Editor—Neurofibromatosis 2 (NF2) is an autosomal dominant disorder that is characterised by benign nervous system tumours (such as vestibular schwannomas (VSs), intracranial meningiomas, and spinal tumours) and other abnormalities.1Somatic mosaicism (the presence of a mutation, deletion, or chromosomal abnormality in a subpopulation of somatic cells) is thought to be relatively common in NF2, affecting perhaps 15% of sporadic cases.2 3

There can be considerable clinical variability in mosaics because somatic mutation can occur at different stages of the postzygotic cell lineage. Evans et al 2 reported the degree of mosaicism for five NF2 patients. Two patients with an estimated <10% of peripheral lymphocytes withNF2 mutations had ages of onset of symptoms of 41-48 years, while three patients with an estimated 21-44% of affected cells had ages of onset of symptoms of 21-28 years. This is consistent with a relationship between degree of mosaicism and disease severity, although there are too few patients to draw firm conclusions.

Few NF2 somatic mosaics have been reported, and clinical and molecular differences between mosaics and sporadic non-mosaic NF2 patients have not been quantified. To examine this question, we compared somatic mosaic and sporadic non-mosaic NF2 patients selected from NF2 populations in the United Kingdom (341 patients)4 5 and Germany (118 patients).6 7 The study groups included 13 previously identified somatic mosaics2 3 and 86 sporadic non-mosaic NF2 patients, all of whom had head and spine gadolinium enhanced magnetic resonance imaging. Sporadic NF2 patients were defined as non-mosaic if they had identified germlineNF2 mutations with normal strength gel bands. In theory, somatic mosaic patients could have near normal band strength, but in practice band strength will be reduced when <50% of peripheral lymphocytes have NF2mutations.2 Clinical data were not available in sufficient detail to determine if the distribution of lesions was non-uniform, which could result from somatic mosaicism.

The covariates that were examined were age at onset of symptoms, age at diagnosis, number of VSs (none or unilateral versus bilateral), presence and number of intracranial meningiomas, presence of spinal tumours, and germline NF2 mutation type (frameshift or nonsense versus other identified mutations). In univariate analyses, the two tailed t test was used for continuous variables and the two tailed Fisher exact test for binary variables. Multiple logistic regression was then used to examine the association between somatic mosaic status and covariates that differed between groups; interaction terms of age with number of tumours were also considered.

The characteristics of somatic mosaics and sporadic non-mosaics are compared in table 1. Age at onset of symptoms, age at diagnosis, and the distribution of germline NF2 mutation types were significantly different between the two groups. On average, somatic mosaics were 7.8 years younger than non-mosaics at onset of symptoms (p=0.012) and 8.9 years younger at diagnosis (p=0.013). Nonsense or frameshift mutations were identified in 92.3% of somatic mosaics, compared to 62.8% of non-mosaics (p=0.055). The prevalence of no VSs or unilateral VSs was higher in mosaics than in non-mosaics (30.0% versus 12.9%), but this difference was not statistically significant. The prevalence and number of other central nervous system tumours were similar in the two groups.

Since age at onset of symptoms and age at diagnosis were highly correlated (r²=0.65, p<0.001), age at diagnosis was used in the multiple logistic regression model because tumour burden was evaluated at diagnosis. In the multiple logistic regression model, age at diagnosis and germline NF2 mutation type were significantly associated with somatic mosaic status and number of VSs was of marginal statistical significance. Interaction terms were not statistically significant. Patients with frameshift or nonsense mutations had 23.1-fold greater odds of somatic mosaicism than patients with other types of identified mutations (95% confidence interval 1.7-316.4, p=0.019). This may be a bias that results from milder disease in somatic mosaic patients and more severe disease in NF2 patients with nonsense or frameshift mutations, so that somatic mosaic NF2 patients with nonsense or frameshift mutations are more likely to be clinically detected as having NF2.2 3

The odds of somatic mosaicism increased 11.1-fold per decade increase in age at diagnosis (95% confidence interval 10.3-11.8, p=0.004). Patients with no VSs or unilateral VSs were 7.1-fold more likely to be mosaic (95% confidence interval 1.0-53.7, p=0.056). Considering two sporadic NF2 patients (one 40 years old at diagnosis with a unilateral VS and another 20 years old at diagnosis with bilateral VSs, both with non-truncating mutations), the former patient has 157.6-fold greater odds of being a somatic mosaic than the latter patient.

This study is based on relatively few mosaic patients, but if substantiated by other studies, these findings may be a useful adjunct in identifying somatic mosaics. For example, sporadic NF2 patients who are >40 years old at diagnosis with no VSs or unilateral VSs could be evaluated for somatic mosaicism. This is an unusual clinical presentation in NF2, occurring less frequently than the estimated 15% prevalence of somatic mosaicism in sporadic NF2 cases. In the United Kingdom NF2 patient series, nine of 206 sporadic patients (4.4%) meet this description. These patients include only one previously identified somatic mosaic; the rest have unidentified germlineNF2 mutations. The eight patients with unidentified mutations have ages at onset of symptoms ranging from 29-45 years (median 36 years) and ages at diagnosis ranging from 42-59 years (median 48 years). We are currently analysing tumours from these patients to determine the proportion that are mosaic.