Tear Exchange under Hydrogel Contact Lenses: Methodological Considerations

Abstract

purpose. To characterize hydrogel lens tear exchange and to apply an optimized method to compare tear exchange of a marketed hydrogel lens to that measured with a prototype silicone hydrogel lens.

methods. Fluorophotometry and a nonpenetrating tracer (70-kDa FITC-dextran) were used with a single extended-wear soft contact lens (EWSCL) material on 11 subjects to characterize tear-exchange kinetics. Twenty to 30 measurements were obtained over a 30-minute period to allow accurate modeling and estimation of the several tear-exchange parameters. Calculated values included tear-replenishment rate (TRR), elimination rate (ER), and the time for 95% of the signal to be eliminated (T ₉₅). Major experiments were (1) comparison of ER under controlled and physiological conditions, (2) comparison of right and left eyes, (3) repeatability of ER and T ₉₅ on five occasions, and (4) comparison of a marketed lens (oxygen permeability [Dk] 28 × 10⁻⁹ [cm/sec][ml O₂/ml mm Hg]) to a prototype silicone hydrogel lens (Dk 140 × 10⁻⁹ [cm/sec][ml O₂/ml mm Hg]).

results. Tracer elimination behind a hydrogel contact lens (CL), up to 30 minutes after insertion, was optimally described by double-exponential kinetics. Physiological ER (5–30 minutes after CL insertion) was optimally described by single-exponential kinetics. Overall, physiological ER was 8.8% ± 3.8% per minute, and T ₉₅ was 31.0 ± 16.1 minutes (n = 76 and 72 determinations, respectively). Differences between right and left eyes in ER and T ₉₅ were not significant at the 0.05 level. No difference in ER or T ₉₅ was found between habitual and controlled blinking. Mean TRR was 0.67% ± 0.26% per blink (n = 11 determinations). No differences were shown between ER or T ₉₅ measurements over time. A prototype highly oxygen-permeable silicone hydrogel lens showed higher ER than did a marketed hydrogel lens (P < 0.01).

conclusions. Estimates of postlens tear exchange using a slit lamp fluorophotometer are similar to previously reported rates using similar fluorophotometric techniques. Fluorescent decay behind a hydrogel lens is most precisely described using a double-exponential curve equation and tear exchange may be described using ER, TRR and T ₉₅, although the T ₉₅ may be the least reliable of these measures. The technique appears capable of discriminating between lens types.

Tear exchange behind flexible contact lenses (CLs) was historically considered to have little importance, because early studies demonstrated that tear exchange is unlikely to contribute greatly to improved corneal oxygenation. ^{1 2 3} Currently, however, improved materials are available that have overcome hypoxia related to flexible lens wear, yet complications remain. ⁴  

Tear exchange appears to be important in preventing the accumulation of debris, cellular material, and metabolites behind a soft lens. ⁵ Mertz and Holden ⁶ presented the first direct evidence that the postlens debris obtained from a patient with an inflammatory reaction from an extended-wear lens was composed of mucus, epithelial cells, and neutrophils. They speculated that the presence of the debris might instigate an inflammatory reaction that becomes acute if the debris remains in situ for a sustained period. ⁶ More recently, the increased risk of corneal infection associated with overnight lens wear has again been reported, ⁷ confirming that there has been no reduction in the risk associated with hydrogel lenses over the past 10 years. ^{8 9} Limited tear exchange has been cited as a key factor in the pathogenesis of CL-related infection. ⁷  

Recent in vitro and ex vivo data studies of bacterial virulence suggest that postlens tear stagnation, perhaps related to inadequate tear exchange, may predispose the cornea to Pseudomonas aeruginosa infection. ^{10 11} Cytotoxic strains of P. aeruginosa cause epithelial cell disruption and death with prolonged cell contact. Other strains of P. aeruginosa cause epithelial cell invasion after 3 hours of contact. It is conceivable that impaired tear exchange delays the removal of infected corneal epithelial cells and provides the basis for bacterial invasion of corneal tissue. 

In summary, tear exchange may have relevance to successful flexible lens wear in the absence of hypoxia, although direct evidence of causal relationships to infection and inflammation has not been found. Before examination of these relationships, accurate and reliable techniques for measuring tear exchange are required. 

