Age-related findings on MRI in neurofibromatosis type 1
T2 hyperintensities (T2H) on MRI are the most common CNS lesions in individuals with neurofibromatosis type 1 (NF1).
The aim was to determine the frequency, signal characteristics and localization of T2H at different ages. In addition, we examined the sensitivity of different MR imaging sequences in detecting these lesions.
Materials and methods
We studied prospectively a cohort of children, adolescents and young adults with NF1 using T2-volume (T2-V) and conventional MRI sequences. Lesions were designated as either discrete or diffuse, and the region of signal abnormality was recorded. A total of 103 patients were studied (age range 8.0–25.4 years, mean 13.9 years).
The frequency, size, and intensity of T2H decreased with age in the basal ganglia (BG) and the cerebellum/brainstem (CB/BS). The majority of thalamic and CB/BS lesions were diffuse. Of the total cohort, 80% had diffuse bilateral hippocampal hyperintensities and 18.4% had hemispheric lesions best demonstrated on FLAIR; there was no significant difference in the frequency or signal intensity of hemispheric lesions with age.
Lesions in the cerebral hemispheres and hippocampus imaged by MR do not change in prevalence over time, suggesting a different pathological basis from the lesions in the in BG and CB/BS that resolve with age. FLAIR and T2-V sequences are more sensitive in detecting lesions than standard T2-weighted sequences.
KeywordsNeurofibromatosis type 1 MRI Hippocampus
Neurofibromatosis type 1 (NF1) is an autosomal dominant neurocutaneous disorder with an estimated prevalence of 1 in 3,500 . The NF1 gene [2, 3, 4] is a tumour suppressor gene and encodes the protein neurofibromin, which is expressed in neurons and glial tissue in the central and peripheral nervous system [5, 6, 7]. NF1 is characterized by aberrant cell growth and differentiation of neuronal and pigmentary cell lines resulting in neurofibromas, café au lait spots, and Lisch nodules. Alterations in the growth and maturation of CNS tissue results in CNS tumours such as optic pathway gliomas , macrocephaly associated with an increase in the ratio of grey matter to white matter , and increased surface area of the corpus callosum .
The most common CNS lesions in NF1 are seen by MRI on T2-weighted (T2-W) images as areas of hyperintensity (prolongation of T2). These lesions are usually isointense on T1-weighted (T1-W) images, they exert no mass effect, there is no surrounding oedema, and they do not enhance following administration of contrast material . The T2 hyperintensities (T2H) are reported to occur in 60–70% of children with NF1  (although figures vary from study to study depending on the age range of the patients and the imaging techniques used). They are also referred to as hamartomas  or ‘UBOs’ (unidentified bright objects)  and are most commonly seen in the globus pallidus, brainstem and thalamus . The pathology underlying these lesions is unclear. Early autopsy studies of brains from adults with NF1 found evidence of focal heterotopia, disordered cortical architecture with random orientation of neurons and focal proliferation of glial cells [15, 16]. These observations laid the basis for the assumption that the neuropathology of NF1 is characterized by dysplasia or hamartomatous changes. Neither of these studies correlated the histological findings with neuroimaging and there is only one study that has done so . In this study it was concluded that the high signal intensity lesions on MRI may represent increased fluid within the myelin associated with hyperplastic or dysplastic glial proliferation.
Cross-sectional data suggest that the number of T2H decrease with age and that the lesions become rare in the third decade [10, 12, 17]. Hyman et al.  studied 32 patients with NF1 prospectively over an 8-year period and demonstrated a significant decrease in size, number and intensity of T2H.
It is not known whether the same neuropathology underlies all T2H. There is variability in the site of the T2H as well as variation in their signal intensity ranging from discrete hyperintensities to more diffuse lesions . Some lesions have T1-related changes [14, 19, 20, 21] and there is occasionally contrast enhancement with gadolinium . Only limited conclusions can be drawn from published studies due to the small number of patients studied , retrospective study design [10, 17, 22] and the inclusion of patients with other CNS and ocular pathology that may have biased the studies with an increased likelihood of coexistent T2H . Individual studies have used different imaging sequences and have not taken into account the ability of more sensitive MRI sequences to detect signal abnormalities. In addition, in no study have the parameters been defined for defining what constitutes an MRI T2H, i.e. whether diffuse areas of T2H are the same as discrete and well-circumscribed lesions.
