- Email: [email protected]
Magnetic resonance spectroscopy in childhood brainstem tumors
Magnetic Resonance Spectroscopy in Childhood Brainstem Tumors Richard G. Curless, MD*, Brian C. Bowen, MD, PhD†, Padrip M. Pattany, PhD†, Renato Gonik, MD*, and Deborah L. Kramer, MD‡ Five children with brainstem tumors and two control patients had magnetic resonance spectroscopy studies of the brainstem. Two of the malignant tumor patients had magnetic resonance spectroscopy studies before and after radiation therapy. The third was irradiated 14 years earlier but developed new symptoms and a new brainstem lesion on MRI. Magnetic resonance spectroscopy demonstrated a different degree of malignancy between the old and new lesion. The fourth patient had magnetic resonance spectroscopy of a chronic, large pontine lesion 6 years after diagnosis and radiation. The spectral pattern suggested a low degree of malignancy. The fifth patient had neurofibromatosis type 1 with brainstem lesions. Magnetic resonance spectroscopy suggested neoplastic tissue of low malignancy. These results suggest that magnetic resonance spectroscopy offers additional information for anticipating the degree of anaplasia in children with brainstem tumors. © 2002 by Elsevier Science Inc. All rights reserved. Curless RG, Bowen BC, Pattany PM, Gonik R, Kramer DL. Magnetic resonance spectroscopy in childhood brainstem tumors. Pediatr Neurol 2002;26:374-378.
Introduction Brainstem tumors represent approximately 10% of brain neoplasms in children, and less than 50% of children with such tumors survive 2 years after diagnosis regardless of the therapeutic modality [1]. High-dose chemotherapy with stem cell rescue has not been successful [2]. In another study, 130 children with brainstem gliomas were treated with hyperfractionated radiotherapy [3]. Only nine of these children achieved a long-term survival. In most cases, magnetic resonance imaging (MRI) technology obviates the need for a biopsy to confirm the
From the Departments of *Neurology; †Radiology; and ‡Pediatrics; University of Miami School of Medicine; Miami Beach, Florida.
374
PEDIATRIC NEUROLOGY
Vol. 26 No. 5
presence of a tumor. The accurate determination of malignancy relies on the clinical course. In this study, MRI and localized proton spectroscopy are combined in a single examination in an attempt to characterize the morphology and potential aggressiveness of brainstem lesions in five children. Two of the patients (Patients 1 and 2) had a rapidly progressive clinical course, and two patients (Patients 4 and 5) had an indolent clinical pattern. Patient 3 was a 14-year-old with a stable tumor who developed a satellite lesion of high malignancy. A few examples of spectra obtained from pediatric brainstem tumors have appeared in the literature [4]. Serial spectra from the same brainstem tumor have not been reported, and there have been no published reports correlating the spectral results with the clinical course.
Methods The magnetic resonance study was performed on a 1.5-torr clinical scanner and consisted of an initial imaging component followed by single-voxel 1H spectroscopy. Axial and sagittal T1-weighted spin-echo images (TR ⫽ 650 ms, TE ⫽ 20 ms, number of signal averages (NSA) ⫽ 1), T2-weighted fast spin-echo axial images (TR ⫽ 3555 ms, TE ⫽ 96 ms, NSA ⫽ eff/1}, and fast FLAIR axial images (T1 ⫽ 2000 ms, TR ⫽ 7155 ms, TE ⫽ 112 ms, NSA ⫽ 1) were obtained with the same slice thickness (5 mm), gap (1 mm), field of view (230 mm), and matrix (256 ⫻ 256). Postcontrast (gadolinium pentetate dimeglumine, 0.1 mmol/kg) T1-weighted axial, sagittal, and coronal images were also obtained. For the purpose of voxel placement, additional coronal and sagittal T1weighted, spin-echo pilot images (TR ⫽ 200 ms, TE ⫽ 20 ms, NSA ⫽ 1) at 5-mm thickness, no gap, 300-mm field of view, and 128 ⫻ 256 matrix were acquired. In all five patients, localized 1H spectra were obtained from a 4.5-mL voxel (1.5 cm ⫻ 1.5 cm ⫻ 2cm) located within the lesion that was visualized on the T2-weighted images. In Patient 3, MRI of the brainstem and midbrain revealed two areas with different characteristics: a pontine lesion that had not been detected on previous magnetic resonance images and a predominantly midbrain lesion that was unchanged in appearance. Because the pontine lesion was new and accounted for the new clinical symptoms, it was considered as a separate entity from the midbrain lesion. Thus six spectra were obtained from six lesions in the five
Communications should be addressed to: Dr. Curless; 5950 La Gorce Drive; Miami Beach, FL 33140. Received June 12, 2001; accepted December 5, 2001.
