Purpose and Experimental Design: The role of RASSF1A has been elucidated recently in regulating apoptosis and cell cycle progression by inhibiting cyclin D1 accumulation. Aberrant RASSF1A promoter methylation has been found frequently in multiple adult cancer types. Using methylation-specific PCR and reverse transcription-PCR, we investigated epigenetic deregulation of RASSF1A in primary tumors, adjacent nontumor tissues, secondary metastases, peripheral blood cells, and plasma samples from children with 18 different cancer types, in association with their clinicopathologic features.
Results: Regardless of the tumor size, ubiquitous RASSF1A promoter methylation was found in 67% (16 of 24) of pediatric tumors, including neuroblastoma, thyroid carcinoma, hepatocellular carcinoma, pancreatoblastoma, adrenocortical carcinoma, Wilms’ tumor, Burkitt’s lymphoma, and T-cell lymphoma. A majority (75%) of pediatric cancer patients with tumoral RASSF1A methylation was male. Methylated RASSF1A alleles were also detected in 4 of 13 adjacent nontumor tissues, suggesting that this epigenetic change is potentially an early and critical event in childhood neoplasia. RASSF1A promoter methylation found in 92% (11 of 12) of cell lines largely derived from pediatric cancer patients was significantly associated with transcriptional silencing/repression. After demethylation treatment with 5-aza-2′-deoxycytidine, transcriptional reactivation was shown in KELLY, RD, and Namalwa cell lines as analyzed by reverse transcription-PCR. For the first time, RASSF1A methylation was detected in 54% (7 of 13), 40% (4 of 10), and 9% (1 of 11) of buffy coat samples collected before, during, and after treatment, correspondingly, from pediatric patients with neuroblastoma, thyroid carcinoma, hepatocellular carcinoma, rhabdomyosarcoma, Burkitt’s lymphoma, T-cell lymphoma, or acute lymphoblastic leukemia. Concordantly, RASSF1A methylation was found during treatment in plasma of the same patients, suggesting cell death and good response to chemotherapy.
Conclusions: RASSF1A methylation in tumor or buffy coat did not correlate strongly with age, tumor size, recurrence/metastasis, or overall survival in this cohort of pediatric cancer patients. Of importance, epigenetic inactivation of RASSF1A may potentially be crucial in pediatric tumor initiation.
Childhood neoplasms are biologically different from adult tumors, in that childhood neoplasms resemble the embryonic precursors of the cell types they arise, similar to undifferentiated cells appearing during normal embryonic development (1 , 2) . Genetic or epigenetic alterations in pediatric tumors may potentially lead to arrested differentiation or dedifferentiation. The biological characteristics of pediatric malignancies, such as the ability to proliferate, invade, migrate, and exhibit differential sensitivity to cytotoxic agents, may provide insights into normal cellular and developmental processes. Pediatric tumors are very often highly invasive, metastasizing early in the course of the development, but they are very responsive to current therapies (1 , 2) .
In contrast to inconsistent chromosomal aberrations found in various adult cancer types, recurring cytogenetic abnormalities are observed in pediatric cancers (1 , 2) . As opposed to adult epithelial tumors, pediatric solid tumors possessing only a few genetic mutations can develop after short latent periods (2 , 3) . The methylation patterns in adult tumors have been studied extensively, and each tumor type appears to have a distinct methylation profile (4) . However, little has been known about the methylation profiles in childhood malignancies.
RAS plays an important role in the signal transduction from cell surface receptors to an array of intracellular signaling pathways. Mutations leading to constitutive activation of RAS are commonly found in human cancers (4 , 5) . RAS binds and activates a diverse array of effectors and mediates tumor suppressive effects in addition to oncogenic effects (6) . Activated RAS mediates the induction of DNA synthesis (7) , tumorigenic transformation (8) , metastasis/invasion (9) , reduction of growth factor dependence (10) , loss of contact inhibition (11) , inhibition of terminal differentiation (12) , and resistance to apoptosis (13) . On the other hand, RAS can induce growth inhibitory effects, such as senescence (14) , necrosis (15) , apoptosis (16) , and terminal differentiation.