We have performed neuroimaging of a large cohort of children, adolescents and young adults with NF1, ascertained prospectively in our clinic. We have performed a cross-sectional analysis of MRI findings to determine the frequency, signal characteristics and localization of T2H at different ages. In addition we examined the relative sensitivity of different MR imaging modalities in detecting these lesions including conventional T2-W imaging, FLAIR and T2-volume (T2-V) sequences.
Materials and methods
The patients were ascertained from the neurofibromatosis clinic at the Children’s Hospital at Westmead in Sydney, Australia. All members of the cohort were recruited as part of a study of both cognition and MRI features of NF1. All individuals were examined by a neurologist and satisfied the diagnostic criteria for NF1. Patients were excluded if there was a history of CNS tumour, or epilepsy. None of the patients had significant visual or hearing impairment. A total of 103 patients were recruited. This study was approved by the Ethics Committees of the Children’s Hospital at Westmead and the United States Army Research and Materiel Command. Some of the data from 27 patients have been published previously as part of our study of the natural history of cognitive deficits in NF1 .
The MRI examinations were performed on a magnet operating at 1.5 T (ACS-NT; Philips, The Netherlands). MRI sequences were sagittal T1, axial T2, axial FLAIR, T2V, coronal FLAIR and axial T1-W imaging following intravenous administration of gadopentetate dimeglumine at 0.1 mmol/kg (Magnevist, Berlex Laboratories, Wayne, N.J.). Typical imaging parameters were TR/TE 500/15 ms for axial T1-W images before and after contrast enhancement, TR/TE 3,600/120 ms for axial T2-W sequences, and TR/TE/TI 7,000/130/2,200 ms for axial FLAIR. T2-V parameters were TR/TE 7,000/110 ms. The images were obtained with 5-mm thick sections with a 1-mm spacing, 220-cm field of view, and a 512×256 matrix, except for T2-V where 100 sections were obtained each 1.60 mm in thickness, giving an almost contiguous appearance.
The MRI examinations were reported independently by the radiologist on duty at the time of performance of the MRI. The scans were then reported by a radiologist (A.S.) and a neurologist (D.G.) together. Two neurologists (D.G. and K.N.) reported the scans together, 1 year after the initial analysis. The 1-year interval reduced the likelihood of bias of previous interpretation by the principal investigator, D.G. At the time of the second report D.G. was blinded to the result of the first reporting session. D.G. observed that the variability between the two reporting sessions, although not formally calculated, was low; however a further consensus meeting was convened with D.G., K.N. and A.S. so agreement was reached on lesion presence or absence and signal characteristics. K.N. and A.S. were blinded to the results of other reports except at the consensus meeting. D.G. and A.S. were blinded to the clinical history other than NF1. K.N. had prior knowledge of the clinical history but not of the results of detailed cognitive testing.
This study includes MRI data on 103 patients (49 males and 54 females) with an age range of 8.0 to 25.4 years, mean 13.9 years. For the purpose of cross-sectional analysis the cohort was divided into five groups of roughly equal size: group 1 8–9 years (n=18), group 2 10–11 years (n=21), group 3 12–13 years (n=27), group 4 14–19 years (n=22), group 5 20–25 years (n=15).
Interpretation of T2-W images has been the basis of the majority of MRI studies in NF1, and it was this sequence that was studied in detail to assess the frequency, size and intensity of lesions (T2H). Each region in each individual was noted for the presence or absence of T2H either discrete or diffuse, and a percentage occurrence of T2H was calculated for each age cohort. The percentage of individuals with discrete T2H only, diffuse T2H only and both discrete and diffuse T2H in the BG was calculated for each cohort. The total number of lesions (discrete or diffuse) on standard T2-W sequences was recorded.