© 2002 by Elsevier Science Inc. All rights reserved. PII S0887-8994(01)00418-0 ● 0887-8994/02/$—see front matter
Figure 1. Fluid-attenuated inversion recovery axial images from Patient 3 (TR ⫽ 6000, T1 ⫽ 2000, effective echo time ⫽ 128 ms, echo train length ⫽ 8) are presented. (A) Axial image at the level of the mid to lower pons reveals the voxel encompassing the right pontine lesion. (B) Axial image at the level of the upper pons indicates no signal abnormality, indicating a distinct separation between the pontine lesion (A) and the midbrain lesion (C). (C) Axial image located at the level of the midbrain illustrates the voxel located within the midbrain lesion.
patients. For the two volunteers, spectra were obtained from a 4.5-mL voxel that was positioned within the brainstem. Water-suppressed spectra were acquired using a point resolved spectroscopy (PRESS) sequence (TR ⫽ 1500 ms, TE ⫽ 135 ms), 256 accumulations, bandwidth ⫾ 1000 Hz, and 2048 data points. The spectral analysis routine involved eddy-current correction, zero-filling to 8 K data points, and Gaussian-Lorentzian filtering in the time domain, followed by global phase correction using unsuppressed reference water signal, linear phase correction, and baseline fitting in the frequency domain. The spectra were fitted using a seventh-order polynomial function to approximate the baseline and a fourth-order Gaussian-Lorentzian function to approximate the resonance lines or peaks. Peak assignments were as follows: N-acetyl (representing the combined resonances of N-acetylaspartate and N-acetylaspartylglutamate) at 2.01 ppm, total creatine (representing the sum of creatine and phosphocreatine) at 3.02 ppm, choline (representing the combined resonances of small molecular weight choline-containing compounds, primarily phosphocholine and glycerophosphocholine) at 3.22 ppm, and lactate at 1.33 ppm (center of the inverted doublet). From the fitted spectra, the ratio of the area under the peak of interest to the area under the creatine peak (e.g., N-acetyl/creatine and choline/creatine peak ratios) was calculated. Magnetic resonance spectroscopic imaging was not used in this study. It has been shown to identify regions of viable cancer in heterogeneous lesions with a broad range of histopathologic findings [5]. However, the technique has rarely been applied to the evaluation of brainstem tumors primarily because the larger volumes sampled in magnetic resonance spectroscopic imaging acquisitions are more prone to artifacts from susceptibility effects and bone marrow fat signal intensity at the skull base than localized, single-voxel acquisitions [6,7].
Patient 1 A 9-year-old male presented with a 3-week history of double vision, left-sided weakness, difficulty swallowing, and mild headaches. There was no associated vomiting, fever, or personality change. The past medical history, review of systems, and family history failed to reveal any neurologic problems that would relate to the present illness. Mother died of human immunodeficiency virus-related problems including a “brain tumor.” Examination revealed a right sixth nerve palsy. The results of the remainder of the cranial nerve examination were normal. The motor examination revealed a mild left hemiparesis. Both plantar responses were flexor. There were no cerebellar abnormalities, and general cortical function was normal. Computed tomography and MRI scans revealed a large pontomedullary lesion with nodular enhancement. A neurosurgical consultant suggested therapy without a biopsy because of the presumptive diagnosis of a pontine glioma. Two milligrams of intravenous dexamethasone were administered every 6 hours, and magnetic resonance spectroscopy (MRS) was obtained before treatment with fractional irradiation with a total dose of 5,400 cGy. Four months later, a right lower motor neuron seventh nerve palsy was first noted, and a second MRS was obtained. One month later, he developed more severe dysphagia, facial weakness, and difficulty walk-
ing. MRI findings revealed an increased size of the pontine tumor. Chemotherapy was initiated with etoposide, but he died 8 months after diagnosis. Postmortem examination was denied.
Patient 2 A 3-1/2-year-old female was admitted on the same day as Patient 1. Two weeks earlier she was first noted to have a mild left-sided weakness. Over the next 12 days she developed increasing left hemiparesis, began drooling, and developed difficulty pronouncing words. Examination revealed a mild left hemiparesis manifested by difficulty walking and left hyperreflexia. Both plantar responses were flexor. There was no cranial nerve abnormality, and her behavior was normal. MRI revealed a large, nonenhancing pontine mass. Intravenous dexamethasone was administered, and a MRS was obtained before radiation therapy with a total dose of 5,440 cGy in 160-cGy fractions. One month later, examination revealed a significant improvement with only a mild left hemiparesis. A second MRS was obtained 4 months after the initial study. Five months after irradiation therapy she developed a right sixth nerve palsy. MRI revealed an enlarging mass. At 8 months, her left hemiparesis became worse. A third MRS was obtained. Etoposide was administered, but she died 2 months later. Postmortem examination was not allowed.
Patient 3 An 18-year-old male was diagnosed with a brainstem tumor at 4 years of age. Symptoms at that time included dysarthria and right hemiparesis. MRI revealed a tumor of the left pons, midbrain, and thalamus with obstructive hydrocephalus. After a ventriculoperitoneal shunt, he was treated with hyperfractionated radiotherapy with a total dose of 6,140 cGy. Speech improved, but the hemiparesis persisted. No new symptoms or MRI findings appeared until the patient was 18 years of age, when he presented with a rapidly progressive left hemiparesis, dysphagia, and dysarthria. MRI revealed no change in the lesions originally identified 14 years earlier, although a new lesion was found in the right pons (Fig 1). MRS of the old and new lesions was obtained (Fig 2). The patient was treated with procarbazine, vincristine, and lomustine. He expired several months later. A postmortem examination was not permitted.