Loss of heterozygosity of chromosome 3p21.3 is one of the most frequent alterations in solid tumors (17 , 18) . Located within this 3p21.3 locus, RASSF1 encodes a novel RAS effector, which has been identified recently as a tumor suppressor of many different cancer types (19, 20, 21) . The RASSF1 gene has two CpG islands within two known promoters controlling gene expression (19) . RASSF1 encodes two major transcripts, 1A and 1C, by alternative promoter usage and alternative RNA splicing. RASSF1A and 1C transcripts have four common exons, which encode a COOH-terminal RAS association domain (19 , 22) . RASSF1A and RASSF1C have PEST sequences with a serine residue as a putative phosphorylation target for ataxia-telangiectasia-mutation (23) . As differed from RASSF1C, RASSF1A has an NH2-terminal SH3 domain and a putative cysteine-rich diacylglycerol/phorbolester-binding domain (24) . RASSF1A inactivation can be a tumorigenic mechanism distinct from the oncogenic activation of RAS signaling. Loss of RASSF1A expression may shift the balance of RAS activities toward a growth-promoting effect (25) .
Frequent RASSF1A promoter methylation has been observed recently in tumor types with uncommon RAS mutations and associated with transcriptional silencing of RASSF1A (18 , 19 , 26 , 27) . In fact, RASSF1A blocks cell cycle progression from G1 phase to S phase by controlling the entry at the retinoblastoma restriction point and inhibiting cyclin D1 protein accumulation at the post-transcriptional level (28) . RASSF1A has been implicated in suppressing tumorigenesis in vitro and in vivo (19) . Reactivation of RASSF1A transcription in lung carcinoma cells reduced colony formation, suppressed cell growth dependent or independent of anchorage, and inhibited tumor formation in nude mice (19) . Oncogenic RAS does not alter RASSF1A-induced growth inhibitory effects in an immortalized cell line, but the effects of RASSF1A are dominant to oncogenic RAS in human mammary epithelial cells (28) . Thus, loss of RASSF1A can be a determining step for oncogenic transformation without RAS-activating mutations.
In the present study, we analyzed RASSF1A promoter methylation in a wide range of pediatric tumors and cell lines derived from pediatric cancer patients. The association between promoter methylation and transcriptional repression was studied in the cell lines before and after demethylation treatment. In relation to cancer progression, we studied whether this epigenetic alteration could be detected in blood cells or plasma samples collected from pediatric cancer patients before, during, and after treatment. Moreover, the clinical relevance of aberrant RASSF1A methylation was investigated among the pediatric cancer patients studied.
MATERIALS AND METHODS
Profile of Pediatric Cancer Patients.
With written consent and ethics approval, we recruited a total of 35 pediatric cancer patients, who suffered from neuroblastoma (n = 8), medulloblastoma (n = 1), primitive neuroectodermal tumor (n = 1), thyroid carcinoma (n = 2), hepatocellular carcinoma (n = 3), Langerhans cell histiocytosis (n = 2), desmoplastic small round cell tumor (n = 1), pancreatoblastoma (n = 1), adrenocortical carcinoma (n = 1), Wilms’ tumor/nephroma (n = 2), ovarian dysgerminoma (n = 1), rhabdomyosarcoma (n = 3), Burkitt’s lymphoma (n = 3), T-cell lymphoma (n = 1), or acute myeloid leukemia/acute lymphoblastic leukemia/chronic myeloid leukemia (n = 5). With curative intent, these patients underwent surgical resection, chemotherapy, radiotherapy, peripheral blood stem cell or bone marrow transplantation.
Pediatric Tumors, Adjacent Nontumor Tissues, and Secondary Metastases.
