NEUROBLASTOMA IS a common solid tumor of

Transcription

1 Circulating Neuroblastoma Cells Detected by Reverse Transcriptase Polymerase Chain Reaction for Tyrosine Hydroxylase mrna Are an Independent Poor Prognostic Indicator in Stage 4 Neuroblastoma in Children Over 1 Year By Susan A. Burchill, Ian J. Lewis, Keith R. Abrams, Richard Riley, John Imeson, Andrew D.J. Pearson, Ross Pinkerton, and Peter Selby Purpose: In this prospective, multicenter study, the independent prognostic power of neuroblastoma cells detected by reverse transcriptase polymerase chain reaction (RT-PCR) for tyrosine hydroxylase (TH) mrna was evaluated. Patients and Methods: The clinical significance of disease detected by RT-PCR in peripheral blood from children at diagnosis was compared with established prognostic markers [ie, age, lactate dehydrogenase (LDH), neuron-specific enolase, ferritin, and MYCN gene amplification] by multivariate analysis. The value of disease detection by RT-PCR during treatment and follow-up was also examined. Results: TH mrna was detected in peripheral blood from 33 of 49 (67%) children with stage 4 neuroblastoma > 1 year old at diagnosis and was a significant predictive factor for overall survival [hazard ratio (HR) NEUROBLASTOMA IS a common solid tumor of childhood showing a wide range of clinical behavior, from localized tumors in children with a good prognosis to highly metastatic aggressive tumors with unfavorable outcomes. The presence of disseminated disease is a poor prognostic indicator, especially in children 1 year old. 1-3 As with many cancers, metastases in children with neuroblastoma are routinely identified by clinical evidence of spread using imaging studies and detection of tumor cells in bone marrow by histologic examination. We have previously described the amplification of tissue-specific mrna using reverse transcriptase polymerase chain reaction (RT- PCR) for the detection of minimal disease burden. 4 This approach assumes that nonhemopoietic cells are not normally present in peripheral blood and requires identification of a target RNA expressed in tumor cells but not hemopoietic cells. Because catecholamines are produced by 98% of all neuroblastomas, the first enzyme in the catecholamine synthesis pathway, tyrosine hydroxylase (TH), was used as an mrna target for the detection of neuroblastoma cells by RT-PCR. 5-8 Other targets for the detection of neuroblastoma cells by RT-PCR, including PGP 9.5, 9 NFM and SYN, 10 MAGE and GAGE, 11 have been evaluated, but TH mrna appears to be the single most useful target. 12 Using this method, it is possible to detect one to 10 neuroblastoma 2.40, 95% confidence interval (CI) 1.19 to 4.84, P.014) and event-free survival (HR 2.09, 95% CI 1.06 to 4.17, P.034) in a multivariate analysis. Detection of disease in blood from clinically disease-free children was related to increased risk of death (HR 2.54, 95% CI 1.42 to 4.55, P.0014). Conclusion: TH mrna in peripheral blood of children with neuroblastoma is a poor prognostic indictor, reflecting the propensity for dissemination via the bloodstream. When combined with a serum LDH > 1500 IU/L, this is the most powerful poor prognostic model at diagnosis for children > 1 year old with stage 4 disease. The detection of TH mrna in peripheral blood from clinically disease-free children is related to increased risk of relapse and death. J Clin Oncol 19: by American Society of Clinical Oncology. cells spiked into normal blood cells. 5,6 This level of detection is an order of magnitude more sensitive than conventional immunocytochemical techniques. 13 TH mrna has not been detected in normal peripheral blood. 