Neurofibromatosis-1 (NF1 [MIM 162200]) is a common autosomal dominant disorder with an incidence of 1 in 2500–3000 births and a prevalence of 1 in 4000 1 . NF1 is associated with increased morbidity and mortality rates 2 3 . In spite of tremendous progress in understanding and treating NF1, there continues to be inconsistency, if not confusion about the various types of neurofibromas. Our interest in this article is to sensitize clinicians to NF1 neurofibromas that are not obvious by inspection and/or palpation of the skin. That is, the focus is on internal neurofibromas. An internal neurofibroma simply is a neurofibroma that is not appreciated by physical examination. Part of the problem is that some neurofibromas that may have internal components may also be apparent externally. In particular, both the large diffuse plexiform neurofibroma and the large nodular plexiform neurofibroma, as described by Riccardi 4 , corresponding to Masson’s diffuse neurofibroma and encapsulated neurofibroma, respectively, as utilized by Tucker and coworkers 5 6 , can have both external and internal components. In the present article we are reserving the term, internal neurofibroma, for those neurofibromas not having an apparent external component. By definition, both cutaneous neurofibromas and subcutaneous neurofibromas are excluded from being internal neurofibromas. Most specifically, the term, plexiform neurofibroma, does not on its own afford distinction from or inclusion in what we are specifying here as an internal neurofibroma.

Internal neurofibromas are among the main causes of life-threatening events in patients with NF1 4 7 . Ten percent of internal neurofibromas undergo transformation to malignant peripheral nerve-sheath tumors (MPNSTs) 8 , a leading cause of death in adults with NF1 2 3 . Clinical indicators of MPNST are persistent or increasing pain, enlargement of the tumor, and neurological deficiencies 9 . The diagnosis is often delayed, especially for deep diffuse plexiform or internal neurofibromas as imaging studies are not performed routinely as part of the follow-up of patients with NF1 10 but instead are ordered only as clinically indicated 11 . The mean age at diagnosis of MPNST is younger in patients with NF1 than in unaffected individuals 8 12 13 14 . MPNSTs may develop before 30 years of age 8 12 and even in childhood 13 . Moreover internal neurofibromas grow faster in young patients 15 . In previous work, we developed a simple scoring system (the NF1Score) that accurately predicted the presence of internal neurofibromas at risk for transformation to MPNST in adults with NF1 16 . Four characteristics were independently associated with internal neurofibromas: presence of subcutaneous neurofibromas, fewer than 6 café-au-lait spots, absence of cutaneous neurofibromas, and age ≤30 years. However, the NF1Score was developed in patients aged 17 years or older. Consequently, a specific study was needed in pediatric patients. The clinical expression of NF1 varies across the life span in a given patient 11 17 . For instance, café-au-lait spots are present within the first year of life and cutaneous or subcutaneous neurofibromas are usually present by adolescence 1 17 18 .

The purpose of this study was to identify clinical characteristics associated with internal neurofibromas in children with NF1. Such clinical characteristics could serve to ensure the early identification of children with NF1 who require particularly close monitoring for internal neurofibromas.

In this study, we identified easily recognizable clinical characteristics associated with internal neurofibromas in children with NF1. By multivariate analysis, age, xanthogranulomas, and presence of both subcutaneous and plexiform neurofibromas were independently associated with internal neurofibromas.

NF1 has been reported to be associated with a 15-year decrease in life expectancy 2 . We recently reported overall excess mortality in a cohort of 1895 patients with NF1 compared to the general population in France 3 . Excess mortality occurred among NF1 patients aged 10 to 20 years (Standard Mortality Ratio, SMR, 5.2; 95% CI, 2.6-9.3; P < 10^-4) and 20 to 40 years (SMR, 4.1; 95% CI, 2.8-5.8; P < 10^-4). MPNSTs were the main cause of death (60%) 3 . Thus, the main concern in the long-term clinical management of patients with NF-1 is the identification of patients at high risk for MPNSTs developed from preexisting internal neurofibromas 2 . Although no effective treatment is available for inoperable MPNSTs, several novel targeted treatments may hold promise 20 . We therefore developed a clinical score, the NF1Score, for predicting the presence of internal neurofibromas in adults with NF1 16 .