A total of 39 surgically resected specimens, including 24 primary tumors, 13 matched adjacent nontumor tissues, and two secondary metastases, were collected from 24 pediatric patients with neuroblastoma (n = 8), thyroid carcinoma (n = 2), hepatocellular carcinoma (n = 2), Langerhans cell histiocytosis (n = 2), desmoplastic small round cell tumor (n = 1), pancreatoblastoma (n = 1), adrenocortical carcinoma (n = 1), Wilms’ tumor/nephroma (n = 2), rhabdomyosarcoma (n = 1), Burkitt’s lymphoma (n = 3), or T-cell lymphoma (n = 1).
Peripheral Blood Cells and Plasma Samples from Pediatric Cancer Patients.
Forty-nine buffy coat samples (n = 34) and plasma samples (n = 15) were collected before, during, and after treatment from 23 pediatric cancer patients, who suffered from neuroblastoma (n = 5), medulloblastoma (n = 1), primitive neuroectodermal tumor (n = 1), thyroid carcinoma (n = 1), hepatocellular carcinoma (n = 2), adrenocortical carcinoma (n = 1), ovarian dysgerminoma (n = 1), rhabdomyosarcoma (n = 3), Burkitt’s lymphoma (n = 2), T-cell lymphoma (n = 1), or acute myeloid leukemia/acute lymphoblastic leukemia/chronic myeloid leukemia (n = 5). As a control, 20 buffy coat samples and 20 plasma samples were collected from 20 pediatric patients with no cancer.
Cell Lines Derived from Pediatric Cancer Patients.
Neuroblastoma (SK-N-AS, SK-N-DZ, SK-N-SH, SK-N-MC, and KELLY), hepatocellular carcinoma (Hep3B), hepatoblastoma (HepG2), rhabdomyosarcoma, Burkitt’s lymphoma (JIYOYE, DAUDI, Namalwa), and papillary thyroid carcinoma (K1) cell lines were purchased from the American Type Culture Collection (Manassas, VA). All of the cell lines, except K1, were derived from pediatric cancer patients. SK-N-AS was cultured in DMEM supplemented with 4.5 grams/liter glucose and 0.1 mm nonessential amino acids. SK-N-DZ was cultured in DMEM supplemented with 4.5 grams/liter glucose and 1 mm sodium pyruvate (Life Technologies, Inc., Grand Island, NY). SK-N-SH and SK-N-MC were cultured in Eagle’s Minimal Essential Medium supplemented with 0.1 mm nonessential amino acids and 1 mm sodium pyruvate (Life Technologies, Inc.). K1 was cultured in DMEM:Ham’s F12:MCDB 105 in 2:1:1 proportions (Life Technologies, Inc., Sigma Chemical Co., St. Louis, MO). HepG2 and Hep3B were cultured in DMEM medium. RD was cultured in DMEM supplemented with 2% nonessential amino acids and 2% vitamins (Life Technologies, Inc.). The remaining cell lines were cultured in RPMI 1640 (Life Technologies, Inc.). All media were also supplemented with 2 mm glutamine, 100 units/ml penicillin, 100 μg/ml streptomycin, and 10% fetal bovine serum (Life Technologies, Inc.).
KELLY, RD, Namalwa, and DAUDI cell lines were treated with 3 μm 5-aza-2′-deoxycytidine (Sigma Chemical Co.) for 3–10 days in the corresponding growth media (29) .
Total genomic DNA was extracted from cancer cell lines, primary tumors, nontumor tissues, and secondary metastases from pediatric cancer patients using the QIAamp Tissue Kit (Qiagen, Hilden, Germany). Total genomic DNA was extracted from blood cells and plasma samples using the QIAamp Blood Kit (Qiagen).
Bisulfite Modification and Methylation-Specific PCR for Analyzing the Promoter Region.