There are few complete reports of postlens tear exchange, particularly in relation to current hydrogel materials. ^{2 3 12} Sorensen et al. ¹³ examined tear flow in the presence of a hydrogel lens, but did not characterize the kinetics of tracer elimination from beneath the material. In addition, tear exchange repeatability and reproducibility remain largely undocumented, and several methodological questions persist. 

In two early reports, a fluorescent tracer (Fluorexon; molecular weight 710; Dioptic Laboratories, Markham, Ontario, Canada) was used to estimate tear exchange. ^{2 3} In subsequent studies, the tracer size was increased to 4.4 kDa (FITC-dextran). ¹² Fluorexon is absorbed into a hydrogel lens within 10 minutes, ¹⁴ which likely occurs during the 8- to 12-minute measurement time of the experiments. ² Moreover, Fluorexon probably penetrated the cornea, because much larger fluorescent compounds such as FITC-dextrans have demonstrated rabbit scleral penetration in vitro. ^{15 16} Tissue penetration overestimates the elimination rate (ER) by as much as 25% compared with a nonpenetrating dye. ¹⁷  

Another consideration may be the manner in which the decay data are collected. Polse et al., ² Wagner et al., ³ and McNamara et al. ¹² used a controlled blink rate to calculate the tear-replenishment rate (TRR, denoted as percentage of volume exchange per blink) based on concepts put forth by Cucklanz and Hill in 1969. ^{18 19} Although the rate used, 15 blinks per minute, is physiological, it is not clear whether controlling the blink rate affects tear exchange under typical wearing conditions under which wearers blink at will. 

With the renewed interest in accurately quantifying postlens tear exchange, the purpose of this study was to characterize the behavior of a nonpenetrating fluorescent tracer behind the lens. The investigation reported herein included an examination of decay kinetics and the development of useful models, comparison of tear mixing under controlled and noncontrolled blinking conditions, and a comparison of tear exchange using an optimized approach between a current hydrogel lens and a prototype highly oxygen-permeable silicone hydrogel lens. 

Methods

Instrumentation

A photographic biomicroscope (model FS-2; Nikon Corp., Tokyo, Japan) was selected as the base instrument for the fluorophotometer system. This system has excellent optics and a built-in beam splitter for convenient switching to a photomultiplier from the normal observation arrangement. The intensity of the slit lamp illumination was stabilized using an optical feedback system to maintain constant voltage. One filter in the light tower filter wheel was replaced with a custom fluorescein excitation filter (490RDF10; λmax 488 nm, full width at half maximum [FWHM] 10 nm; Omega Optics, Brattleboro, VT). The emitted light was transmitted through the standard instrument optics onto a side-on photomultiplier tube (PMT) mounted on the camera port (PMT housing, model 70680; PMT, model 77349; Oriel Corp., Stratford, CT). The emitted light passed through a 10-mm diameter black metal mask (to reduce stray light) and an emission filter (530RDF40,λ max 530 nm, FWHM 30 nm, Omega Optics). The PMT output was directed to an amplifier (model 7070, Oriel Corp.) and the final signal fed to a standard strip chart recorder (Fig. 1) . 

For study measurements, the biomicroscope magnification was set at 30×, using a beam dimension of 1.5 mm wide by 6.0 mm high. This resulted in a sampling area presented to the PMT of 3.8 mm². Combined with the lamp angle of 50°, the 30× magnification allowed a minimal depth measurement diamond. The amplifier was checked over the range of settings to× 10⁻⁵ from ×10⁻⁸ and found to be linear. In terms of sensitivity, our experimental data were at signal amplitudes of at least two orders of magnitude above the equivalent amplitude of the background fluorescence attributable to the cornea. Whereas background signal attributable to the crystalline lens is higher than the cornea, due to the small measurement diamond achievable, the small depth of focus (the difference in focus between the tear film and the stroma is easily discernible), and the obliquity of the microscope and lamp angles with our configuration, we were able to effectively eliminate background signal from the crystalline lens. The dark current of the system (as measured with illumination lamp turned off) has an equivalent amplitude that is at least two orders of magnitude below the measurement amplitude. Hence, we are confident that our measurements were not influenced by any lack of sensitivity of the system. 

