The study presents the first molecular evidence of inactivation of both copies of the NF1 gene in a typical superficial spreading melanoma of a patient with neurofibromatosis type 1. The paternal allele bears the germline mutation NF1 c. 5556 G/A as the father of the patient is the disease carrier. The maternal allele is deleted in the tumor tissue as demonstrated by loss of heterozygosity of a microsatellite marker located within the NF1 gene. No LOH of the analyzed microsatellite markers could be detected in the tissue surrounding the melanoma. The inactivation pattern of the NF1 gene in the analyzed melanoma represents the classical double-hit inactivation of a tumor suppressor gene as postulated by Knudson in 1971 . Deletion of the NF1 gene in NF1-associated melanoma suggests that the NF1 genetic background promoted melanoma genesis. This assumption is strengthened by the observation that the melanoma occurred at the unusually young age of 15 years and by the detection of the gene deletion in an early disease stage melanoma. It has been reported previously that melanoma in NF1 tends to arise at younger age  and mutation detection in an early tumor stage is a stronger indication of a causative role of the detected mutation than mutation detection in a late stage malignant tumor.
Despite the molecular findings supporting a melanoma promoting genetic background in the analyzed NF1 case, the question remains why melanoma incidence is not markedly elevated in NF1. This apparent inconsistency may be explained by the melanocyte abnormalities found in NF1 patients, by melanoma histogenesis and by divergent tumor evolution pathways in malignant melanoma.
Increased macular pigmentations visible in café-au-lait-macules (CALMs) and in axillary freckling are among the first clinical signs of NF1 . It has been hypothesised that formation of CALMs might be caused primarily by the effect of neurofibromin haploinsufficiency on melanocyte differentiation and pigmentation and less by an effect on cell proliferation. It could be demonstrated that neurofibromin colocalizes with melanosomes  and melanocyte density seems only moderately and inconsistently elevated in NF1-associated CALMs [10, 27]. Furthermore, cell studies with melanocytes of NF1 patients have not detected enhanced RAS-GTP-levels, which would have been indicative of an enhanced melanocyte proliferation .
On the other hand, the round shape of CALMs is compatible with a monoclonal melanocyte expansion within the skin and analysis of X-inactivation pattern indeed suggests a monoclonal origin . Nevertheless, the monoclonal origin of CALMs may be explained sufficiently by melanocyte differentiation during embryogenesis and does not require enhanced melanocyte proliferation. Most knowledge on melanocyte differentiation during embryogenesis has been obtained from the murine system. Melanoblasts, which are the melanocytes precursors, migrate in mice from the neural crest dorsolaterally and enter the skin where they proliferate clonally and finally differentiate into mature skin melanocytes [30, 31]. One may assume that modulation of pigmentation by neurofibromin haploinsufficiency during regular melanoblast expansion in the skin leads to formation of monoclonal and round shaped CALMs.
The lack of NF1 allele loss in CALMs  and the observation that melanomas arising in NF1 patients do not demonstrate a preferential association with CALMs  further support the view that neurofibromin does not seem to control proliferation of mature skin melanocytes. Knowledge is still incomplete on homeostasis and regeneration of skin melanocytes but in the murine system, hair follicle melanocytes are provided by melanocyte stem cells located in the lower permanent portion of the hair follicle  and a similar system of melanocyte stem cells may be responsible for melanocyte regeneration in humans as well. As CALMs patterns appear stable after the first decade it is not probable that neurofibromin haploinsufficiency may strongly affect skin melanocyte regeneration.
Assuming that neurofibromin does not control proliferation of mature skin melanocytes or putative melanocyte stem cells, one may not expect that the NF1 genetic background enhances melanoma development from these cells.
A significant portion of melanomas do not arise in normally pigmented skin but develop from melanocytic nevi which are present at birth or are acquired later in life . The true identity of the precursor cells of melanocytic nevi remains elusive but it is assumed that congenital melanocytic nevi (CMN) develop during embryogenesis from neural crest derivatives which retain a certain ability to segregate into different neural crest lineages as CMN may contain cells displaying neurogenic differentiation patterns [34, 35]. Likewise, acquired melanocytic nevi are thought to origin from melanocyte precursors which show wider differentiation plasticity than mature skin melanocytes [36, 37]. It can be speculated that melanocytic nevi displaying a neurogenic differentiation type may retain neurofibromin control of cell proliferation and may, therefore, be susceptible to malignant transformation in a NF1 genetic background. Nine of the 37 NF1-associated melanoma cases reported in the literature were associated with giant congenital melanocytic nevi  which supports the hypothesis that nevi might be the primary source of NF1-associated melanomas. Likewise, the melanoma of our patient most probably developed from an acquired melanocytic nevus. The importance of a neurogenic differentiation pattern for NF1-associated melanoma genesis is further demonstrated by the detection of frequent NF1 allele loss in desmoplastic neurotropic melanoma  which may display neural features and markers [38, 39].
The existence of distinct and divergent tumor evolution pathways in malignant melanoma  may serve as a complementary explanation for the apparent lack of a grossly elevated melanoma incidence in NF1. It could be demonstrated that more than 80% of all melanomas contain mutations in BRAF or N-RAS . Moreover, BRAF mutations could be detected in 82% of all melanocytic nevi . As neurofibromin functions as a negative regulator of the RAS-RAF-MEK-ERK-pathway , selective pressure for NF1 loss should be absent in premalignant melanocytic proliferations harbouring activating mutations of BRAF or N-RAS which represent the majority of all acquired melanocytic nevi.
Taken together, neurofibromin function in mature skin melanocytes, melanoma histogenesis as well as genetic tumor evolution pathways suggest that neurofibromin loss or haploinsufficiency may enhance melanoma risk only in minor subset of all melanocytic cells which may progress to melanoma. Nevertheless, congenital or acquired melanocytic nevi could share an enhanced melanoma risk in NF1 patients.
The presented data were generated by PCR analysis of microdissected tissue obtained from a thin superficial spreading malignant melanoma which was formalin-fixed and paraffin-embedded. Molecular analysis of this material faces specific problems. It has been observed, that formalin fixation leads to a bias when microsatellite alleles are detected and quantified by conventional PCR amplification . When two alleles differing in size are present in formalin-fixed tissue, microsatellite PCR tends to amplify preferentially the allele of smaller molecular size. This may lead to false positive detection of loss of heterozygosity (LOH). A second technical limitation is related to the low number of PCR-amplifiable DNA sequences within DNA obtained from formalin-fixed tissue. It could be demonstrated that the fraction of PCR-amplifiable DNA can be as low as ≈1 to 3600 . Especially when using microdissected formalin-fixed tissue, only few target molecules may be present for microsatellite PCR and allele ratios may not be determined correctly due to stochastic error. Both problems can be circumvented by using digital PCR for quantification of microsatellite allele ratios. Digital PCR seeks to amplify allele sequences starting from one target molecule. In case of heterozygous microsatellite DNA, each individual and successful PCR amplification should begin with one molecule of the allele of lower molecular size or with one molecule of the allele of higher molecular size. As allele ratios are calculated based on the number of positive PCR amplifications for each allele, differences between allele amplification efficiencies do not bias allele quantification. Moreover, digital PCR reports the number of amplifiable alleles contained in the analyzed DNA solution which allows correct statistical interpretation of the results. In the case of the analyzed NF1-associated melanoma, 29 alleles could be detected in DNA obtained by microdissection of approx. 3500 cells. The paternal allele was present in 79% of all PCR amplifications and LOH detection was statistically significant based on the sequential probability ratio test.