Patients and methods

Patients were selected from a large ongoing collection of clinical data at the Reading Centre of Moorfields Eye Hospital. Inclusion criteria were: a clinical diagnosis of AMD, both manifest and resolved soft macular drusen, a level of fixation and general fitness sufficient to perform the full test sequence. Patients with exudative late-stage AMD in the eye studied were excluded.

Imaging

Standard 30° stereo colour images (SCI) of the fundus, centred on the fovea (Field 2) were recorded digitally. FA imaging was performed using a cSLO (Heidelberg Retina Angiograph 2, Heidelberg Engineering, Dossenheim, Germany). The fundus was illuminated using a solid-state laser (488 nm), and induced fluorescence was recorded through a long-pass filter with a short-wavelength cut-off at 500 nm. The image of a 30°×30° retinal area centred on the fovea was recorded digitally at 1536×1536 pixel (5 μm/pixel) and 256-level grey-scale resolution. To reduce random noise, a master image was produced by averaging 16 individually recorded images for each eye in the study. Lens opacities—where present—were in all cases diffuse and permitted adequate imaging of FA.

Phenotyping

Detailed phenotyping was performed by grading based on Field 2 colour and FA images, according to the system defined in the International Classification for ARM and AMD (IC),5 independently by two certified graders masked to the identity of the patients and the main aim of the study. These gradings were compared for (both inter- and intraobserver) reliability. A final copy was created by adjudication, and changes over time in the final copy were analysed. In cases where the simultaneous presence of AMD, disappearance of drusen and absence of exudative disease were confirmed, the patient was invited for functional testing.

Psychophysical testing

For psychophysical testing, fine matrix mapping (FMM) was selected due to its superior spatial resolution and its ability to measure wide ranges of sensitivity changes. The full test sequence included a visual acuity test, a standard visual field examination using the 30–2 program of the Humphrey Field Analyzer (HFA, Carl Zeiss, Welwyn Garden City, UK), a photopic FMM test, pupil dilatation (with 1.0% tropicamide and 2.5% phenylephrine), dark adaptation for 45 min, a scotopic 30–2 visual-field test, scotopic FMM and finally the determination of the location and stability of fixation.

Fine-matrix mapping

FMM was performed using a modified HFA. Test flashes were positioned over the retinal area of interest, including both drusen and areas where the resolution of drusen was confirmed by grading. The coordinates of four interlaced 5×5 grids (25 locations with 2° intervals) were entered in the “Custom Grid” feature of the perimeter. Each grid was offset relative to the other grids by 1° on the x, y or both axes. Data thus obtained were merged into a single 9°×9° matrix of 100 test locations with 1° intervals. A standard (Goldmann) size 3 stimulus was used. Photopic testing was done using a white stimulus with a background illuminance of 31.5 apostilbs, and scotopic measurements were done using a blue stimulus with no background illumination. Fixation stability was monitored through an infrared camera. Detection threshold sensitivity was expressed in decibels and thresholds as log units. The maximum stimulus illuminance of 10 000 apostilbs was interpreted as 0 dB.

Data processing

FMM thresholds were processed by (3×3) Gaussian filtering to improve repeatability.12 The attenuation of the original signal and loss of detail inherent in all smoothing methods were insignificant in our case and outweighed by the benefits with respect to noise reduction (repeatability).13 For each patient, sensitivity data were analysed intraindividually as well as compared with age-matched normal data (collected under identical conditions) that have been published,141516 through point-by-point subtraction of averaged normal values at corresponding coordinates within the FMM matrix. Filtered data were used to calculate the mean and the maximum threshold elevation from baseline as well as elevation relative to averaged threshold levels of the normal control group. Interpolated threshold values at 0.25° intervals were used to generate a contour plot showing luminance sensitivity gradients across the matrix. Three-dimensional threshold profiles were also generated. Composite images containing overlays of aligned sequential colour fundus and FA images and sensitivity threshold plots were analysed for correlations between fundus appearance, FA and retinal sensitivity. The total area covered by drusen at baseline and at FMM testing was measured for each zone within the IC grading grid.5 Areas with disappearing drusen, persistent drusen and drusen-free areas were identified, and associated levels of sensitivity loss were compared. Software used include: Adobe Photoshop v7 with the “FoveaPro” plug-in, ImageJ v1.38x, Microsoft Excel and SPSS v11.

