A Quantitative Assessment of the Burden and Distribution of Lisch Nodules in Adults with Neurofibromatosis Type 1

Abstract

Purpose.: The presence of two or more Lisch nodules (melanocytic hamartomas of the iris) is one of seven diagnostic criteria for neurofibromatosis type 1 (NF1), a common monogenic disorder of dysregulated neurocutaneous growth. The hypothesis that Lisch nodules arise secondary to exposure to ultraviolet (UV) radiation from sunlight was investigated.

Methods.: Lisch nodule burden was mapped and quantified in the irides of 77 adults with NF1. Lifetime sunlight (UV radiation) exposure was inventoried, NF1 neurocutaneous severity determined, and two NF1 mutations predictive of severity selectively genotyped.

Results.: There was high interindividual variability in Lisch nodule burden. Lisch nodules were primarily located in the inferior hemifield (half) of the iris, regardless of its color (P = 3.0 × 10⁻²⁰). Light irides harbored significantly more Lisch nodules than dark irides (P = 4.8 × 10⁻⁵). There was no statistically significant correlation of Lisch nodule burden to lifetime sunlight exposure “dose” or NF1 neurocutaneous severity.

Conclusions.: The difference in Lisch nodule burden between the superior and inferior iris hemifields is most likely due to the sunlight-shielding effects on the superior half by periocular structures. The difference in Lisch nodule burden between light and dark irides is probably due to the photoprotective effects of pigmentation. The genes underlying the control of iris color may thus be viewed as modifiers of severity of Lisch nodule burden in NF1. Given the role of UV radiation and, presumably, DNA damage in Lisch nodule pathogenesis, “benign tumor of the iris,” not “hamartoma,” may be a better descriptor.

Neurofibromatosis type (NF) 1 is a common monogenic disorder of neurocutaneous tissue growth that arises secondary to mutations in the tumor suppressor gene NF1. ^1,2 It features an autosomal dominant pattern of inheritance. Individuals with NF1 typically have an increased predisposition to a variety of benign and malignant tumors and develop neurofibromas, café-au-lait macules, and axillary and inguinal freckling. Variable expressivity, even within families, is common in NF1 and is often attributed to genetic modifiers. ^3,4 The NF1 genotype does not predict severity, with the exception of a 3-bp in-frame deletion in exon 17 (c.2970-2972 delAAT: absence of dermal neurofibromas) ⁵ and microdeletion of NF1 (high neurofibroma burden, dysmorphic features). ⁶  

The NIH Consensus Criteria for the clinical diagnosis of NF1 include the presence of two or more Lisch nodules, traditionally described as “iris hamartomas.” ⁷ They tend to appear early in life ⁸ and only rarely in individuals without the disorder. ⁹ Under slit lamp magnification, Lisch nodules appear as well-defined, gelatinous, dome-shaped elevations of the iris surface. Aside from their usefulness in diagnosis, they have no known clinical consequence. It is controversial whether Lisch nodule burden predicts ^10,11 or does not predict ⁸ cutaneous severity in NF1. By adulthood, Lisch nodules are observed in more than 90% of individuals with NF1. ^8,10–16  

Little is known about the molecular pathogenesis of Lisch nodules. Two reports advocate a melanocytic origin of Lisch nodules. ^17,18 A third study describes the presence of “pigmented cells, fibroblast-like cells and mast cells” in Lisch nodules in a pattern akin to a neurofibroma. ¹⁹ The authors note that certain ultra-structural details in the pigmented cells and the presence of mast cells are evidence of a “Schwannian origin” of Lisch nodules. 

Ultraviolet (UV) radiation from sunlight is hypothesized to be causative in the pathogenesis of Lisch nodules. ^12,20 The unequal distribution of Lisch nodules in the iris is hypothesized to be secondary to the relative shielding of the superior hemifield (half) of the iris from sunlight-derived UV radiation (hereafter, “UV radiation”) by the eyelid and eyebrow. ^12,21  

The role of UV radiation in Lisch nodule pathogenesis is controversial ²² and has not been formally evaluated. In this report, we quantitatively determined Lisch nodule burden (number of Lisch nodules in the irides) and distribution in a large cohort of adults with NF1. We hypothesized a priori that the inferior hemifield of the iris harbors a greater number of Lisch nodules than the superior one, due to the shielding effects of periocular structures. ^23–25 Second, we examined the effects of iris color (dark versus light) on Lisch nodule burden. Third, we collected quantitative estimates of ocular UV radiation exposure and explored its effects on Lisch nodule burden. Fourth, we investigated the relationship of Lisch nodule burden with café-au-lait macule and neurofibroma burden, two measures of NF1 neurocutaneous severity. Fifth, we examined differences in Lisch nodule burden between the left and right irides. Last, we determined the frequency of certain NF1 genotypes (specifically, c.2970-2972 delAAT and NF1 microdeletion) in our population and evaluated their effects, if any, on Lisch nodule burden. 