5,6 From the Candlelighter s Children s Cancer Research Laboratory, Imperial Cancer Research Fund Cancer Medicine Research Unit and Department of Paediatric Oncology, St James s University Hospital, Leeds; Department of Epidemiology and Public Health, University of Leicester, Leicester; Department of Child Health, Royal Victoria Infirmary, Newcastle Upon Tyne; and Department of Paediatric Oncology, The Royal Marsden NHS Trust, Sutton, United Kingdom. Submitted May 10, 2000; accepted November 20, Presented on behalf of the United Kingdom Children s Cancer Study Group. Supported by grants from the Candlelighter s Trust and the Northern and Yorkshire Regional Health Authority. We are indebted to the clinical and nursing staff of the United Kingdom Children s Cancer Study Group centers for collecting samples for this study and the United Kingdom Children s Cancer Study Group Data Center for providing clinical information. Address reprint requests to SA Burchill, Imperial Cancer Research Fund Cancer Medicine Research Unit, St James s University Hospital, Beckett Street, Leeds LS9 7TF, United Kingdom; s.a.burchill@ leeds.ac.uk by American Society of Clinical Oncology X/01/ Journal of Clinical Oncology, Vol 19, No 6 (March 15), 2001: pp

2 1796 BURCHILL ET AL Although RT-PCR increases the sensitivity of neuroblastoma cell detection, 5,6 its clinical value is difficult to assess from current literature, with reported frequencies in peripheral blood from children with stage 4 disease at diagnosis varying from 25% 8 to 100%. 6 In this study, the clinical power of low-level disease detection by RT-PCR for TH mrna in peripheral blood from children with neuroblastoma has been evaluated through a multicenter, qualitycontrolled blind study [United Kingdom Children s Cancer Study Group (UKCCSG)study number NB9305]. The independent prognostic value of RT-PCR for TH mrna in peripheral blood compared with more conventional markers of disease status in neuroblastoma has been evaluated by a multivariate analysis using Cox proportional hazards regression model. PATIENTS AND METHODS Cell Lines and Preparation of Cell Spikes The neuroblastoma IMR-32 cell line was used as a positive control in this study. These cells were grown in Dulbecco s Modified Eagle Medium RPMI1640 media plus 10% fetal calf serum. Positive controls of one and 10 cells added to whole blood from healthy volunteers by micromanipulation were used to control for sensitivity and reproducibility of tumor-cell detection. Clinical Samples Blood samples were collected from an unselected cohort of children diagnosed with neuroblastoma between December 1993 and December 1998 attending five UKCCSG Centers (Glasgow, Leeds, Manchester, Newcastle On Tyne, Sutton). Blood samples (2 2 ml) were taken into ethylenediaminetetracetic acid (20 mmol/l) or ethylenediaminetetracetic acid (20 mmol/l) plus Ultraspec RNA (8 ml; Biogenesis, Bournemouth, UK) containing tubes and frozen at -80 C until required for RNA extraction. Samples were taken from children at diagnosis (n 79), during treatment, at the end of treatment, and when children were clinically disease free on follow-up. In the follow-up group, blood samples from an additional 33 children were analyzed. The total number of children studied on follow-up was therefore 112. Blood samples were also taken from some children with stage 4 disease at the time of clinical relapse (n 24). Normal blood samples from 30 age-matched children attending St James s University Hospital for routine surgical operations were collected as negative controls. At the end of treatment, children were defined as clinically disease free based on the absence of an undetectable mass following metaiodobenzylguanidine scan or x-ray and ultrasound and normal urinary catecholamines. Institutional ethical approval was obtained, and parental consent was given for all children from whom blood was taken. Neuroblastoma staging was defined according to International Neuroblastoma Staging System 14 and urinary catecholamine levels were measured as previously described. 