Patient 4 A 7-year-old female was diagnosed with a large pontine tumor at 3-1/2 years of age. She presented with a left hemiparesis and a marked left sixth nerve palsy and was treated with steroids and irradiation. The neurologic symptoms resolved during steroid therapy except for a mild left sixth nerve palsy. The current examination revealed only the unchanged left sixth nerve abnormality. MRI findings demonstrated a 4-cm pontine mass. A MRS of the lesion was obtained.
Curless et al: MRS in Brainstem Tumors 375
Figure 3. Point resolved spectroscopy spectrum obtained at TE ⫽ 135 ms from the brainstem of Patient 6 (control subject) is illustrated.
Patient 7 (Control Subject) Magnetic resonance imaging was requested to evaluate this 5-year-old male with developmental delay and a congenital left hemiplegia. The study revealed a right open lip schizencephaly. The brainstem was normal. A brainstem MRS was obtained.
Results
Figure 2. Point resolved spectroscopy spectra obtained at TE ⫽ 135 ms from the voxels illustrated in Figure 1 (Patient 3) are presented. (A) Right pontine lesion is depicted. (B) Midbrain lesion is demonstrated.
Patient 5 A MRI was obtained on a 13-year-old male with neurofibromatosis type 1. He had multiple caf, au lait spots, Lisch nodules (iris hamartomas), mild mental retardation, macrocephaly, generalized epilepsy, and no focal neurologic findings. The MRI revealed multiple high-signal lesions in the basal ganglia and pons. A lesion in the left midbrain and adjacent cerebellar peduncle had a higher than usual signal intensity with some mass effect. There was no enhancement. A MRS was obtained.
Patient 6 (Control Subject) Magnetic resonance imaging was obtained on this 4-year-old male because of developmental delay and aggressive behavior. The study results were normal. A brainstem MRS was also obtained (Fig 3).
376
PEDIATRIC NEUROLOGY
Vol. 26 No. 5
Based on the clinical evaluation, physical examination, and hospital course after the initial MRS study, the lesions in the five patients were divided into two groups: group II lesions were associated with an indolent presentation and course, and group I lesions were associated with more aggressive clinical features. The peak area ratios for N-acetyl/creatine and choline/creatine are shown in Table 1. For the first MRS study, the N-acetyl/creatine mean values for the patients with brainstem tumor are approximately 40% (indolent lesion) to 50% (aggressive lesion) of the control values. The choline/creatine values differ considerably for the indolent and the aggressive lesions, being increased by approximately 10% for the former and 110% for the latter, compared with control values. In Patient 1, with a clinically aggressive tumor, two successive spectroscopy studies were performed. The second study was obtained 4 months after the first study. The N-acetyl/creatine decreased by approximately 17%, and the choline/creatine decreased by approximately 37% between the two studies. The patient’s neurologic condition had been stable until the end of that 4-month interval. During the first 4-5 months after the onset of treatment, the patient demonstrated neurologic improvement and subsequently deteriorated. In Patient 2, with clinically aggressive tumor, three spectroscopy studies were performed. The second and third studies were obtained 4 and 8 months, respectively,
Table 1. Metabolite levels (peak area ratios relative to Cr) in brainstem lesions
Patient Group I 1 2 3a Mean Group II 3b 4 5 Mean Control 6 7 Mean
First Visit NA/Cr Cho/Cr
1.25 0.71 1.12 1.02
3.64 2.77 2.65 3.02
0.41 0.87 1.16 0.81
1.53 1.2 1.92 1.55
2.13 2.07 2.1
1.32 1.5 1.41
Second Visit NA/Cr Cho/Cr
1.03 0.59
2.29 1.47
Third Visit NA/Cr Cho/Cr
0.68
1.98
Abbreviations: Cho/Cr ⫽ Choline/creatine NA/Cr ⫽ N-acetyl group/creatine
after the first study. The N-acetyl/creatine decreased by approximately 16% between the first and second studies and then increased by approximately 14% between the second and third studies. The choline/creatine decreased by approximately 47% between the first and second visits but then increased by approximately 35% between the second and third visits. This patient demonstrated initial clinical improvement but then deteriorated by the time of the third study. Patient 3 has a clinically active lesion, which we have labeled 3a, and an old indolent lesion, which we have labeled 3b. In Patient 3, the smaller pontine lesion (group I, lesion 3a) had a N-acetyl/creatine value that was more than twice that of the larger midbrain lesion (group II, lesion 3b). The choline/creatine value was approximately 73% greater than that of the midbrain lesion. Concerning the lactate resonance, the control patients had lactate/creatine values of ⫺0.21 (Patient 6) and ⫺0.11 (Patient 7). The only brainstem lesion with a lactate/ creatine value exceeding control values was the lesion observed in Patient 2 who demonstrated lactate/creatine values of ⫺0.45 (at the first visit) and ⫺0.33 (at the third visit). These ratios are negative because the lactate peak is inverted at TE ⫽ 135 ms. Discussion The accuracy of MRI studies, the high frequency of malignant tissue in childhood brainstem tumors, and the surgical risk have greatly reduced the justification for biopsy. MRS of cerebellar tumors in children revealed that benign tumors have a relatively low N-acetyl/choline ratio and a high lactate level [8]. The more malignant tissue demonstrated a lower N-acetyl/choline ratio and no additional elevation of lactate. Brainstem tumors were not