Bisulfite modification of genomic DNA would convert unmethylated cytosine residues into uracil residues (30 , 31) . Conversely, methylated cytosine residues would remain unmodified. Thus, methylated and unmethylated DNA sequences would be distinguishable by using sequence-specific PCR primers (30 , 31) . Bisulfite modification was conducted using the CpGenome DNA Modification Kit (Intergen Co., Purchase, NY). Bisulfite-modified DNA was amplified using primers specific for the methylated sequence, 5′-GTGTTAACGCGTTGCGTATC-3′ and 5′-AACCCCGCGAACTAAAAACGA-3′. All bisulfite-modified DNA samples were also amplified using primers specific for the unmethylated sequence, 5′-TTTGGTTGGAGTGTGTTAATGTG-3′ and 5′-CAAACCCCACAAACTAAAAACAA-3′. PCR was conducted for 35 or 55 cycles using the GeneAmp DNA Amplification Kit and AmpliTaq Gold polymerase (Applied Biosystems, Perkin-Elmer, Foster City, CA). The optimized thermal profile included initial denaturation at 95°C for 12 min, followed by 35 or 55 cycles of 95°C for 45 s, 60°C for 45 s, 72°C for 1 min, and a final extension at 72°C for 10 min. Each sample was analyzed in triplicate. PCR products were analyzed by agarose gel electrophoresis and ethidium bromide staining.
After washing in PBS and centrifugation, the cell pellet was resuspended in 0.5 ml of guanidinium thiocyanate solution. Total RNA was extracted using a single-step method (32) .
Total RNA (2 μg) was denatured at 65°C for 2 min and annealed with 1 μg of random primers (Invitrogen, Carlsbad, CA) at 37°C for 10 min (33) . cDNA was synthesized at 37°C for 1 h using 200 units of Moloney murine leukemia virus reverse transcriptase (Invitrogen). β2-microglobulin cDNA was amplified as a control to ensure that a similar amount of high-integrity RNA was reverse transcribed in each reaction (33) . PCR was conducted using primers specific for RASSF1A, 5′-CAGATTGCAAGTTCACCTGCCACTA-3′ and 5′-GATGAAGCCTGTGTAAGAACCGTCCT-3′. The optimized thermal profile included initial denaturation at 95°C for 12 min, followed by 40 cycles of 95°C for 1 min, 65°C for 1 min, 72°C for 1 min, and a final extension at 72°C for 10 min. Each sample was analyzed in triplicate. PCR products were analyzed by agarose gel electrophoresis and ethidium bromide staining.
Association of RASSF1A promoter methylation in tumor/buffy coat with gender, age, tumor size, recurrence/metastasis, and overall survival was analyzed using Fisher’s exact test and Kaplan-Meier Log-rank test.
Ubiquitous RASSF1A Promoter Methylation in Pediatric Tumors, Adjacent Nontumor Tissues, and Secondary Metastases.
RASSF1A promoter methylation was found in 67% (16 of 24) of primary pediatric tumors of eight cancer types (Fig. 1A)⇓ , including neuroblastoma (seven of eight), thyroid carcinoma (two of two), hepatocellular carcinoma (two of two), pancreatoblastoma (one of one), adrenocortical carcinoma (one of one), Wilms’ tumor (one of one), Burkitt’s lymphoma (one of three), and T-cell lymphoma (one of one). In addition, aberrant RASSF1A promoter methylation was found in both lung metastases from the two hepatocellular carcinoma patients studied. Moreover, methylated RASSF1A alleles were detected in 4 of 13 nontumor tissues, which were resected from 2 hepatocellular carcinoma patients, 1 Wilms’ tumor patient, and 1 adrenocortical carcinoma patient (Fig. 1B)⇓ . However, this epigenetic change was not detected in blood cells from 20 pediatric patients with no cancer. Unmethylated RASSF1A alleles were detected in all of the nontumor tissues, tumors, and secondary metastases from pediatric cancer patients and blood cells from 20 pediatric patients with no cancer.
RASSF1A Methylation Patterns in Peripheral Blood Cells and Plasma Samples from Pediatric Cancer Patients Before, During, and After Treatment.