Methods

Lisch Nodule Evaluation and Determination of NF1 Neurocutaneous Severity

In a natural history study of neurofibromatosis type 1, we evaluated 77 postpubertal individuals from 56 families who met the NIH consensus criteria for NF1. ⁷ Each individual was evaluated by a single physician (DRS). Both eyes were digitally photographed with a slit lamp under consistent lighting conditions (ReSeeVit 5V; Veatch Ophthalmic Instruments, Tempe, AZ). For each eye, a map of the distribution and number of Lisch nodules was sketched during the evaluation. Care was taken to note Wolfflin-Kruckman spots, iris mammillations, and flat iris lesions such as freckles and nevi. During a history and physical examination, a semiquantitative determination of the number dermal neurofibromas (0, 1–10, 11–100, 101–500, 500+) and a count and distribution of café-au-lait macules greater than 1.5 cm (using a Wood's lamp and ruler) was made. The study was approved by the institutional review board of the National Human Genome Research Institute and adhered to the tenets of the Declaration of Helsinki. After explanation of the nature and possible consequences of the study, informed consent was obtained from all participants. 

Determination of Iris Color

The color of each iris was determined by comparing it to three photographic standards obtained from the Beaver Dam Eye study. ²⁶ The three standards varied from little iris pigment (blue-gray) to significant iris pigment (medium to dark brown). We compared each iris to the two extreme standards to define it as “light” (blue-gray to hazel) or “dark” (brown). Three masked observers independently examined each iris with disagreements settled by consensus. 

Mapping Lisch Nodule Distribution

We used NIH Image software (ImageJ ver. 1.40 for Macintosh OS X; provided in the public domain by the National Institutes of Health, Bethesda, MD and available at http://rsbweb.nih.gov/ij/index.html) to process the digital slit lamp images. The pupil and corneal light reflection spots (if any) and surrounding cornea were removed, leaving solely the iris for analysis. The Walter band-pass filter (ImageJ plug-in: http://rsb.info.nih.gov/ij/docs/source/ij/plugin/filter/FFTFilter.java.html) was then applied to the image to smooth and equalize the shading. Another ImageJ plug-in, Threshold Color (http://rsb.info.nih.gov/ij/plugins/rgb-measure-plus.html) was then used to filter out all non-Lisch nodule color. The position of each Lisch nodule was verified with the map made at the time of slit lamp photography. Nodule location (on a Cartesian plane, relative to the pupil center), area, distance from pupil center, and maximum nodule diameter were recorded. The resultant image was a map of Lisch nodule silhouettes and permitted precise assignment of each nodule into one of the four quadrants of each iris: inferonasal, inferotemporal, superonasal, or superotemporal. The images were also compiled into stacks. The z-Project function in ImageJ enabled projection of the stacks to render a collective map of Lisch nodule distribution and density. The color of the map was calibrated to reflect the density of Lisch nodules in a given area. 

Quantification of Environmental UV Radiation Exposure

Participants completed a questionnaire that sought to quantify environmental UV radiation exposure (Supplementary Fig. S1) using both quantitative measures (years) and five-point Likert scales (1, never; 3, occasionally; 5, frequently). We measured both easily recalled objective data (e.g., age, places lived) and more subjective data (recall of frequency of sunglasses use and time spent outdoors). Specifically, we inventoried the length of time lived in various locales since birth (years), time spent wearing eyeglasses or contact lenses (years), frequency of sunglasses use in the past 10 years (Likert scale), amount of time spent outdoors between 10 AM and 3 PM in the past 10 years (Likert scale), amount of time spent outdoors in spring and summer (April-August) for work, home, and recreation (Likert scale), and history of intense UV radiation-exposure activities such as spring skiing, ice fishing or over-water fishing. ^25,27–32 From the list of places lived, we calculated a weighted lifetime average latitude and elevation. The percentage of lifespan spent wearing eyeglasses and/or contact lenses was calculated. Eyeglasses and contact lenses filter UV radiation and thus decrease exposure to ocular tissues. Sunglasses may paradoxically increase UV radiation exposure to ocular tissues. ²⁵ The five-point Likert scales for time spent outdoors for work, home, and recreation were summed as the outdoor score (range, 3–15). 

Statistical Analysis and Control of Multiple Hypothesis Testing

Before collecting our data on Lisch nodule distribution, iris color, and UV radiation exposure, we a priori formulated five primary and 33 secondary hypotheses about Lisch nodule distribution, pathogenesis and natural history (Tables 1, 2). The prior formulation of primary hypotheses prevented the statistically deleterious effects of uncontrolled a posteriori data exploration. For each individual, we summed the number of Lisch nodules from the right and left irides. We used nonparametric tests (the signed rank and Kruskal-Wallis tests of location in paired and unpaired data, respectively, and the Spearman rank correlation coefficient), which do not assume normality and are conservative. 

Table 1.