evaluated in this study or in an earlier review of MRS in childhood brain tumors [9]. Patients 1 and 2 had the clinical and radiographic diagnosis of a pontine tumor with anaplastic characteristics. MRS was obtained on each patient before and after radioactive therapy. MRS in the third patient revealed two brainstem lesions with different MRS characteristics. The lesion that had been present for many years appeared much less anaplastic. In Patients 4 and 5 the spectra were appropriate for a nonmalignant tumor, which agreed with the clinical picture. Elevated lactate/creatine was not observed in the brainstem lesions in either group, except in Patient 2 at the time of the first and third visits. This finding suggests that necrosis with anaerobic glycolysis is not a prominent feature of these lesions. Table 1 details the spectra results on all seven patients. The clinical course and magnetic resonance spectral pattern of Patients 3b, 4, and 5 suggested a benign or indolent lesion. Patient 3 has an old, indolent lesion (lesion 3b) and a new clinically active lesion (lesion 3a). The indolent cases have significant reduction in N-acetyl/creatine (compared with the control subjects) and mild elevation in choline/creatine in Patients 3b and 5. The clinical radiographic picture of these three lesions supports a sustained indolent pattern in Patients 3b and 4, whereas Patient 5 could demonstrate a premalignant state. The low N-acetyl/ creatine in Patient 5 and the diagnosis of neurofibromatosis type 1 should alert the physicians to the necessity of a prolonged follow-up. Hopefully Patient 4 has a lifetime benign lesion. The three examples of clinically aggressive tumors, Patients 1, 2, and 3a, demonstrate a greater elevation in choline/creatine and a more modest reduction in N-acetyl/ creatine. In addition, the three studies of Patient 2 add support to the greater significance of the alteration in choline/creatine as an indicator of malignancy. It is noteworthy that at the time of the second study on Patient 2 there was definite clinical improvement. At that time, the choline/creatine had fallen to 1.47 compared with 2.77 4 months earlier. When the third study was completed, new neurologic deficits had developed. The choline/creatine had climbed to 1.98. It is also noteworthy that although the second choline/creatine in Patient 1 suggested some reduction in anaplasia, clinical improvement did not develop. In summary, more aggressive brainstem tumors have larger values of choline/creatine, as has been reported for pediatric cerebellar tumors. This finding is best explained by the more rapid cellular proliferation and accompanying membrane synthesis occurring in these tumors compared with normal tissue or tissue that no longer has much neuronal activity. In addition, the amount of decrease in N-acetyl/creatine within brainstem tumors is not related to the clinical aggressiveness. It appears that the replacement of neurons by tumor cells occurs to various extents in both indolent and aggressive tumors [5]. The serial studies in
Curless et al: MRS in Brainstem Tumors 377
Patient 2 suggest that choline/creatine values more closely reflect the clinical condition, and that N-acetyl/creatine changes lag behind the choline/creatine and clinical changes. Perhaps rapid cellular proliferation affects the clinical neurologic examination before the loss of neurons is sufficiently great to be detected by MRS. Based on the results of this study, we conclude that MRS in childhood brainstem tumors enhances the diagnostic capability of routine MRI studies. In some cases, MRS indicating low malignancy will support a decision to delay radiation therapy. Such a delay would be particularly helpful in young children who have a high risk for radiation damage. In situations such as neurofibromatosis type 1, the level of malignancy may be higher than anticipated resulting in expeditious initiation of therapy that might otherwise be delayed.
References [1] Shiminski-Maher T. Brain stem tumors in childhood: Preparing patients and families for long- and short-term care. Pediatr Neurosurg 1996;24:267-71.
378
PEDIATRIC NEUROLOGY
Vol. 26 No. 5
[2] Dunkel IJ, O’Malley B, Finlay JL. Is there a role for high-dose chemotherapy with stem cell rescue for brain stem tumors of childhood? Pediatr Neurosurg 1996;24:263-6. [3] Freeman CR, Bourgoin PM, Sanford RA, Cohen ME, Friedman HS, Kun LE. Long term survivors of childhood brain stem gliomas treated with hyperfractionated radiotherapy. Clinical characteristics and treatment related toxicities. Cancer 1996;77:555-62. [4] Tzika AA, Vigneron DB, Ball WS, Dunn RS, Kirks DR. Localized proton MR spectroscopy of the brain in children. J Magn Reson Imaging 1993;3:719-29. [5] Dowling C, Bollen AW, Noworolski SM, et al. Preoperative proton MR Spectroscopic imaging of brain tumors: Correlation with histopathological analysis of resection specimens. Am J Neuroradiol 2001;22:604-12. [6] Barker PB. Fundamentals of clinical MRS. Syllabus, Magnetic resonance spectroscopy: Clinical applications and research frontiers. 8th Scientific Meeting of the International Society for Magnetic Resonance in Medicine, Denver, 2000:400-8. [7] Speilman D. Spectroscopic imaging techniques. Syllabus, Magnetic resonance spectroscopy. Clinical applications and research frontiers. 8th Scientific Meeting of the International Society for Magnetic Resonance in Medicine, Denver, 2000:416-22. [8] Wang Z, Sutton LN, Canaan A, et al. Proton MR spectroscopy of pediatric cerebellar tumors. Am J Neuroradiol 1995;16:1821-33. [9] Sutton LN, Wang Z, Gusnard D, et al. Proton magnetic resonance spectroscopy of pediatric brain tumors. Neurosurgery 1992;31: 195-202.