Fluorescent intensities were measured as deflection distances from the hard copy of the chart recorder output. These distances were converted into arbitrary fluorescence units, such that all intensities were standardized to the 10⁻⁸ amplifier setting. Intrinsic or autofluorescence intensities were subtracted from all postinstillation values before normalization. Normalization of fluorescent intensity data was performed to obviate variation due to tracer concentration and the anatomic variation between subjects. Peak fluorescent intensities were obtained in all cases immediately after lens insertion. Commercial graphing software (Kaleidagraph, ver. 3.01; Synergy Software, Reading, PA) was used to analyze the 20 to 30 data points available for each decay curve. 

Tracer

Although several potential fluorescent tracers may be acceptable for tear-exchange investigation, FITC-dextran was selected because its spectral characteristics are similar to sodium fluorescein. Key attributes of the tracer are its stability, lack of protein affinity, and safety. ^{20 21 22} In addition, this tracer was known to have no scleral or corneal penetration. ^{16 23 24}  

FITC-dextrans have been widely used in research on the circulatory system and have been shown to be stable in vitro and in vivo. ²⁰ Plasma disappearance curves suggest little or no protein binding, because FITC-dextran leaves plasma at rates close to those of similar-size unlabeled dextran ²⁰ and inulin. ²¹ Further evidence that there was no protein binding was found in an investigation of labeled and unlabeled dextran permeation across the blood–lymph barrier, wherein no difference in transport rate was observed. ²¹ FITC-dextrans appear to be safe for in vivo use, because direct injection of 25% concentration in microlymphography studies have shown no adverse reactions. ²²  

An FITC-dextran of molecular weight 70,000 (Sigma Chemical Co., St. Louis, MO) at 0.1% wt/vol concentration was used as the tracer in these studies. The tracer was dissolved into borate-buffered saline at pH 7.0 and sterile filtered before use. Microbiologic and toxicological data were obtained for these preparations before human use. These assays demonstrated no bacterial growth and no cytotoxicity. Viscosity of the 0.1% wt/vol FITC-dextran formulation was compared with that of the buffered saline alone and was found to be identical (i.e., the viscosity was approximately 1.0 mPa/sec at a shear rate of 19.2 seconds and temperature of 25°C). 

Subject Selection

Eleven nonhabitual CL wearers were selected to participate in the study. Subjects had no preexisting ocular disease, reported no eye symptoms, and were using no topical or systemic medication. All subjects had normal tear film by slit lamp microscopy and had a noninvasive tear break-up time of at least 15 seconds. Central corneal curvature and corneal topography were measured in all subjects, and only those with normal topographic maps and less than 0.75 D of corneal astigmatism were included. The subject demographic data are shown in Table 1 . 

This investigation was conducted in accordance with the tenets of the Declaration of Helsinki. Written consent was obtained after explanation of the study procedure. University of New South Wales Ethics Committee approval for the procedures was also obtained before the investigation. 

Study Design

The sample size required was estimated based on pilot study data. A minimum sample size of 11 subjects was established based on estimates of type I error α = 0.05, for a power of 80%, with five repeated measures, assuming a null hypothesis that there was no difference between days. Eleven subjects were recruited and measurements were taken on six occasions. The eye initially measured was randomly selected, and five repeatability measurements were performed at the same time of day to mitigate potential effects of diurnal variation. ²⁵ Bilateral measurements were made on two occasions: once for the comparison of right and left eye data and on a second occasion for the comparison of controlled blinking with physiological blinking rate. On a final occasion, subjects wore a prototype lens of extended-wear design. 

Procedures

Baseline intrinsic corneal fluorescence measurements were determined for the right and left eyes. For the methods characterization, all subjects were fitted with disposable etafilcon A lenses (Vistakon; Johnson & Johnson, Jacksonville, FL) of power −3.00 D, back optic zone radius of 8.80 mm, diameter of 14.0 mm, and material oxygen permeability of 28 × 10⁻⁹ (cm/sec)(ml O₂/ml mm Hg). A prototype lotrafilcon silicone hydrogel lens (CibaVision, Duluth, GA), of power −1.00 D, back optic zone radius of 8.8 0 mm, diameter of 14.00 mm, and material oxygen permeability of 140 × 10⁻⁹ (cm/sec)(ml O₂/ml mm Hg), was fitted for the final series of measurements. Two microliters of the fluorescent tracer was applied to the back surface of the CL before insertion. The lens was applied directly to the corneal apex with minimal manipulation, and the central corneal fluorescence was measured at the following intervals: immediately after insertion, every 30 seconds until 5 minutes, every minute until 10 minutes, and every 2 minutes until 30 minutes after insertion. The fluorescence intensity was measured for a minimum of 3 seconds at each interval. The lens was removed after 30 minutes, and the central corneal fluorescence intensity was remeasured after buffered saline irrigation to verify the absence of corneal penetration. Subjects were asked to report their subjective comfort with the CL for each experiment, using a 0 to 100 scale, where 0 was“ intolerable” and 100 was “could not be felt.” Habitual blink rates with lenses were estimated on a separate occasion at which subjects naïve to the purpose of the experiment were examined by slit lamp microscope under low illumination and low magnification. The number of complete blinks only was recorded. 

