Patient 3. Axial proton density images. Left, At age 7 years there is bilateral high-intensity signal lesions in the globus pallidus. Right, At age 11.5 years the lesions are now completely resolved. Clinically asymptomatic at each study.
Patient 4. Axial proton density images. Left, At age 9.5 years there is bilateral high-intensity signal lesions in the globus pallidus. Right, At age 13 years there is marked reduction in previously noted lesions. Clinically asymptomatic at each study.
Patient 7. Axial proton density images. Left, At age 5 years there is middle cerebellar peduncle and pontine high-intensity signal lesions with minimal mass effect and normal clinical examination results. Middle, At age 8 years there is increased signal intensity and mass effect of enlarging previously identified masses with concurrent clinically correlated cranial nerve and cerebellar dysfunction on examination. Right, At age 10 years there is increased mass effect and lesion bulk since initial imaging. Clinical examination findings remained stable until the time of this image when symptoms increased, prompting debulking of pilocytic astrocytoma.
Composite of all the lesions plotted by number and indexed size per patient, by age at study, for each anatomical region and by total. Oldest patient (47 years) with late-onset cystic astrocytoma was omitted from the plots.
DiMario FJ, Ramsby G. Magnetic Resonance Imaging Lesion Analysis in Neurofibromatosis Type 1. Arch Neurol. 1998;55(4):500–505. doi:10.1001/archneur.55.4.500
To define the evolution of identified high-signal brain parenchymal lesions on magnetic resonance imaging (MRI) studies in patients with neurofibromatosis type 1 (NF-1).
A cohort of patients with NF-1 who underwent MRI were identified prospectively and their imaging studies analyzed.
All referred patients with NF-1 (as defined by National Institutes of Health consensus criteria), who had undergone imaging with MRI were eligible. Of 123 patients with NF-1 whose conditions were evaluated, 30 patients had undergone 59 MRIs. There were 22 males and 8 females, aged 1 to 53 years with mean age of 12.5 years. Two groups of patients were identified, those with brain lesions (WBL) and those with no brain lesions. All initial and subsequently obtained MRIs from the WBL group were analyzed and tallied for number, size, and location of lesions over serial studies.
Of the 19 patients with WBL, lesions were in hemispheres in 19 patients, and in the brainstem and the cerebellum in 10 patients each, respectively. Lesions were located in the cerebellum and globus pallidus most often (87 of 129 lesions). Of the patients with WBL having serial studies, a total of 97 lesions equaling 197 units (mean, 2.03 units per lesion) were identified at initial study. Follow-up evaluation (interval, 0.5-4.5 years; mean, 2.3 years), showed a decrease in both total number of lesions (68 [−29%]) and size (132 units; mean, 1.86 units per lesion [−33%]). Importantly, brainstem lesions increased in both number (+36%) and size (+6.4%) over the same intervals in 7 of 13 patients with WBL studied serially, whereas hemispheric and cerebellar lesions were more evanescent.
High-signal T2 lesions on MRI in patients with NF-1 evolve over time. The evolution of the NF-1 lesion is region specific and may relate to preferential region-specific effects of the NF-1 gene product.
NEUROFIBROMATOSIS type 1 (NF-1) is a relatively common (incidence 1 in 3000) autosomal dominant disorder.1 The diagnosis is based on an individual demonstrating at least 2 of the following 7 clinical criteria: 6 or more café au lait spots (pubertal dependent size dimension); 2 or more neurofibroma, or a plexiform neurofibroma, axillary or inguinal freckling; distinctive osseous lesions; optic nerve glioma; more than 2 iris Lisch nodules; or a first-degree relative with NF-1.1 A mutation in chromosome 17 (NF-1 gene) encodes for the protein neruofibromin.1,2 Its function is related to intracellular signal transduction.2
Among patients with NF-1, a number of common brain lesions are identified using magnetic resonance imaging (MRI) and include tortuous optic nerves, optic glioma, Arnold-Chiari malformation, aqueductal stenosis, and discrete T2high-signal intensity abnormalities within the basal ganglia, cerebrum, cerebellum, and brainstem.3- 7 These latter lesions in particular are often asymptomatic in contrast to more definitive parenchymal mass lesions.8 The discrete high-signal lesions are identified in as many as 60% of patients with NF-1 who undergo MRI.4,5,7 They have been postulated to represent foci of neural dysplasia, heterotopia, low-grade glioma, and dysmyelination. Limited tissue specimens have confirmed several of these histologic findings in addition to spongiotic change.9,10
We prospectively reviewed serial MRI scans from patients with NF-1 whose conditions were evaluated in our Neurogenetics Clinic and quantified the process of lesion evolution in these patients to examine the hypothesis that the NF-1 gene product may produce region-specific effects within the central nervous system.
