Journal of Neuro-Ophthalmology, December 2002, Volume 22, Issue 4

Contrast Letter Acuity as a Measure of Visual Dysfunction in Patients with Friedrich Ataxia

BACKGROUND: Friedreich ataxia is a progressive neurodegenerative disorder affecting afferent cerebellar pathways and other neuronal systems, including afferent visual pathways. A systematic clinical outcome measure for examination of visual dysfunction in Friedreich ataxia has not been identified. We sought to identify a simple, reliable method for assessing clinical and subclinical visual dysfunction in patients with Friedreich ataxia. METHODS: Contrast letter acuity was measured binocularly in Friedreich ataxia patients and age-matched visually asymptomatic volunteers (control group) using the Low-contrast Sloan Letter Charts at three different low-contrast levels (5.0%, 1.25%, and 0.6%). Binocular high-contrast visual acuity (100% level) was also determined for each participant. RESULTS: Despite equal median binocular high-contrast visual acuities between the two groups, patients with Friedreich ataxia had significantly lower (worse) Low-contrast Sloan Letter Chart scores compared with controls, particularly at the lowest contrast levels (1.25% and 0.6%). Ambulation status significantly predicted Low-contrast Sloan Letter Charts scores in linear regression models accounting for patient age, suggesting a potential complementary role for Low-contrast Sloan Letter Chart testing in the assessment of disease status as well as visual function in Friedreich ataxia. CONCLUSIONS: This study demonstrates that Low-contrast Sloan Letter Chart testing may provide a useful clinical outcome measure for Friedreich ataxia and other neuro-ophthalmologic disorders.

ORIGINAL CONTRIBUTION Contrast Letter Acuity as a Measure of Visual Dysfunction in Patients with Friedreich Ataxia David R. Lynch, MD, PhD, Jennifer M. Farmer, MS, Dustin Rochestie, BS, and Laura J. Balcer, MD, MSCE Background: Friedreich ataxia is a progressive neurode­generative disorder affecting afferent cerebellar pathways and other neuronal systems, including afferent visual path­ways. A systematic clinical outcome measure for examina­tion of visual dysfunction in Friedreich ataxia has not been identified. We sought to identify a simple, reliable method for assessing clinical and subclinical visual dysfunction in patients with Friedreich ataxia. Methods: Contrast letter acuity was measured binocularly in Friedreich ataxia patients and age- matched visually asymptomatic volunteers ( control group) using the Low-contrast Sloan Letter Charts at three different low- contrast levels ( 5.0%, 1.25%, and 0.6%). Binocular high- contrast visual acuity ( 100% level) was also determined for each participant. Results: Despite equal median binocular high- contrast vi­sual acuities between the two groups, patients with Frie­dreich ataxia had significantly lower ( worse) Low- contrast Sloan Letter Chart scores compared with controls, particu­larly at the lowest contrast levels ( 1.25% and 0.6%). Am­bulation status significantly predicted Low- contrast Sloan Letter Charts scores in linear regression models account­ing for patient age, suggesting a potential complementary role for Low- contrast Sloan Letter Chart testing in the as­sessment of disease status as well as visual function in Friedreich ataxia. Conclusions: This study demonstrates that Low- contrast Sloan Letter Chart testing may provide a useful clinical out­come measure for Friedreich ataxia and other neuro-ophthalmologic disorders. ( JNeuro- Ophthalmol 2002; 22: 270- 274) Supported by grants NS01789 and EY00351 from the National Insti­tutes of Health, a grant from the Muscular Dystrophy Association, and a Beeson Scholar Award. Departments of Neurology ( DRL, DR, LJB), Pediatrics ( DRL, DR), Medicine ( JMF), and Ophthalmology ( LJB), University of Pennsylvania and The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania. Address correspondence to David R. Lynch, MD, PhD, Division of Neuroscience Research, Children's Hospital of Philadelphia, 502 Abramson Building, Philadelphia, PA 19104- 4318, USA; E- mail: lynch@ pharm. med. upenn. edu Friedreich ataxia ( FRDA) is an autosomal recessive de­generative disorder characterized