Data Modeling

Preliminary evaluation suggested that a custom exponential curve fit using either single- or double-exponential equations would most accurately describe the data. An approach using the philosophy of fewest parameters with physical significance was adopted. Curve fits were evaluated on the basis of maximizing the R ² and χ² statistics. Curve fits were performed for both 0- to 30-minute and 5- to 30-minute data sets for each subject and each repeat. The equations used were as follows. 

Single exponential:  

\[Y{=}A\mathrm{e}^{\mathrm{-}t/{\tau}}\]

For the decay data, where Y = fluorescent intensity, we would expect A = 1, because the intensity data were normalized by dividing by the peak response; t is the elapsed time and τ is the lifetime of the decaying species in minutes. The reciprocal of τ is the fractional loss per minute (k) or the ER. 

Double exponential:  

\[Y{=}f\mathrm{e}^{\mathrm{-}t/{\tau}1}{+}(1-f)\mathrm{e}^{\mathrm{-}t/{\tau}2}\]

Elapsed time is shown by t, in minutes; f is the rapidly decaying fraction; τ1 is the shorter lifetime related to the rapid-elimination constant (k1 = 1/τ1); (1 − f) is the slower decaying fraction; 1/τ2 is the longer lifetime related to the slower, more physiological ER constant (k2 = 1/τ2). 

The general curve fit option in the graph software (Kaleidagraph; Synergy Software) uses the Levenberg-Marquardt fitting algorithm to minimize the sum of the squared errors of prediction. It allows weighting input, allowable error specification (set at 0.1%) and input of partial differentials to guide the program through parameter space for greater accuracy. 

Parameters of the approximate order of magnitude were input to guide the program through the iterative process. Decisions were made concerning the accuracy of each of the curve fits by examination of the individual parameter errors and the χ² (goodness of fit) and R ² (coefficient of determination values). Initial guesses were altered or the equation type (single to double) was altered until no further improvement in fit could be attained. 

Based on the best-fit approach, the ER (percentage per minute), the T ₉₅ (time for 95% of the dye to be eliminated) and the TRR per blink were calculated for each data set as appropriate: (1) ER per minute was estimated as 1/τ1 or 1/τ2 for single and double-exponential curve fits, respectively. (2) T ₉₅ was estimated from the curve-fitting coefficients derived from the 0- to 30-minute post–lens-insertion data set. The time taken for 95% of the dye to be eliminated—that is, for a fluorescent intensity of 0.05 to be reached from the normalized data, was estimated from the derived curve equation. (3) TRR is the fractional volume exchanged per blink and was estimated using the curve-fitting coefficients derived from the 5- to 30-minute post–lens-insertion data set. The curve equation was used to derive the fluorescence intensity at 5 minutes (C _o ) and at 30 minutes (C _n ) after insertion, and TRR was calculated as follows, where n is the number of blinks ¹⁸ :  

\[TRR{=}1-\sqrt[n]{C_{n}/C_{o}{\times}100}\]

Repeated-measures ANOVA was used to compare the differences between repeat measurements of ER and T ₉₅. Differences between the left and right eyes, pre- and posttrial fluorescence, and controlled and habitual blink rate and between lens designs were evaluated using a paired or grouped t-test, as appropriate. The relationship between tear exchange and lens comfort was examined using the Pearson parametric correlation test. 

Results

General Results

For the methods optimization, a total of 77 tear-exchange determinations were made (five repeats on the same eye of 11 subjects and fellow eye determinations during two of the repeatability measurements). Of the 77 measurements, 74 could be optimally fitted using either the single- (17 measurements, 23%) or the double- (57 measurements, 77%) exponential equation. The multiple correlation coefficients (R) were 0.90 or greater for 73 of the 74 acceptable fits. Three of the complete data sets could not be well fitted with either the single- or double-exponential least-squares model, although only one of these could not be fitted with a single-exponential model for the 5- to 30-minute ER estimation. 

