Patients Say the Darndest Things
IER Matrix Update: Adding Another Silicone Hydrogel
Two hour exposure to silicone hydrogel lens/solution combinations does not predict three-month clinical findings.
By Nicole Carnt, BOptom (Hons); Vicki Evans, PhD, BOptom; Brien A. Holden, PhD, DSc, OAM; Thomas J. Naduvilath, PhD; Daniel Tilia, BOptom (Hons), MOptom; Eric B. Papas, PhD, BSc, MCOptom, DipCL; & Mark D.P. Willcox, PhD
|Ms. Carnt is a senior research optometrist at the International Clinical Trials Centre for the Institute for Eye Research, Sydney, Australia.|
Dr. Evans is Head of Clinical Research at the International Clinical Trials Centre for the Institute for Eye Research, Sydney, Australia.
Prof. Holden is scientia professor of the University of New South Wales, chief executive officer of the Institute for Eye Research, deputy CEO and director of commercialization of Vision CRC, executive chair of the International Centre for Eyecare Education and executive chair of Optometry Giving Sight.
Dr. Naduvilath is a biostatistician and clinical database manager at the Institute for Eye Research, Sydney, Australia.
Mr. Tilia is a research optometristat the International Clinical Trials Centre for the Institute for Eye Research, Sydney, Australia, part-time clinical supervisor at the School of Optometry and Vision Science, University of New South Wales, and works part-time in private practice.
Dr. Papas is executive director of Research and Development at the Institute for Eye Research and Vision Cooperative Research Centre in Sydney, Australia.
Prof. Willcox is a professor at School of Optometry and Vision Science, University of New South Wales, chief scientific officer of the Institute for Eye Research and executive director of Science and Core Capabilities of the Vision CRC, Sydney, Australia.
The first IER Matrix Study report that appeared in the September 2007 issue of Contact Lens Spectrum detailed the corneal staining results from approximately 640 patient lens/solution combinations in 16 daily wear studies of four commercially available silicone hydrogel lenses with four contact lens solutions, each over a three-month period. This update adds data from the studies carried out with Night & Day (CIBA Vision) and discusses the contentious issue of the validity of a two-hour exposure test.
The IER Matrix Study reported on solution-induced corneal staining (SICS), which we defined as punctate fluorescein staining associated with the lens care regimen affecting at least four of five areas of the cornea, presenting as either diffuse (spread over the cornea) or as a peripheral/annular band around the limbus. We termed it solution-induced corneal staining rather than lens/solution 'combination' staining because this type of staining was virtually absent with the same silicone hydrogels when tested with hydrogen peroxide. It therefore appeared to be a multipurpose solution (MPS) related phenomenon, occurring to a greater or lesser extent depending on the care solution/silicone hydrogel combination used.
Institute for Eye Research (IER) researchers have previously shown that SICS is associated with a three-times increased risk of corneal inflammation and decreased comfort. Further publications will detail the results of all the variables collected in the IER Matrix Study as well as further risk factors and associations.
The Night & Day lens studies were completed with 40 patients in each of the four lens/solution combinations, contributing an additional 160 patient lens/solution combinations, bringing the total of patient lens-wearing combinations to approximately 800. Table 1 presents the results.
Andrasko has garnered significant attention with his Staining Grid data as well as his statement that the IER Matrix Study results support the validity of Staining Grid data as a clinical predictor. We will evaluate the data from the two studies and demonstrate why we feel this may not be the case.
IER Matrix IER Matrix methodology for the Night & Day cells was the same as we reported in the previous study. We determined the percentage of patients who had SICS per month during the first three months of wear for each lens/solution combination. The 50 percent confidence intervals for the aggregate of all patients was calculated and lens/solution combinations were colored by which quartile they fell into, with orange designating the highest, yellow the middle two quartiles and green the lowest.