 

Primary Hypotheses Investigating the Relationship of Sunlight-Derived UV Radiation Exposure and Lisch Nodule Burden in Individuals with NF1

Number	Hypothesis	Dataset	Test	P	Bonferroni P	Confidence Interval (95% Simultaneous)
1	The number of Lisch nodules in the inferior hemifield of the iris is greater than the number of nodules in the superior hemifield of the iris.	All individuals (n = 77)	Signed rank	6.1 × 10−21	3.04 × 10−20	10.79 to 22.82
2	Individuals with light-colored irides harbor a greater number of Lisch nodules than do individuals with dark irides.	Dark irides: n = 28	Kruskal-Wallis	9.6 × 10−6	4.8 × 10−5	−66.16 to −22.14
		Light irides: n = 49
3	The number of Lisch nodules in the nasal hemifield of the iris is less than the number of nodules in the temporal hemifield. (See also hypotheses 14a and 14b, Table 2.)	All individuals (n = 77)	Signed rank	NS	NS	−2.34 to 1.88
4	Individuals who have a more southerly weighted latitude of residence have more Lisch nodules. (See also hypotheses 15a and 15b, Table 2.)	All individuals with latitude data (n = 75)	Spearman correlation	NS	NS	−0.27 to 0.32
5	Lisch nodule burden will increase with age. (See also hypotheses 16a and 16b, Table 2.)	All individuals (n = 77)	Spearman correlation	NS	NS	−0.28 to 0.30

Table 2.

 

Secondary Hypotheses Investigating the Role of Certain Behaviors and Geography in Lisch Nodule Burden, Correlation of Lisch Nodule Burden with Neurocutaneous Severity, Lisch Nodule Burden in Certain Iris Quadrants, and the Sex of Individuals with NF1

Number	Hypothesis	Dataset	Test	P
6a	Individuals who have a greater weighted elevation of residence will have more Lisch nodules.	All individuals with elevation data (n = 74)	Spearman correlation	NS
6b	Individuals with dark irides who have a greater weighted elevation of residence will have more Lisch nodules.	All individuals with elevation data and dark irides (n = 26)	Spearman correlation	NS
6c	Individuals with light irides who have a greater weighted elevation of residence will have more Lisch nodules.	All individuals with elevation data and light irides (n = 48)	Spearman correlation	NS
7a	Individuals with a greater percentage of life spent wearing eyeglasses or contact lenses will have fewer Lisch nodules.	All individuals with eyeglass/contact lenses data (n = 73)	Spearman correlation	NS
7b	Individuals with dark irides and a greater percentage of life spent wearing eyeglasses or contact lenses will have fewer Lisch nodules.	All individuals with eyeglass/contact lenses data (n = 27)	Spearman correlation	NS
7c	Individuals with light irides and a greater percentage of life spent wearing eyeglasses or contact lenses will have fewer Lisch nodules.	All individuals with eyeglass/contact lenses data (n = 46)	Spearman correlation	NS
8a	Individuals with more frequent sunglasses use will have fewer Lisch nodules.	All individuals with sunglasses data (n = 73)	Spearman correlation	NS
8b	Individuals with dark irides and more frequent sunglasses use will have fewer Lisch nodules.	All individuals with sunglasses data (n = 27)	Spearman correlation	NS
8c	Individuals with light irides and more frequent sunglasses use will have fewer Lisch nodules.	All individuals with sunglasses data (n = 46)	Spearman correlation	NS
9a	Individuals with greater time spent outdoors between 10 AM and 3 PM will have more Lisch nodules.	All individuals with time outside data (n = 73)	Spearman correlation	NS
9b	Individuals with dark irides and greater time spent outdoors between 10 AM and 3 PM will have more Lisch nodules.	All individuals with time outside data (n = 27)	Spearman correlation	NS
9c	Individuals with light irides and greater time spent outdoors between 10 AM and 3 PM will have more Lisch nodules.	All individuals with time outside data (n = 46)	Spearman correlation	0.02
10a	Individuals with a higher cumulative outdoor score will have more Lisch nodules.	All individuals with outdoor score data (n = 72)	Spearman correlation	NS
10b	Individuals with dark irides and a higher cumulative outdoor score will have more Lisch nodules.	All individuals with outdoor score data (n = 27)	Spearman correlation	NS
10c	Individuals with light irides and a higher cumulative outdoor score will have more Lisch nodules.	All individuals with outdoor score data (n = 45)	Spearman correlation	NS
11a	Lisch nodule burden will positively correlate with severity of café-au-lait macule burden.	All individuals with CALM burden data (n = 76)	Spearman correlation	NS
11b	In individuals with dark irides, Lisch nodule burden will positively correlate with severity of café-au-lait macule burden.	All individuals with CALM burden data (n = 27)	Spearman correlation	NS
11c	In individuals with light irides, Lisch nodule burden will positively correlate with severity of café-au-lait macule burden.	All individuals with CALM burden data (n = 48)	Spearman correlation	NS
12a	Lisch nodule burden will positively correlate with severity of neurofibroma burden.	All individuals with cutaneous neurofibroma burden data (n = 71)	Spearman correlation	NS
12b	In individuals with dark irides, Lisch nodule burden will correlate with severity of neurofibroma burden.	All individuals with cutaneous neurofibroma burden data (n = 25)	Spearman correlation	NS
12c	In individuals with light irides, Lisch nodule burden will correlate with severity of neurofibroma burden.	All individuals with cutaneous neurofibroma burden data (n = 43)	Spearman correlation	0.08
13a	Lisch nodule burden will correlate with a history of ever participating in the intense UV-exposure activity of spring snow skiing.	All individuals with activity data (n = 72)	Kruskal-Wallis	NS
13b	Lisch nodule burden will correlate with a history of ever participating in the intense UV-exposure activity of ice fishing.	All individuals with activity data (n = 72)	Kruskal-Wallis	NS
13c	Lisch nodule burden will correlate with a history of ever participating in the intense UV-exposure activity of ocean, bay, or lake fishing.	All individuals with activity data (n = 72)	Kruskal-Wallis	NS
14a	The number of Lisch nodules in the superior nasal hemifield of the iris is less than the number of nodules in the superior temporal hemifield.	All individuals (n = 77)	Signed rank	0.02
14b	The number of Lisch nodules in the inferior nasal hemifield of the iris is less than the number of nodules in the inferior temporal hemifield.	All individuals (n = 77)	Signed rank	0.02 (t = −1.83)
15a	Individuals with dark irides who have a more southerly weighted latitude of residence have more Lisch nodules.	All individuals with latitude data and dark irides (n = 27)	Spearman correlation	NS
15b	Individuals with light irides who have a more southerly weighted latitude of residence have more Lisch nodules.	All individuals with latitude data and dark irides (n = 48)	Spearman correlation	NS
16a	Lisch nodule burden will increase with age in individuals with dark irides.	All individuals with dark irides (n = 28)	Spearman correlation	NS
16b	Lisch nodule burden will increase with age in individuals with light irides.	All individuals with dark irides (n = 49)	Spearman correlation	NS
17a	Median Lisch nodule burden is not influenced by the sex of the person.	All individuals (n = 77)	Kruskal-Wallis	NS
17b	In individuals with light irides, median Lisch nodule burden is not influenced by the sex of the person.	All individuals with light irides (n = 28)	Kruskal-Wallis	NS
17c	In individuals with dark irides, median Lisch nodule burden is not influenced by the sex of the person.	All individuals with dark irides (n = 49)	Kruskal-Wallis	NS