All patients whose conditions were evaluated in the Neurogenetics Clinic at the University of Connecticut Health Center in Farmington were included in this review. Patients with a diagnosis of NF-1 using National Institutes of Health consensus criteria were selected. The MRI scans from each patient who had undergone this examination were reviewed. The clinical indications for MRI in these patients were varied and included exclusion of optic nerve glioma, evaluation of headache complaints, abnormal findings on neurologic examination, and obtained prior to referral. Sedation was routinely employed in children younger than age 5 years and only when specifically required otherwise.
Spin echo MRI scans were obtained using a 0.3-T Fonar System (Fonar Corp, Melville, NY) obtaining dual echo proton density axial and T1-weighted sagittal sections in 40 studies. The routine short echo time/repetition time (TR/TE) pulse sequence was 400 to 800/16-30 milliseconds (TR range/TE range) in T1-weighted images, and 2000/30 to 85 milliseconds (TR/TE range) in dual echo proton density images and T1. Ten additional studies were with a 1.5-T system (General Electric System, GE Medical Systems, Westborough, Mass), with pulse sequences of 500 to 800/20 milliseconds (TR range/TE) in T1-weighted images, and 2000/30 to 85 milliseconds (TR/TE range) in T2-weighted images. Two other studies were done with a 0.5-T unit (Picker International Inc, Roslyn Heights, NY) with pulse sequences of 500 to 750/20 milliseconds (TR range/TE) for T1-weighted images and a TR/TE of 2700/100 milliseconds for T2-weighted images. An average imaging time of 30 to 40 minutes varied from unit to unit depending on the number of imaging planes, image quality, etc. Contrast enhancement with gadolinium was used in the majority of patients. Results of noncontrast studies were analyzed for this study.
Each set of images was systematically reviewed for the presence of abnormal T2-weighted signal lesions, the presence of multiplicity and/or bilaterality, and the presence of mass effect. Two groups of patients were identified, those with brain lesions (WBL) and those with no brain lesions (NBL). All initial and subsequent images from the WBL group were analyzed for number, location (brainstem [BS] including diencephalic structures, cerebellum [CB], and hemispheres [H]), and size of lesions. Maximal diameter measurements of high-signal lesions were made independently by us to the nearest 0.5 mm in the axial plane and a consensus mean measurement was recorded. Individual lesions were measured once even if present on more than 1 contiguous image slice. Interslice gap width was 5 mm or less in all images. Measurements were categorized first by size (defined as small, ≤5 mm; moderate, >5 mm but ≤15 mm; and large, >15 mm), and indexed using the formula: Total index units = (number of large lesions×3) + (number of moderate lesions×2) + (number of small lesions×1). Left and right sides of the 3 anatomical brain regions were tallied separately then combined. Total lesion number and unit size per subject were then examined over serial examinations. Clinical records of all patients whose conditions were evaluated in which the findings from physical, neurologic, and ophthalmologic examinations were delineated were reviewed.
Of 123 patients with NF-1 whose conditions were evaluated, 30 had undergone 59 MRI scans and were available for review. There were 22 males and 8 females (aged 1-53 years, with a mean age of 12.5 years) with 41 of 59 studies in patients younger than 12 years and 18 of 59 studies in patients older than 12 years. There were 50 WBL studies of which 36 were in patients younger than 12 years and 14 in patients older than 12 years. Fifteen patients (14 WBL; 1 NBL) had multiple MRI examinations (mean of 3 examinations per patient) and 15 patients (10 NBL; 5 WBL) had single studies (Table 1 and Table 2).
