Stargardt disease (STGD), first described in 1909 by Karl Stargardt,³⁵ is by far the most common form of juvenile macular degeneration. Although the incidence of one in 10,000 is frequently cited, precise estimates of prevalence are not available. STGD is characterized by discrete yellowish deposits within the posterior pole. One variant of STGD, fundus flavimaculatus, was described by Franceschetti and Francois in 1963.¹⁹ The term was used to describe patients who had white pisciform flecks throughout the fundus but who typically retained good visual acuity until later in life. Molecular studies have more recently shown that autosomal recessive forms of fundus flavimaculatus, STGD, and some more widespread forms of cone-rod dystrophy (CRD) and retinitis pigmentosa (RP) are all associated with mutations in the ABCA4 (formerly ABCR) gene. A rare dominant form of macular disease with phenotypic similarity to STGD has been related to mutations in ELOVA4, a gene that is believed to be related to long-chain fatty acid metabolism.⁴⁰

STGD typically begins in the first or second decade of life. Presenting symptoms include visual acuity that cannot be corrected to 20/20. Typically, there is a fairly rapid decline in acuity during the teenage years, with final acuity of 20/200 to 20/400 by adulthood.¹⁷ The full-field ERG is useful for ruling out more widespread forms of retinal degeneration.⁴ As shown in figure 62.1, responses are usually within the normal range in children with the disease. The cone electroretinogram (ERG) to 31-Hz flicker typically lies toward the lower limit of normal, and the cone b-wave implicit time is usually longer than mean normal but still within the normal range.²² Older patients with extensive macular generation may show subnormal cone and rod amplitudes (see figure 62.1), but the magnitude of loss is roughly predictable from the extent of macular degeneration. An important characteristic of STGD is that cone b-wave implicit time remains borderline normal despite advanced disease. This is an important prognostic indicator, since patients who retaining normal or near normal b-wave implicit times are likely to retain useful peripheral function throughout life.⁶ As we shall see, this discriminates patients with STGD from those with CRD, who, despite having similar gene mutations, nevertheless have a distinctly different visual prognosis.

Figure 62.1.  

Representative full-field ERGs from a normal subject (left column) and two patients with STGD at different stages of disease (middle and right column). Response amplitudes are borderline reduced in the older patient, but b-wave implicit times are at the upper end of normal.

While the full-field ERG is useful for discriminating the more localized pathology of STGD from widespread forms of retinal degeneration, it is of little value in the early detection of macular disease and for following patients in longitudinal studies and clinical trials. In evaluating the possible retinal basis for reduced acuity in a child, it is necessary to obtain a focal or multifocal ERG. The focal ERG should be conducted with direct visualization of the fundus to ensure that the response originates from the area of interest.³⁴ Typically, in evaluating an acuity loss, the region of interest is the fovea. With the stimulus as small as 4 degrees, the stimulus is flickered at a frequency that is higher than the rod fusion frequency. The Maculoscope^TM, for example, based on the work of Sandberg et al.,³⁴ employs a spot flickering at 42Hz within a more intense, steady surround. The sensitivity and utility of this test for documenting the retinal basis of acuity loss have been reviewed previously.⁷ It is generally thought that the foveal response drops below the lower normal amplitude limit when visual acuity is 20/50 or less owing to macular degeneration.¹⁵ In STGD, the amplitude may be below the lower limit of normal before substantial loss of acuity, making this an important prognostic test.^11,15,26

The multifocal ERG (mfERG) adds the capability for simultaneously measuring retinal function at dozens of locations throughout the macula.³⁸ With the recent development of the fundus camera-based stimulus delivery system, it is now possible to monitor fundus position while testing. This is particularly important for patients with STGD, who may use a preferred eccentric locus of fixation whenever possible. In practice, it is useful to assess the fixation behavior of the patient through a fundus camera prior to mfERG testing. It is then possible to correct for eccentric fixation during testing. Maintaining the position of the optic disk helps to ensure that the stimulus pattern is centered on the fovea. Figure 62.2A shows the fundus of the left eye of a 20-year-old female with STGD. Characteristic flecks (lipofuscin) are scattered throughout the posterior pole but are not present in the macula, which has an atrophic appearance (more evident on fluorescein angiography). The mfERG from the same eye is characteristic of responses in a patient with recently diagnosed STGD (figure 62.2B). Despite only a modest reduction in acuity, responses from the central 10 degrees are severely reduced in amplitude, while responses from outside the macula are normal. The pattern of loss in the mfERG corresponds to that seen in the visual field (figure 62.2C).

Figure 62.2.  

A, Fundus photo from left eye showing lipofuscin accumulation within the posterior pole. B, mfERG from same patient showing selective loss of responses from the central 10 degrees.

Figure 62.2.  

(continued) C, Humphrey static perimetric fields from central 24 degrees (left) and 10 degrees (right) showing loss of sensitivity corresponding to mfERG regional loss.