 Parametric tests (paired t-test, 2-sample t-test, and Pearson's correlation coefficient) were also performed as a confirmation (data not shown). NS, not significant.

For the five primary hypotheses, we used the Bonferroni principle to construct conservative approximate 95% simultaneous confidence intervals about the five primary estimated effects by erecting a 99% individual confidence interval about each one. To determine the confidence intervals, we used Fisher's approximation with Pearson's correlation coefficient (hypotheses numbers 4 and 5) and standard methods for paired differences (hypotheses 1 and 3) and differences between groups (hypothesis 2). These intervals allow us to reject, at an experiment-wise level of 5%, any set of hypothesized effect values that does not lie completely within the five 99% confidence intervals. 

Our dataset is powered to detect strong to moderate effects in a correlation analysis. To determine power in our five primary and selected secondary hypotheses, we calculated that a true correlation of 0.39 or better was needed to guarantee at least 80% power (if n > 70, α = 0.01, two-tailed). In addition, for the comparison of Lisch nodules in nasal and temporal hemifields, we calculated that a true hemifield difference of least 2.7 nodules is needed, assuming an SD of 7, a value derived from our data. 

To control for the effects of multiple hypothesis testing, we applied the conservative Bonferroni principle to the nominal probabilities determined for the five primary hypotheses. The 33 secondary hypotheses are exploratory and investigate (1) the role of certain behaviors and geography (eyeglasses/sunglasses use, time spent outside, elevation) on Lisch nodule burden, (2) the correlation of Lisch nodule burden with neurocutaneous severity (neurofibroma and café-au-lait macule burden), (3) Lisch nodule burden in certain iris quadrants and, (4) the role of the sex of the individual in Lisch nodule burden. Since no multiple testing correction was applied, caution is necessary in interpreting a significant secondary hypothesis. 

We used the following model to distinguish contributions to Lisch nodule burden variance between patients from contributions between eyes within patients:  

 

where x = Lisch nodule burden for a given eye; i = 1,…, n = 77 people; j = 1 (left eye) or 2 (right eye); μ = overall mean Lisch nodule burden; var(τ) = interindividual Lisch nodule variance; and var(ε) = variance due either to eyes within individuals, or to stochastic noise. We modeled the percentage of Lisch nodule variance due to the person as 100 · var(τ)/[var(τ) + var(ε)]. 