Staining Grid Study Data The Andrasko Staining Grids (www.staininggrid.com) are derived from the average percentage area of the cornea that exhibits staining for the worst eye after an overnight soak and two-hour wear of each lens/solution combination. However, you can obtain the percentages of patients affected (as opposed to the area affected) in the Andrasko two-hour studies by clicking on each cell in the Andrasko grid on the Staining Grid site.
Table 1. IER Matrix including Night & Day lenses (percent of patients per month).
Results and Discussion
1. Updated Results Table 1 presents the IER Matrix updated results. The Acuvue Oasys (Vistakon) with Aquify (CIBA) rate was updated from 2.5 percent in the first report to 2.6 percent in this report because of the availability of additional visit data. The Oasys with Opti-Free Express (Alcon) rate now appears orange (highest quartile), whereas in the original IER Matrix staining table it was yellow (middle two quartiles). This is because both the overall rate for all patients and confidence intervals changed slightly with the addition of the Night & Day results.
2. Night and Day Night & Day lenses are in the lowest quartile with Clear Care (CIBA) and Aquify and in the highest quartile range for both Opti-Free products, along with PureVision (Bausch & Lomb) and Oasys.
One potentially important factor related to SICS is that lens wettability and deposition seem to affect surface adsorption of solution components. A recent analysis of the IER Matrix data showed that patients who had experienced SICS had poorer lens surface characteristics at dispensing and at the time of the SICS event compared to patients who had not shown signs of this type of corneal interaction. Willcox et al (2007) showed that a higher level of cholesterol and protein has been extracted from PureVision lenses compared to other lens types in the grid, which may suggest a hypothesis that the SICS response is partly associated at least with the type and amount of lens surface deposition.
3. Andrasko Percentage Area versus Andrasko Percentage of Patients Affected Table 2 shows the difference between the Andrasko staining percentage area and the Andrasko percentage of patients who developed more than 10 percent area of staining.
Presenting the data as the percentage of patients affected puts the solutions in very different categories from those of the percentage area grid promoted by Andrasko. For the same average percentage corneal area affected, the percentage of patients affected can vary significantly. For example, examining the five combinations in Table 2 for which the area of the cornea affected was 3 percent*, the percentage of affected patients varied up to 10 percent. In fact, Andrasko's percentage of patients affected averages 2.6 times higher numerical values than the Andrasko area affected (Table 2). Likewise, a seemingly small difference in the percentage of cornea affected may correspond to substantial disparities in the number of patients affected. For example, this is evident with Opti-Free Express, for which the range of area affected changed by only 3 percent across all lens types but the percentage of patients affected varied up to 13 percent. Another example is Oasys with Opti-Free Replenish, which shows 5 percent area affected versus 13.3 percent of patients affected.
Because the Andrasko Staining Grid shows the percentage area staining rather than the percentage of patients affected, it 'masks' lens/solution combinations that produce corneal effects in a significant percentage of people.
4. IER (Three Months) versus Andrasko (Two Hours) Percentage of Patients Affected Table 3 compares the percentage of patients who developed staining for the Staining Grid two-hour result and the IER Matrix three-month result using the 50 percent confidence interval analysis for the two studies.
The Andrasko two-hour data in Table 3 over represents Aquify and under represents the Opti-Free products, particularly Opti-Free Express, compared to the three-month data of the IER Matrix. This may be because a high percentage of the Staining Grid subjects were previous 'Opti-Free survivors.'
Table 2. Andrasko Staining Grid percentage area versus percentage of patients affected (from www.staininggrid.com).
Table 3. IER Matrix versus Staining Grid for percentage of patients affected. Colors indicate 50 percent confidence interval analysis applied to each study with the lowest green, middle two quartiles yellow and highest orange.
5. Problems With the Andrasko Two-Hour "Prediction" of the Percentage of Patients That Will be Affected Andrasko recently highlighted the percentage of patients who had more than 10 percent area of staining in the two-hour studies and compared it to the percentage of patients per month experiencing SICS in the IER Matrix and claimed that the two hour studies could predict the IER Matrix long-term staining results.