There were no significant differences in the frequency of hemispheric lesions among groups 1–5. Six patients with hemispheric T2H had previous neuroimaging (intervals of 7–8 years). In four of the six patients, the lesions were not present on the first scan. However FLAIR was not performed on the initial scans, and this appeared to be the most sensitive sequence for detecting the lesions. Seven patients (ages 8, 9, 10, 12, 13, 13, and 21 years of age, respectively) with cortical or subcortical lesions had further follow-up MRI over 1–2 years during the period of this study, and no interval change was detected.
We report the neuroimaging findings in the largest prospective MRI study of individuals with NF1 and demonstrate the diverse nature of T2H in both their anatomical location and intensity. BG T2H are more likely to be discrete whereas lesions in the thalamus, brainstem and cerebellum are rarely discrete and discerning the limits of individual lesions in these sites is difficult. The number and intensity of T2H diminish with age in the BG, cerebellum, brainstem and thalamus. Hemispheric and hippocampal lesions may appear over time; this suggests that they have a different pathogenetic basis from classic BG, cerebellar, brainstem and thalamic T2H. The signal characteristics on T2 imaging may be the only common feature shared by BG and the lesions of other sites in NF1.
A weakness of our study was that our patient population had a significant selection bias; the patients were cooperative and were able to undergo nonsedated procedures. Younger patients, patients with severe cognitive impairment, epilepsy, optic pathway tumours and other tumours were excluded from the analysis; T2H occur in almost all patients with NF1 and CNS tumours . Our radiological findings thus represent the milder end of the clinical spectrum of NF1. We undertook a blinded method of reporting and a consensus view. In this study, however, interrater reliability of detection of signal change was not tested.
Of our cohort, 18.4% had lesions in the cerebral hemispheres that were predominantly cortical or subcortical. These lesions characteristically have high signal on T2-W imaging, and are most easily visualized on FLAIR. These lesions are heterogeneous and appear to be distinct from the lesions in the BG, brainstem, cerebellum and thalamus. The lesions did not change over a follow-up period of 12–24 months in seven patients. The hemispheric lesions were present in all age groups and did not decrease in prevalence with age. The absence of such findings in a previous study  may reflect differences in imaging techniques, as FLAIR appears to be more sensitive for detecting hemispheric lesions; however it is also possible that these lesions had developed over time. The pathological correlate of these lesions is unknown. We speculate that they represent glial heterotopia described in the early studies of Rosman and Pearce . The presence of T2 signal abnormality, however, would be atypical for heterotopic lesions. The cortical lesions may represent areas of disordered proliferation or increased astrogliosis, such as has been observed in autopsy brain specimens from patients with NF1 . Others have observed malformations of cortical development (MCD) in individuals with NF1 and severe mental retardation and epilepsy ; this may represent a rare and extreme end of the spectrum of cortical dysplasia in NF1. The diagnosis of epilepsy was an exclusion criterion in our study. So by definition the lesions in our study were not clinically epileptogenic; however the relationship between cognitive impairment and cortical lesions and NF1 needs to be evaluated.
We observed a high prevalence of signal change within the hippocampus with 80% of the cohort demonstrating DBHH. A previous study looking specifically at the use of FLAIR in NF1 has shown signal change within the hippocampus . We observed that there was a spectrum of signal change in the hippocampi with some individuals having minimal diffuse signal change on T2-W images without changes on FLAIR. This led to some discordance amongst investigators, and hence we assigned DBHH to those individuals who had signal change on both T2-W and FLAIR sequences. One weakness of our study was that we did not have an age-matched control group for comparison, so we could not specifically comment on the relevance in our study population of the finding of minimal signal change in the hippocampal structures on T2-W images alone; however the reporting radiologist and the investigating radiologist deemed these appearances to be within normal limits. Some individuals demonstrated signal change on FLAIR not just restricted to the hippocampi but also of the parahippocampal structures, amygdala and extension into the insula. The involvement of limbic structures has not been previously highlighted in human studies in NF1. Interestingly, in the mouse model of NF1, defects in behaviour and learning are thought to be suggestive of disordered hippocampal function, either due to abnormal function due to unregulated ras (regulator protein) activity during hippocampus-dependent learning, or to abnormal neurofibromin-mediated signalling in both the adult and developing hippocampus .