The exponential signal decay obtained with a nonpenetrating tracer instilled into the base curve of the CL is illustrated in Figure 2 . The decay followed first-order kinetics, similar to that reported for normal tear flow. ^{13 26} In the majority of determinations, the fluorescent signal decayed rapidly, probably due to reflex tearing after lens insertion, and required rapid measurement during the initial 5 minutes after insertion to capture the decay kinetics. Data collection speed with this system was limited to measurements every 30 to 45 seconds, which appears adequate, given the accuracy of the regression fits to these data. 

Analysis of the data from 5 to 30 minutes revealed that the double-exponential fit restricted to these time points gave no advantage and a single exponential fit was used for all ER calculations. The mean ER (± SD) for the entire data set was 8.8% ± 3.8% (n = 76) and mean T ₉₅ was 31.6 ± 16.1 minutes (n = 72; Table 2 ). 

Corneal dye penetration was examined in nine subjects by irrigating the ocular surface after tear-exchange determination. Pre- and posttrial corneal fluorescence levels were compared. The ratio of pre- to posttrial fluorescence was 1.07 ± 0.09, indicating that on average the posttrial fluorescence was 7% less than at baseline. This fluctuation was within the measurement variability of the technique. Moreover, there was no statistical difference between the pre- and posttrial fluorescence values (paired t-test, P = 0.08). 

Laterality Effects

Both eyes of all subjects were measured similarly on a single day to assess potential laterality effects. The first eye to be tested was randomly selected. Table 3 shows the mean ER and T ₉₅ values for each eye. Differences between eyes were not significant for either variable. Mean ERs (n = 11) were 9.1% ± 3.6% per minute in the right eye (T ₉₅ 29.9 ± 11.5 minutes) and 8.3% ± 3.6% per minute (T ₉₅ 36.2 ± 19.1 minutes) in the left eye. 

Effect of Controlling Blink Rate

ER, T ₉₅, and TRR are shown for the comparison of habitual blinking rate (19 ± 7 times/min) and a controlled blink rate of 15 times/min in Table 4 . Differences in ER between habitual and forced blinking were not significant for either ER (forced blinking 9.4% ± 4.0%, n = 11; habitual blinking 10.2% ± 4.9%, n= 10; paired t-test, P = 0.68) or T ₉₅ (forced blinking, 28.6 ± 21.3 minutes; habitual, 25.5 ± 13.8 minutes; paired t-test, P = 0.71, n = 9 for both). TRR was 0.62% ± 0.26% per blink (n= 11). 

Repeated-Measures Analysis

Repeated-measures data for both ER and T ₉₅ were seen to meet the assumptions for ANOVA, so that the data exhibited a normal distribution and the samples showed similar variance. Using repeated-measures ANOVA, there were no significant differences in ER or T ₉₅ between visits (P > 0.05). Figure 3 illustrates the individual ER values for each subject and the group mean. Between-subjects repeatability for ER was 2.91% and within-subject was 0.8%. Coefficient of repeatability was 1.6% for ER. 

Figure 4 demonstrates the individual T ₉₅ values for each subject and the group mean. Between-subjects repeatability was 9.9 minutes and within-subject was 4.4 minutes. Coefficient of repeatability for T ₉₅ was 8.6 minutes. 

Lens Comfort

Lens comfort scores ranged from 40 to 100, and there was no significant correlation between poor lens comfort and a higher ER (R ² = 0.018, P > 0.05). 

Comparison of Lens Designs

The 5- to 30-minute ER was used as the optimized measure to compare the prototype extended-wear soft contact lens (EWSCL) design with the marketed lens. Significantly higher tear exchange was found for the prototype lotrafilcon design (14.2% ± 3.9%/min, n = 7) compared with the etafilcon A design (8.8% ± 3.8%/min, n = 11; P < 0.01). 

Slightly but not significantly greater primary gaze movement was measured with the prototype EWSCL (0.3 ± 0.2 mm) than with the etafilcon A lens (0.2 ± 0.1 mm, P > 0.05). Lens comfort was slightly but not significantly poorer with the prototype EWSCL than the etafilcon A lens (mean comfort score, 75 vs. 83, P > 0.05). 