Introduction Brainstem tumors represent approximately 10% of brain neoplasms in children, and less than 50% of children with such tumors survive 2 years after diagnosis regardless of the therapeutic modality [1]. High-dose chemotherapy with stem cell rescue has not been successful [2]. In another study, 130 children with brainstem gliomas were treated with hyperfractionated radiotherapy [3]. Only nine of these children achieved a long-term survival. In most cases, magnetic resonance imaging (MRI) technology obviates the need for a biopsy to confirm the
From the Departments of *Neurology; †Radiology; and ‡Pediatrics; University of Miami School of Medicine; Miami Beach, Florida.
374
PEDIATRIC NEUROLOGY
Vol. 26 No. 5
presence of a tumor. The accurate determination of malignancy relies on the clinical course. In this study, MRI and localized proton spectroscopy are combined in a single examination in an attempt to characterize the morphology and potential aggressiveness of brainstem lesions in five children. Two of the patients (Patients 1 and 2) had a rapidly progressive clinical course, and two patients (Patients 4 and 5) had an indolent clinical pattern. Patient 3 was a 14-year-old with a stable tumor who developed a satellite lesion of high malignancy. A few examples of spectra obtained from pediatric brainstem tumors have appeared in the literature [4]. Serial spectra from the same brainstem tumor have not been reported, and there have been no published reports correlating the spectral results with the clinical course.
Methods The magnetic resonance study was performed on a 1.5-torr clinical scanner and consisted of an initial imaging component followed by single-voxel 1H spectroscopy. Axial and sagittal T1-weighted spin-echo images (TR ⫽ 650 ms, TE ⫽ 20 ms, number of signal averages (NSA) ⫽ 1), T2-weighted fast spin-echo axial images (TR ⫽ 3555 ms, TE ⫽ 96 ms, NSA ⫽ eff/1}, and fast FLAIR axial images (T1 ⫽ 2000 ms, TR ⫽ 7155 ms, TE ⫽ 112 ms, NSA ⫽ 1) were obtained with the same slice thickness (5 mm), gap (1 mm), field of view (230 mm), and matrix (256 ⫻ 256). Postcontrast (gadolinium pentetate dimeglumine, 0.1 mmol/kg) T1-weighted axial, sagittal, and coronal images were also obtained. For the purpose of voxel placement, additional coronal and sagittal T1weighted, spin-echo pilot images (TR ⫽ 200 ms, TE ⫽ 20 ms, NSA ⫽ 1) at 5-mm thickness, no gap, 300-mm field of view, and 128 ⫻ 256 matrix were acquired. In all five patients, localized 1H spectra were obtained from a 4.5-mL voxel (1.5 cm ⫻ 1.5 cm ⫻ 2cm) located within the lesion that was visualized on the T2-weighted images. In Patient 3, MRI of the brainstem and midbrain revealed two areas with different characteristics: a pontine lesion that had not been detected on previous magnetic resonance images and a predominantly midbrain lesion that was unchanged in appearance. Because the pontine lesion was new and accounted for the new clinical symptoms, it was considered as a separate entity from the midbrain lesion. Thus six spectra were obtained from six lesions in the five
Communications should be addressed to: Dr. Curless; 5950 La Gorce Drive; Miami Beach, FL 33140. Received June 12, 2001; accepted December 5, 2001.
© 2002 by Elsevier Science Inc. All rights reserved. PII S0887-8994(01)00418-0 ● 0887-8994/02/$—see front matter
Figure 1. Fluid-attenuated inversion recovery axial images from Patient 3 (TR ⫽ 6000, T1 ⫽ 2000, effective echo time ⫽ 128 ms, echo train length ⫽ 8) are presented. (A) Axial image at the level of the mid to lower pons reveals the voxel encompassing the right pontine lesion. (B) Axial image at the level of the upper pons indicates no signal abnormality, indicating a distinct separation between the pontine lesion (A) and the midbrain lesion (C). (C) Axial image located at the level of the midbrain illustrates the voxel located within the midbrain lesion.