by ataxia, dysarthria, sensory loss, diabetes, and cardiomyopathy ( 1,2). Patients with FRDA have genetic abnormalities in a gene now known as FRDA; 95%> of patients carry two copies of an expanded intronic GAA repeat, whereas 5%> carry an ex­panded GAA repeat on one allele in conjunction with a point mutation in FRDA on the other allele. The length of the shorter GAA repeat inversely correlates with age of on­set in large series of patients ( 2,3). The gene product ( frataxin) involved in FRDA is localized to the mitochon­drion, and its absence causes mitochondrial iron accumula­tion ( 4). This suggests possible therapies for FRDA, but there is presently no widely accepted measure for longitu­dinal clinical outcomes assessment in patients with FRDA. Analysis of candidate clinical outcome measures is neces­sary for use in long- term clinical trials. Afferent and efferent visual dysfunctions are frequent manifestations of FRDA. Most patients with FRDA have normal high- contrast visual acuities, but a subpopulation ( 5- 30%) may have mildly decreased visual acuity, usually to the level of 20/ 30 ( 1,5- 9). In most cases, this is believed to reflect optic neuropathy, and pathologic abnormalities have been found in the lateral geniculate nuclei of patients with FRDA. A few patients have severe visual loss from optic neuropathy ( 10,11). A variety of eye movement ab­normalities, including fixation instability, also cause visual dysfunction in patients with FRDA ( 12- 14). Therefore, a combination of afferent and efferent visual abnormalities is likely to contribute to the neurologic dysfunction of FRDA, and a clinical outcome measure that captures all aspects of FRDA- related visual loss is needed. In contrast to classically designed mechanisms for as­sessing ataxia, visual function can be quantitatively as­sessed using convenient, readily available clinical outcome measures. Recently, contrast letter acuity testing has been explored as a potential quantitative measure for detecting visual dysfunction in patients with multiple sclerosis, a dis­order characterized by functional limitations and neuro­logic progression analogous to FRDA ( 15). The purpose of the current study was to examine the capacity for binocular Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 270 J Neuro- Ophthalmol, Vol. 22, No. 4, 2002 CONTRAST LETTER ACUITY IN FRIEDREICH ATAXIA JNeuro- Ophthalmol, Vol. 22, No. 4, 2002 contrast letter acuity testing, using Low- contrast Sloan Letter Charts ( LCSLC) to identify visual dysfunction in patients with FRDA. Lo accomplish this, LCSLC scores for FRDA patients were compared with those of age-matched visually asymptomatic volunteers ( controls). Lhe relation of LCSLC scores to high- contrast visual acuity and to other aspects of neurologic function in FRDA was also determined. METHODS Selection of Participants Study protocols were approved by the Institutional Review Board at the University of Pennsylvania. Thirty-two adult patients with genetically confirmed ( n = 30) or clinically diagnosed FRDA ( n = 2) were invited at the time of their clinical evaluation to participate in testing using the Low- contrast Sloan Letter Charts ( LCSLC) as described previously; these charts have gray letters of progressively smaller sizes on a white background ( 15- 18). Visually asymptomatic volunteers ( controls), who were category-matched for median age with FRDA patients, also were in­vited to participate. Volunteers were selected from among staff and faculty members of the University of Pennsylvania and were eligible if they had no known history of neuro­logic or ophthalmologic disease other than refractive error. Contrast letter acuity testing Binocular contrast letter acuity testing was performed as described previously using LCSLC ( Precision Vision, LaSalle, IL) ( 15). The LCSLC have a standardized format based on that of the Bailey- Lovie and ETDRS ( Early Treat­ment Diabetic Retinopathy Study) visual acuity charts, the standard charts used for acuity measurement in ophthalmol­ogy clinical trials ( 16- 18). For this study, the 5%, 1.25%, and 0.6% contrast charts were used. High- contrast visual acuity was also tested using the 100% contrast chart; bin­ocular Snellen