Fixation stability

A major factor influencing FMM test reliability is the stability of fixation. For the measurement of fixation stability, a customised cSLO was used in a way similar to a technique described previously.17 Values thus obtained reflect eye movements and are used to calculate the bivariate contour ellipse area (BCEA), which is the area of an ellipse on the retinal surface within which the centre of the target was imaged 68% of the time. BCEA is a standardised measure which provides a means for quantification and comparison of fixation stability. Smaller BCEA values correspond to more precise fixation.

The study protocol adhered to the tenets of the Declaration of Helsinki and was approved by the local ethics committee. Before inclusion, written, informed consent was obtained from each participating patient after explanation of the nature of the study. Maximum retinal irradiance of lasers used was well below the limits established by the American National Standards Institute (ANSI Z136.1; 1993) and other international standards.

Results

Sequential colour fundus and FA images of 960 patients were screened for disappearing drusen. Soft drusen regression was detected in 34 cases (25 spontaneous, nine following prophylactic laser treatment). Nineteen patients met all inclusion criteria, and 14 (10 female, four male) agreed to participate in the study, ranging in age from 52 to 84 years (median age was 72 years). The baseline of the study was defined as the earliest date when both clinical data and fundus SCI were available, and the endpoint as the date on which the patient was last seen clinically. The mean follow-up period was 5.9 years (ranging from 2.8 to 14.4 years).

Phenotype

The predominant phenotype in the study eye at baseline was “soft drusen” in all cases. Disappearance of drusen was followed by NV in one and GA in three cases, while in 10 cases no indications of end-stage disease were seen in the colour fundus images. In most cases, parallel to fading drusen, new drusen in other locations appeared and grew in size and confluence. New drusen tended to form at increasingly peripheral locations relative to the fovea (table 1). Repeated appearance of drusen in the same retinal location was not seen. Representative images are shown in fig 1.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Representative images from selected cases. Columns: (A) the fundus at baseline; (B, C) colour and FA images taken during fine matrix mapping (FMM) testing, white rectangles mark the placement of respective FMM test grid; (D, E) surface plots of photopic and scotopic FMM thresholds. Rows 1–2 and 4 (patients 3, 6 and 11) illustrate the typical picture seen in the majority of patients. Row 3 shows a case (patient 7) where areas with regressed drusen appear normal in the colour fundus image but show a speckled variation in FA. Row 5 shows a case where progression to atrophy was noted (patient 2).

Autofluorescence

Fundus autofluorescence associated with drusen varied from decreased to increased, and no good correspondence was detectable. FA corresponding to areas with disappearing drusen in the absence of pigmentary changes was normal in seven cases. In two patients increased FA was seen, in one case in an area adjacent to the junctional zone of a GA, in the other adjacent to a large crystalline druse. One other patient showed widely varying levels of FA in connection with regressed drusen (fig 1, row 3). GA was associated with decreased FA centrally and increased FA along the boundaries. Crystalline drusen showed decreased FA, while granular hyperpigmentation showed increased FA.

Functional characteristics

Best corrected visual acuities in the study eye assessed at the time of FMM testing ranged from 6/12 to 6/5 (median = 6/6). All patients had less than two lines’ loss in BCVA compared with the baseline value. Fixation stability was good in all cases (table 2). Two patients (2 and 4), both with end-stage disease, showed significant (more than two lines) deterioration in BCVA at the endpoint as compared with the baseline value. FMM showed generalised threshold elevation relative to normal controls under photopic and scotopic conditions (table 2). Scotopic sensitivity loss exceeded photopic loss in all cases (fig 2). Scotopic loss over areas with drusen or regressed drusen did not differ significantly from that over non-drusen areas (p = 0.289 and p = 0.989 respectively, ANOVA; fig 3). Elevated scotopic thresholds were seen to be associated with GA, crystalline drusen and coarse granular hyperpigmentation, all in connection with abnormal FA. Photopic thresholds showed little topographic variation except in areas with GA.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Photopic versus scotopic sensitivity loss of the 14 patients in the study.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Regional differences in scotopic sensitivity loss at retinal locations with (C) drusen, (B) regressed drusen and (A) areas with normal appearance.

