Fig 1.
A) The fovea and central visual axis (orange arrow) were approximated on the mid-transverse axial slice by measuring exactly 2.5 disc diameters (DD) temporal to the optic nerve. B) This approximation was verified by superimposing the same measurements on RetCam fundus photography where anatomic landmarks (fovea, optic disc) can be directly visualized.
Fig 2.
Points along the tumor margin were selected in axial (A), sagittal (C) and coronal (E) planes, and point locations were measured as angle of eccentricity relative to the central visual axis (orange arrow) or polar angle relative to the vertical meridian. Each point was then plotted on a polar coordinate graph of the retinal surface based on its measured retinal location (B, D, H). The final tumor margin was interpolated along the measured points (I, J). Funduscopic images were used to verify that the tumor graph was consistent with the actual lesion (K).
Table 1.
Descriptive characteristics of the mapped patients and eyes in relation to those eligible and those unmapped.
Fig 3.
Overview of eligible, mapped, and unmapped patients and eyes.
A) The distribution of age at diagnosis. Heavy bar indicates the median, vertical extent of the box indicates the interquartile range, and the whiskers indicate the full range. B) The wide bars indicate the relative proportions of the eyes in each group. The diagram at right summarizes study exclusion criteria and the number of patients and eyes excluded for each one.
Fig 4.
The distribution of age at diagnosis and tumor size varied between patients with germline or somatic mutations.
A) Age at diagnosis and B) tumor area (expressed as % of the mapped retinal area) as a function of mutation type. C) Parallel coordinate plot illustrating the relationship among age at diagnosis, mutation type, and tumor area for all mapped tumors. Color indicates the age quartile. Tumor area quartiles: μ = smallest, S = small, M = medium, L = large. D) The cumulative tumor burden was mapped over the retina by superimposing the mapped areas of all tumors enrolled in this study. Tumors in the left eye were inverted around the nasal-temporal axis to preserve spatial symmetry with right eye tumors. The color legend indicates the number of tumors which were mapped over a specific point.
Fig 5.
A) Spatial distribution of tumor centroids rendered in the azimuthal-equidistant projection. Plotting symbol color indicates age at diagnosis quartile, and symbol size indicates tumor area quartile, and symbol shape indicates mutation type and disease laterality. B) Distribution of all tumor centroids with respect to polar angle. The optic disk was located at 88 degrees. C) Box-and-whisker plots for tumor centroid eccentricity as a function of mutation type (germline vs. somatic) by age at diagnosis (quartiles, as indicated by color) and by tumor area (quartiles with μ = smallest, S = small, M = medium, L = large).
Fig 6.
The tumor centroid distribution rendered in the azimuthal-equidistant projection by (A) tumor area quartiles and (B) mutation type. C) Scatter plot of tumor centroid eccentricity vs. tumor area. The range of each tumor area quartile is indicated on the x-axis. The black curve shows the approximate maximum eccentricity for a tumor of a given size (see text). Plotting symbol color indicates age at diagnosis quartile (magenta, yellow, blue, green) and shape indicates mutation type and disease laterality.
Fig 7.
Tumor centroid distribution in each age quartile.
A) Spatial point process model with significant clustering of tumor centroids in different parts of the retina as a function of age at diagnosis. The color code indicates the density of tumor centroids relative to the density expected for a completely random spatial distribution of the same number of tumors on the mapped retinal surface. B) Polar plot of tumor centroids in each age quartile with lines connecting multiple tumors within an eye. C) Mapped tumors within each age quartile are superimposed over one another to map the cumulative tumor burden for each quartile. The color code indicates the fraction of the tumors in each age quartile that overlap at a given location.
Fig 8.
Distribution of approximated locations for all 50 unmapped tumors with respect to each patient age quartile.
With increasing age at diagnosis, there was a significant increase in tumor frequency in the inferior quadrants while the number of tumors noted in superior quadrants decreased. No significant effect was noted in variation of tumor occurrence between nasal and temporal quadrants with respect to age.