equivalents were recorded. Patients were po­sitioned 2 meters from the charts and tested using standard­ized protocols based on ETDRS visual acuity testing ( 15,19- 20). Using these protocols, patients were asked to identify single letters, starting on the top line of each chart, reading left to right. Letter- by- letter scoring ( number of let­ters identified correctly per chart) was used, resulting in a letter score ( maximum 60) for each chart ( 21- 22). Uniform, overhead lighting ( 80- 100 foot candles on white back­ground of charts) was used for each testing session and ac­complished using standard fluorescent office lighting. Par­ticipants wore their usual corrective lenses during testing. Binocular testing was used in this study to minimize patient fatigue and to best capture visual function as is present for daily activities. Data Analysis The number of letters identified correctly on each chart was recorded ( maximum 60/ chart). An aggregate score including the three low- contrast charts and the high-contrast visual acuity chart was also calculated. The binocu­lar Snellen visual acuity equivalent was determined based on the lowest line of the high- contrast ( 100%) chart for which the participant identified three or more letters cor­rectly. Comparisons of LCSLC scores between FRDA and control groups were performed using the Wilcoxon rank sum test. Demographic data and disease features for patients with FRDA were recorded, including length of the shorter GAA repeat, ambulation status, age, and disease duration. GAA repeat lengths were available in 29 patients as deter­mined by commercial DNA analysis. In one patient, GAA repeat length was presumed to be identical to the geneti­cally confirmed sibling. A second patient, who carried a C- terminal point mutation, was treated as having a repeat length of 1000, consistent with the expected severity of C-terminal point mutations ( 23). Ambulation status was cat­egorized based on whether the patient walked unassisted, used an assistive device ( cane or walker), or used a wheel­chair. Spearman rank correlations were used to assess the relation of LCSLC scores to FRDA disease features such as triplet repeat length, age, and disease duration. Linear re­gression analysis was used to further examine the relation of LCSLC scores to ambulation status while accounting si­multaneously for patient age. RESULTS Demographic characteristics of FRDA patients and visually asymptomatic volunteers ( controls) were com­pared. Age was similar between the groups ( for patients: median = 25.5 years, range = 10- 57 years; for controls, me­dian = 31 years, range = 24- 50 years; P = 0.15 by Wilcoxon rank sum test). Among patients with FRDA, the length of the shorter triplet repeat correlated significantly with age of onset ( rs = - 0.73, P < 0.0001), consistent with previ­ous reports ( 2,3). Median binocular high- contrast Snellen visual acuity equivalents ( 100% contrast level) did not differ between the FRDA and control groups ( median Snellen equivalents = 20/ 16 in both groups). LCSLC scores, however, were sig­nificantly lower in the FRDA group for the aggregate letter score ( LCSLC + high- contrast acuity, P < 0.0001, Wil­coxon rank sum test) and at each low- contrast level ( 5% chart: P = 0.0006; 1.25% chart: P < 0.0001; 0.6% chart: P < 0.0001) ( Fig. 1, Table 1). The magnitude of the differences between groups increased with progressively lower con­trast levels. We then assessed correlations between LCSLC scores and FRDA disease features. In linear regression Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 271 JNeuro- Ophthalmol, Vol. 22, No. 4, 2002 Lynch et al. u U ! 