patients. For the two volunteers, spectra were obtained from a 4.5-mL voxel that was positioned within the brainstem. Water-suppressed spectra were acquired using a point resolved spectroscopy (PRESS) sequence (TR ⫽ 1500 ms, TE ⫽ 135 ms), 256 accumulations, bandwidth ⫾ 1000 Hz, and 2048 data points. The spectral analysis routine involved eddy-current correction, zero-filling to 8 K data points, and Gaussian-Lorentzian filtering in the time domain, followed by global phase correction using unsuppressed reference water signal, linear phase correction, and baseline fitting in the frequency domain. The spectra were fitted using a seventh-order polynomial function to approximate the baseline and a fourth-order Gaussian-Lorentzian function to approximate the resonance lines or peaks. Peak assignments were as follows: N-acetyl (representing the combined resonances of N-acetylaspartate and N-acetylaspartylglutamate) at 2.01 ppm, total creatine (representing the sum of creatine and phosphocreatine) at 3.02 ppm, choline (representing the combined resonances of small molecular weight choline-containing compounds, primarily phosphocholine and glycerophosphocholine) at 3.22 ppm, and lactate at 1.33 ppm (center of the inverted doublet). From the fitted spectra, the ratio of the area under the peak of interest to the area under the creatine peak (e.g., N-acetyl/creatine and choline/creatine peak ratios) was calculated. Magnetic resonance spectroscopic imaging was not used in this study. It has been shown to identify regions of viable cancer in heterogeneous lesions with a broad range of histopathologic findings [5]. However, the technique has rarely been applied to the evaluation of brainstem tumors primarily because the larger volumes sampled in magnetic resonance spectroscopic imaging acquisitions are more prone to artifacts from susceptibility effects and bone marrow fat signal intensity at the skull base than localized, single-voxel acquisitions [6,7].
Patient 1 A 9-year-old male presented with a 3-week history of double vision, left-sided weakness, difficulty swallowing, and mild headaches. There was no associated vomiting, fever, or personality change. The past medical history, review of systems, and family history failed to reveal any neurologic problems that would relate to the present illness. Mother died of human immunodeficiency virus-related problems including a “brain tumor.” Examination revealed a right sixth nerve palsy. The results of the remainder of the cranial nerve examination were normal. The motor examination revealed a mild left hemiparesis. Both plantar responses were flexor. There were no cerebellar abnormalities, and general cortical function was normal. Computed tomography and MRI scans revealed a large pontomedullary lesion with nodular enhancement. A neurosurgical consultant suggested therapy without a biopsy because of the presumptive diagnosis of a pontine glioma. Two milligrams of intravenous dexamethasone were administered every 6 hours, and magnetic resonance spectroscopy (MRS) was obtained before treatment with fractional irradiation with a total dose of 5,400 cGy. Four months later, a right lower motor neuron seventh nerve palsy was first noted, and a second MRS was obtained. One month later, he developed more severe dysphagia, facial weakness, and difficulty walk-
ing. MRI findings revealed an increased size of the pontine tumor. Chemotherapy was initiated with etoposide, but he died 8 months after diagnosis. Postmortem examination was denied.
Patient 2 A 3-1/2-year-old female was admitted on the same day as Patient 1. Two weeks earlier she was first noted to have a mild left-sided weakness. Over the next 12 days she developed increasing left hemiparesis, began drooling, and developed difficulty pronouncing words. Examination revealed a mild left hemiparesis manifested by difficulty walking and left hyperreflexia. Both plantar responses were flexor. There was no cranial nerve abnormality, and her behavior was normal. MRI revealed a large, nonenhancing pontine mass. Intravenous dexamethasone was administered, and a MRS was obtained before radiation therapy with a total dose of 5,440 cGy in 160-cGy fractions. One month later, examination revealed a significant improvement with only a mild left hemiparesis. A second MRS was obtained 4 months after the initial study. Five months after irradiation therapy she developed a right sixth nerve palsy. MRI revealed an enlarging mass. At 8 months, her left hemiparesis became worse. A third MRS was obtained. Etoposide was administered, but she died 2 months later. Postmortem examination was not allowed.
Patient 3 An 18-year-old male was diagnosed with a brainstem tumor at 4 years of age. Symptoms at that time included dysarthria and right hemiparesis. MRI revealed a tumor of the left pons, midbrain, and thalamus with obstructive hydrocephalus. After a ventriculoperitoneal shunt, he was treated with hyperfractionated radiotherapy with a total dose of 6,140 cGy. Speech improved, but the hemiparesis persisted. No new symptoms or MRI findings appeared until the patient was 18 years of age, when he presented with a rapidly progressive left hemiparesis, dysphagia, and dysarthria. MRI revealed no change in the lesions originally identified 14 years earlier, although a new lesion was found in the right pons (Fig 1). MRS of the old and new lesions was obtained (Fig 2). The patient was treated with procarbazine, vincristine, and lomustine. He expired several months later. A postmortem examination was not permitted.
Patient 4 A 7-year-old female was diagnosed with a large pontine tumor at 3-1/2 years of age. She presented with a left hemiparesis and a marked left sixth nerve palsy and was treated with steroids and irradiation. The neurologic symptoms resolved during steroid therapy except for a mild left sixth nerve palsy. The current examination revealed only the unchanged left sixth nerve abnormality. MRI findings demonstrated a 4-cm pontine mass. A MRS of the lesion was obtained.
Curless et al: MRS in Brainstem Tumors 375
Figure 3. Point resolved spectroscopy spectrum obtained at TE ⫽ 135 ms from the brainstem of Patient 6 (control subject) is illustrated.