NF1 Genotyping: c.2970-2972 delAAT and NF1 Microdeletion

We sought evidence of the 3-bp in-frame deletion in exon 17 (c.2970-2972 delAAT) in 10 individuals with NF1 with no palpable dermal neurofibromas. Genotyping was performed as per Upadhyaya et al. ⁵ Briefly, a 386-bp fragment was amplified with forward (5′-ATTTGGCTCTATGCCTGTGG-3′) and reverse (5′-CACACCCTAGTTTGTGTGCAG-3′) primers. Purified amplicons were sequenced with forward and reverse sequencing primers and analyzed (Lasergene software; DNAStar, Madison, WI). In 12 individuals from 12 families with high neurofibroma burden (500+), we sought evidence of the NF1 microdeletion by assessment of loss of heterozygosity (LOH) by single nucleotide polymorphism (SNP) genotyping. We selected 19 SNP genotyping assays (TaqMan; Applied Biosystems, Inc. [ABI] Foster City, CA) with an SNP minor allele frequency of ∼0.25 or greater spanning ∼1.5 Mb of the NF1 locus (Table 3). To aide in haplotype construction, we genotyped all available family members from the 14 sample and control pedigrees (44 individuals). The reactions were performed with a real-time PCR system (model 7500; ABI) according to the manufacturer's protocol. Genotype data were assessed by Merlin software for Mendelian inconsistencies. ³³ Haplotypes of the NF1 locus were constructed by hand and confirmed by Merlin. For those individuals with noninformative SNP genotyping, we performed multiplex ligation probe analysis (MLPA; Holland MRC, Amsterdam, The Netherlands) of the NF1 locus according to the manufacturer's instructions. 

Table 3.

 

List of Genotyped SNP Loci Used in the Determination of NF1 Microdeletion Status

dbSNP Name	ABI Assay Name*	Gene Symbol	Chr. 17 Base Position	Distance to Upstream SNP (bp)
rs7219456	C_403750_10	EFCAB5	25,398,238	0
rs8072345	C_7911284_10	BLMH	25,628,415	230,177
rs9913237	C_2980234_10	CPD	25,785,065	156,650
rs170053	C_1083154_10	TBC1D29	25,917,209	132,144
rs1123232	C_11619935_10	NF1	26,448,115	530,906
rs2012581	C_2263017_10	NF1	26,584,926	136,811
rs4795587	C_32294423_10	NF1	26,612,043	27,117
rs8080679	C_2193426_10	NF1	26,626,736	14,693
rs4795593	C_2193435_10	NF1	26,637,726	10,990
rs9902893	C_2533273_10	NF1	26,649,764	12,038
rs2040792	C_11941361_10	NF1	26,652,675	2,911
rs7505	C_7562823_1_	NF1	26,668,978	16,303
rs7226006	C_2533298_10	NF1	26,674,782	5,804
rs2057769	C_11941321_10	NF1	26,715,494	40,712
rs1800845	C_7562857_10	NF1	26,727,564	12,070
rs178858	C_1083716_10	RAB11FIP4	26,797,877	70,313
rs11657523	C_2181826_10	RAB11FIP4	26,838,188	40,311
rs12937238	C_2181841_10	RAB11FIP4	26,846,982	8,794
rs2343245	C_16210542_10	RAB11FIP4	26,859,343	12,361

 Chromosome coordinates from dbSNP Build 129.

*  Applied Biosystems, Inc., Foster City, CA.

Results

Range of Age, Ethnicity, NF1 Neurocutaneous Severity, UV Radiation Exposure and Iris Color

The 77 participants ranged in age from 15.5 years to 76 years, with an average age of 37 years and a median age of 34 years. There were 28 males and 49 females and 18 parent–child pairs and 7 sibling pairs. Seven (9%) people noted African ancestry, and two (2.6%) people noted Asian ancestry. A spectrum of severity of the neurocutaneous manifestations of NF1 was represented. Accurate counts of café-au-lait macule burden in 76 (99%) of 77 participants were available and ranged from 4 to 34 café-au-lait macules greater than 1.5 cm (mean: 16, median 14). We assigned a semiquantitative assessment of dermal neurofibroma burden to 71 (92%) of 77 participants. Of the assessed 71 participants, 11% had no apparent dermal neurofibromas, 34% had 1 to 10, 23% had 11 to 50, 6% had 51 to 100, 11% had 101 to 500, and 15% had 500 or more. 

We obtained completed UV radiation exposure questionnaires from 73 (95%) of 77 participants. Two females (one with dark irides and one with light irides) were lost to follow-up. For two other participants (a brother and sister, both with light irides), we obtained data only on places lived from a family member. Participants spent most of their lives in the United States, a fact reflected in the range of weighted lifetime average latitude (20.7°–47.6° north) and elevation (sea level to >5000 feet). For the 73 participants with completed questionnaires, 16 (22%) never wore eyeglasses or contact lenses. Of the remaining 78% of participants, use of eyeglasses/contact lenses ranged from less than 1% to greater than 95% of lifespan. A wide range of frequency of sunglasses use in the past 10 years (four with no data; range 1–5; mean 3.4), amount of time spent outdoors between 10 AM and 3 PM in the past 10 years (four with no data; range 1–5; mean 3.76), outdoor scores and history of intense UV radiation exposure (five with no data; range 3–15; mean 9.09) activities were observed. 