60- 100% contrast 5% contrast ED 1.25% contrast 0.6% contrast FRDA Control FRDA Control FbDA FIG. 1. Low- contrast Sloan Letter Chart scores in patients with Friedreich ataxia compared with visually asymptomatic vol­unteers ( controls). Patients with Friedreich ataxia and a control group of visually asymptomatic volunteers were tested us­ing LCSLC at 100% ( high- contrast acuity), 5%, 1.25%, and 0.6% contrast levels. Pa­tients had significantly lower ( worse) Low-contrast Sloan Letter Charts scores at each contrast level. models accounting for patient age ( Wald test), ambulation status predicted LCSLC scores to a significant degree ( Fig. 2, Table 2). Specifically, wheelchair use was associated with a lower LCSLC score for each chart and for the aggre­gate score [ LCSLC + high- contrast acuity, R2 = 0.38, P = 0.0015; 100% chart ( high- contrast acuity), R2 = 0.42, P = 0.0011; 5% chart, R2 = 0.25, P = 0.020; 1.25% chart, R2 = 0.41, P = 0.0010; 0.6% chart, i? 2 = 0.23, P = 0.041]. Values for R2 in these regression models correspond to relatively high R values for human disease ( R = 0.50- 0.65); R2 indi­cates the proportion of the variability in LCSLC scores that is explained by ambulation status and age. In contrast, pa­tient age correlated minimally and nonsignificantly with high- contrast acuity and LCSLC scores ( rs = - 0.26 for ag­gregate score; rs = - 0.30 for 100% chart; rs = - 0.14 for the 5% chart; rs = - 0.25 for the 1.25%; rs = - 0.20 for the 0.6%). The length of the patient's shorter triplet repeat demon­strated no correlation with LCSLC scores ( rs = - 0.0078 for aggregate LCSLC + high- contrast acuity score, all P values > 0.5). Disease duration correlated moderately with LCSLC scores ( rs = - 0.42 for aggregate score, P = 0.017; rs = - 0.47 for 100%, P = 0.0073; rs = - 0.24 for 5%, P = 0.19; rs = - 0.40, P = 0.026 for 1.25% chart; rs = -. 39, P = 0.028 for 0.6% chart). This demonstrates that LCSLC scores are more likely to reflect levels of disability and neurologic im­pairment ( ambulation and disease duration) rather than bio­chemical or genetic features ( such as triplet repeat length). DISCUSSION The results of this investigation demonstrate that con­trast letter acuity as measured using the Low- Contrast Sloan Letter Charts ( LCSLC) provides a useful measure for assessment of visual dysfunction in FRDA. LCSLC scores were significantly lower among patients with FRDA com­pared with age- matched, visually asymptomatic volunteers. Among FRDA patients, LCSLC scores also correlated with disease duration and were predicted significantly by ambu­lation status in linear regression models accounting simul­taneously for age. LCSLC scores did not correlate with trip­let repeat length and correlated minimally with age. This result suggests that visual dysfunction in FRDA, as mea­sured by the LCSLC, is associated with progressive features of the disease. LCSLC testing therefore demonstrates excellent potential as a candidate visual outcome measure in FRDA, analogous to its proposed role in multiple sclerosis ( 15). The ability of LCSLC scores to discriminate between FRDA patients and controls was greatest for the 5% charts and 1.25% charts, suggesting that LCSLC may be most use­ful at these contrast levels. Conversely, binocular high-contrast visual acuity scores ( 100% chart) were similar between FRDA patients and controls, emphasizing that tra­ditional measures of visual acuity may not entirely capture visual dysfunction in patients with FRDA. Patients and control participants in this study wore their habitual distance correction ( glasses or contact lenses) for LCSLC testing and did not receive detailed refractions. Although differences in LCSLC scores between the FRDA and control groups at low- contrast levels could potentially reflect differences in access to routine ophthalmologic care, the following evidence suggests that the observed dif­ferences are likely to reflect underlying visual pathway dis­ease in FRDA: 1) median Snellen equivalents ( used for TABLE 1. Median low- contrast Sloan letter chart scores of visually asymptomatic volunteers ( controls) and Friedreich ataxia ( FRDA) patients: no. ( range) Low- contrast Sloan letter chart scores Subjects 100% contrast 5% contrast 1.25% contrast 0.6% contrast Aggregate score Control FRDA 60 ( 58- 60) 59 ( 39- 60) 56 ( 47- 60) 48 ( 17- 60) 42 ( 27- 53 30 ( 0- 49) 32( 0- 38) 190( 134- 207) 10 ( 0- 44) 148 ( 58- 212) Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 272 © 2002 Lippincott Williams & Wilkins CONTRAST LETTER ACUITY IN FRIEDREICH ATAXIA JNeuro- Ophthalmol, Vol. 22, No. 4, 2002 100% contrast o o UN. Assist WC 5% contrast - D • • a • n a s 1 UN. Assist. WC 1.25% contrast • 1 I 1 1 I rTTTTTi G a B l § B • • n 0.6% contrast UN. Assist. WC FIG. 2. Relation of ambulation status to