15 Additional biologic investigations were in line with UKCCSG recommendations [ie, lactate dehydrogenase (LDH), neuron specific enolase, ferritin, and MYCN gene amplification]. All children received similar treatment regimens determined by age and stage of disease at diagnosis. All clinical and biologic data were held at the UKCCSG Data Center, University of Leicester, UK. Treatment Children were treated according to national and international protocols in place during the study period. Children with stage 1 and 2a disease were treated according to low-stage neuroblastoma study group guidelines, which advised surgery where feasible but with strict indications for chemotherapy and radiotherapy depending on tumor site or symptoms. Children with stage 2b disease were either treated with surgery alone or with chemotherapy (OJEC, which included vincristine, carboplatin, etoposide, cyclophosphamide). Stage 3 disease was treated according to the European Neuroblastoma Study Group guidelines (ENSG 9) with chemotherapy (OJEC), surgery and radiotherapy when required. Children 1 year old with stage 4 disease were treated on ENSG 5 protocols with vincristine, cyclophosphamide, etoposide, carboplatin, and cisplatin, delivered as either standard therapy (OPEC/ OJEC) or as dose intensified-therapy (rapid COJEC). Primary tumor surgery and high-dose chemotherapy with melphalan and hemopoietic stem-cell rescue were recommended for all children with stage 4 disease. Infants with stage 4 or 4s disease were treated according to ENSG 8 (infant protocol). Following relapse, treatment was tailored on an individual basis, with children receiving palliative care alone, high-intensity chemotherapy regimens or, in some cases, new phase 1 or 2 agents. RNA Extraction Total cellular RNA was extracted from IMR-32 cells, the whole blood from healthy volunteers or children with neuroblastoma using Ultraspec RNA, as previously described. 5 Blood samples frozen without Ultraspec RNA were taken from the freezer directly into Ultraspec RNA and allowed to thaw in this solution to reduce degradation of RNA by RNAses. The amount of recovered RNA, and its purity, was measured by reading the optical density at 260nm and 280nm. The quality of isolated RNA for RT-PCR was confirmed by amplification of the housekeeping gene, 2 microglobulin. The total amount of RNA used in each RT-PCR was 5 g. Contaminating DNA was removed either by treating with DNAse 5 or by isolating poly A RNA Dynal beads; 16 both methods were equally sensitive when analyzing 5 g of total RNA. Reverse Transcriptase-Polymerase Chain Reaction RT-PCR for TH was performed as previously described. 5 For each sample, an RT-negative control (RT enzyme absent) was included. Water controls in which all components of the RT-PCR reaction were present, except RNA, were also included. RNA from the IMR-32 cell line and IMR-32 cell spikes into whole blood (see above) were included as positive controls. The identity of RT-PCR products was confirmed by either Southern blotting or liquid hybridization with an 32 P end-labeled oligonucleotide probe specific for TH. 5 For Southern blotting and liquid hybridization, RT-PCR products of an unrelated gene were included in the analysis to confirm specificity of hybridization. Statistical Methods Differences in all-cause mortality of children with stage 4 disease 1 year old between potential prognostic groups were assessed univariately using Kaplan-Meier survival curves and the Log-Rank test. The prognostic significances of the various factors were assessed simultaneously via a Cox proportional hazards regression model using a forward selection procedure with variables entering the model if changes in minus twice the log-likelihood were statistically significant

3 NEUROBLASTOMA SIGNIFICANCE DETECTED BY RT-PCR Table 1. Stage Association of Stage With TH mrna in Peripheral Blood Taken at Diagnosis From Children With Neuroblastoma Total at the 5% significance level. In addition, the relationship between event-free survival, TH mrna in peripheral blood, and the various biologic factors was examined. Statistical analyses were performed using SAS (SAS Version 6.12: copyright 1989 to 1996 by SAS Institute Inc., Cary, NC, USA) and S-Plus software (S-plus Version 3.4: Copyright 1988 to 1996 by Mathsoft Inc, Cambridge, MA, USA) packages. RESULTS TH mrna Detected No. of Children TH mrna not Detected year old year old s RT-PCR Detection of Neuroblastoma Cells Following RT-PCR for TH mrna in IMR-32 