Detection of Mitochondrial DNA Mutations in the Tumor and Cerebrospinal Fluid of Medulloblastoma Patients1

[CANCER RESEARCH 63, 3866–3871, July 15, 2003] Advances in Brief

Lee-Jun C. Wong,2 Maria Lueth,2 Xiao-Nan Li, Ching C. Lau,3 and Hannes Vogel Institute for Molecular and Human Genetics, Georgetown University, Washington, DC, 20007 [L-J. C. W., M. L.]; Division of Hematology/Oncology, Department of Pediatrics, Baylor College of Medicine, Houston, Texas 77030 [X-N. L., C. C. L.]; and Department of Pathology, Stanford University Medical Center, Stanford, California 94305 [H. V.]

Abstract require respiratory chain activity and can occur independently of the mitochondrial apoptosis pathway (6). Because the net growth of Medulloblastoma is the most common malignant brain tumor in chil- medulloblastomas represents a balance between cell division and dren. Although the prognosis has improved considerably, leptomeningeal apoptosis, the role of mitochondria in energy metabolism and apo- spread of the tumor remains a significantly negative predictor of survival. ptosis is likely to be of importance in understanding both the biology Mitochondrial DNA (mtDNA) mutations have been detected in many types of human tumors but not in medulloblastomas. Using temporal of medulloblastoma and as a potential target of therapy. 4 temperature gradient gel electrophoresis, we have analyzed the entire In an increasing list of diverse neoplasms, mtDNA alterations have mitochondrial genome in 15 cases of medulloblastoma and the corre- been documented (7). These have proven to be of interest as tumor sponding cerebrospinal fluid (CSF) samples in 8 of 15 cases. Six of 15 cases markers and as evidence that they may play a role in neoplastic (40%) showed at least one mtDNA mutation in each of the tumors. A total cellular adaptation to altered energy requirements. The nuclear ge- of 18 somatic mtDNA mutations was detected with one of the tumors nome has also been investigated in cancer in relation to mitochondrial having 11 mutations, of which 9 were novel. Seven of 8 CSF samples that physiology upon observing that mutations in genes encoding struc- were analyzed showed mtDNA mutations. One patient who showed per- tural subunits of complex II are present in autosomal dominant he- sistent mtDNA mutation in the CSF collected at the end of therapy when reditary paraganglioma (8). These mutations lead to a defect in oxy- there was no evidence of disease had a relapse 5 months later. In contrast, gen sensing and the response to hypoxia, functionally resulting in loss patients whose end-of-therapy CSF samples that showed either no detectable mtDNA mutation or different mutations from that of the tumor of the tumor suppressor’s role for complex II (9). Nonsense and continue to be disease free. Our results demonstrate that mtDNA muta- germ-line mutations involving complex II subunits have also been tions are frequently found in medulloblastomas. The mtDNA alterations reported in familial and sporadic pheochromocytoma (10–12). These detected in CSF may be used as sensitive markers to monitor disease observations lead to the present study in which the complete mito- progression and predict relapse. chondrial genome sequences in 15 cases of medulloblastoma are compared with those in blood and CSF from these patients. The Introduction possibility of using mtDNA alterations in CSF to monitor disease recurrence in medulloblastoma patient was also explored. Medulloblastoma is the most common pediatric malignant brain tumor involving the cerebellum. The pathogenesis of medulloblas- Materials and Methods toma is unknown in most cases, although certain germ-line and acquired genetic abnormalities are associated with a susceptibility to Patients. Tumor tissues, blood, and CSF samples were collected and medulloblastoma in subsets of patients. Cytogenetic abnormalities banked with informed consent through an Institutional Review Board- include isochromosome 17q in ϳ50% of cases. Losses of genetic approved protocol at Baylor College of Medicine and affiliated hospitals. material have also been documented from chromosomes 1q (1) and Tumor tissues were collected at the time of surgery and were snap frozen in Ϫ 10q (2) in 20–40% of medulloblastomas. Mutations of the human liquid nitrogen within 30 min after resection and stored at 80°C. Blood samples were collected at the time of surgery and before any additional homologue of the Drosophila patched gene (PTCH), as well as other therapy. DNA was extracted from blood samples immediately after collection. members of the sonic hedgehog pathway have also been reported in CSF samples were collected at various times during routine staging work-up 10–15% of sporadic medulloblastomas (3). In addition, the adenom- and therapy according to the treatment protocol. CSF was spun at 1000 rpm for atous polyposis coli gene and its corresponding WNT signaling path- 10 min immediately after collection and the supernatant frozen and stored at way have been implicated in ϳ13% of sporadic medulloblastomas (4). Ϫ80°C in 200-␮l aliquots. Treatment of all patients involved in this study, MYC amplification has been associated with large cell medulloblas- except patients 116 and 118, was according to the same in-house protocol for tomas (5). medulloblastoma, which included craniospinal radiation and chemotherapy. Mitochondrial abnormalities in medulloblastoma have not been Patient 116 expired before receiving any therapy because of rapid progression specifically investigated beyond their morphological features. A study of disease after initial surgery. Patient 118 received only chemotherapy and no radiation therapy because of his young age. Tissues from 15 patients, 12 male of the role of mitochondria in in vitro studies of apoptosis in medul- and 3 female, were used in this study with the age at diagnosis ranging from loblastoma cell lines indicates that ceramide-induced apoptosis may 14 months to 14 years (mean of 6.25 years). DNA Isolation. DNA was extracted from frozen tumor tissues with Trizol Received 2/4/03; accepted 5/20/03. reagent (Invitrogen, Carlsbad, CA) following the manufacturer’s instruction. The costs of publication of this article were defrayed in part by the payment of page Briefly, after the aqueous phase containing RNA has been removed, DNA was charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. precipitated with 100% ethanol, washed twice with 0.1 M sodium citrate in 1 Supported, in part, by NIH Grant CA87327, Muscular Dystrophy Association grant, 10% ethanol, and once with 75% ethanol. After the DNA pellet was dried and United States Army Grant DAMD17-01-1-0258 (to L-J. C. W.), and Fleming and redissolved in 8 mM NaOH, 0.1 M HEPES was added to adjust the pH to 8.4. Davenport Award (to C. C. L.). 2 These authors contributed equally to this study. 3 To whom requests for reprints should be addressed, at Texas Children’s Hospital, 4 The abbreviations used are: mtDNA mitochondrial DNA; CSF cerebrospinal fluid; 6621 Fannin Street, MC 3-3320, Houston, TX 77030. Phone: (832) 824-4543; Fax: TTGE temporal temperature gradient gel electrophoresis; ND4, NADH dehydrogenase (832) 825-4038; E-mail: [email protected]. subunit 4. 3866

