Cellular heterogeneity is a well-known complication in the analyses of NF1-associated tumors. By contrast, heterogeneity at the molecular level in NF1-associated tumors has scarcely been addressed. In this study, we set out to determine the extent of molecular heterogeneity within and between 10 MPNSTs, derived from 10 unrelated NF1 patients, by determining the differences in the levels of LOH at the NF1, TP53, RB1, CDKN2A, and PTEN gene loci. The results of this study indicated that 70% of the MPNST tumors studied exhibit molecular heterogeneity between sections of the same tumor sample. This heterogeneity was especially evident in the case of those sections from the same tumor which were anatomically adjacent to each other prior to dissection but which had nevertheless been found to differ with respect to the degree of LOH. Indeed, some sections were found to be entirely devoid of LOH for all five gene loci analyzed but were located beside sections exhibiting complete LOH for one or more of these genes (e.g., T196.22 section A4). Prior to dissection, in a number of the MPNST tumor sections under study, there were clearly defined centrally located tumor portions (T196.22 subsection 5 A to E, T516 section B, T518 section B). These areas corresponded to the sections in which different levels of LOH and subsequently TP53 staining were identified (T196.22 section B5, T196.22 section C5, T196.22 section E5; Figure 2).
The observation that TP53 LOH correlates directly with the p53 expression, as demonstrated here, has been observed in previous studies [37, 38]. However, the precise relationship between TP53 LOH and p53 protein expression remains unclear since some studies have identified p53 expression in tumors which have no TP53 LOH. This may be explicable in terms of the presence of two TP53 alleles, one mutant and the other wild-type in a given cell type, resulting in the production of both wild-type and mutant p53 protein. As a consequence, p53 may be detected by IHC even in the absence of p53 function. This suggests that in the context of evaluating tumor heterogeneity, IHC analysis is unlikely to be as reliable as the other molecular genotyping methods. IHC analysis could therefore be replaced by more accurate methods including AQUA [39, 40] and tissue analysis with multiplex quantum dots (QD)  to yield a digital map of molecular and cellular heterogeneity to improve the sensitivity of detection and the prediction of a therapeutic response.
p53 is associated with malignant transformation in NF1-associated tumors  and LOH of the TP53 gene was identified in 5 of the 10 tumors under study (Table 1, Additional file 2: Supplementary Table S1). TP53 has also been found to manifest in intra-tumoral molecular heterogeneity with respect to its mutation in other tumor types, including breast cancer [41, 42]. In pancreatic cancer, molecular heterogeneity is evident in cells with different capacities for initiating metastasis  suggesting that molecular heterogeneity may well prove to be the rule rather than the exception. If the molecular heterogeneity identified in this set of tumors was to emerge as relevant to MPNST development, we may have to revise our view not only of MPNST tumor biology, but also of the basic processes underlying MPNST tumorigenesis.
It might be assumed that, owing to the large size of some of the MPNST tumors, they would be divisible into a larger number of sections thereby allowing molecular heterogeneity to be assessed more clearly. However, the size of the tumor was not found to correlate with the level of molecular heterogeneity detected. It is, however, possible that intra-tumoral molecular heterogeneity could be related in some way to the grade of tumor, at least for those tumor samples analyzed here. A larger study is clearly warranted in order to determine whether these results could be replicated in a larger set of NF1-associated MPNSTs.
The pathological diagnosis of an MPNST is usually held to represent the ‘gold standard’ for the purposes of analysis and currently relies on the examination of not just one but a number of different sections. The results of this study, and from other previous studies on solid tumors [44–46], are broadly illustrative of the importance of careful dissection in the analysis of large tumors and suggest that in the interest of diagnostic accuracy, molecular analysis should be performed on several tumor sections alongside a pathological diagnosis. The clear implication for those studies that involve microarray analysis is that replicates across several sections would be advisable.
The results of this study therefore have important implications for molecular studies of NF1-associated tumor specimens. For example, although molecular techniques currently employed in mutation detection in large solid tumors are adequate for identifying and characterizing the underlying molecular and genetic aberrations, the potential for molecular heterogeneity means that a single dissected piece of tumor should not be assumed to be representative of the tumor as a whole; as a consequence, some somatic mutations may well be missed.
Although MPNSTs only develop in approximately 15% of NF1 patients, they represent a frequent cause of lethal progression of the NF1 phenotype. The prognosis for individuals diagnosed with an MPNST is usually very poor; 5-year survival rates in patients with advanced non-resectable and/or metastatic MPNSTs that have exhibited a limited response to chemotherapy are in the order of 20% to 50%, while 10-year survival rates are as low as 7.5% . Treatment options for MPNSTs are currently rather limited, and complete surgical excision with clear margins is the recommended therapy for MPNSTs. A larger study is required in order to determine the full extent of molecular heterogeneity within MPNSTs in NF1 patients. However, such a study will be laborious and time-consuming to set up especially as MPNSTs are quite rare.
Genomic instability and high intra-tumoral genetic heterogeneity may synergize so as to accelerate the evolutionary processes within the tumor leading to the development of resistance to cytotoxic and targeted anticancer drugs. Improvements clearly need to be made to the treatment regimes for patients with MPNSTs. The results from this study indicate that while drugs can be developed in vitro and in vivo animal studies that would be capable of targeting the genes involved in the genesis of MPNSTs, the efficacy of these drugs is likely to be somewhat limited unless the cellular, molecular, and architectural heterogeneity of the tumor are considered alongside the tumor microenvironment.
This study represents the first systematic analysis of molecular heterogeneity in MPNSTs associated with NF1. The molecular heterogeneity evident at a number of different gene loci indicates that there is an urgent need not only for the integration of molecular and morphological biomarkers in cancer diagnosis, but also for the development of specific treatments for NF1-associated MPNSTs.