Among 34 buffy coat samples collected before, during, and after treatment, RASSF1A promoter methylation was detected in 12 samples from pediatric patients with neuroblastoma (n = 4), thyroid carcinoma (n = 2), hepatocellular carcinoma (n = 2), rhabdomyosarcoma (n = 1), Burkitt’s lymphoma (n = 1), T-cell lymphoma (n = 1), or acute lymphoblastic leukemia (n = 1; Fig. 2⇓ ). Before treatment, 54% (7 of 13) of buffy coat samples from pediatric cancer patients showed RASSF1A promoter methylation. During treatment, 40% (4 of 10) of the buffy coat samples analyzed showed this epigenetic alteration. Concordant with the positive results on the buffy coat from 2 of these 4 patients, RASSF1A promoter methylation was also found during treatment in plasma of a neuroblastoma patient and a hepatocellular carcinoma patient (Fig. 3)⇓ . This epigenetic change was not detectable in the remaining 13 plasma samples collected before (n = 6), during (n = 3), or after (n = 4) treatment from the pediatric cancer patients studied. Only one of six buffy coat samples collected 1 month after treatment possessed the same alteration, whereas none of five buffy coat samples collected 2 months after treatment showed RASSF1A promoter methylation. As a control, methylation-specific PCR was also performed on buffy coat and plasma samples from 20 pediatric patients with no cancer, and RASSF1A promoter methylation was not detected. However, unmethylated RASSF1A alleles were detected in all of the buffy coat and plasma samples from the 20 noncancer pediatric patients and all of the 23 pediatric cancer patients studied.
RASSF1A Promoter Methylation in Nearly All of the Cell Lines Derived from Pediatric Cancer Patients.
RASSF1A promoter methylation was found in 92% (11 of 12) of cell lines largely derived from pediatric cancer patients, including neuroblastoma (SK-N-AS, SK-N-DZ, SK-N-SH, SK-N-MC, and KELLY), hepatocellular carcinoma (Hep3B), hepatoblastoma (HepG2), rhabdomyosarcoma, Burkitt’s lymphoma (DAUDI, Namalwa), and papillary thyroid carcinoma (K1) cell lines (Fig. 4A)⇓ . Only JIYOYE cell line did not possess RASSF1A promoter methylation. SK-N-AS, SK-N-DZ, SK-N-SH, SK-N-MC, KELLY, Hep3B, HepG2, and RD cell lines showed complete RASSF1A methylation. Unmethylated RASSF1A alleles were only detected in JIYOYE, DAUDI, and K1 and were barely detectable in Namalwa (Fig. 4B)⇓ .
Reactivation of RASSF1A Transcription by Demethylation Treatment.
RASSF1A promoter methylation correlated with transcriptional silencing or repression in KELLY, RD, HepG2, and Namalwa cell lines (Fig. 5)⇓ . JIYOYE cell line containing solely unmethylated RASSF1A alleles showed RASSF1A mRNA. KELLY, RD, Namalwa, and DAUDI cell lines were treated with 5-aza-2′-deoxycytidine to examine the relationship between demethylation and transcriptional reactivation. RASSF1A transcription was examined by reverse transcription-PCR before and after treatment with 5-aza-2′-deoxycytidine. RASSF1A transcripts were not detected in KELLY and RD but were minimally detectable in Namalwa before 5-aza-2′-deoxycytidine treatment (Fig. 5)⇓ . After demethylation treatment, reactivation of RASSF1A transcription was seen in all of the three cell lines. These results confirmed that transcriptional silencing or repression was directly associated with RASSF1A promoter methylation. Conversely, DAUDI cell line possessing both methylated and unmethylated RASSF1A alleles showed RASSF1A mRNA before and after demethylation treatment.
Correlation of RASSF1A Methylation with Clinicopathologic Profile of Pediatric Cancer Patients.