Patient 7 (Control Subject) Magnetic resonance imaging was requested to evaluate this 5-year-old male with developmental delay and a congenital left hemiplegia. The study revealed a right open lip schizencephaly. The brainstem was normal. A brainstem MRS was obtained.
Results
Figure 2. Point resolved spectroscopy spectra obtained at TE ⫽ 135 ms from the voxels illustrated in Figure 1 (Patient 3) are presented. (A) Right pontine lesion is depicted. (B) Midbrain lesion is demonstrated.
Patient 5 A MRI was obtained on a 13-year-old male with neurofibromatosis type 1. He had multiple caf, au lait spots, Lisch nodules (iris hamartomas), mild mental retardation, macrocephaly, generalized epilepsy, and no focal neurologic findings. The MRI revealed multiple high-signal lesions in the basal ganglia and pons. A lesion in the left midbrain and adjacent cerebellar peduncle had a higher than usual signal intensity with some mass effect. There was no enhancement. A MRS was obtained.
Patient 6 (Control Subject) Magnetic resonance imaging was obtained on this 4-year-old male because of developmental delay and aggressive behavior. The study results were normal. A brainstem MRS was also obtained (Fig 3).
376
PEDIATRIC NEUROLOGY
Vol. 26 No. 5
Based on the clinical evaluation, physical examination, and hospital course after the initial MRS study, the lesions in the five patients were divided into two groups: group II lesions were associated with an indolent presentation and course, and group I lesions were associated with more aggressive clinical features. The peak area ratios for N-acetyl/creatine and choline/creatine are shown in Table 1. For the first MRS study, the N-acetyl/creatine mean values for the patients with brainstem tumor are approximately 40% (indolent lesion) to 50% (aggressive lesion) of the control values. The choline/creatine values differ considerably for the indolent and the aggressive lesions, being increased by approximately 10% for the former and 110% for the latter, compared with control values. In Patient 1, with a clinically aggressive tumor, two successive spectroscopy studies were performed. The second study was obtained 4 months after the first study. The N-acetyl/creatine decreased by approximately 17%, and the choline/creatine decreased by approximately 37% between the two studies. The patient’s neurologic condition had been stable until the end of that 4-month interval. During the first 4-5 months after the onset of treatment, the patient demonstrated neurologic improvement and subsequently deteriorated. In Patient 2, with clinically aggressive tumor, three spectroscopy studies were performed. The second and third studies were obtained 4 and 8 months, respectively,
Table 1. Metabolite levels (peak area ratios relative to Cr) in brainstem lesions
Patient Group I 1 2 3a Mean Group II 3b 4 5 Mean Control 6 7 Mean
First Visit NA/Cr Cho/Cr
1.25 0.71 1.12 1.02
3.64 2.77 2.65 3.02
0.41 0.87 1.16 0.81
1.53 1.2 1.92 1.55
2.13 2.07 2.1
1.32 1.5 1.41
Second Visit NA/Cr Cho/Cr
1.03 0.59
2.29 1.47
Third Visit NA/Cr Cho/Cr
0.68
1.98
Abbreviations: Cho/Cr ⫽ Choline/creatine NA/Cr ⫽ N-acetyl group/creatine
after the first study. The N-acetyl/creatine decreased by approximately 16% between the first and second studies and then increased by approximately 14% between the second and third studies. The choline/creatine decreased by approximately 47% between the first and second visits but then increased by approximately 35% between the second and third visits. This patient demonstrated initial clinical improvement but then deteriorated by the time of the third study. Patient 3 has a clinically active lesion, which we have labeled 3a, and an old indolent lesion, which we have labeled 3b. In Patient 3, the smaller pontine lesion (group I, lesion 3a) had a N-acetyl/creatine value that was more than twice that of the larger midbrain lesion (group II, lesion 3b). The choline/creatine value was approximately 73% greater than that of the midbrain lesion. Concerning the lactate resonance, the control patients had lactate/creatine values of ⫺0.21 (Patient 6) and ⫺0.11 (Patient 7). The only brainstem lesion with a lactate/ creatine value exceeding control values was the lesion observed in Patient 2 who demonstrated lactate/creatine values of ⫺0.45 (at the first visit) and ⫺0.33 (at the third visit). These ratios are negative because the lactate peak is inverted at TE ⫽ 135 ms. Discussion The accuracy of MRI studies, the high frequency of malignant tissue in childhood brainstem tumors, and the surgical risk have greatly reduced the justification for biopsy. MRS of cerebellar tumors in children revealed that benign tumors have a relatively low N-acetyl/choline ratio and a high lactate level [8]. The more malignant tissue demonstrated a lower N-acetyl/choline ratio and no additional elevation of lactate. Brainstem tumors were not