Discussion

This study presents normative, reproducibility, and comparative data on tear exchange behind a hydrogel CL estimated from tracer decay data obtained using a convenient fluorophotometric technique. Tracer decay behind a soft lens measured with a nonpenetrating tracer from 0 to 30 minutes after insertion was usually best described by a double-exponential curve-fitting approach. In contrast, data between 5 and 30 minutes after insertion were adequately described using a single-exponential curve fitting. The biexponential nature of the complete decay suggests a rapid tear exchange rate during the first 5 minutes after insertion, followed by a slower basal decay. The initial rapid decay may be due to reflex tearing on lens insertion. However, initial (subjective) lens comfort recorded on a 0 to 100 scale, for which 100 corresponded to “cannot be felt,” did not correlate with ER. 

These decay kinetics have not previously been presented for hydrogel lenses, although similar decay behavior has been reported for tear flow with both fluorescent ^{26 27 28} and γ-emitting tracers. ¹⁷ Accurate curve fitting based on the complete kinetics can have a significant impact on parameters such as the T ₉₅ calculated from the data. For example, the data in Figure 2 were best described by the double-exponential fit, giving a T ₉₅ of 27.3 minutes. If the same data were fitted with a single-exponential curve (Y = 0.870e^−t/4.8139;χ ² = 0.17, R = 0.96), the T ₉₅ would be 13.7 minutes—approximately a 50% decrease. 

The mean ER in this study was 8.8% ± 3.6% per minute for 76 determinations, which is around half of the normal, non-CL physiological ER of 16% to 18% per minute. ^{13 24 26 29} Both ER (percentage per minute) and TRR (percentage volume exchange per blink) data are similar to those in studies reported previously, using fluorophotometry to estimate postlens tear exchange, in which single exponential equations were used in curve fitting. ² Mean TRR in the present study was 0.62% ± 0.26% per blink, which is comparable to a previous report of 0.5% per blink (95% confidence interval [CI], 0.39–0.62), ³⁰ using an identical lens. 

The prototype silicone hydrogel lens demonstrated an ER of approximately 14%/minute, compared with the marketed etafilcon design with an ER of approximately 9%. Mechanical characteristics of the lens material have been recognized to influence tear exchange behind hydrogel lenses. ³¹ The modulus of rigidity, defined as the measurement of the resistance to deformation of a material under compression, of the lotrafilcon lens has been measured at 1.2 MPa, ³² compared with 0.26 MPa for etafilcon A. ³¹ The modulus of elasticity, defined as the measurement of the resistance to deformation of a material under tension (stretch) may be up to 20 times higher in silicone hydrogel materials than in conventional hydrogels. ³¹ Both increased material modulus and elasticity are likely to result in higher blink-induced lens movement. However, in the present study, mean lens movement for the lotrafilcon lenses was 0.3 mm, and for the etafilcon A lenses, 0.2 mm; and differences in lens movement between designs in this group were not significant. In a recent study, ³³ tear exchange was shown to be related to the degree of lens movement on blink and for each 1-mm increase in lens movement, TRR increased 1.05% per blink. 

Apart from lens movement, lens comfort is also likely to affect tear exchange. ³³ Slightly increased lens movement and reduced comfort associated with the lotrafilcon lens are unlikely to have been responsible for differences in ER of the magnitude described in the present study. We hypothesize that material characteristics play a major role in tear ERs in the open-eye environment. 

In this study tear ERs were similar for the right and left eyes, and there were no differences in repeated measures of ERs within this subject group. The variances for repeated measurements were similar on each occasion for ER and TRR. Within-subject variations due to measurement variations and physiological fluctuations were low, despite the inherent variability in tear turnover, even in the absence of CL wear. ¹⁷ As expected, the main variability in the current measurements was attributed to the between-subjects variability. Similarly, large individual variations in physiological tear turnover have been documented. ²⁵ The coefficient of repeatability data will assist in gauging meaningful differences between population groups, by using this technique. These data suggest that the ER technique offers reasonable discriminatory ability and the reproducibility of this technique, defined as interlaboratory agreement using similar techniques, ³⁴ is acceptable. 