DNA was extracted from blood leukocytes using the Wizard kit (Promega, from homoplasmic wild-type to homoplasmic variant. The T15904C Madison, WI) and stored at 4°C. Total DNA was quantified using fluorescent in tumor 124 was a change from homoplasmic variant in blood back Hoechst dye H33258 with Dyna Quant 200. DNA was diluted to 5 ng/␮ltobe to homoplasmic wild-type in tumor. Five changed from heteroplasmic used in PCR reactions. CSF was heated at 95°C for 5 min. Two ␮l of heated in blood to homoplasmic in tumor, 1 changed from homoplasmic in CSF were used for PCR without DNA extraction. blood to heteroplasmic in tumor, and 4 were heteroplasmic in both Mutational Analysis of the Entire Mitochondrial Genome. Thirty-two blood and tumor but with a quantitatively different proportion of the pairs of overlapping primers were used to amplify the entire mitochondrial genome (13). The DNA fragments vary in size from 306 to 805 bp with an mutant mtDNA. Three tumors have one somatic mtDNA mutation, 2 average of 594 bp. The amplified fragments had ϳ70-bp overlap at each end. have two mutations, and 1 (tumor 124) has 11 mutations. Because of The position and the sequence of the PCR primers, and the PCR and TTGE the numerous somatic mtDNA mutations in tumor 124, sample mis- conditions have recently been published (13). Briefly, the DNA template, after handling was suspected. To verify that blood and tumor DNA were the initial denaturation at 94°C for 5 min, was amplified with 35 cycles of 45 s indeed from the same patient, 4 polymorphic markers: the short each at 94°C for denaturation, 55°C for reannealing, and 72°C for extension, tandem repeat in intron 3 of the phenylalanine hydroxylase gene and completed by 4 min of extension at 72°C. PCR products were denatured (chromosome 12), the CTG repeats of the myotonin protein kinase at 95°C for 30 s and slowly cooled to 45°C for a period of 45 min at a rate of gene (disease gene for myotonic dystrophy, chromosome 19), the 1.1°C/min. The reannealed homoduplexes and heteroduplexes were kept at CAG repeats of the androgen receptor gene (X chromosome), and of 4°C until loading onto the gel. TTGE analysis was performed on a Bio-Rad spinocerebellar ataxia type 3 (chromosome 4) were analyzed. The D-Code apparatus (Bio-Rad Laboratories, Hercules, CA). Polyacrylamide (ac- identity of the blood specimen was found to match the corresponding rylamide:Bis ϭ 37.5:1) gels were prepared in 1.2ϫ TAE buffer containing 6 M Ͼ urea. Five ␮l of denatured and reannealed PCR products were loaded onto the tumor tissue at 98% probability. gel. Electrophoresis was carried out at 145 V for 4–5 h at a constant temper- Germ-Line Sequence Variations. DNA fragments showing dif- ature increment of between 1–2°C/h (13, 14). The temperature range was ferent banding patterns between blood and tumor tissues on TTGE determined by computer simulation (MacMelt software; Bio-Rad Laborato- analysis are indications of somatic mutations. During the sequence ries). The gels were stained with 2 mg/liter ethidium bromide for 5 min and analysis to identify the somatic mutations (Table 1A), it was noted that imaged with a digital CCD gel documentation system. On TTGE analysis, a there were numerous germ-line sequence variations when the se- single bandshift represents a homoplasmic DNA alteration and a multiple- quence of normal tissue (blood) was compared with that of the banding pattern represents a heteroplasmic mutation (14). PCR products from published Cambridge sequence. A total of 72 distinct germ-line vari- blood and tumor tissues of the same patient were analyzed side-by-side. Any ations was identified from the sequenced fragments. These do not DNA fragments showing different banding patterns between the blood and tumor samples were sequenced to identify the exact mutations. represent all of the sequence variations in the specimens analyzed Sequencing Analysis. The confirmation of mutation was performed by because only the DNA regions that show somatic mutations by TTGE direct DNA sequencing of the purified PCR product using the original PCR were sequenced. Thirteen of these variations are novel (Table 