three studies strongly suggest that the decisive factor driving the majority of referrals for cornea transplant surgery was the level of contact lenses fitting skill present at the referring offices. Moving forward from when these studies were published to current times, it is quite possible that the percentage of surgical referrals that are able to be refit with rigid contact lenses has decreased somewhat due to improvements in keratoconic lens designs and rigid contact lens materials. However, it is also likely that there still could be considerable improvement in the keratoconic rigid contact lens fitting skill in the eye care professions. One of the pieces of information that could improve practitioner confidence and fitting success is the ability to more accurately choose the initial rigid lens base curve based on objective information like keratometry or corneal topography. Unfortunately, the guidance in the literature does not inspire much confidence. A search for assistance in the above three papers finds two recommendations, but no supporting evidence or references. Smiddy et al8 started their fits with a base curve equal to the flatter keratometric reading, but offered no references or data to support this choice. Belin et al.6 stated that their “initial lens selection was guided by the keratometry reading and the configuration of the cone”, but did not elaborate. Fowler et al.7 suggested starting with a base curve equal to the flat K measure, but offered no supporting evidence of their recommendation. Looking at a sample of other significant papers on the topic of fitting keratoconic corneas, we find a similar quality of guidance. In 1972 Soper and Jarrett9 published their influential early work describing the Soper trial lens set. They only stated that the first trial lens is “one which relates to the degree of conical progression”, without further elaboration or support. In 1978 Caroline, McGuire, and Doughman10 first described the McGuire fitting set and recommended the first base curve should equal the average keratometry reading, but again offered no data or references. Korb and colleagues’ influential 1982 paper11 primarily suggested that flat fitting rigid contact lenses tend to promote corneal scarring. In this paper they recommended first base curve that was equal to the steep keratometric reading with no supporting data or reference. In 1997 Mandell12 published an update to the keratoconus chapter in his well regarded book entitled “Contact Lens Practice”. He stated that videokeratography or keratometry can serve as a guide to picking the first trial lens, but did not elaborate. In absence of topographical or keratometric data, Mandell recommended to simply start flat. Looking at this list, it is easy to understand why a novice fitter of keratoconic corneas might desire better guidance.

In the late 1990s two papers13, 14 endeavored to add more science to the task of picking the first trial lens base curve. Edrington and CLEK study group13 fit 53 eyes in 30 patients. Their Methods section stated that their study used the average keratometry reading to pick the first base curve, but they also offered no prior evidence, rational or references related to this choice. However, they improved upon the papers cited above by presenting the keratometric readings and final base curve fit on each of the eyes in their Table 2. Unfortunately, the only analysis reported on this data was to average the keratometric and base curve data. They found that the average flat K was 46.18, the average steep K was 49.74, and the average final base curve was 48.94. While this analysis loosely supports their method of using the average K reading to pick the first trial lens base curve, it assumes that the relationship between the keratometric readings and final base curve is linear with a slope of one without any supporting analysis. We will show later in our paper that this assumption is likely not correct. Szczotka and Thomas14 studied 28 keratoconic eyes previously fit with rigid contact lenses by the first author. Their corneal topographer was an EyeSys2000. They used linear regression analysis to look at the relationship between final base curve dependant variable and the independent variables axial steep SIM-K, axial apex curvature, instantaneous steep SIM-K, and instantaneous apex curvature to learn which topographical variable best predicted final base curve. The authors presented r2 values that ranged from .6949 for base curve versus instantaneous steep SIM-K to .8838 for the base curve versus instantaneous apex curvature. Szczotka and Thomas concluded that axial steep SIM-K best predicted the RGP base curve if all of the cones are analyzed together. If the peripheral cones are separated from the central cones then the best predictors of final base curve were the axial apex curvature and axial steep SIM-K respectively. However, their r2 values are very high and a look at their method section yields a potential explanation. They did corneal topography 15 minutes after removing the already fit and adapted-to rigid contact lens. There is a possibility that the patients’ rigid contact lenses exerted some molding influence that, at least partially, imposed the radius of the base curve on the keratoconic corneas and, therefore, artificially improved their r2 values. In addition, their r2 values were presented without the associated regression formulas. Without the formulas, their readers are not informed about how to scale the observed measurements to choose the first rigid lens base curve. In an effort to provide the novice rigid contact lens fitter with better guidance on how to pick the initial rigid contact lens base curve, we compiled pre-fit corneal topography data from a large keratoconic population and compared it to the patients’ final best-fit base curves.

Methods A retrospective chart review was done on keratoconic patients fit with rigid corneal contact lenses at an academic Contact Lens Clinic in the previous 5 years. The research followed the tenets of the Declaration of Helsinki and was approved by the local Institutional Review Board. Pre-fit Keratron Scout or Tomey TMS-1 topography data and the final best-fit base curve were gathered on 200 consecutive keratoconic eyes. Cones with the clear majority of the apex falling within the central 3mm ring by visual inspection of the axial map were classified as central. All cones that could not be clearly classed as central cones were categorized as peripheral cones. We gathered topographic variables that included flat and steep simulated keratometry readings, flat and steep normalized curvatures, Maloney best-fit sphere, irregularity index, and asymmetry index. Multivariate and univariate linear regression analyses yielded a formula and R2 value that described the relationship between the pre-fit topographical variables and the final best-fit base curve. Our fitting methods and biases The results of a study like this will depend considerably on the design of the back surface of the rigid contact lenses used to fit the patients. Therefore, a brief description of the contact lens fitting methods and biases that are used in our clinic will help the reader evaluate their ability to generalize our results to their practices. Our fitting biases are: 1) We used spherical base curve lenses to produce a three-point-touch fit with minimal apical clearance or very mild central touch. This fitting technique has been previously discussed12, 15. Our keratoconic trial lens set had base curves that changed in one diopter increments. Occasionally lenses were ordered for mild keratoconic eyes (BC