Internal neurofibromas grow faster in young children than in adults 15 . Volumetric MRI has been used to assess the growth rate of internal neurofibromas in 49 NF1 patients (median age, 8.3 years; range, 3.3-25) with a median follow up of 34 months. The growth rate per year of internal neurofibromas was significantly greater in the younger patients (<8.3 years) than in the older patients (21.1% versus 8.4%, P = 0.001). In addition, if effective treatments are found, they will be more likely to prevent the growth of internal neurofibromas than to reduce their size 24 . Therefore, the benefits will probably be greatest if the treatments are given at the time of most rapid internal neurofibroma growth, that is, in childhood or adolescence. Therefore, the risk factors identified in our study will help to improve the clinical management of children with NF1.

Many factors support the internal validity of our study. First, all patients had a definitive diagnosis of NF1. Selection bias seems unlikely, as the prevalence of NF1 characteristics were consistent with those reported previously in other populations of NF1 patients 25 26 . Our study provided a good external validation as it was performed using two NF1 populations, from France and North America, respectively. However, our patients were recruited at hospitals and may therefore have had greater disease severity compared to the overall NF1 population. An important strength of our study is the accuracy of the information on internal neurofibroma status obtained by using MRI or CT. However, the absence of routine MRI or CT in the study population may constitute a limitation of our study. MRI or CT was performed only when routine imaging studies (chest radiograph and abdominal sonogram) or clinical symptoms suggested the presence of internal neurofibromas. Half our patients did not undergo MRI or CT. This may have resulted in verification bias. Finally, to validate the final model, we used multiple imputation analysis to deal with the missing values. Multiple imputations allow individuals with incomplete data to be included in analyses and improve the validity of the results. The empirical rule of entering only one variable per 10 events in a multivariate analysis model was followed 27 .

Internal neurofibromas were present in 35 (9.8%) patients, in keeping with previous data. In a study of 53 children with NF1 who underwent MRI of the entire spine, 7 (13.2%) patients had internal spinal neurofibromas 28 . In our study, older age was independently associated with internal neurofibromas. Excess risk of developing internal neurofibromas seems to occur between the adolescence and the age of to 30 in NF1 patients. Moreover internal neurofibromas increased during adolescence and developed earlier in females, in agreement with previous data 29 . Changes in steroid hormone production may affect the NF1 phenotype. Immunostaining studies have provided support for this hypothesis by identifying the progesterone receptor in neurofibromas 30 . Estrogen and progesterone increased the growth rate of MPNSTs xenografts in mice 31 . Plexiform and subcutaneous neurofibromas were also independently associated with internal neurofibromas in our study, in accordance with earlier work in both adults and children 7 32 . In the MRI study of the spine in children with NF1, characteristics found more often in patients with than without internal neurofibromas included scoliosis (71.4% vs. 30.4%), subcutaneous neurofibromas (71.4% vs. 39.1%), and plexiform neurofibromas (28.6% vs. 8.7%); none of these differences was statistically significant, (P = 0.08, P = 0.22, and P = 0.17, respectively), possibly because statistical power was limited 28 . Furthermore, patients with subcutaneous neurofibromas were at higher risk for mortality in two different NF1 populations, from France 33 and North America 21 , respectively. Finally, xanthogranulomas were independently associated with internal neurofibromas in our study. NF1 children with xanthogranulomas are at increased risk for juvenile chronic myelogenous leukemia 34 . Although it should be noted that this study included a small number of patients and has not been confirmed since. To our knowledge, xanthogranulomas have not been previously reported to be associated with internal neurofibromas.

In sum, we identified easily recognizable clinical characteristics that were associated with internal neurofibromas in children with NF1. These risk factors could be used to ensure the early identification of NF1 patients who require particularly close monitoring for internal neurofibromas. In the future, this approach may allow the initiation of treatments at the time of fastest growth of internal neurofibromas, when they are most likely to be effective.

aOR: Adjusted odds ratio; CI: Confidence interval; CT: Computed tomography; MPNST: Malignant peripheral nerve sheath tumor; MRI: Magnetic resonance imaging; NF-1: Neurofibromatosis 1; SC-NF(s): Subcutaneous neurofibroma(s); SD: Standard deviation.

SBG, PW, DR, SF, OC, SB, LVA, VMR, SHR, ES have non-financial interests that may be relevant to the submitted work.

SBG and PW made substantial contributions to the study conception and design. SBG, PW, SF, SHR, VMR, LVA, ES, DR, and SB made substantial contributions to data acquisition or analysis and to interpretation of the data. ES performed the data analysis under the supervision of SBG. SBG, PW, and ES were involved in drafting the manuscript or revising it critically for important intellectual content. All authors gave final approval of the version submitted for publication.

Grant from the programme hospitalier de recherche clinique (PHRC, AOM 02 108/P 020906), French Ministry of Health.