1B), and primers and a BigDye terminator cycle sequencing kit (Perkin-Elmer Life 59 (data not shown) of them have been previously recorded in the Sciences, Boston, MA) on an ABI 377 (Applied Biosystems, Foster City, CA) Mitomap database. Eleven of 13 novel variations occurred only once automated sequencer. The results of DNA sequence analysis were compared (Table 1B), whereas 24 of 59 reported polymorphisms occurred mul- with the published Cambridge sequence using Mac Vector 7.0 (Oxford Mo- tiple times. All germ-line variations are silent nucleotide substitutions lecular Ltd., Oxford, United Kingdom) software. Sequence variations found in or alterations in noncoding D-loop region or tRNA genes. both tumor and blood mtDNA were scored as germ-line variations. Each was mtDNA Mutations in CSF. CSF was available from 10 patients, then checked against the Mitomap database.5 Those not recorded in the including the 6 who harbored somatic mtDNA mutations in their database were categorized as novel mtDNA variations, and those appeared in the database were reported as polymorphisms. Any DNA sequences that were tumors. We were able to amplify the regions containing the mutations different between tumor and matched blood mtDNA were scored as somatic in 7 of them, 5 of which were from cases with mtDNA mutations in mutations. mtDNA from CSF was sequenced and compared with those from the tumor. The CSF sample of patient 126 failed to amplify. CSF of the blood and tumor of the same individual. patients 143 and 146 was not investigated because there was no mutation found in the tumor. Table 2 summarizes the age of diagnosis, RESULTS staging, clinical status, and somatic mtDNA mutations in tumor and CSF. The average age of diagnosis of the patients with somatic Somatic mtDNA Mutations in Medulloblastomas. TTGE analy- mtDNA mutations is higher (8.5 year) than that (5.1 year) of those sis revealed numerous homoplasmic and heteroplasmic somatic who do not have mtDNA mutations. The proportion of tumors having mtDNA mutations in medulloblastomas. Fig. 1A shows three ho- somatic mtDNA mutation is independent of the metastatic status of moplasmic mtDNA alterations: G7337A, T7389C, and G7521A, the patient, both at 40%. within a DNA fragment. Change from homoplasmy in blood to With patient 122, the same T152C somatic mutation found in tumor heteroplasmy in tumor is illustrated in Fig. 1B. Fig. 1C depicts a was also detected in CSF sample (Fig. 1C). The T152C mutation was change in the degree of heteroplasmy detectable by both TTGE and the only mutation found in the tumor and was in heteroplasmic state sequencing. in blood, tumor, and CSF, but the proportion of mutant was lowest in Overall, a total of 18 somatic mtDNA mutations were detected blood and highest in CSF (Fig. 1C). The CSF sample of this patient (Table 1A) in 6 of 15 medulloblastomas (40%). Among these muta- was taken 1 month after completion of all therapy at which time there tions, 5 (29.4%) were insertions or deletions in np 303–315 polyC was no evidence of disease based on magnetic resonance imaging tract region. The remaining 13 somatic mutations were single-base studies, as well as CSF cytology. However, this patient subsequently substitutions (72.2%). Eleven (61%) mutations are in the D-loop developed recurrent disease 5 months later. These results suggest that region, 3 in tRNA (G7521A, T15904C, and A15937G) and 4 in mtDNA mutation in CSF can be detected long before the recurrence mRNA (protein coding region; Table 1A). Two of the protein coding of disease. Presumably, cells resistant to therapy continued to divide region mutations are silent mutations and the other two are missense and the mutant mtDNA progressively accumulated. mutations: Y496H in cytochrome c oxidase subunit I (COX I) and Three regions, SD, TP, and D-loop, of CSF mtDNA of patient 124 L96P in ND4. Among the 18 somatic mutations, 8 were changing were amplified. This patient harbored a total of 11 somatic mtDNA mutations in the tumor, two of which, T15904C and A15937G, were 5 Internet address: http://www.mitomap.org. novel (Table 2). Only the T15904C mutation in tRNA threonine was 3867