Mutations in the ABCA4 gene were identified as a cause of STGD in 1997.^2,3 It is now thought that all cases of autosomal recessive STGD are due to ABCA4 mutations.²⁰ The protein encoded by the ABCA4 gene is called rim protein (RmP) because it was initially described in frog rod outer segment rims.³⁰ RmP is a member of the adenosine triphosphate–binding cassette (ABC) transporter superfamily.³ Because it produces increased ATPase activity from RmP in vivo, a likely substrate of RmP is all-trans-retinal.³⁶ On the basis of findings in the abcr−/− mouse model, in which RmP is completely absent, Travis and colleagues have proposed a model for RmP (figure 62.3A).³⁹ According to the model, RmP participates in the metabolism of vitamin A in the photoreceptors after exposure to light. After exposure of rhodopsin to light (hv), all-trans-retinal is released within the outer segment disk. All-trans-retinal combines with phosphatidylethanolamine (PE) normally present in the disk membranes. The all-trans retinal-PE complex is called N-retinylidine-PE (N-ret-PE). The RmP protein normally transports the N-ret-PE out of the disk, where all-trans-retinal is reduced to all-trans retinol (atROL) and eventually reconverted back to 11-cis-retinal within the RPE.

Figure 62.3.  

Model for the function of RmP (ABCR) protein in disk membranes. A, Wild-type, in which ABCT is a transporter (flippase) for N-ret-PE. B, abcr−/− mouse (and patients with reduced flippase activity). N-ret-PE trapped in the disk combines with a second molecule of all-trans-retinal to produce A2PE-H₂. A2PE-H₂ is ultimately hydrolyzed to form A2E. Many of these reactions occur in the RPE after disks containing the excessive trapped A2PE-H₂ are shed as part of the normal phagocytotic process. The A2E accumulates as lipofuscin in the RPE and may ultimately damage intracellular membranes and destroy the overburdened RPE cells within the macula. A2E: N-retinylidene-N-retinyl-ethanolamine; A2PE-H₂: N-retinylidene-N-retinyl-PE; atRAL: all-trans-retinal; atRDH: all-trans-retinal dehydrogenase; atROL: all-trans-retinol; ops: opsin; PE: phosphatidylethanolamine; PM: plasma membrane. (From Weng J, Mata NL, Azarian SM, Tzekov RT, Birch DG, Travis GH: Insights into the function of Rim protein in photoreceptors and etiology of Stargardt's disease from the phenotype in abcr knockout mice. Cell 1999; 98:13–23.) (See also color plate 31.)

With missing or defective RmP, N-ret-PE accumulates within the intradiskal space (figure 62.3B). The consequences of this are far reaching. One consequence is that “naked” opsin may be activated by the excess levels of free all-trans-retinal within the intradiskal space (ops/atRAL). This activation is believed to produce a noisy receptor in the dark, leading to an equivalent background and consequently, to delays in dark adaptation. These delays have been reported in patients with STGD after exposure to adapting light that bleaches a substantial fraction of the visual pigment.¹⁶ This delay in dark adaptation parallels that found in both homozygote³⁹ and heterozygote²⁴ abcr knock-out mice.

A second consequence of the buildup of N-ret-PE is the combining of a second molecule of all-trans-retinal with N-ret-PE to produce N-retinylidene-N-retinyl-PE (A2PE-H₂). A2PE-H₂ is ultimately hydrolyzed to form N-retinylidene-N-retinyl-ethanolamine (A2E). Many of these reactions occur in the RPE after disks containing the excessive trapped A2PE-H₂ are shed as part of the normal phagocytotic process. The A2E accumulates as lipofuscin in the RPE and may ultimately damage intracellular membranes and destroy the overburdened RPE cells within the macula.¹⁴

Also associated with ABCA4 mutations and therefore part of the spectrum of STGD is a subset of cone-rod dystrophy, a progressive retinal degeneration that is typically inherited as an autosomal-recessive disease. A common early symptom is decreased visual acuity due to macular degeneration. In fact, young patients with CRD form may be thought to have STGD because of the similarity in appearance, but with time, it develops into a more progressive disorder. The severe visual loss manifests as a posterior pole cellophane maculopathy and expanding central scotoma or widespread posterior pole flecks, which then degenerate, leaving an atrophic macular scar. Both forms show the dark choroid effect outside the macular areas that may show hyperfluorescence due to the central degeneration. Patients with CRD have characteristic changes in the full-field ERG that include delayed cone b-wave implicit times.⁵ A subset of patients with CRD shows a prolonged time course of dark adaptation following a bleach.¹⁸ CRD is distinguished from RP on the basis of visual acuity, fundus appearance, ERG findings, and the absence of night blindness as a presenting symptom. In a large prospective study of 100 patients with either CRD or RP, it was shown that the rate of rod ERG loss was significantly lower in CRD than in RP.⁹ Moreover, the rate of rod loss in CRD was similar to the rate of cone loss. This is quite different from RP, in which rod ERG function is lost approximately three times faster than cone function.⁹ In addition, the patterns of visual field loss⁸ and ERG loss¹⁰ are different in CRD and RP. Thus, when ABCA4 mutations take the CRD pathway, it is a clinically distinct retinal disorder from STGD and RP that has widespread involvement of both cone and rod photoreceptors.