cells, the expected single band of 180 bp was identified in as little as 10 pg of total RNA (results not shown). In cell-spiking controls, it was possible to detect a single IMR-32 cell in 2 ml of whole human blood (Fig 1). The amplified products were shown to be TH mrna by Southern blot and direct sequence analysis (results not shown). Consistent with previous results, TH mrna was not detected in blood samples from normal healthy controls (results not shown). Analysis of Blood From Children With Neuroblastoma At Diagnosis. Blood samples were analyzed from 79 children at diagnosis. The median age of children at the time of diagnosis was 2.49 years, with an interquartile range of 1.37 years to 4.12 years (a minimum age of 0.11 years and a maximum age of years). The median follow-up of all 79 children was 1.73 years (20.76 months), with an interquartile range of 0.92 years to 3.28 years (a minimum of 0.24 years and a maximum of 7.40 years). Of the 79 children, 56 were stage 4, 11 were stage 3, six were stage 2, three were stage 1, and three were stage 4s. All children had elevated urinary catecholamines at the time of diagnosis. All children with stage 1, 2, 3, or 4s disease were negative for TH mrna, with the exception of one child with stage 3 disease who on re-examination was found to have low-level (less than 5%) bone marrow involvement and one child with stage 4s disease (Table 1). More than 70% of children presented at diagnosis with stage 4 disease (56 of 79). Of 1797 these children, 73% (41 of 56) had disease of the bone marrow, 77% (43 of 56) disease of the bone, and 88% (49 of 56) were 1 year old. There was a positive correlation between the detection of TH mrna in peripheral blood by RT-PCR and disease of the bone marrow (P.0001) and the bone (P.001) detected by conventional imaging. In the children with stage 4 disease 1 year old, TH mrna was detected in peripheral blood from 67% (33 of 49)(Fig 1). The presence of TH mrna in peripheral blood at diagnosis in children 1 year old with stage 4 disease carried more than a two-fold increased risk of death (Fig 2a; Table 2). In addition, an elevated serum LDH carried an 11-fold increase in risk, and children between 2 and 3 years old had a four-fold increase in risk compared with children between 1 and 2 years old (Table 2). Using a Cox proportional hazards regression model, the best model for prediction of prognosis includes both TH and LDH status at diagnosis (Table 3); the presence of TH mrna and LDH 1,500 IU/L at diagnosis indicates a statistically significant worse prognosis (in terms of mortality) compared with that of children who are TH mrna negative and have an LDH 1,500 IU/L. The MYCN status of tumors and serum neuron specific enolase or ferritin level at diagnosis were not statistically significant predictors of prognosis in stage 4 children 1 year old (Table 2). TH mrna was detected in six of seven children with stage 4 disease 1 year old; two of these children are alive and well, but four have died. The presence of TH mrna in peripheral blood in the children with stage 4 disease 1 year old also carried a two-fold decreased risk of event-free survival (Fig 2b; Tables 4 and 5). Again, LDH status and age both had an important effect on event-free survival (Tables 4 and 5). During and at the End of Treatment. Analysis of TH mrna in peripheral blood from children on treatment (n 61) was not predictive of outcome (Cox regression analysis, HR 1.30, 95% CI 0.70 to 2.42, P.40). The presence of TH mrna in peripheral blood once or more than once was taken as a positive result; if TH mrna was never detected in peripheral blood taken from children on treatment, the result was scored as negative. Of the 21 children in whom TH mrna was detected in peripheral blood, 76% (16 of 21) relapsed and died of disease compared with 67% (27 of 40) of the children in whom TH mrna was not detected. In all children with stage 4 disease except five, TH mrna was not detected at the end of therapy. The five children in whom TH mrna was detected at diagnosis, during therapy, and at the end of therapy all died within 7 1 months of diagnosis, consistent with rapidly advancing disease. However, peripheral blood from one child with stage 4s disease was also positive for TH mrna at diagnosis and remained positive during treatment, at this time catecholamine levels