Fig. 1. Detection and identification of somatic mtDNA mutations in tumor and CSF of patients with medulloblastoma by TTGE and sequencing. A, detection of homoplasmic mtDNA mutations in the SD region of patient 124 by TTGE. Sequencing analysis revealed multiple homoplasmic nucleotide substitutions in tumor mtDNA. B, detection of heteroplasmic mtDNA mutation in the ND4 region of patient 126 by TTGE. Sequencing analysis revealed a quantitative heteroplasmic change of T11046C. C, detection of mtDNA alterations in D-loop region of patient 122 in blood, tumor, and CSF. Sequencing analysis revealed a change of T152C with an obvious quantitative difference in the degree of heteroplasmy among the three specimens. D, mtDNA mutations in the tRNA threonine gene detected in blood, tumor, and CSF of patient 124. Sequencing analysis revealed two homoplasmic changes T15904C and A15937G in tumor but only the T15904C mutation in CSF.

found in the CSF sample (Fig. 1D). The mutant mtDNA appeared to homoplasmic, this suggests that the original spinal metastases could be homoplasmic. This patient initially presented with metastatic dis- represent an earlier clone of tumor cells that had migrated down the ease involving the spinal cord. The CSF sample was taken 6 months neuraxis and developed independently of the cerebellar tumor. This is after all therapy was completed at which time there was no evidence supported by the other mutation found in the CSF of this patient in the of disease by magnetic resonance imaging and CSF cytology. The 303–309 polyC tract region. It was C8/C7 heteroplasmy in blood, patient has remained disease free after Ͼ40 months of follow-up. homoplasmy C7 in tumor, and C9/C8 heteroplasmy in CSF. Although Unlike in the case of 122 where the mtDNA mutations in the tumor the possibility of normal cell contamination and the random effects of and the CSF were identical and the patient eventually relapsed, in this therapy cannot be ruled out, it is still hard to explain why the CSF and case, at least 7 mtDNA mutations found in the tumor were not tumor share some of the same mutations, especially rare ones like detected in the CSF. These results suggest that residual mtDNA from T15904C. Furthermore, in the case of patient 149 where the CSF was tumor cells can still be found long after therapy has been completed collected before any therapy was given, mtDNA mutations in the when no tumor cells can be detected by conventional methods such as 303–309 polyC tract region were also different between the tumor and imaging and cytology. Furthermore, because only one of several the CSF. Interestingly, patient 149 was diagnosed with nonmetastatic mtDNA mutations found in the tumor was detected in CSF and was disease and the fact that the pretherapy CSF sample contains mtDNA 3868

Table 1 Medulloblastoma Case Gene/ Cambridge nl to tu Amino acid Previously reported in no. Region Somatic mutation sequence patterna change Function tumors Reference A. Somatic mtDNA mutations 3 120 D-Loop 303–309,C9 C9/10 C7 Homo-hetero Conserv. seq block crc,gastric,eso,ov,brca 20, 16, 17 122 D-Loop T152C T Hetero-hetero H-strand origin brca,ov 17, 16 124 D-Loop C151T C Hetero-homo H-strand origin Novel This study 124 D-Loop C182T C Hetero-homo Novel This study 124 D-Loop G246A G Homo-homo mtTF Novel This study binding site 124 D-Loop A297G A Homo-home mtTF binding site Novel This study 3 124 D-Loop 303–309,C8/7 C7 C7 Hetero-homo Conserv. seq. block crc,gastric,eso,ov,brca 20, 16, 17 124 D-Loop G317A G Homo-homo Replication primer Novel This study 124 SD G7521A G Homo-homo tRNA asp Thyroid 124 SD G7337A G Homo-homo S478S COXI Novel This study 124 SD T7389C T Homo-homo Y496H COXII Novel This study 124 TP T15904C C Homo-homo tRNA thr Novel This study 124 TP A15937G A Homo-homo tRNA thr Novel This study 3 126 D-Loop 303–309,C8/9 C9 C7 Hetero-homo Conserv. seq. block crc,gastric,eso,ov,brca 20, 16, 17 126 ND4 T11046C T Hetero-hetero L96P NADH subunit 4 Novel This study 3 128 D-Loop 303–309,C9ϾϾ10 C9Ͼ10 C7 Hetero-homo Conserv. seq. block crc,gastric,eso,ov,brca 20, 16, 17 3 149 D-Loop 303–309,C8/9 C8 C7 Hetero-homo Conserv. seq. block crc,gastric,eso,ov,brca 20, 16, 17 149 COXI T7028C T Hetero-hetero A375A COXI Gene/Region Germ-line mutation Frequency Significance B. Novel germ-line variations