Correlations were analyzed between RASSF1A methylation status and clinicopathologic features, including gender, age, tumor size, recurrence/metastasis, and overall survival (Table 1)⇓ . Of note, 12 of 16 pediatric cancer patients with RASSF1A promoter methylation in tumors were male (Fisher’s exact test, P = 0.058, n = 23). However, RASSF1A promoter methylation in tumors was not significantly associated with age (Fisher’s exact test, P = 0.685, n = 23) or tumor size (Fisher’s exact test, P = 0.361, n = 21). In addition, RASSF1A methylation status in tumor or buffy coat was not associated with metastasis development or tumor recurrence in this cohort of pediatric cancer patients (Fisher’s exact test, P = 0.667, n = 24; P = 1.000, n = 23). During a median follow-up of 19 months for 22 pediatric cancer patients with informative RASSF1A methylation status in tumor, only 2 pediatric cancer patients died of cancer or metastasis, and they both had tumoral RASSF1A methylation. On the other hand, all of the 7 pediatric cancer patients with unmethylated RASSF1A alleles were alive and well. Using Kaplan-Meier Log-rank test, RASSF1A methylation in tumors was not significantly associated with overall survival in this series of pediatric cancer patients (P = 0.302, n = 22).
RASSF1A promoter methylation and transcriptional repression have been found in many different cancer types, including lung and breast cancers (19 , 20 , 26 , 27 , 34, 35, 36, 37, 38, 39, 40) . In the present study, we demonstrated frequent and ubiquitous RASSF1A promoter methylation in childhood neuroblastoma, thyroid carcinoma, hepatocellular carcinoma, pancreatoblastoma, adrenocortical carcinoma, Wilms’ tumor, Burkitt’s lymphoma, and T-cell lymphoma. Furthermore, aberrant RASSF1A methylation was found in secondary lung metastases from hepatocellular carcinoma patients. Thus far, RASSF1A methylation is the first molecular abnormality that is common to a large and diverse group of pediatric tumors. Our findings implicate that aberrant RASSF1A promoter methylation may contribute to the development of a wide range of pediatric tumors.
RAS promotes both cell transformation and death, leading to a hypothesis that signal transduction pathways driving proliferation and death are tightly linked to protect against oncogenic transformation. Of interest, an inverse correlation has been found between oncogenic RAS mutations and RASSF1A promoter methylation. The frequency of RASSF1A methylation (20%) in colorectal cancer with common RAS mutations was lower than the frequencies observed in other tumor types (62–91%) with uncommon RAS mutations (5 , 20 , 26 , 27 , 34, 35, 36, 37 , 41, 42, 43, 44) . This is consistent with the fact that concomitant genetic and epigenetic alterations in the same signaling pathway are rarely observed in a single tumor type. Cancer cell lines harboring oncogenic RAS mutations also showed epigenetic silencing of RASSF1A, suggesting that RAS activation and RASSF1 inactivation are not necessarily mutually exclusive (45) . However, the majority of colorectal cancers with RAS mutations lacked RASSF1A promoter methylation; thus, the coexistence of RASSF1A and RAS alterations might be stochastic events (41) .
Transcriptional repression of RASSF1A was identified in 17% of paired bladder carcinomas and nontumor bladder tissues (43) . In this study, we also found aberrant RASSF1A promoter methylation in nontumor tissues adjacent to hepatocellular carcinoma, adrenocortical carcinoma, or Wilms’ tumor, suggesting that epigenetic alteration of RASSF1A is an early and critical event during childhood neoplasia. Methylation of CDH1, CDH13, and RASSF1A has been demonstrated previously in nonmalignant prostatic tissues (46) , and p16 promoter methylation was detected in preneoplastic lung tissues (47) . Furthermore, minimal RASSF1A promoter methylation has been detected in normal lung, breast, colon and kidney tissues, and normal prostate epithelial cells from cancer patients (20 , 21 , 34 , 36 , 41) . This might be explained by the presence of tumor cells in some “nontumor” tissues, or RASSF1A methylation is indeed an early and premalignant alteration.