On initial study, 63 of 129 lesions were located within the H in 19 patients, 19 of 129 lesions were within the BS in 10 patients, and 47 of 129 lesions were within the CB in 10 patients. The most common lesion locations were the globus pallidus (48 of 129 lesions) and dentate nucleus (39 of 129 lesions). The BS lesions were most commonly identified within the pons (12 of 23 lesions) followed by the midbrain (5 of 23 lesions) and the medulla (2 of 23 lesions).
Of the 19 WBL patients undergoing serial studies, a total of 103 lesions equaling 220 units (mean, 2.01 units per lesion) were identified at initial study. First follow-up evaluation (interval, 0.5-4.5 years; mean, 2.3 years) showed a decrease in both total number (73 lesions [−29%]) and size (132 units [−33%]; mean, 1.81 units per lesion). In WBL patients there were decreases in total number of lesions and size per patient of −20% and −27%, respectively. One patient had lesions resolve completely (Figure 1 and Figure 2). Importantly, BS lesions per patient increased in both number (+30%) and size (+24%) over the same intervals in 7 of 13 patients studied serially, whereas H and CB lesions decreased. True mass effect was evident in lesions confined to the BS and cerebellar peduncles and a single H lesion. There were 3 (16%) of 19 WBL patients who had focal abnormal neurologic examination findings referable to lesion location. No patient with H lesions had abnormalities on neurological examination findings referable to lesion site. Of the subset of patients with BS lesions, 10 (56%) of 19, abnormalities found on neurologic examination results referable to BS dysfunction were found in 2 (20%) of 10 and of the subset of patients with CB lesions, in 10 (56%) of 19; abnormalities on neurologic examination findings referable to CB dysfunction were found in 1 (10%) of 10. Of note was that all 19 WBL patients had lesions in the H either alone or in combination with another location. Seven patients had lesions in the H alone; 8, in all 3 locations (H, BS, and CB); 2, in H and BS; and 2, in H and CB.
Therapy for lesions identified on MRI was required in 3 cases. In each case there was contrast enhancement within the lesion that was distinct from all other lesions. In one case the patient had resection and radiotherapy of a radiographically enlarging, headache-producing, supratentorial mass lesion identified as a grade 3 cystic astrocytoma (Figure 3). A second patient underwent biopsy and radiotherapy of a contiguous lesion extending from the medulla to the thalamus identified as glioblastoma multiforme. The third patient with a multicentric BS tumor required a ventriculoperitoneal shunt, subsequent tumor debulking, and radiotherapy for a pilocytic astrocytoma. The patient with glioblastoma multiforme died 11 months after diagnosis was made.
Brain MRI is not indicated in the routine evaluation of patients with NF-1, especially for those who are asymptomatic. Nonetheless, the use of this imaging modality in the study of neurocutaneous disorders has aided in the identification of both asymptomatic structural abnormalities and a greater definition of the extent of symptomatic structural abnormalities. Thorough evaluation of patients with seemingly clinically homogeneous disease allows for a better understanding of the natural history and a more detailed definition of the phenotype. However, the clinical heterogeneity of NF-1, and the unpredictability of disease manifestations introduces a considerable degree of counseling ambiguity as to anticipated disease complications. In this regard, the clinician is faced with asking whether a particular study will provide important and useful information. Since there is often no clinical neurologic deficit correlated with the presence of these high-signal parenchymal lesions (particularly hemispheric) seen with MRI (84% in this study), defining the natural course of their evolution and biological character would be of value in understanding and interpreting their significance for the individual patient.
In this study we were able to prospectively document the evolution of high-signal brain lesions followed serially with MRI. The majority of these lesions were located within the deep gray structures of the basal ganglia, dentate nuclei, and throughout the BS. Mass effect was predominantly associated with lesions confined to the BS, thalamus, and cerebellar peduncles. T2-weighted imaging demonstrated sequential regression of both lesion number and size, quantified by an indexing system over a mean of 2 to 3 years (Figure 4). While we did not perform volumetric lesion analysis, we were able to track specific lesions conservatively to quantitate longitudinally evolving changes.
Some investigators have found T2-weighted, fluid-attenuated inversion recovery pulse sequences more effective in detecting multiple lesions in patients with NF-1 than the conventional T2-weighted spin echo imaging.11 Our cohort of patients did not undergo study with this technique and we cannot comment on the greater sensitivity of this method.