D-loop C298T 1 Hypervariable Segment 2 D-loop G417A 1 H-strand origin D-loop T507C 1 D-loop A538G 1 mtTF1 binding site D-loop A551G 1 Major H-strand promotor ND2 A4793G 1 ND2, ATA-ATG, M108M Wancy A5581G 1 non coding nucleotide COXI C6296A 1 COX I, CCC-CCA, P149P SD A7771G 1 COX II, GAA-GAG, E62E ND4L C10581T 1 CTA-TTA,L38L ND4L G10688A 2 GTG-GTA,V73V ND4L T10810C 2 CTT-CTC,L17L a nl, normal (blood); tu, tumor; homo, homoplasmic; hetero, heteroplasmic; thr, threonine; Conserv., conserved sequence block; cro, colorectal cancer; eso, esophageal cancer; ov, ovarian cancer; brca, breast cancer; Region: the region is named after the gene it encodes, tRNA gene is named after the single letter of the amino acid it carries (Ref. 13). mutations suggests that there could be micrometastasis present in the medulloblastomas and other tumors (17, 19). Back change from CSF compartment at the time of diagnosis that are beyond the detec- variant in normal tissue to wild-type in tumor tissue has also been tion limit of imaging and cytology. mtDNA mutations in the end-of- observed (Table 1A, T149; Ref. 17). These results demonstrate that therapy CSF samples from the other two nonmetastatic patients who tumorigenesis is a progressive process during which the mutant mi- harbored somatic mtDNA mutations in tumors (patients 120 and 128) tochondria may gain replicative advantage and somatic mutations were also in the np 303–309 polyC mutation hot-spot region and were accumulate (15, 20). different from those of the tumors (Table 2). The remaining two CSF The frequent mutation in np303–309 polyC region probably reflects samples (patients 135 and 140) that had mtDNA mutation were also mitochondrial genomic instability in this hypervariable triple helix collected after the completion of therapy and the corresponding tu- region. To investigate if this is the result of generalized microsatellite mors only had germ-line sequence variations but no somatic muta- instability, we sequenced 10 mtDNA regions containing the poly tions. Both showed the heteroplasmic reversion of the germ-line mononucleotide tracts, and no mutation was detected in these regions. variations to wild-type sequence (A247G in patient 135 and T295C in These results suggest that the instability at np 303–309 is probably not patient 140) in the CSF. The CSF of patient 140 harbored a second because of generalized microsatellite instability but because it locates heteroplasmic reversion of germ-line variation to wild-type. All 5 at the relatively unstable triple strand displacement loop region. The nonmetastatic patients and 1 metastatic patient that had different somatic mtDNA mutations observed in tumors are not PCR artifacts mtDNA mutations between the tumor and the CSF remain clinically for we have analyzed mtDNA in at least 50 muscle/blood specimens well without evidence of disease after follow-up of 27–46 months of individuals of various ages and somatic mtDNA mutations have not (median, 35 months). The single patient that showed accumulation of been found. Our observations substantiate the concept that genomic the same mtDNA mutation in CSF as was found in the tumor (patient instability in tumors should include the mitochondrial genome. 122) had recurrent disease. mtDNA Mutation Spectrum. The spectrum of somatic mtDNA mutations in medulloblastoma was compared with that of neurofi- Discussion broma type 1, glioblastoma, and breast cancer (Table 3). The inser- Somatic mtDNA Alterations in Medulloblastomas. The pres- tions and deletions at np 303–309 polyC regions occurred most ence of somatic mtDNA mutations in tumor tissue is a general frequently, consistent with the hot spot for mitochondrial genomic phenomenon in the neoplastic process (7, 15–18). Most reports de- instability in all tumors. If counting only the distinct somatic mtDNA scribed somatic mtDNA changes from homoplasmic wild-type in mutations, 71% (10 of 14) were novel in medulloblastoma (Table 3). normal tissue to homoplasmic mutant in the tumor tissue. We ob- All of the somatic mtDNA mutations in neurofibroma type 1 and 79% served changes from homo- to homo-, homo- to hetero-, hetero- to of the somatic mtDNA mutations in breast cancer are in the D-loop homo-, and hetero- to hetero- with different percentage of mutant in region (17, 19), but only 50% of the distinct somatic mtDNA muta- 3869