Low- contrast Sloan Letter Chart scores in patients with Friedreich ataxia. UN, unas­sisted ambulation; Assist, assisted ambu­lation ( cane/ walker); WC, wheelchair use. Low- contrast Sloan Letter Charts scores on each chart were predicted by ambulation status. UN. Assist. WC nonresearch refraction techniques) were equal between the FRDA and control groups in this study; 2) in an ongoing cross- sectional study of visual outcome measures in mul­tiple sclerosis ( 24), LCSLC scores at low- contrast levels were significantly lower ( worse) among patients than among age- matched controls, even after detailed refrac­tions. Future studies of LCSLC testing in patients with FRDA and other neuro- ophthalmologic disorders could in­clude detailed refractions and an assessment of efferent vi­sual abnormalities. Whereas ambulation status predicted LCSLC scores in linear regression models accounting for age, there was substantial variability in scores within each ambulation cat­egory ( Fig. 2). This indicates that LCSLC testing in patients with FRDA may capture aspects of disease that are not en­tirely captured by ambulation status alone. LCSLC testing is also a convenient yet cost effective measure to add to clini­cal outcome scales in FRDA. The charts are portable ( 14x14 inches), and testing may be administered by nonphysician professionals within a 10- to 15- minute time period. Visual dysfunction has been noted as an aspect of neurologic dysfunction in FRDA. However, traditional as­sessments note largely subclinical visual loss with abnor­malities detected by electrophysiological testing in up to 90% of patients ( 1,5- 9). Based on studies of patients with optic neuropathy and severe visual loss, the sites of afferent visual dysfunction in FRDA are thought to be the optic nerve and pregeniculate visual pathways. Since efferent ( ocular motor) pathways are also affected in FRDA, visual dysfunction as captured by binocular LCSLC testing may also result from convergence or fixation abnormalities. However, whether the site of the abnormality is afferent or efferent, the correlation with ambulation status and disease duration demonstrates that LCSLC capture a component of progressive disability in FRDA. A clear correlation of vi­sual dysfunction with disease duration and disability has not always been noted in previous studies on vision in FRDA ( 7,9). LCSLC may therefore be sensitive to disease progression even in patients whose ambulation status can no longer be measured. The lack of correlation between shorter triplet repeat length and LCSLC scores seems paradoxical. This discrep­ancy most likely reflects a selection bias in our patient population because the most severely affected individuals ( those with longest triplet repeat length and long disease durations) may not seek ongoing medical care at tertiary care centers and therefore are not included in our study. In our patient cohort, shorter triplet repeat length was high­ly inversely correlated with disease duration ( data not shown). Because disease duration affects LCSLC scores, this observation may explain the lack of correlation of the GAA triplet repeat length with visual function in our patients. The results of this investigation provide quantitative evidence that a measure of visual function, specifically the LCSLC, should be considered foruse in future clinical trials of therapies for FRDA. The usefulness of LCSLC testing for longitudinal assessment of patients with neurologic TABLE 2. Median low- contrast Sloan letter chart scores of Friedreich ambulation status: no. ( range) Ambulation status Unassisted Assisted Wheelchair 100% contrast 60 ( 54- 60) 60 ( 55- 60) 52 ( 39- 60) ataxia patients categorized by Low- contrast Sloan letter chart scores 5% contrast 51 ( 44- 59) 48 ( 37- 60) 36 ( 17- 60) 1.25% contrast 39 ( 25^ 19) 30 ( 22^ 18) 12 ( 0- 37) 0.6% contrast 16 ( 4- 30) 21 ( 0^ 14) 0 ( 0- 26) Aggregate score 164( 132- 189) 147( 131- 212) 97 ( 58- 178) Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 273 JNeuro- Ophthalmol, Vol. 22, No. 4, 2002 Lynch et al. disorders, including FRDA, remains to be determined and is under examination in ongoing multiple sclerosis clinical trials. Future studies of FRDA patients with the