The sample size required for statistical validity was estimated from the present data. The presumed experiment is a randomized crossover comparison of two hydrogel lenses. The sample size was estimated based on the ER parameter, a two-sided paired t-test (α = 0.05, power = 80%) using the pooled SD of the differences in ER between the first and second repeatability determinations (5.72). The results are plotted in Figure 5 and suggest that a sample size of 31 or 19 subjects is required for a clinically significant change in ER of 3% or 4% per minute, respectively. ER in this study showed moderate repeatability; however, this sample-size analysis demonstrates that the technique is sufficiently precise to determine differences in tear ER between groups of subjects wearing different types of lenses. 

Sang and Maurice ³⁵ have pointed out that in the absence of a CL, mixing of a small drop of dye with the tear fluid takes several minutes, because of the spreading of the tracer, and this effect may confound measures of ER and tear volume. In the present study, in 75 of 77 data sets, peak fluorescence intensity was obtained from the central cornea within 20 to 30 seconds of lens insertion. We believe that directly applying the tracer to the corneal apex and taking subsequent measurements at the same position minimizes the effect of drop mixing. However, if mixing takes several minutes, the variability of early tear fluorescence readings means that the initial 5-minute data required for estimation of T ₉₅ may be imprecise. The T ₉₅ data reported in this study (mean 31.6 ± 16.1 minutes, n = 62) are almost identical with previously published data. ¹² However, estimation of T ₉₅ is based on the regression equation rather than on measured data. Given that the measurements at early time points are likely to show variation due to reflex tearing and tear-tracer mixing, the T ₉₅ estimate of tear exchange is somewhat imprecise. The ER and TRR parameters for tear exchange may be more reliable. For these reasons, we suggest these parameters as the preferred measures for tear ER. 

ER is also affected by penetration of tracer into the CL and ocular tissue. Ocular penetration of sodium fluorescein artificially elevated ER by approximately 25% compared with a large-molecular-weight nonpenetrating fluorescent tracer. ¹⁷ In the present study, no elevated corneal fluorescence was measured, suggesting that no penetration of the 70-kDa FITC-dextran occurred, thus obviating this potential source of error. 

Lens-related factors, such as design; fitting relationship between the lens, cornea, and lids; postlens tear volume; lens diameter; and material modulus and elasticity are likely to influence tear exchange. However, the present study was undertaken to validate a method of measurement rather than to assess the impact of different lens and fitting relationship parameters. The effect of the lens-cornea–bearing relationship, based on the difference in radius between the lens base curve and the central keratometry, was shown not to alter TRR in a small group of subjects. ³  

Controlling the blink rate in this study did not modify the ER compared with the habitual blink rate; however, only one rate was tested in this study. The habitual blink rate reported here (19 ± 7 per minute) is similar to that previously reported during soft CL wear ³⁶ (20 ± 1 per minute). One previous study ³⁷ demonstrated a marginal increase in corneal oxygenation (0.7%) when blink rate increased from 20 to 60 times per minute. This increase was attributed to increased tear exchange. 

In addition to fluorophotometry, other techniques and tracers have been used to estimate tear ERs. Gamma scintigraphy, which uses a gamma emitter such as technetium-99m and a gamma camera to record radioactivity, has also been used to monitor post–hydrogel-lens tear elimination. ¹³ This method is limited by the size of the gamma emitter, the lack of mobility of the gamma camera, and the restriction to two-dimensional observation. Another issue for scintigraphy is sensitivity. Approximately a 1-log-unit range of detection is afforded by scintigraphy over the observation period. ³⁸ In contrast, fluorescein techniques typically demonstrate a 2- to 3-log range of signal detection. ³⁹  

A further approach to the measurement of tear ER uses calibrated microspheres in a range of sizes from 6 to 40 μm in diameter applied to the back surface of a hydrogel lens. The rate of exchange of microspheres is estimated by imaging residual particles over time. ⁴⁰ A linear reduction in small (6-μm) particles over time was demonstrated. This technique may predict the rate of postlens debris removal. 

The precision of fluorophotometric techniques may be limited by the long duration of measurement, the small area sampled, and the large depth of focus, depending on the instrumentation. The slit lamp technique described herein allows short measurements (2- to 3-second duration) to be taken at any interval over a wide area of the central precorneal tear film. The alignment technique is repeatable, allowing precise repositioning to the same ocular site, and the depth of focus is small, which minimizes the detection of extraneous fluorescence. The use of large-molecular-weight tracers eliminates confounding fluorescence due to lens or corneal penetration. 