4 1798 BURCHILL ET AL Fig 1. Detection of TH mrna in sequential blood samples taken throughout the course of disease in two children > 1 year old with stage 4 neuroblastoma.rt ve reverse transcriptase enzyme present, RT -ve reverse transcriptase enzyme absent. TH tyrosine hydroxylase mrna, 2m 2 microglobulin mrna (house-keeping gene). w water negative control. c 1 1 IMR-32 cell and c 2 10 IMR-32 cells spiked into 2 ml of whole blood (positive controls). M molecular weight markers. (A) Positive at diagnosis (1), negative 4 weeks into therapy (2), and subsequently throughout follow-up (3-8). A blood sample taken at suspected but unproven clinical relapse (4) was negative. This child is currently alive and well with no evidence of disease (current follow-up time of 6 years 6 months from initial diagnosis). (B) Positive at diagnosis (1) and 4 weeks into therapy (2), negative off treatment (3 5 and 4 7 weeks), positive at 13 weeks (5) and at 12 months (6), negative at 13 months (7) and positive preclinical relapse (8 14 months), at the time of suspected relapse (9 17 months) and at the time of clinical relapse that did not respond to further treatment (10 19 months). This child died of disease 22 months after initial diagnosis. Fig 2. The presence of TH mrna in peripheral blood from children with stage 4 neuroblastoma > 1 year old associated with (A) increased risk of death (overall survival) and (B) reduced probability of event-free survival. remained elevated. This child received no therapy and 6 years later is well with normal urinary catecholamine levels. Clinically Disease-Free Children, Off Treatment. The detection of TH mrna in peripheral blood from clinically disease-free children off treatment (n 112) (Fig 1b) appears to detect clinically significant disease and identify those who are destined to relapse (Cox regression analysis, HR 2.54, 95% CI 1.42 to 4.55 P.0014). Off-treatment blood samples were taken from clinically disease-free children on follow-up, postsurgey, postautograft, and post peripheral blood stem-cell harvesting. In children where multiple off-treatment samples were analyzed, the relationship between increased risk of death and the first positive TH mrna result was analyzed. Fifty-four percent (61 of 112) of children on follow-up had peripheral blood samples positive for TH mrna. Seventy-nine percent (48 of 61) of these children with TH mrna in peripheral blood subsequently relapsed and died of disease compared with 31% (16 of 51) in whom TH mrna was not detected. TH mrna was detected in peripheral blood 4 1 months (range 1 to 11 months) before clinical relapse and up to 12 months before death. TH mrna was also detected in 83% (20 of 24) of peripheral blood samples taken from children with stage 4 disease at the time of final clinical relapse with disease that did not subsequently respond to therapy. DISCUSSION This multicenter, prospective, long-term clinical outcome study is the first of its kind to demonstrate prognostic

5 NEUROBLASTOMA SIGNIFICANCE DETECTED BY RT-PCR Table 2. Prognostic Significance of TH mrna Detected in Peripheral Blood and Other Common Variables at Diagnosis for Overall Survival Analyzed by Univariate Analysis in Children Over the Age of 1 Year With Stage 4 Disease No. of Patients Variable Alive Dead Total Hazard Ratio 95% CI P TH Negative Positive NSE status 100 ng/ml ng/ml Missing LDH status 1500 IU/L IU/L Missing Ferritin status 150 ng/ml ng/ml Missing Age 1-2 years years years years MYCN Negative Positive Missing Overall NOTE. TH mrna status in peripheral blood, LDH, and age all have an important effect on survival. significance of disease detected in peripheral blood by RT-PCR for a tissue-specific target within tumor stage. We have also shown a correlation between disease stage and the presence of tumor cells in peripheral blood, as previously described in other cancer types. 