Discussion

Regression of macular soft drusen has been described in clinical2418 as well as histopathological studies.319 Gass noted that drusen may fade and disappear, leaving only an irregular mottling of the RPE, while visual acuity may not be affected.3 Gass also observed that most cases of GA follow the fading of drusen or the collapse of a serous detachment of the RPE.3 Drusen disappearance is thus believed to be associated with subsequent degeneration and atrophy of the RPE and photoreceptors.3 It was however noted in large population-based studies that in some patients, drusen may also regress without residual signs.2418 The prognostic implications of true drusen regression are unknown. From the clinical aspect, it raises the possibility that arrested progression or even regression of the disease process may exist naturally. In this study, 10 out of 14 patients showed no ophthalmoscopic indications of manifest or incipient end-stage disease in the fundus following drusen regression. However, in nine, parallel to regression, new drusen in other locations appeared and grew in size and confluence, signifying the continued activity of the disease, with a tendency toward the periphery. The mean retinal sensitivity relative to normal controls was reduced in all patients tested, in both the light and dark adapted states, with significantly higher loss under scotopic conditions. This observation confirms that in AMD, rods are at increased risk for degeneration, and function loss occurs before progression to the late clinical stage.920 Earlier psychophysical studies found that rod-mediated sensitivity declines faster with age than photopic sensitivity, and patients with early AMD have a significantly lower mean central scotopic sensitivity than age-matched controls. Also, in most AMD patients, mean scotopic sensitivity loss exceeded mean photopic sensitivity loss, and the peak deficit in scotopic sensitivity was within 9° of fixation, corresponding to the parafovea.20 Histopathological studies show direct correlates to these findings. In early AMD, a preferential loss of macular rod photoreceptors was demonstrated, with the greatest loss occurring in the parafovea. In late AMD, this leads to a reversal of the macular (9:1) rod predominance seen in the young.9 The rod system also shows altered kinetics with ageing and in AMD.6 One possible explanation for this preferential vulnerability of rods is provided by the retinoid deficiency hypothesis.9 RPE cells are responsible for the transport of nutrients from the choriocapillary circulation to the photoreceptors, as well as for the recycling of the end products of POS disc degradation. 11-Cis retinal (a derivative of vitamin A) is essential for the regeneration of photoreceptor pigment after bleaching by light as well as for photoreceptor survival. In AMD, diffuse sub-RPE deposits may act as a diffusion barrier between the choriocapillaris and the RPE,8 thereby disrupting transport across BrM and leading to a local scarcity of 11-cis-retinal. Vitamin A deprivation is a known cause of outer segment degeneration and photoreceptor death. Cones have an additional retinoid delivery pathway involving Müller cells and possibly the neurosensory retina, and may thus be less vulnerable to reduced transport across the Bruch membrane.921 Although generalised sensitivity loss was measured in all our patients, topographic variation in scotopic (and photopic) sensitivity loss over drusen relative to areas with normal appearance was not significant. This confirms earlier observations that the presence of macular soft drusen seems to have little effect on the local sensitivity of affected retinal areas,2223 with the exception of large, soft foveal drusen. These may be regarded as small RPE detachments and show mildly reduced photopic and considerably reduced scotopic sensitivity as well as increased FA.23 It appears reasonable that the impact of focal barriers such as drusen on RPE metabolism and photoreceptor function may be more limited than that of diffuse deposits. We also found that scotopic sensitivity over areas with regressed drusen was not substantially different from that over unaffected retinal areas; thus we did not find any functional evidence for manifest or incipient photoreceptor atrophy. In diseases where the RPE is primarily affected, the presence of abnormal FA may be an early sign of progression. Areas with increased FA were found to precede the development of new or the enlargement of existing atrophic patches.24 Decreased FA over the drusen overall may also reflect incipient atrophy.25 In eyes with soft drusen but without apparent GA, if focal decreased FA is present, patches with increased FA in areas adjacent to rather than corresponding to drusen may be markers for progression to atrophy.26 In this study, autofluorescence of retinal areas with normal appearance in the colour image following drusen regression was in most cases normal, and none of the above FA patterns associated with impending atrophy was seen. Thus, although the generalised disease persists, retinal sensitivity and FA, two direct and sensitive measures of retinal health, do not indicate that disappearance of drusen is necessarily followed by local function loss and atrophy.

In summary, although the limitations of this study in terms of subject numbers and follow-up time as well as the lack of psychophysical testing before the drusen disappeared may not permit definitive conclusions, our results do indicate that macular soft drusen may fade or disappear without any detectable ophthalmoscopic, FA or psychophysical signs of local dysfunction or incipient atrophy. This phenomenon suggests that some normal-appearing eyes at the present time may have had some features of AMD in the past. This is a potential source of misclassification and needs to be remembered in epidemiological studies investigating the natural history of the disease as well as in clinical trials that evaluate the efficacy of possible therapies.18 It is not known to what extent focal and diffuse sub-RPE deposits coincide and whether the regression of drusen is accompanied by the regression of diffuse deposits. The prognosis for cases with true regression of drusen compared with those without needs to be considered in future studies on AMD.