To evaluate this claim, we obtained the Andrasko percentage of two-hour patients affected from the Staining Grid Web site and plotted that against the percentage of IER patients affected over three months (Figure 1). One data point from Table 3, PureVision with Aquify, was a clear outlier in the Andrasko data (reported to affect 63 percent of people, nearly three times the number of people actually affected in the IER Matrix Study). Statistically, this data point was also a significant outlier for the regression (p<0.05), so we removed this point from the analysis.
We believe the claim that the Staining Grid data is a good predictor of the IER Matrix findings is incorrect for the following reasons:
• While the R Square value for the equation in Figure 1 indicates a correlation, the precision around the estimated values determines the equation's usefulness as a future predictor. In Figure 1, the relative precision ranges from ±22 percent (acceptable) to ±220 percent (poor) for the Staining Grid rates. Nearly half the Andrasko data points lie outside the 95 percent confidence interval.
Figure 1. Correlation of Staining Grid and IER Matrix data excluding one outlier (PureVision with Aquify). The orange line indicates where the data would lie if the Staining Grid two-hour data accurately correlated with IER Matrix three-month data.
• The most obvious prediction failures are the four cases in which the Andrasko result was zero whereas IER rates ranged up to 6 percent as well as the six cases in which the IER data was approximately 6 percent to 7 percent and the Staining Grid varied from 0 percent to 17 percent.
Some of the factors that may lead to a higher IER rate of SICS (for three months compared to the Staining Grid two hours) include cumulative dosage and deposits.
Another likely factor biasing the Staining Grid data for a lower result is the solution experience of Andrasko's subjects. Lower Staining Grid SICS rates at two hours (compared to the IER studies) would be obtained if patients have already adapted or were pre-selected by use of a particular lens care product. For example, if Staining Grid patients who routinely use Opti-Free products were exposed to the same product in a two-hour trial, their response may be skewed in a favorable direction as they are essentially successful Opti-Free survivors. Andrasko does not report on protecting against this bias. Conversely, 88 percent of patients in the IER Matrix study were new to contact lens wear or used a different lens care solution prior to participating in a block of the matrix.
Night & Day lenses behave similarly in the IER Matrix to Acuvue Oasys and O2Optix, falling into the lowest quartiles for Aquify and Clear Care and in higher quartiles for Opti-Free products.
We view the percentage area of corneal staining results from the Staining Grid with caution, as they underestimate the more widespread influence some lens/solution combinations have. Our analysis indicates that presenting results of the percentage of patients affected is a more appropriate and preferred method.
The cumulative effect of solution usage as well as the interaction with deposits, lens surface wetting and lens case interactions (factors missed in the Staining Grid study) are variables associated with the long-term SICS response and should not be overlooked when assessing SICS rates.
It's also likely that the differences among the performances of lens types with various solutions depend on the uptake and release kinetics of the solutions within the different polymers as well as the surface adsorption of solution components. Further study of the IER Matrix data will likely help identify such key factors.
While minimal corneal staining often occurs in contact lens wear, moderate amounts of controllable staining should always be avoided, especially given the links that have been established between SICS and low-grade corneal inflammation and discomfort.
A need clearly exists for further investigation into the mechanism of SICS. In the meantime, we believe clinicians should be aware that our analysis demonstrates that two-hour percentage area studies do not reliably predict the long-term SICS responses for the majority of lens and solution combinations evaluated. We believe the IER Matrix data is more clinically relevant and represents a truer, longer-term indication of the 'real world' clinical situation. CLS
For references, please visit www.clspectrum.com/references.asp and click on document #148.
*In the published March 2007 issue of Contact Lens Spectrum, this sentences incorrectly said �3 percent or less.� This was a request from the authors.
Contact Lens Spectrum, Issue: March 2008