evaluated in this study or in an earlier review of MRS in childhood brain tumors [9]. Patients 1 and 2 had the clinical and radiographic diagnosis of a pontine tumor with anaplastic characteristics. MRS was obtained on each patient before and after radioactive therapy. MRS in the third patient revealed two brainstem lesions with different MRS characteristics. The lesion that had been present for many years appeared much less anaplastic. In Patients 4 and 5 the spectra were appropriate for a nonmalignant tumor, which agreed with the clinical picture. Elevated lactate/creatine was not observed in the brainstem lesions in either group, except in Patient 2 at the time of the first and third visits. This finding suggests that necrosis with anaerobic glycolysis is not a prominent feature of these lesions. Table 1 details the spectra results on all seven patients. The clinical course and magnetic resonance spectral pattern of Patients 3b, 4, and 5 suggested a benign or indolent lesion. Patient 3 has an old, indolent lesion (lesion 3b) and a new clinically active lesion (lesion 3a). The indolent cases have significant reduction in N-acetyl/creatine (compared with the control subjects) and mild elevation in choline/creatine in Patients 3b and 5. The clinical radiographic picture of these three lesions supports a sustained indolent pattern in Patients 3b and 4, whereas Patient 5 could demonstrate a premalignant state. The low N-acetyl/ creatine in Patient 5 and the diagnosis of neurofibromatosis type 1 should alert the physicians to the necessity of a prolonged follow-up. Hopefully Patient 4 has a lifetime benign lesion. The three examples of clinically aggressive tumors, Patients 1, 2, and 3a, demonstrate a greater elevation in choline/creatine and a more modest reduction in N-acetyl/ creatine. In addition, the three studies of Patient 2 add support to the greater significance of the alteration in choline/creatine as an indicator of malignancy. It is noteworthy that at the time of the second study on Patient 2 there was definite clinical improvement. At that time, the choline/creatine had fallen to 1.47 compared with 2.77 4 months earlier. When the third study was completed, new neurologic deficits had developed. The choline/creatine had climbed to 1.98. It is also noteworthy that although the second choline/creatine in Patient 1 suggested some reduction in anaplasia, clinical improvement did not develop. In summary, more aggressive brainstem tumors have larger values of choline/creatine, as has been reported for pediatric cerebellar tumors. This finding is best explained by the more rapid cellular proliferation and accompanying membrane synthesis occurring in these tumors compared with normal tissue or tissue that no longer has much neuronal activity. In addition, the amount of decrease in N-acetyl/creatine within brainstem tumors is not related to the clinical aggressiveness. It appears that the replacement of neurons by tumor cells occurs to various extents in both indolent and aggressive tumors [5]. The serial studies in
Curless et al: MRS in Brainstem Tumors 377
Patient 2 suggest that choline/creatine values more closely reflect the clinical condition, and that N-acetyl/creatine changes lag behind the choline/creatine and clinical changes. Perhaps rapid cellular proliferation affects the clinical neurologic examination before the loss of neurons is sufficiently great to be detected by MRS. Based on the results of this study, we conclude that MRS in childhood brainstem tumors enhances the diagnostic capability of routine MRI studies. In some cases, MRS indicating low malignancy will support a decision to delay radiation therapy. Such a delay would be particularly helpful in young children who have a high risk for radiation damage. In situations such as neurofibromatosis type 1, the level of malignancy may be higher than anticipated resulting in expeditious initiation of therapy that might otherwise be delayed.
References [1] Shiminski-Maher T. Brain stem tumors in childhood: Preparing patients and families for long- and short-term care. Pediatr Neurosurg 1996;24:267-71.
378
PEDIATRIC NEUROLOGY
Vol. 26 No. 5
[2] Dunkel IJ, O’Malley B, Finlay JL. Is there a role for high-dose chemotherapy with stem cell rescue for brain stem tumors of childhood? Pediatr Neurosurg 1996;24:263-6. [3] Freeman CR, Bourgoin PM, Sanford RA, Cohen ME, Friedman HS, Kun LE. Long term survivors of childhood brain stem gliomas treated with hyperfractionated radiotherapy. Clinical characteristics and treatment related toxicities. Cancer 1996;77:555-62. [4] Tzika AA, Vigneron DB, Ball WS, Dunn RS, Kirks DR. Localized proton MR spectroscopy of the brain in children. J Magn Reson Imaging 1993;3:719-29. [5] Dowling C, Bollen AW, Noworolski SM, et al. Preoperative proton MR Spectroscopic imaging of brain tumors: Correlation with histopathological analysis of resection specimens. Am J Neuroradiol 2001;22:604-12. [6] Barker PB. Fundamentals of clinical MRS. Syllabus, Magnetic resonance spectroscopy: Clinical applications and research frontiers. 8th Scientific Meeting of the International Society for Magnetic Resonance in Medicine, Denver, 2000:400-8. [7] Speilman D. Spectroscopic imaging techniques. Syllabus, Magnetic resonance spectroscopy. Clinical applications and research frontiers. 8th Scientific Meeting of the International Society for Magnetic Resonance in Medicine, Denver, 2000:416-22. [8] Wang Z, Sutton LN, Canaan A, et al. Proton MR spectroscopy of pediatric cerebellar tumors. Am J Neuroradiol 1995;16:1821-33. [9] Sutton LN, Wang Z, Gusnard D, et al. Proton magnetic resonance spectroscopy of pediatric brain tumors. Neurosurgery 1992;31: 195-202.