use of standardized clinical outcome measures, detailed refrac­tions, and quantitative assessment of efferent visual abnor­malities will be crucial in determining the extent and nature of visual dysfunction in FRDA in the broader context of neurologic impairment. ACKNOWLEDGMENTS This work was supported by grants NS01789 and EY00351 from the National Institutes of Health, a grant from the Muscular Dystrophy Association, and a Beeson Scholar Award. REFERENCES 1. Harding AE. Friedreich's Ataxia: A clinical and genetic study of 90 families with an analysis of early diagnostic criteria and intrafamil-ial clustering of clinical features. Brain 1981; 104: 589- 620. 2. Campuzano V, Montermini L, Molto MD, et al. Friedreich's ataxia: autosomal recessive disease caused by an intronic GAA triplet re­peat expansion. Science 1996; 271: 1423- 7. 3. Durr A, Cossee M, Agid Y, et al. Clinical and genetic abnormalities in patients with Friedreich's ataxia. New Engl J Med 1996; 335: 1169- 75. 4. Priller J, Scherzer CR, Faber PW, et al. Frataxin gene of Frie­dreich's Ataxia is targeted to mitochondria. Ann Neurol 1997; 42: 265- 269. 5. Carroll WM, Kriss A, Baraitser M, et al. The incidence and nature of visual pathway involvement in Friedreich's ataxia. A clinical and visual evoked potential study of 22 patients. Brain 1980; 103: 413- 34. 6. Fehrenbach RA, Wallesch CW, Claus D. Neuropsychologic find­ings in Friedreich's ataxia. Arch Neurol 1984; 41: 306- 8. 7. Wenzel W, Camacho L, Claus D, et al. Visually evoked potentials in Friedreich's ataxia. AdvNeurol 1982; 32: 131- 9. 8. Bird TD, Crill WE. Pattern- reversal visual evoked potentials in the hereditary ataxias and spinal degenerations. Ann Neurol 1981 ; 9: 243- 50. 9. Livingstone IR, Mastaglia FL, Edis R, et al. Visual involvement in Friedreich's ataxia and hereditary spastic ataxia. A clinical and vi­sual evoked response study. Arch Neurol 1981; 38: 75- 9. 10. McCormack ML, Guttmann RP, Schumann M, et al. Frataxin point mutations in two patients with Friedreich's ataxia and unusual clini­cal features. JNeurolNeurosurgPsychiatry 2000; 68: 661^ 1. 11. Givre SJ, Wall M, Kardon RH. Visual loss and recovery in a patient with Friedreich ataxia. JNeuro- Ophthalmol 2000; 20: 229- 33. 12. Ciuffreda KJ, Kenyon RV, Stark L. Eye movements during reading: further case reports. Am JOptom Physiol Opt 1985; 62: 844- 52. 13. Furman JM, Perlman S, Baloh RW. Eye movements in Friedreich's ataxia. Arch Neurol 1983; 40: 343- 6. 14. Kirkham TH, Guitton D, Katsarkas A, et al. Oculomotor abnormali­ties in Friedreich's ataxia. Can J Neurol Sci 1979; 6: 167- 72. 15. Baker LJ, Baier ML, Pelak VS, et al. New low- contrast vision charts: reliability and test characteristics in patients with multiple sclerosis. Mult Scler 2000; 6: 163- 71. 16. Bailey IL, Lovie JE. New design principles for visual acuity letter charts. Am J Optom Physiol Opt 1976; 53: 740- 5. 17. Ferris FL, Kassoff A, Bresnick GH, et al. New visual acuity charts for clinical research. Am J Ophthalmol 1982; 94: 91- 6. 18. Ferris FL, Bailey I. Standardizing the measurement of visual acuity for clinical research studies: guideline from the Eye Care Technol­ogy Forum. Ophthalmology 1996; 103: 181- 2. 19. Blackhurst DW, Maguire MG. Reproducibility of refraction and vi­sual acuity measurement under a standard protocol. The Macular Photocoagulation Study Group. Retina 1989; 9: 163- 9. 20. Klein R, Klein BE, Moss SE, et al. Inter- observer variation in re­fraction and visual acuity measurement using a standardized proto­col. Ophthalmology 1983; 90: 1357- 9. 21. Bailey IL, Bullimore MA, Rassch TW, et al. Clinical grading and the effects of scaling. Invest Ophthalmol Vis Sci 1991 ; 32: 422- 2. 22. Bailey IL. New procedures for detecting early vision losses in the elderly. Optom Vis Sci 1993; 70: 299- 305. 23. Cossee M, Durr A, Schmitt M, et al. Friedreich's ataxia: point mu­tations and clinical presentation of compound heterozygotes. Ann Neurol 1999; 45: 200- 6. 24. Baker LJ, Baier ML, Kooijmans M, et al. Contrast letter acuity as a candidate visual outcome measure for the MS Functional Compos­ite. MultScler 2001 ; 7: S 104. Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 274 © 2002 Lippincott Williams & Wilkins