There were 49 (64%) participants with light irides (age range, 15.5–76 years; mean, 38 years) and 28 (36%) participants with dark irides (age range, 16–59 years; mean, 35 years). Of the 28 individuals with dark irides, 8 were of African or Asian ancestry. One man with African and Caucasian ancestry had light irides. There were no examples of heterochromia. 

Unequal Distribution of Lisch Nodule Burden in Superior Versus Inferior Iris Hemifields but Not in Temporal versus Nasal Iris Hemifields

Lisch nodules were observed in 73 (95%) of 77 participants. All participants had clear images from both irides. Of the four individuals without observed Lisch nodules, three had dark irides (31-year-old woman, 24-year-old man, and 35-year-old-man) and one had light irides (48-year-old woman). Of the five primary hypotheses subjected to correction for multiple testing (Table 1), results for hypothesis 1 (number of Lisch nodules in the inferior hemifield of the iris is greater than the number of nodules in the superior hemifield of the iris) was statistically significant (Bonferroni adjusted P = 3.0 × 10⁻²⁰). Figure 1 maps the distribution of 2826 Lisch nodules from both irides of all participants. To investigate the Coroneo effect (focusing of UV radiation on the medial limbus by the cornea ³⁰ ), we examined differences in the quantity of Lisch nodules in temporal and nasal hemifields (hypothesis three) but found no statistically significant evidence of unequal distribution. An unequal distribution of Lisch nodules was observed between the superonasal and superotemporal quadrants (Table 2, secondary hypothesis 14a, signed rank test nominal P = 0.02). An unequal distribution of Lisch nodules between the superior quadrants was also evident (Fig. 1). The distribution of Lisch nodules between the inferonasal and inferotemporal quadrants (Table 2, secondary hypothesis 14b) was of borderline significance (signed rank test nominal P = 0.02), but in the opposite direction. However, since tests of secondary hypotheses were not a priori subject to Bonferroni correction, their significance, if any, must be interpreted with caution. 

Figure 1.

 

Map of Lisch nodule distribution in the right and left irides. Data are from 73 adults with NF1 and Lisch nodules. Data from both dark and light irides are included. Nodules are displayed as silhouettes of their digital images.

Lisch Nodule Burden in Light Irides versus Dark Irides

Results for hypothesis 2 (light irides harbor a greater number of Lisch nodules than do dark irides) was statistically significant (Bonferroni corrected P = 4.8 × 10⁻⁵). The mean number of Lisch nodules in individuals with dark and light irides was 8.6 and 52.8, respectively (two-sample t-test P < 0.0001). The median number of Lisch nodules in individuals with dark and light irides was 4 and 35, respectively (Kruskal-Wallis P = 0.0002). Figures 2 and 3 map the distribution of Lisch nodules from both eyes of participants with dark and light irides, respectively. In both light and dark irides, the number of Lisch nodules in the inferior hemifield of the iris was significantly greater than the number of nodules in the superior hemifield of the iris (Bonferroni corrected P = 0.004, dark irides; Bonferroni corrected P < 0.0001, light irides). 

Figure 2.

 

Map of Lisch nodule distribution in the right and left irides. Data are from 25 individuals with NF1 and Lisch nodules who have dark irides. Nodules are displayed as silhouettes of their digital images.

Figure 3.

 

Map of Lisch nodule distribution in the right and left irides. Data are from 48 individuals with NF1 and Lisch nodules who have light irides are included. Nodules are displayed as silhouettes of their digital images.

Correlation of Lisch Nodule Burden with Age, Weighted Lifetime Latitude of Residence, and Other Measures of Differential UV Radiation Exposure

The two primary hypotheses investigating Lisch nodule burden and differential amounts of UV radiation exposure in our sample (Table 1, hypotheses 4 and 5: southerly latitudes and increasing age) were not significant. Results for nearly all the secondary hypotheses (Table 2) exploring the relationship of Lisch nodule burden and age and various measures of UV radiation exposure, even after adjustment for iris color, were nonsignificant. One of the secondary hypotheses had a marginally significant nominal probability (9c: individuals with light irides and greater time spent outdoors between 10 AM and 2 PM will have more Lisch nodules; nominal Pearson P = 0.06, nominal Spearman P = 0.02). However, after Bonferroni correction the findings for this hypothesis were not significant. 