Table 2 mtDNA alterations in blood, tumor, and CSF of patients with medulloblastoma Clinical Case no. Gender Age at DX Staging status Sequence in blood Sequence in tumor Sequence in CSF 110 Female 8 years Ma Dead Cambridge seq. Cambridge seq. Not investigated 112 Male 5 years M NED Cambridge seq. Cambridge seq. Not investigated 114 Male 14 years NM NED Cambridge seq. Cambridge seq. Not investigated 116 Male 30 months NM Dead Cambridge seq. Cambridge seq. Not investigated 118 Male 14 months NM Dead Cambridge seq. Cambridge seq. Not investigated 120 Female 4 years NM NED 303–309,C9 (hm) 303–309,C9ϾϾ10(ht) 303–309,C8ϾϾ9(ht) 122 Male 6.5 years M AWD T152 T152C,TϽC(ht) T152C,TϽϽC(ht) 124 Male 6.5 years M NED C151 C151T No amplification C182 C182T No amplification G246 G246A G246 A297 A297G A297 303–309,C8ϾϾ7(ht) 303–309,C7(hm) 303–309,C9ϾϾ8(ht) G317 G317A G317 G7521 G7521A G7521 G7337 G7337A G7337 T7389 T7389C T7389 T15904 T15904C T15904C A15937 A15937G A15937 126 Male 12 years NM AWD T11046C(ht) T11046C(ht) No amplification 303–309,C8ϾϾ9(ht) 303–309,C8(hm) No amplification 128 Male 11 years NM NED 303–309,C9ϾϾ10(ht) 303–309,C9Ͼ10(ht) 303–309,C9(hm) 135 Male 9 years NM NED G247A(hm) G247A(hm) A247G,GϾϾA(ht) Cambridge seq. Cambridge seq. 303–309insC(ht) 140 Male 3 years NM NED C295T(hm) C295T(hm) T295C,CϾϾT(ht) 310insT(hm) 310insT(hm) T310C,CϾT(ht) 143 Male 5 years NM NED Cambridge seq. Cambridge seq. Not investigated 146 Male 3 years M NED Cambridge seq. Cambridge seq. Not investigated 149 Female 11 years NM NED 303–309,C8ϾϾ9(ht) 303–309,C8(hm) 303–309,C9ϾϾ8(ht) T7028C,TϾϾC(ht) T7028C,CϾϾT(ht) T7028 a M, Metastatic disease at the time of diagnosis; NM, nonmetastatic at the time of diagnosis; NED, no evidence of disease; AWD, alive with disease; Cambridge seq., cambridge sequence; hm, homoplasmic; ht, heteroplasmic.

tions of medulloblastoma are in the D-loop, which is similar to that of Somatic mtDNA Mutations in CSF as Potential Markers for glioblastoma multiforme (56%; Ref. 18). The average number of Disease Prognosis. Somatic mtDNA mutations were detectable in the somatic mtDNA mutations/tumor is around one to three. Kirches et al. CSF of patients with medulloblastoma with or without metastasis. (18) reported 19 somatic mtDNA mutations found in a single glio- There are three potential sources of mutated mtDNA in the CSF of blastoma. One of our patients harbored 11 somatic mtDNA mutations. these patients: the intracranial tumor; spinal metastases; and nontumor Significance of Somatic mtDNA Mutations. Mutations in coding cells in the spinal compartment that have been mutated by the therapy. region may play a role in the pathogenesis of tumors. The G7521A One would predict that the probability of detecting mtDNA mutations mutation in tRNA aspartate changes a GT base pairing to AT base in the CSF would be higher in the metastatic cases than the nonmeta- pairing at the amino acyl stem region. This alteration perhaps makes static cases. Unfortunately, with the small sample size in this study the stem region more stable and allows the tRNA to work more and the fact that most of the CSF samples were collected at the end of efficiently. Similarly, the T15904C mutation in tRNA threonine at the therapy, we were unable to confirm this prediction. Interestingly, all loop region and the A15937G mutation at the first bp next to the loop of the CSF mutations in the nonmetastatic cases involved the np may also affect tRNA structure and stability. Two missense muta- 303–309 polyC tract region, whereas none of the metastatic cases had tions, Y496H in cytochrome c oxidase and L96P in NADH subunit 4, CSF mutations in that region. Because all of the CSF samples except involved in the substitution of the hydrophobic aromatic tyrosine with 2 (patients 146 and 149) were collected either right after radiation positively charged histidine, and the hydrophobic leucine residue with therapy (patient 135) or 1–9 months after the completion of all ␣ helix destabilizing secondary amino acid proline, are nonconserved therapy, one would also predict that if the mutation in CSF is identical changes that are expected to bring about significant structural/func- to that found in the original tumor, it could indicate the presence of tional alteration. Mutations in noncoding region, although not directly resistant tumor cells and therefore might predict relapse of disease as involved in amino acid change, are located in transcription factor seen in patient 122. Likewise, if the CSF mutation reverts back to that binding sites, replication primer sites, origin of replication, or con- in normal tissue, it would signal the absence of tumor cells and served sequence blocks. Alterations in these positions may affect therefore predict a disease free status as in the case of patient 124. mtDNA replication, transcription, and overall expression. Finally, for CSF mutation that is found in neither the normal nor