RASSF1A promoter methylation was detected in 36–100% of cell lines derived from lung cancer, breast cancer, or ovarian cancer (26 , 27) . Consistently, 92% (11 of 12) of cell lines largely derived from pediatric cancer patients, including neuroblastoma, hepatoblastoma, hepatocellular carcinoma, rhabdomyosarcoma, and Burkitt’s lymphoma cell lines, showed RASSF1A promoter methylation in correlation with transcription silencing or repression. The absence of unmethylated RASSF1A alleles and RASSF1A mRNA in KELLY, RD, and HepG2 cell lines suggest homozygous inactivation by biallelic methylation. After demethylation treatment with 5-aza-2′-deoxycytidine, RASSF1A expression was reactivated in KELLY, RD, and Namalwa. In fact, RASSF1A blocked cell cycle progression from G1 phase to S phase and post-transcriptionally inhibited cyclin D1 accumulation in cancer cell lines (28) , implicating the biological and physiological relevance of RASSF1A in cell cycle regulation by inducing G1 arrest at the retinoblastoma restriction point and inhibiting cell growth or suppressing tumorigenicity.
Methylation and loss of heterozygosity are the major loss-of-function mechanisms for RASSF1A inactivation (19) . RASSF1A mutations appear to be rare in human cancers (18 , 20 , 26) . Consistent with Knudson’s two-hit hypothesis, 77% of small cell lung cancers with 3p21.3 allelic loss also had RASSF1A methylation (25 , 26) . Similarly, clear cell renal cell carcinomas with RASSF1A methylation also demonstrated 3p21 allelic loss, but some tumors with 3p21 allelic loss did not show RASSF1A methylation (37) . These results suggest that RASSF1A inactivation by two hits (methylation and 3p loss) is a critical step in tumorigenesis. However, concurrent loss of heterozygosity and methylation may be stochastic events during tumor initiation. According to Knudson’s two-hit model, the first “hit” is often a point mutation, small deletion, or epigenetic event, which is followed by the second chromosomal loss or loss of heterozygosity (25) . Most likely, the first hit in RASSF1A during carcinogenesis is epigenetic silencing, which may be followed by the second 3p loss.
RASSF1A methylation was rarely found in non-small cell lung, ovarian, and cervical cancer with 3p21.3 loss (26) . Frequent epigenetic inactivation of RASSF1A occurred in papillary renal cell carcinomas, despite rare 3p21.3 allelic loss (37) . In line with this reciprocal relationship, there is increasing evidence supporting that RASSF1A haploinsufficiency by either promoter methylation or 3p21 allelic loss might promote tumorigenesis without the need for a “second hit” (18) . In addition, growth suppression experiments indicate that even partial RASSF1A repression may provide selective growth advantage (36) .
For the first time, we demonstrated frequent RASSF1A promoter methylation in buffy coat from pediatric patients with neuroblastoma, thyroid carcinoma, hepatocellular carcinoma, rhabdomyosarcoma, Burkitt’s lymphoma, T-cell lymphoma, or acute lymphoblastic leukemia. RASSF1A promoter methylation was shown in 54%, 40%, and 9% of buffy coat samples collected from pediatric cancer patients before, during, and after treatment, correspondingly. Of note, most of the pediatric cancer patients studied underwent chemotherapy with or without surgical resection. Concordantly, RASSF1A promoter methylation was found during chemotherapy in plasma of 2 patients with neuroblastoma or hepatocellular carcinoma, possessing the identical epigenetic alteration in buffy coat, suggestive of cell death and good response to chemotherapy. RASSF1A methylation was only found in 1 of 11 buffy coat samples after treatment but not in plasma of the same patient, suggesting poor response to treatment. Indeed, this patient had developed metastasis. Conversely, RASSF1A methylation was not detected in the remaining three plasma samples collected after treatment or the corresponding buffy coat samples, and these patients did not have recurrence or metastasis.