Correlation of Lisch Nodule Burden with Neurofibroma or Café-au-Lait Macule Burden

Results for the secondary hypotheses exploring the correlation of café-au-lait macule burden (11a, 11b, and 11c) and dermal neurofibroma burden (12a, 12b, and 12c) with Lisch nodule burden were nonsignificant (Table 2). Lisch nodule burden in the eight participants who lacked palpable dermal neurofibromas varied, depending on iris color: in the three individuals with dark irides, Lisch nodule burden ranged from 1 to 4; in the four individuals with light irides, the burden ranged from 7 to 64. Similarly, of the 11 participants with an estimated 500 or more dermal neurofibromas, five had dark irides; the Lisch nodule burden in these individuals ranged from 0 to 20. In the six individuals with light irides, the burden ranged from 16 to 167. In participants at the extremes of dermal neurofibroma burden, Lisch nodule burden was primarily determined by iris color. 

Lisch Nodule Burden in Left Iris Predicts Burden in Right Iris

To investigate the degree to which Lisch nodule burden in one iris predicts the burden in the contralateral iris, we determined the Pearson and Spearman correlation coefficients. The Pearson correlation between left iris and right iris was 0.96 (P < 0.0001) and the Spearman correlation was 0.91 (P < 0.0001). The mean number of Lisch nodules in the right and left irides was 18.4 and 18.3, respectively (not significant). Lisch nodule burden in the left iris is plotted against burden in the right iris in Figure 4. To understand the contribution of interindividual variance and between-iris variance in the determination of Lisch nodule burden, we estimated these values to be: Var (τ) = 608.8 and Var(ε) = 28.8. That is, approximately 95% of the total variance of Lisch burden is due to differences between individuals; Lisch nodule burden between irides differs little within an individual. 

Figure 4.

 

Lisch nodule burden in the left iris correlated highly with the burden in the right iris.

The NF1 c.2970-2972 delAAT Mutation and NF1 Microdeletion

Only one individual of the 10 samples sequenced for the NF1 c.2970-2972 delAAT mutation harbored the mutation. The participant, a 44-year-old female with hazel (light) irides, had a total of 11 Lisch nodules in her eyes (30th centile for Lisch nodule burden among individuals with light irides). In this one individual, the NF1 c.2970-2972 delAAT mutation does not preclude Lisch nodule development. 

In five individuals, all or nearly all the genotyped NF1 SNP loci were heterozygous; an NF1 microdeletion was deemed unlikely. The remaining seven samples were noninformative at all or nearly all the genotyped NF1 SNP loci. In these seven samples plus a control and an additional sample not subject to SNP genotyping, a microdeletion of the NF1 gene was excluded by multiplex ligation probe analysis MLPA. 

Discussion

Our data show that most Lisch nodules are located in the inferior hemifield of the iris, regardless of its color. This finding is consistent with the effects of differential UV radiation exposure. Second, we showed that light irides harbor significantly more Lisch nodules than dark irides. Third, for modest effect sizes and regardless of iris color we did not find evidence of significant correlation between Lisch nodule burden and two measures of neurocutaneous severity in NF1: café-au-lait burden and neurofibroma burden. This result is in contrast to those in earlier reports, ^10,11 in which the investigators did not control for iris color. Fourth, for modest effect sizes and regardless of iris color we observed no statistically significant correlation between Lisch nodule burden and various measures of UV radiation exposure. Fifth, the one example of the NF1 c.2970-2972 delAAT mutation in our population was not associated with an absence of Lisch nodules. Last, Lisch nodule burden in one iris was highly predictive of Lisch nodule burden in the contralateral iris. An analysis of variance components found high interindividual, but not intereye, variance in Lisch nodule burden in our population. Taken together, our data suggest that NF1-haploinsufficiency and ocular UV radiation exposure are, in most individuals, necessary and sufficient for the pathogenesis of Lisch nodules. However, the ultimate burden of Lisch nodules is modulated by other nonrandom factors. 

The most important of these factors is iris color, which in humans is primarily determined by the quantity and ratio of eumelanin (dark brown to black) and pheomelanin (reddish pink). ^34–36 Melanins protect against the adverse effects of UV radiation by serving as a physical photoscreen and as a free radical scavenger. Melanins are produced by the melanocytes of the iris stroma from which Lisch nodules are thought to arise. ^17,18 The antioxidant properties of eumelanin are superior to those of pheomelanin. ³⁵ On exposure to UV radiation, pheomelanin can generate hydroxyl radicals and superoxide anions and thus may be phototoxic itself. ³⁶ Consistent with this observation, individuals with light irides are at greater risk for uveal melanoma, which like Lisch nodules also has an inferior hemifield predilection. ^35,37  