17,18 For the first time, this study demonstrates the detection of clinically relevant disease by RT-PCR for a tissue-specific target in apparently disease-free children on follow-up before relapse. We have shown that RT-PCR of peripheral blood from children with neuroblastoma for TH mrna identifies those with an unfavorable prognosis by association with the disease stage at diagnosis and within the advanced stage group. Furthermore, the presence of TH mrna detected by RT-PCR in peripheral blood may improve the accuracy of staging for some children, as demonstrated in one child with stage 3 disease who was shown by RT-PCR to have disease of the bone marrow. However, further studies are required in children with stage 1, 2, or 3 disease. In univariate and multivariate analysis, the presence of TH mrna in peripheral blood from children with stage 4 disease 1 year old at diagnosis was a 1799 Table 3. Multivariate Analysis of Measured Variables Using Cox Proportional Hazards Regression Model in Children Over the Age of 1 Year With Stage 4 Disease at Diagnosis for Overall Survival Variable 2LL Change P Null model TH status, / NSE status, 100/ 100 ng/ml LDH status, 1500/ 1500 IU/L Ferritin, 150/ 150 ng/ml Age, 2-3/3-5/5 /1-2 years MYCN, / LDH TH LDH NSE LDH ferritin LDH age LDH MYCN LDH TH NSE LDH TH ferritin LDH TH age LDH TH MYCN Variable Hazard Ratio 95% CI P Summary of LDH TH model TH status, / LDH status, 1500/ 1500 IU/L NOTE. The best model for prognostic significance at diagnosis is one that includes both LDH and TH status. significant predictor of event-free survival and overall survival. In a multivariate analysis, the presence of TH mrna in peripheral blood and a serum LDH 1500 IU/L was the most powerful poor prognosis model in terms of all cause mortality. However, in this study, the presence of TH mrna does not predict for a difference in overall survival at 5 years. The number of patients surviving at 5 years is very small (reducing the power of statistical analysis at this time), reflecting the aggressive nature of stage 4 neuroblastoma. Despite this, overall and event-free survival remains significantly better in children without TH mrna in peripheral blood. This information may allow selection of children for improved treatment strategies. The detection of tumor cells in peripheral blood from clinically disease-free children is a powerful tool to predict those most likely to relapse. Detection of TH mrna in peripheral blood was made up to 11 months before catecholamine levels increased and clinical relapse occurred. Whether treatment of low-level disease detected preclinical relapse will offer a survival advantage is not known. For most children, this will only have clinical impact when new treatment strategies have been defined, although the method may be useful to evaluate the efficacy of such new treatments. The presence of TH mrna in peripheral blood of children 1 year old at the end of therapy was rare but associated with rapidly progressing disease and a high

6 1800 BURCHILL ET AL Table 4. Prognostic Significance of TH mrna Detected in Peripheral Blood and Other Common Variables at Diagnosis Analyzed by Univariate Analysis in Children Over the Age of 1 Year With Stage 4 Disease Variable No. of Children Alive Event/Death Total Hazard Ratio 95% CI P TH Positive Negative NSE status 100 ng/ml ng/ml Missing LDH status 1500 IU/L IU/L Missing Ferritin status 150 ng/ml ng/ml Missing Age 1-2 years years years years MYCN Negative Positive Missing Overall NOTE. TH mrna status in peripheral blood, LDH status, and age all have an important effect on survival. Table 5. Multivariate Analysis of Measured Variables Using Cox Proportional Hazards Regression Model in Children Over the Age of 1 Year With Stage 4 Disease at Diagnosis for Event-Free Survival Variable 2LL Change P Null model TH status, / NSE status, 100/ 100 ng/ml LDH status, 1500/ 1500 IU/L Ferritin, 150/ 150 ng/ml Age, 2-3/3-5/5 /1-2 years MYCN, / LDH TH LDH NSE LDH ferritin LDH age LDH MYCN LDH TH NSE LDH TH ferritin LDH TH age LDH TH MYCN Variable Hazard Ratio 95% CI P Summary of LDH TH model TH status, / LDH status, 1500/ 1500 IU/L NOTE. The best model for prognostic significance is one that includes both LDH and TH status. relapse rate. In most children, TH mrna was not detected 8 to 10 weeks after the start of therapy, which compares well with the clearance of tumor cells from peripheral blood reported using immunocytology. 