Despite the distinctive characteristics of each phenotype, mutations in the human ABCA4 gene that cause STGD have been implicated in a subset of patients with recessive RP and recessive CRD.^13,23,27,37 As in STGD, one consequence of the defect in RmP is the accumulation of all-trans-retinaldehyde within the rod outer segment disks and the production of a persistent “equivalent background” due to transient accumulation of the “noisy” photoproduct. This equivalent background is believed to be a major factor causing the slowed time course of adaptation in patients with ABCA4 mutations. The time course of dark adaptation following a photobleach is shown in figure 62.4 for a patient with CRD. Also shown is the average time course (±1 standard deviation) based on the 15 control subjects. Compared to normal, the time course of recovery in the patient with CRD is remarkably slow. Whereas the average control subject returns to within 0.2log unit of the prebleach (fully dark-adapted) threshold by 25.4 minutes, it took 59 minutes for this patient with CRD to return to within 0.2log unit of the preexposure value. The median recovery time for 11 patients with CRD associated with ABCA4 mutations of 41.6 minutes was significantly longer (t = −4.38, p < .001) than the 25.4 minutes required for the average control subject.¹² Similar delays in the time course of dark adaptation have been reported previously in a subset of patients with CRD.¹⁸ Also similar to this phenotype in CRD patients is the delayed recovery of rod sensitivity following light exposure in mice homozygous³⁹ and heterozygous²⁴ for a null mutation in the abcr gene.

Figure 62.4.  

Time course of dark-adaptation in patient with CRD. Open circles are measured thresholds, solid line is best fit of linear component dark adaptation model. Also shown is the best-fit function for values from 15 normal subjects. The gray area shows ±1 standard deviation. Thresholds were followed until they returned to within 0.2log unit of the patient's prebleach threshold (dashed lines).

After 30 minutes of dark adaptation following a photobleach, thresholds for normal subjects were at their prebleach values, while thresholds for the majority of patients with CRD remained elevated. To determine whether the persistent elevation was associated with an equivalent or noisy background, pupil size was measured after 30 minutes in the dark.³⁰ Patients with CRD and associated ABCA4 mutations had significantly smaller pupil diameters than did normal subjects 30 minutes following a bleach, and the test eye pupil diameter (OS) was consistently smaller than the chemically dilated pupil diameter (OD). This phenotype is apparently due to the accumulation of all-trans-retinaldehyde, which interacts with opsin apoprotein to form a “noisy” photoproduct.^21,23,37 Loss of this transport activity also results in the accumulation of toxic bis-retinoids in the retinal pigment epithelium,^24,25 which may predispose to photoreceptor degeneration.

Research with patients and with animal models of STGD and CRD may also shed light on the mechanisms involved in age-related macular degeneration (AMD). In analyzing data from over 1700 patients with AMD at several centers, Allikmets found that particular ABCA4 alleles raise the risk of AMD.¹ The two most commonly associated variants were G1961E and D2177N. The increase in risk is between threefold and fivefold. Consistent with a role for ABCA4 mutations in heterozygote carriers of ABCA4 are the findings from of transgenic mice. Abcr+/− (abca4+/−) heterozygous mice accumulate A2E in the RPE at a rate approximately intermediate between wild-type and homozygeous mice.²⁴ Delays in recovery from a bleach are also present but are less severe than those in the abcr−/− mice. Interestly, delayed dark adaptation has also been reported in AMD.^28,29

It is interesting that a single gene, ABCA4, can be associated with STG, CRD, RP, and, to some extent, AMD. It has been proposed that the severity of disease is related to the amount of residual RmP activity in a given patient.¹³ Thus, a heterozygote carrying one mutant ABCA4 allele may be at risk for AMD, the degree of risk being related to the severity of the allele. A patient with two mild to moderate mutant alleles would have SRGD. CRD or RP would be the result of inheriting two severe mutant alleles. Although of heuristic value, this model is clearly an oversimplification, and exceptions to this scheme have been documented.¹² Patients with CRD may have the same allele as patients with STGD, even within the same family, and there is no obvious difference in predicted RmP activity that would explain whether a patient has CRD or RP. At the present time, molecular biology appears to have a limited diagnostic role in ABCA4 mutations. Electrophysiology continues to be the technique of choice for determining whether a patient with an ABCA4 mutation has STGD, CRD, or RP. Clearly, the implications of these phenotypic distinctions are enormous for visual prognosis.

Our rapidly evolving understanding of the etiology of STGD is already leading to suggestions for clinical trials to arrest visual loss. On the basis of work in abcr−/− mice, which show less A2E buildup when kept in the dark,³⁹ it seems prudent to recommend that patients with STGD minimize light exposure to the greatest extent practical. Drugs may be available or under development that could inhibit the accumulation of A2E in RPE cells. In this regard, isotretinoin was recently reported to be effective in limiting A2E accumulation in abcr−/− mice.³¹ Finally, drugs and gene therapy have the potential to stimulate under active RmP activity.