NF1 patients with germ line NF1 splice-site mutations were found to have an increased tendency to develop MPNSTs and CNS gliomas by comparison with patients harboring other mutation types, e.g. microinsertions, microdeletions, nonsense and missense mutations. Up to 30% of all NF1 patients carry germ line NF1 splice-site mutations, and such mutations often lead to the synthesis of a truncated form of neurofibromin [4, 16]. In the present NF1 patient cohort, however, only 23 individuals (15%) carried splice-site mutations. This comparatively low frequency could be explained by the fact that the current analysis was performed on genomic DNA (rather than cDNA), and some of the underlying splicing mutations (resulting from deep intronic mutations) could have been missed if they did not occur in the immediate vicinity of a splice site. Moreover, some of the missense or nonsense mutations identified could have exerted an effect on splicing. Despite the small patient number, our analyses identified a significant correlation between the presence of a splice-site mutation and the occurrence of neoplasms (p = 0.006), mostly brain and connective tissue tumors. This initial finding clearly merits further clinical investigation in a much larger mutation-known NF1 patient population. A multicenter study and prospective data collection, based on comprehensive clinical examination, would be necessary to substantiate this finding. If our initial conclusion is supported by further analysis, it could imply that the wider application of RNA-based analyses in NF1 patient screening (in order to detect splice-site mutations with high efficiency)  might be helpful in predicting which patients would benefit the most from regular clinical surveillance.
Of all the characteristic features of NF1, the presence of benign and malignant neoplasm contributes the most to the morbidity and mortality experienced by many NF1 individuals; indeed, the overall cancer risk associated with NF1 has been reported to be higher than in the general population in several NF1 studies [5–7], owing to markedly increased risks of brain and connective tissue tumors. Indeed, malignancy, alongside other tumor-related neurological complications, is the most frequent cause of death in NF1 patients, resulting in a reduced life expectancy of some 8 to 15 years as compared to the general population [5, 7, 27]. The predominance of cancer among causes of death in NF1 is consistent in most studies [5–7, 34, 37] and contributes disproportionately to the mortality in younger age groups (less than 40 years). This increased risk of malignancy in NF1 individuals is a cause of major concern for families and is difficult to address during genetic counseling.
If our finding of an association between the possession of an NF1 splice-site mutation and the likelihood of tumor occurrence is borne out by future work, this group of patients should undergo rigorous clinical surveillance in order to facilitate early diagnosis and permit early management of any tumor-related complications as they arise. Although the types of tumors observed in our patient cohort were heterogeneous, they were similar to the spectrum of tumors reported in many NF1 patients, being mainly brain gliomas and MPNSTs [5–7, 38] which comprised almost half of the neoplasms in our cohort. The prevalence of types of tumors other than the brain gliomas and MPNSTs reported in this cohort was very low, as noted in other studies . A moderately increased susceptibility to develop breast cancer in female NF1 patients before the age of 50 years has previously been documented . However, only 2/74 female (2.7%) cases in our cohort developed ductal type breast cancer between the ages of 34 and 41 years. Phaechromocytomas and catecholamine-secreting tumors of the neural crest are rare tumors in the general population and occur in 0.1% to 5.7%  of patients with NF1. In a recent study, a higher prevalence of 14.7% was noted . At least three patients (2%) in our cohort had phaechromocytoma, whereas two patients had medullary thyroid cancer, one of these in association with bilateral phaechromocytoma as part of multiple endocrine neoplasia type 2A (MEN2A). The coexistence of MEN2A and NF1 (the neurocristopathies) has rarely been reported, suggesting that NF1 patients with phaechromocytoma should be thoroughly examined for clinical evidence of other phaechromocytoma-related syndromes. All three cases reported in the literature (including our own case) had identifiable germ line mutations in the NF1 gene, but not in the RET proto-oncogene which is a known cause of MEN2A . Colorectal carcinomas have been previously noted as a cause of death in NF1 . Only one case of adenocarcinoma of the colon at the age of 71 was noted in our cohort and this could be a sporadic occurrence. GIST are mesenchymal tumors of the gastrointestinal tract and are commonly associated with NF1 and usually with a good prognosis [43, 44]. We report two such tumors in our cohort. Glomus tumors are small, benign but painful tumors and are usually associated with NF1 and considered as part of the tumor spectrum of NF1 . Two cases of multiple glomus tumors were reported in our cohort. Rhabdomyosarcomas are pediatric neoplasms which are more common in children with NF1 than in the general population . One case of a child with rhabdomyosarcoma of the prostate was noted in the study cohort. Multiple subcutaneous lipomas  have been noted in one patient with NF1. However, solitary lipomas are also frequent in the general population, and the presence of these isolated lipomas in four of our patients could be purely coincidental. The presence of xanthogranulomas in pediatric NF1 cases may be related to the possible development of JMML , but none of the patients in our cohort had JMML. The prevalence of xanthogranulomas in the general population of normal children is quite high (1% to 2%), but these are rarely associated with systemic manifestations .
Our observation of a gender effect in NF1 has not been previously reported; despite the very similar sex distribution in our patient cohort, a disproportionate number of females exhibited learning difficulties (p = 0.022). Female NF1 patients were also found to be significantly more likely to manifest SCNFs (p = 0.009). This observation may relate to the differential effect that female steroid hormones appear to have on the development of CNFs and SCNFs, as previously noted in the peri-pubertal and pregnant subpopulations of affected NF1 individuals [49, 50]. Peripheral neurofibromas have also been reported to express progesterone receptors [49, 50] and this could explain the observed gender effect. SCNFs rarely transform into malignant tumors, but their presence could be an expression of a more aggressive disease [5, 51]. A disproportionate number of SCNFs in a cohort of NF1 female patients could indicate a worse prognosis in females in comparison to males, as indicated in some studies [7, 34, 51].