Of particular interest, RASSF1A promoter methylation was detected in blood cells from neuroblastoma and Burkitt’s lymphoma patients who did not possess the identical methylation abnormality in tumors. These circulating tumor cells may represent more malignant clones in blood. These circulating tumor cells with RASSF1A promoter methylation and reduced expression might possess the antiapoptotic mechanism, in that RASSF1A can normally interact with RAS-binding NORE1 protein, proapoptotic protein kinase MST1, and activated RAS to induce apoptosis (48) . This may also explain why RASSF1A methylation was not detected in plasma of all of the pediatric cancer patients before treatment, implicating no signs of cellular apoptosis. According to these results, RASSF1A promoter methylation may prove useful as an important molecular marker for early detection of a diverse group of childhood cancers, as has been demonstrated in human cancers with other epigenetic alterations as tumor markers (49, 50, 51) . Furthermore, RASSF1A may potentially be a novel therapeutic target for pharmacological re-expression using demethylating drugs. A subset of pediatric tumors may well be characterized by a CpG island methylator phenotype. Additional studies are required to determine whether epigenetic deregulation is one of the major pathways contributing to pediatric tumorigenesis.
Clinically, it is still controversial whether aberrant RASSF1A methylation in human cancer is associated with clinicopathologic features. RASSF1A inactivation is common in grade I tumors; RASSF1A methylation may thus be an early event during breast cancer progression (20) . No statistically significant association has been found between RASSF1A methylation and gender, age, Tumor-Node-Metastasis pathological stage, or tumor histology in patients with non-small cell lung or bladder cancer (27 , 43) . RASSF1A methylation was not associated with gender or Dukes’ stage in colorectal cancer patients (41) . In our cohort of pediatric cancer patients, those with tumoral RASSF1A promoter methylation were predominantly male. Consistent with the lack of correlation between RASSF1A methylation and tumor stage or survival in neuroblastoma patients (18) , there was no relationship between RASSF1A promoter methylation in tumor or buffy coat and other clinicopathologic parameters, including age, tumor size, recurrence/metastasis, or overall survival in this series of pediatric cancer patients. The lack of association between RASSF1A methylation and clinicopathologic characteristics may support the notion that this frequent and ubiquitous epigenetic alteration may potentially be a very early and critical event deregulating apoptosis and cell cycle progression in childhood neoplasia (28) . However, further investigation of a larger cohort of pediatric cancer patients is required to confirm the clinical relevance of RASSF1A methylation.
Conversely, grade III renal cell carcinomas showed a higher rate of RASSF1A promoter methylation than grades I and II tumors (35) . Down-regulation of RASSF1A correlated with advanced tumor stage and grade in patients with bladder or gastric cancer (38 , 43) . Moreover, RASSF1A inactivation was more frequent in advanced and poorly differentiated tumors than in early stage well-differentiated or moderately differentiated tumors (38) . In prostate cancer, RASSF1A promoter methylation was also associated with high GS grade (46) . In addition, RASSF1A hypermethylation was associated with poor prognosis in patients with bladder or lung cancer (27 , 44) . Apparently, epigenetic inactivation of RASSF1A may play different roles during the progression of different childhood and adult cancer types. In particular, RASSF1A promoter methylation may possibly be associated with pediatric tumor initiation in an early stage. The tumor suppressor function of RASSF1A is implicated based on the fact that chromosome transfer of 3p fragment suppressed tumorigenicity (52) . Frequent RASSF1A inactivation in multiple pediatric cancer types may provide new opportunities to develop anticancer drugs to reactivate RASSF1A expression and, hence, downstream tumor suppressive effects of RASSF1A.
We thank Kathryn S. E. Cheah, Mai Har Sham, and Kwok Ming Yao for continual support.
Grant support: Research Grants Council Grant No. HKU7484/03M from the Hong Kong Research Grants Council and Research Grants Council Direct Allocation No. 10204245 from the University of Hong Kong.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Requests for reprints: Ivy H. N. Wong, at the Department of Biochemistry, 3/F, Laboratory Block, Faculty of Medicine Building, 21 Sassoon Road, The University of Hong Kong, Hong Kong Special Administrative Region. Phone: (852) 2819 9472; Fax: (852) 2712 2719; E-mail: firstname.lastname@example.org.
- Received March 12, 2003.
- Revision received September 19, 2003.
- Accepted October 23, 2003.