Our findings agree with prior studies that found that transient lesions occur often within the brain parenchyma, particularly the basal ganglia.7,12 This evolution is most pronounced on T2-weighted imaging but is also demonstrated on T1-weighted imaging in a minority of patients (35/195).7,12- 14 The time course of evolution is more prolonged and appears more specific for basal ganglia lesions.12 This gradual evolution is particularly notable when all regions are cumulatively analyzed (Table 2). Of importance, however, is the regional variation encountered when specific brain structures are evaluated separately (Table 1). Because of this transient radiological behavior, lesions would be likely associated with dysregulation in the quantity or function of the NF-1 gene cellular product, an altered rate of product maturation, or a disturbance in product signal transduction.2 It has been speculated that myelin maturation may be a candidate end point of neurofibromin effect.12 Included among other effects under consideration are local cellular proliferation, fibrosis, neuronal alterations, focal edema formation, and the development of microcalcification.
The vacuolar (spongiotic) changes in myelin found on autopsy material from T2-weighted signal lesions reported by DiPaolo et al9 were likely water-containing vacuoles. Additional findings from this series of 3 patients included microcalcifications within basal ganglia, perivascular schwannosis, and anaplastic glioma from a diffusely infiltrating BS mass.9 These later findings suggest the possibility that there may be differential effects of the NF-1 gene product (neurofibromin) on different central nervous system locations. In our current series, lesions of the BS, thalamus, and cerebellar peduncles were more likely to evolve into true masses. These lesions were heterogeneous in their radiological behavior. The greater likelihood of these lesions to enlarge over time suggests that their intrinsic cause may be somewhat different than that of the high-signal lesions seen elsewhere. Whether the NF-1 gene product acts preferentially differently in various brain regions remains to be proven. One might postulate that cell growth and proliferation in response to NF-1 gene product effect is brain region specific.
In support of this hypothesis are recent, limited, functional imaging data as they pertain to MRI-identified, T2-weighted lesions. Positron emission tomographic scanning using fludeoxyglucose F 18 in 4 patients with NF-1 demonstrated widespread hypometabolism concurrent with normal metabolism in the high-signal lesions of the globus pallidus and internal capsule.15 However, a thalamic lesion studied showed profoundly reduced metabolism.15 Findings from a recent series of proton magnetic resonance spectroscopy studies of T2-weighted lesions in patients with NF-1 were reported as normal.16,17 These lesions (hamartomas) and intervening unaffected brain tissue were compared with that of normal controls and tissue-documented glioma.17 Examination findings disclosed notable differences between the high-signal lesions in patients with NF-1 and various grades of gliomas but no differences with normal controls.17 This was the case in patients with NF-1 with and without associated learning disability.17 Lesion sites chosen for study were the middle cerebellar peduncle (n=8), globus pallidus (n=2), and thalamus (n=1).17 However, the regions were not separately compared with corresponding normal regions.
Initial correlational studies found no impact on cognitive function attributed to the presence or absence of these high-signal lesions on MRI. 4,18 More recent studies, however, have revealed more compelling evidence of a relationship between cognitive impairments and the presence of these lesions in children with NF-1 compared with those children with NF-1 but without these lesions. 19,20 When sibling control design studies have been used, the number of lesions rather than lesion location have had significant correlation (P<.001) with lowered full-scale IQ test results.19 Multiple regression analysis of a large sibling control study (19 sibling pairs), demonstrated that the multiplicity of high-signal lesion locations and not the total lesion volume or mode of inheritance of NF-1 (sporadic vs familial) accounts for the largest percentage of IQ reduction.21
With further study of the brain using the advent of functional imaging techniques, greater insight into the biological behavior and physiological impact on the brain of these high-signal MRI lesions will hopefully be gained.
Accepted for publication September 12, 1997.
Presented in part at the 24th Annual Meeting of the Child Neurology Society, Baltimore, Md, October 26, 1995.
Corresponding author: Francis J. DiMario, Jr, MD, Department of Pediatrics, Division of Neurology, Connecticut Children's Medical Center, 282 Washington St, Hartford, CT 06106.