2 The detection of TH mrna in peripheral blood from children on treatment was not of prognostic significance. Many studies have shown that RT-PCR for tissue-specific gene expression in blood or bone marrow identifies clinically relevant disease in a number of different cancers by association with stage. 17,18 However, multicenter prospective clinical outcome studies have previously only been described when RT-PCR has been used to detect minimal residual disease (MRD) in hematological malignancies using DNA targets for the detection of disease. The most powerful of these is a prospective study of 240 children with acute lymphoblastic leukemia, where the presence of MRD detected by RT-PCR was associated with a five- to 10-fold higher relapse rate than that seen in children in whom MRD was not detected. 19 Because therapy for children with neuroblastoma is increasingly tailored for an individual based on risk factors assessed at the time of diagnosis, it is important to determine which risk factors are most informative by multivariate analysis. In this study, the detection of circulating tumor cells in children with advanced stage disease at diagnosis was one of three significant poor prognostic factors, in addition to age and elevated serum LDH levels. 23 Recent studies have also shown the presence of metastasis determined by imaging and histologic examination in neuroblastoma differs with age and correlates with event-free survival, 24 consistent with the findings of our RT-PCR study. However we found no significant correlation between MYCN amplification and other prognostic biologic tumor features with the presence of TH mrna detected by RT-PCR. This is in contrast to the correlation between the presence of metastatic disease detected using more conventional screening methods. 24 However it is consistent with the hypothesis that RT-PCR detection of disease in peripheral blood is of additional prognostic significance to current methods for assessment of disease status and predicting outcome, reflecting distinct biologic features of the disease. Failure to demonstrate a correlation between MYCN gene amplification and outcome most likely reflects the number of patients studied; a large, multicenter study has recently shown MYCN gene amplification is a poor prognostic indicator in children with neuroblastoma. 26 The significance of TH mrna detected in peripheral blood, even allowing

7 NEUROBLASTOMA SIGNIFICANCE DETECTED BY RT-PCR for the relatively small number of children studied here, implies this is a particularly strong prognostic factor. It will be important in future studies to examine the independent 1801 prognostic significance of TH mrna in peripheral blood at diagnosis with additional biologic markers such as deletion of 1p 29 or changes in 17q Rogers DW, Treleaven JG, Kemshead JT, et al: Monoclonal antibodies for detecting bone marrow invasion by neuroblastoma. J Clin Pathol 42: , Moss TJ, Sanders DG: Detection of neuroblastoma cells in blood. J Clin Oncol 8: , Moss TJ, Reynolds CP, Sather HN: Prognostic value of immunocytological detection of bone marrow metastases in neuroblastoma. N Engl J Med 324: , Smith B, Selby P, Southgate J, et al: Detection of melanoma cells in peripheral blood by means of reverse transcriptase and polymerase chain reaction. Lancet 338: , Burchill SA, Bradbury FM, Selby P, et al: Early clinical evaluation of neuroblastoma cell detection by reverse transcriptase-polymerase chain reaction (RT-PCR) for tyrosine hydroxylase mrna. Eur J Cancer 31: , Miyajima Y, Kato K, Numata S, et al: Detection of neuroblastoma cells in bone marrow