To investigate the high interindividual variability (var(τ)) in Lisch nodule burden, we examined the role of two important NF1 genotypes and lifetime UV radiation exposure. The NF1 c.2970-2972 delAAT mutation and NF1 microdeletion were rare in our population; the single example of the ΔAAT mutation did not preclude Lisch nodule development. Quantification of any environmental exposure, such as UV radiation from sunlight, is difficult and is limited by the participant's recall. History of residence (lifetime weighted latitude) and the participant's age, both easily recalled, have a direct correlation with lifetime UV radiation exposure. ³² However, in our study, neither measure correlated with Lisch nodule burden. It is possible that local geographic and atmospheric factors (e.g., ground cover, cloud cover, and stratospheric ozone) affecting albedo exert a greater influence on ocular UV radiation exposure than participant's age and lifetime average latitude. ³⁰ Such factors were not inventoried in our survey. It is also possible that Lisch nodule pathogenesis is akin to cutaneous melanoma in which dose per exposure, not cumulative UV radiation dose, is critical. ³⁶ We asked about a history of spring skiing, ice fishing, and over-water fishing, behaviors known to generate intense UV radiation exposure. ³² No correlation with Lisch nodule burden and these activities was observed, although the number of participants was small. Lastly, the lack of correlation between an environmental exposure and Lisch nodule burden may reflect the limitations of our questionnaire, which has not been validated to gauge its efficacy in determining lifetime UV radiation exposure. 

In skin melanocytes, there is a dose-dependent generation of cyclobutane pyrimidine dimers after exposure to UV radiation. ³⁶ However, there is wide interindividual variability in the rate of DNA repair after the minimum erythema dose of UVA/UVB radiation to skin, even after adjustment for skin phototype. ³⁸ Bi-allelic inactivation of NF1 is observed in many NF1-associated tumors, ^39–41 suggesting that DNA damage and repair are central in the pathogenesis of NF1. Bi-allelic inactivation of NF1 has also been described in melanocytes from café-au-lait macules from two individuals with mosaic NF1. ⁴²  

We are not aware of any investigation demonstrating NF1 bi-allelic inactivation in Lisch nodules. However, NF1 bi-allelic inactivation in melanocytes secondary to UV radiation-induced DNA damage is a plausible, even likely, mechanism for Lisch nodule pathogenesis. Lisch nodules are usually described as “iris hamartomas.” The term “hamartoma” typically applies to an excessive but focal overgrowth of cells and tissues native to the organ in which it occurs. Although the cellular elements are mature, they do not reproduce the normal architecture of the surrounding tissue. They are frequently seen in infancy and childhood. ⁴³ The use of this term should be reconsidered given the histologic and ultrastructural evidence, in a recent report, of fibroblast-like cells and mast cells, in addition to pigmented cells, in Lisch nodules. ¹⁹ The presence of fibroblast-like cells and mast cells, absent in common melanocytic nevi but observed in neurofibromas, suggests that Lisch nodules are more akin to benign tumors like neurofibromas than hamartomas. ¹⁹ These pathologic observations, if replicated, combined with evidence of an increase in Lisch nodule burden in adult life, and not just in childhood, would also be consistent with the behavior of an NF1-associated benign tumor. Given the pathologic evidence, the role of UV radiation and, presumably, DNA damage in their pathogenesis, “benign tumor of the iris” may be a more appropriate description for Lisch nodules. 

The elevated interindividual variance (var(τ)), low between-eye and stochastic variance (var(ε)) and absence of a lifetime UV radiation dose effect highlight the importance of individual, possibly genetic, differences in the determination of Lisch nodule burden. The lack of significant correlation between neurofibroma burden and café-au-lait burden with Lisch nodule burden is consistent with previous studies advocating the existence of trait-specific modifying loci in NF1. ^3,4 Genetic modifiers may explain (1) the high correlation in Lisch nodule burden between eyes, (2) the variability in age of appearance of Lisch nodules, ⁸ (3) the range in burden for a given iris color and 4) the lack, in multiple studies including this one, of Lisch nodules in 5% to 10% of adults with NF1. Genetic modifiers of Lisch nodule burden might (1) influence the efficiency of response to UV radiation-inflicted DNA damage, (2) buffer reactive oxygen species, and/or (3) direct DNA repair. Last, iris color itself is hereditary ⁴⁴ and is of primary importance in the determination of Lisch nodule burden. The genes influencing iris color may themselves be reasonably viewed as genetic modifiers of severity of Lisch nodule burden in NF1. 

Supplementary Materials

Supplementary Figure S1 - (PDF) 

Footnotes

 Supported in part by the Intramural Research Program of the National Human Genome Research Institute, National Institutes of Health. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products or organizations imply endorsement by the U.S. Government.

Footnotes

 Disclosure: S. Boley, None; J.L. Sloan, None; A. Pemov, None; D.R. Stewart, None

Footnotes

 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.

The authors thank David Sliney, PhD, for development of the UV radiation exposure questionnaire; Neal Oden, PhD (Emmes Corp., Rockville, MD), for statistical support; Michelle Bloom for technical assistance; Ludwine Messiaen (University of Alabama, Birmingham) for the MLPA work, Marypat Jones and Ursula Harper (National Human Genome Research Institute, Bethesda, MD) for assistance in genotyping; and Vincent Riccardi for comments.