Table 3 Distribution of novel somatic mtDNA mutations in various tumorsa Glioblastoma Neurofibromatosis Medulloblastoma multiforme type 1 Breast cancer No. of tumors 15 17 44 19 Total no. of distinct mutations 14 18 16 24 No. of novel distinct somatic mtDNA mutations 10 2 5 4 Percentage of novel distinct mtDNA mutations 71 11 31 16.6 No. of distinct mtDNA mutations in D-loop 7 10 16 19 Percentage of distinct mtDNA mutations in D-loop 50 56 100 79 Reference This study 19 To be published 18 a To define somatic mutations as novel or reported, we compared our data with the mitomap.org database as well as with all references listed in this publication. 3870

Downloaded from cancerres.aacrjournals.org on September 28, 2021. © 2003 American Association for Cancer Research. MITOCHONDRIAL DNA MUTATIONS IN MEDULLOBLASTOMA tumor tissues as in the cases of 120 and 128, the origin of these 5. Brown, H. G., Kepner, J. L., Perlman, E. J., Friedman, H. S., Strother, D. R., Duffner, P. K., Kun, L. E., Goldthwaite, P. T., and Burger, P. C. “Large cell/anaplastic” mutated mtDNA is uncertain. Although it could arise from normal medulloblastomas: a Pediatric Oncology Group Study. J. Neuropath. Exp. Neurol., cells in the CSF that have been mutated by radiation and/or chemo- 59: 857–865, 2000. therapy, it is hard to understand how these mtDNA would be able to 6. Poppe, M., Reimertz, C., Munstermann, G., Kogel, D., and Prehn, J. H. Ceramide- induced apoptosis of D283 medulloblastoma cells requires mitochondrial respiratory persist in the CSF long after therapy has been completed, especially if chain activity but occurs independently of caspases and is not sensitive to Bcl-xL the mutated cells are not proliferating as confirmed by the other overexpression. J. Neurochem., 82: 482–494, 2002. surveillance studies. Would these mtDNA eventually disappear from 7. Copeland, W. C., Wachsman, J. T., Johnson, F. M., and Penta, J. S. Mitochondrial DNA alterations in cancer. Cancer Investig., 20: 557–569, 2002. the CSF over time? Such questions can be addressed by conducting a 8. Niemann, S., and Muller, U. Mutations in SDHC cause autosomal dominant para- time series study sampling multiple CSF samples from the same ganglioma, type 3. Nat. Genet., 26: 268–270, 2000. 9. Shoubridge, E. A. Nuclear genetic defects of oxidative phosphorylation. Hum. Mol. patients at various times during long-term follow-up. Lastly, with CSF Genet., 10: 2277–2284, 2001. samples taken before any therapy, it would also be interesting to 10. Gimm, O., Armanios, M., Dziema, H., Neumann, H. P., and Eng, C. Somatic and correlate clinical outcome with the similarity between the mutations in occult germ-line mutations in SDHD, a mitochondrial complex II gene, in nonfamilial pheochromocytoma. Cancer Res., 60: 6822–6825, 2000. the CSF and tumor. Identical mutations in the CSF and tumor would 11. Astuti, D., Latif, F., Dallol, A., Dahia, P. L., Douglas, F., George, E., Skoldberg, F., imply higher turnover of the tumor cells, especially in the nonmeta- Husebye, E. S., Eng, C., and Maher, E. R. Gene mutations in the succinate dehydrogenase subunit SDHB cause susceptibility to familial pheochromocytoma and to static cases. Because apoptosis is used as a marker for higher grade in familial paraganglioma. Am. J. Hum. Genet., 69: 49–54, 2001. medulloblastoma, identical mutations in CSF and tumor would sug- 12. Astuti, D., Douglas, F., Lennard, T. W., Aligianis, I. A., Woodward, E. A., Evans, gest a higher grade tumor and therefore poorer prognosis. Conversely, D. G., Eng, C., Latif, F., and Maher, E. R. Germline SDHD mutation in familial phaeochromocytoma. Lancet, 357: 1181–1182, 2001. divergent mutation patterns could be associated with better prognosis 13. Wong, L-J. C., Liang, M-H., Kwon, H., Park, J., Bai, R., and Tan, D. Comprehensive as seen in patient 149. Ultimately, a prospective study involving a scanning of the whole mitochondrial genome for mutations. Clin. Chem., 48: 1901– larger patient population and serial CSF samples from each patient 1912, 2002. 