and peripheral blood at diagnosis by the reverse transcriptase-polymerase chain reaction for tyrosine hydroxylase mrna. Cancer 75: , Miyajima Y, Horibe K, Fukuda M, et al: Sequential detection of tumour cells in the peripheral blood and bone marrow of patients with stage IV neuroblastoma by the reverse transcription-polymerase chain reaction for TH mrna. Cancer 77: , Kuroda T, Saeki M, Nakano M, et al: Clinical application of minimal residual neuroblastoma cell detection by reverse transcriptasepolymerase chain reaction. J Pediatr Surg 32:69-72, Mattano LA Jr., Moss TJ, Emerson SG: Sensitive detection of rare circulating neuroblastoma cells by the reverse transcriptasepolymerase chain reaction. Cancer Res 52: , Lai PS, Chee S, Chiu LL, et al: Detection of low numbers of neuroblastoma cells in vitro. Ann Acad Med Singapore 26: , Cheung IY, Barber D, Cheung NK: Detection of microscopic neuroblastoma in marrow by histology, immunocytology, and reverse transcription-pcr of multiple molecular markers Clin Cancer Res 4: , Gilbert J, Norris MD, Marshall GM, et al: Low specificity of PGP9.5 expression for detection of micrometastatic neuroblastoma. Br J Cancer 75: , Johnson P, Burchill SA, Selby P: The molecular detection of tumour cells. Br J Cancer 72: , Brodeur GM, Seeger RC, Barrett A, et al: International criteria for diagnosis, staging, and response to treatment in patients with neuroblastoma. J Clin Oncol 6: , Henderson MJ, Heney D, McGinlay JM, et al: Measurement of dopamine, HVA and HMMA in untimed urine samples: Establishment of age-related reference data in children. Ann Clin Biochem 29: , 1992 REFERENCES 16. Burchill SA, Lewis IJ, Selby P: Improved methods using the reverse transcriptase polymerase chain reaction to detect tumour cells. Br J Cancer 79: , Burchill SA, Selby P: Early detection of circulating solid tumour cells using molecular technology. Cancer Topics 10:1-4, Burchill SA, Selby P: Molecular detection of small volume disease. J Pathol 190:6-14, van Dongen JJ, Seriu T, Panzer-Grumayer ER, et al: Prognostic value of minimal residual disease in acute lymphoblastic leukaemia in childhood. Lancet 352: , Breslow N, McCann B: Statistical estimation of prognosis for children with neuroblastoma. Cancer Res 31: , Evans AE, D Angio GJ, Propert K, et al: Prognostic factors in neuroblastoma. Cancer 59: , Joshi VV, Cantor AB, Altshuler G, et al: Age-linked prognostic categorization based on a new histologic grading system of neuroblastomas: A clinicopathologic study of 211 cases from the Pediatric Oncology Group. Cancer 69: , Shuster JJ, McWilliams NB, Castleberry R, et al: Serum lactate dehydrogenase in childhood neuroblastoma: A Pediatric Oncology Group recursive partitioning study. Am J Clin Oncol 15: , DuBois SG, Kalika Y, Lukens JN, et al: Metastatic sites in stage IV and IVS neuroblastoma correlate with age, tumor biology, and survival. J Pediatr Hematol Oncol 21: , Brodeur GM, Seeger RC, Schwab M, et al: Amplification of MYCN in untreated human neuroblastomas correlates with advanced disease stage. Science 224: , Matthay KK, Villablanca JG, Seeger RC, et al: Treatment of high-risk neuroblastoma with intensive chemotherapy, radiotherapy, autologous bone marrow transplantation, and 13-cis-retinoic acid. Children s Cancer Group. N Engl J Med 341: , Kawa K, Ohnuma N, Kaneko M, et al: Long-term survivors of advanced neuroblastoma with MYCN amplification: A report of 19 patients surviving disease-free for more than 66 months. J Clin Oncol 17: , Katzenstein HM, Bowman LC, Brodeur GM, et al: Prognostic significance of age: MYCN oncogene amplification, tumour cell ploidy and histology in 110 infants with stage D(s)neuroblastoma. The Pediatric Oncology Group Experience a Pediatric Oncology Group Study. J Clin Oncol 16: , Caron H: Allelic loss of chromosome 1 and additional chromosome 17 material are both unfavourable prognostic markers in neuroblastoma. Med Pediatr Oncol 24: , Bown N, Cotterill S, Lastowska M, et al: Gain of chromosome arm 17q and adverse outcome in patients with neuroblastoma. N Engl J Med 340: , 1999.