An additional source of confounding fluorescence may be tracer expelled from the postlens tear film and moving into the prelens tear film. Tear flow has been shown to be reduced in the presence of a hydrogel lens in one study in which the tracer was applied to the prelens tear film. ¹³ This phenomenon may underestimate the measured ERs. The magnitude of this effect has not been quantified. However, if both the pre- and postlens fluorescence were being measured simultaneously in the present study, we would expect it to have a consistent effect on the measured ER with similar design lenses. Thus, on a comparative basis, differences in lens behavior can be characterized, as this study has demonstrated. 

In summary, this study has shown that this slit lamp fluorophotometric technique for measuring postlens tear exchange is feasible and convenient and that the results are comparable to those in previous studies. Postlens tracer decay was most precisely described using a double-exponential curve equation in the majority of data sets. The ER may be described using ER (percentage per minute), TRR (percentage per blink), or T ₉₅ (time for 95% of the signal to be eliminated). Regression fits involving early time points after lens insertion may be imprecise, and thus ER or TRR may be more precise estimates of ER than T ₉₅. Within ER data, the major variation in the measurements is due to the between-subject variability. No differences were found between the right and left eyes of subjects and a habitual blink rate did not modify ER. Based on these data, it will be possible to design experiments to establish the role of lens–cornea–lid interactions and lens design and material properties in ER. This will allow rational strategies to be developed to maximize postlens tear exchange. 

 Supported in part by the Australian Federal Government through the Cooperative Research Centres Programme.

 Submitted for publication April 2, 2001; revised July 2, 2001; accepted August 1, 2001.

 Commercial relationships policy: N.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked“ advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Diagram of the fluorophotometer.

Table 1.

 

Subject Demographic Data

N	Age (y ± SD)	Sex (M:F)	Mean Flattest Central Keratometry (D ± SD)	Mean Steepest Central Keratometry (D ± SD)
11	29.1± 4.6	8:3	42.18± 1.09	42.68± 0.96

Figure 2.

 

Fluorescence intensity versus time for subject 4. Data are shown for 0 to 30 minutes after lens insertion and are most accurately described by a double-exponential curve fit. The best-fit equation for these data was Y = 0.64514e^{−t /1.8632} + (1 − f)e^{−t /13.992};χ 2 = 0.055, R = 0.99. The rapid-decay fraction (f) represents 64% of the total fluorescence intensity elimination. The ER for the 5- to 30-minute portion, fit by the single-exponential equation (not shown), was 9.0%/minute. T ₉₅ for this curve fit was 27.3 minutes.

Table 2.

 

Overall Decay Elimination Data

	ER% (n = 76)	T 95 Minutes (n = 72)
Mean± SD	8.8± 3.8	31.6± 16.1
95% CI	7.9–9.7	24.3–38.9

Table 3.

 

Right and Left Eye Data

	Right Eye (n = 10)	Left Eye (n = 10)	P
ER (% per minute)	9.1 ± 3.6	8.3 ± 3.6	0.61
T95 (minutes)	29.9 ± 11.5	36.2 ± 19.1	0.36

 Data are means ± SD.

Table 4.

 

Effect of Controlling Blink Rate

	Controlled Blinking (Rate of 15 Times per Minute)	Habitual Blinking (Mean Rate of 19 ± 7 Times per Minute)	P
ER (% per minute)	9.4 ± 4.0 (n = 11)	10.2 ± 4.9 (n = 10)	0.68
T95 (minutes)	28.6 ± 21.3 (n = 11)	26.5 ± 12.5 (n = 9)	0.71
TRR (% per blink)	0.62 ± 0.26 (n = 11)	—	—

 Data are means ± SD.

Figure 3.

 

Individual ER for each subject. Mean ± SD are shown for each subject; horizontal line: group mean.

Figure 4.

 

Individual T ₉₅ for each subject. Mean ± SD are shown for each subject; horizontal line: group mean.

Figure 5.

 

Sample size effects based on the ER parameter (two-sided paired t-test, α = 0.05, power = 80%) using the pooled SD of the differences in ER between the first and second repeatability determinations (5.72).

The authors thank Reginald Wong for statistical advice and Ron Chatelier, PhD, and Brian Brown, PhD, for helpful comments on the manuscript.