The pathological correlate of T2H remains controversial. In vitro studies have shown the NF1 gene to be a tumour suppressor gene [26, 27]. The expression of the gene product, neurofibromin, is predominantly restricted to neuronal tissue in adults. Part of the protein encoded by neurofibromin shows high sequence homology with the GAP (GTPase activator protein) family of proteins that interact with ras proteins that regulate cell growth and differentiation, therefore acting as a negative regulator of neurotrophin-mediated signalling . Immunohistochemical studies have demonstrated upregulation of glial fibrillary acidic protein (GFAP), leading to the suggestion that this reactive astrogliosis may be an important pathogenetic mechanism in NF1 . On this basis it has been proposed  that T2H in the BG, brainstem, thalamus and cerebellum represent the formation of a chemically abnormal myelin sheath that is subsequently broken down to be replaced by myelin with a more stable form. The “haloing” effect on T1-W images may reflect the presence of more stable myelin being laid down at the centre of a T2H, best seen in the BG. The fact that this observation appears to be unique for the lesions within the BG implies that the stable formation of myelin in the BG may be different from that in other regions.
The diversity of the hemispheric (cortical and subcortical), and hippocampal MRI lesions and the relative lack of change in prevalence with age, suggests that lesions in these regions behave differently to those in the BG and CB/BS.
We observed that the BG have a propensity towards more intense and discrete T2H. Using fine sections and T2-V sequences, we also demonstrated that, with increased resolution, a single T2H can actually be comprised of a number of poorly delineated lesions, i.e. the lesion count is affected by the sensitivity of the imaging technique used. This implies that the absolute number of lesions may not be accurate in published studies whose aim was to correlate the number of T2H with cognitive deficits . The imaging technique used may determine whether a lesion is discrete or diffuse and the discrete nature of a T2H may be related to the intrinsic anatomy of the structure containing the lesion (e.g. globus pallidus) rather than true anatomical or physiological differences. As imaging becomes even more sensitive, the frequency of T2H may approach 100% in the younger age group. If T2H indeed represent a developmental or dysplastic change in the CNS in NF1, then this is likely to be a feature of the disorder in the vast majority of patients. Thus it will be increasingly important to distinguish between lesions in different locations and with different signal characteristics (e.g. intensity, discrete or diffuse distribution) in research studies of the pathogenesis and clinical associations of these MRI lesions.
MRI lesions in individuals with NF1 in the cerebral hemispheres and hippocampus did not change in prevalence over time, suggesting a different pathological basis to lesions in the BG and CB/BS that resolve with age. FLAIR and T2-V were more sensitive in detecting CNS lesions than standard T2-W and the prevalence of T2H in children with NF1 is likely to approach 100% as imaging techniques become more sensitive.
This research was supported by the Department of Defense Neurofibromatosis Research Program, managed by the U.S. Army Medical Research and Materiel Command (USAMRMC; award number DAMD17-00-1-0534). We are grateful to Dr. Sridhar Gibikote for his helpful comments on the significance of the radiological findings and Mrs. Susanne Smith for her administrative support.
- 7.North K, Ratner N (2003) The brain in neurofibromatosis type 1. In: Fisch GS (ed) Genetics and genomics of neurobehavioural disorders in contemporary clinical neurosciences series. Humana Press, Totowa, pp 97–135Google Scholar
- 10.Sevick RJ, Barkovich AJ, Edwards MSB, et al (1992) Evolution of white matter lesions in neurofibromatosis type 1: MR findings. AJNR 159:171–175Google Scholar