14. Chen, T. J., Boles, R., and Wong, L-J. C. Detection of mitochondrial DNA mutations will be necessary to establish the utility of mtDNA mutations as bi- by temporal temperature gradient gel electrophoresis. Clin. Chem., 45: 1162–1167, omarkers for prognostication and early detection of recurrent disease. 1999. 15. Fliss, M. S., Usadel, H., Caballero, O. L., Wu, L., Buta, M. R., Eleff, S. M., Jen, J., and Sidransky, D. Facile detection of mitochondrial DNA mutations in tumors and References bodily fluids. Science (Wash. DC), 287: 2017–2019, 2000. 16. Liu, V. W. S., Shi, H. H., Cheung, A. N. Y., Chiu, P. M., Leung, T. W., Nagley, P., 1. Kraus, J. A., Koch, A., Albrecht, S., Von Deimling, A., Wiestler, O. D., and Pietsch, Wong, L. C., and Ngan, H. Y. S. High incidence of somatic mitochondrial DNA T. Loss of heterozygosity at locus F13B on chromosome 1q in human medulloblas- mutations in human ovarian carcinomas. Cancer Res., 61: 5998–6001, 2001. toma. Int. J. Cancer, 67: 11–15, 1996. 17. Tan, D-J., Bai, R., and Wong, L-J. C. Comprehensive scanning of somatic mitochon- 2. Reardon, D. A., Michalkiewicz, E., Boyett, J. M., Sublett, J. E., Entrekin, R. E., drial DNA mutations in breast cancer. Cancer Res., 62: 972–976, 2002. Ragsdale, S. T., Valentine, M. B., Behm, F. G., Li, H., Heideman, R. L., Kun, L. E., 18. Kirches, E., Krause, G., Warich-Kirches, M., Weis, S., Schneider, T., Meyer-Puttlitz, Shapiro, D. N., and Look, A. T. Extensive genomic abnormalities in childhood B., Mawrin, C., and Dietzmann, K. High frequency of mitochondrial DNA mutations medulloblastoma by comparative genomic hybridization. Cancer Res., 57: 4042– in glioblastoma multiforme identified by direct sequence comparison to blood sam- 4047, 1997. ples. Int. J. Cancer, 93: 534–538, 2001. 3. Zurawel, R. H., Allen, C., Chiappa, S., Cato, W., Biegel, J., Cogen, P., de Sauvage, 19. Lueth, M., Foster, R., Kurtz, A., and Wong, L-J. C. Somatic mitochondrial DNA F., and Raffel, C. Analysis of PTCH/SMO/SHH pathway genes in medulloblastoma. mutations are found in neurofibromatosis type 1. Proc. Am. Assoc. Cancer Res., 43: Genes Chromosomes Cancer, 27: 44–51, 2000. 1484, 2002. 4. Huang, H., Mahler-Araujo, B. M., Sankila, A., Chimelli, L., Yonekawa, Y., Kleihues, 20. Polyak, K., Li, Y., Zhu, H., Lengauer, C., Willson, J. K., Markowitz, S. D., Trush, P., and Ohgaki, H. APC mutations in sporadic medulloblastomas. Am. J. Pathol., 156: M. A., Kinzler, K. W., and Vogelstein, B. Somatic mutations of the mitochondrial 433–437, 2000. genome in human colorectal tumours. Nat. Genet., 20: 291–293, 1998.

3871

Downloaded from cancerres.aacrjournals.org on September 28, 2021. © 2003 American Association for Cancer Research. Detection of Mitochondrial DNA Mutations in the Tumor and Cerebrospinal Fluid of Medulloblastoma Patients

Lee-Jun C. Wong, Maria Lueth, Xiao-Nan Li, et al.

Cancer Res 2003;63:3866-3871.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/63/14/3866

Cited articles This article cites 20 articles, 7 of which you can access for free at: http://cancerres.aacrjournals.org/content/63/14/3866.full#ref-list-1

Citing articles This article has been cited by 7 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/63/14/3866.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/63/14/3866. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.