Purpose: We determined whether quantifying neuroblastoma-associated mRNAs (NB-mRNAs) in bone marrow and blood improves assessment of disease and prediction of disease progression in patients with relapsed/refractory neuroblastoma.

Experimental Design: mRNA for CHGA, DCX, DDC, PHOX2B, and TH was quantified in bone marrow and blood from 101 patients concurrently with clinical disease evaluations. Correlation between NB-mRNA (delta cycle threshold, ΔCt, for the geometric mean of genes from the TaqMan Low Density Array NB5 assay) and morphologically defined tumor cell percentage in bone marrow, 123I-meta-iodobenzylguanidine (MIBG) Curie score, and CT/MRI-defined tumor longest diameter was determined. Time-dependent covariate Cox regression was used to analyze the relationship between ΔCt and progression-free survival (PFS).

Results: NB-mRNA was detectable in 83% of bone marrow (185/223) and 63% (89/142) of blood specimens, and their ΔCt values were correlated (Spearman r = 0.67, P < 0.0001), although bone marrow Ct was 7.9 ± 0.5 Ct stronger than blood Ct. When bone marrow morphology, MIBG, or CT/MRI were positive, NB-mRNA was detected in 99% (99/100), 88% (100/113), and 81% (82/101) of bone marrow samples. When all three were negative, NB-mRNA was detected in 55% (11/20) of bone marrow samples. Bone marrow NB-mRNA correlated with bone marrow morphology or MIBG positivity (P < 0.0001 and P = 0.007). Bone marrow and blood ΔCt values correlated with PFS (P < 0.001; P = 0.001) even when bone marrow was morphologically negative (P = 0.001; P = 0.014). Multivariate analysis showed that bone marrow and blood ΔCt values were associated with PFS independently of clinical disease and MYCN gene status (P < 0.001; P = 0.055).

Conclusions: This five-gene NB5 assay for NB-mRNA improves definition of disease status and correlates independently with PFS in relapsed/refractory neuroblastoma. Clin Cancer Res; 23(18); 5374–83. ©2017 AACR.

Translational Relevance

Quantitation of neuroblastoma-associated mRNA in blood, bone marrow, and peripheral blood stem cells with reverse- transcriptase polymerase chain reaction is sensitive and can provide prognostic information for patients with high-risk neuroblastoma prior to relapse. We provide a systematic comparison of disease burden quantification in relapsed patients by standard evaluations versus a new TaqMan Low Density Array assay for mRNA of five neuroblastoma genes (NB5 assay) and show that the NB5 assay enhances standard evaluations and provides independent prognostic information in this population. This biomarker assay has been incorporated into all NANT consortium therapeutic studies in order to validate the assay and prospectively analyze disease burden assessment and to provide standard clinical disease response testing in the context of uniform therapy in large numbers of patients.

Neuroblastoma is the most common extracranial solid tumor in children, and 45% of patients have high-risk, metastatic tumors (stage 4) when diagnosed (1). Although long-term survival has improved over the past 25 years with multimodal therapy, 50% of patients relapse or have refractory disease (2–5). Patients with relapsed neuroblastoma have a poor prognosis with a 5-year overall survival after relapse of 20% (6, 7). The International Neuroblastoma Risk Group (INRG) reported that time to first relapse, age, stage of disease, and MYCN gene copy number were independently predictive of postrelapse survival (6). For stage 4 disease, only 8% of those with MYCN nonamplified tumors diagnosed after 18 months of age and 4% of those with MYCN amplified tumors survived 5 years after relapse (6).

New therapeutic strategies are needed to improve the outcome of patients with refractory or relapsed neuroblastoma. These likely will be based upon investigations that include next-generation sequencing of tumor cells to identify actionable targets (8), immunological assessments to identify targets for immunotherapy (4), defining targets in the tumor microenvironment that promote growth and resistance to therapy (9–11), and accurately defining disease burden to provide response and prognostic data (12, 13). Assessment of disease burden by quantifying neuroblastoma-associated mRNA in blood and bone marrow with reverse-transcriptase polymerase chain reaction (RT-PCR) assays may contribute to response assessment and prognostication. Studies have been performed at diagnosis or during initial therapy prior to disease progression using RT-PCR and have shown that this sensitive method can provide prognostic information for patients with neuroblastoma (5, 14–21).

Defining disease burden for assessing response and prognosis of metastatic neuroblastoma is challenging due to multiple sites of disease, which include bone marrow, bone, and soft tissue, and to variables impacting quantification of disease with current clinical evaluations (22). For bone marrow, sampling and test sensitivity are important variables. For example, the number of sites sampled can affect results and morphologic examination may not clearly define disease response versus progression. 123I-meta-iodobenzylguanidine (MIBG) radionuclide scans for bone and soft tissue disease can be subject to interobserver variation in scoring (23) and are not readily applicable to quantifying changes in known sites of disease or to defining new sites of very minimal disease. Assessment of soft tissue disease response by CT or MRI imaging using RECIST criteria may be affected by intratumor variables such as tumor cell differentiation and components of the tumor microenvironment.

This study determined whether quantitative definition of disease in bone marrow and blood with a TaqMan Low Density Array (TLDA) assay that detected mRNA of the neuroblastoma-associated genes [chromogranin A (CHGA), doublecortin (DCX), dopadecarboxylase (DDC), paired-like homeobox 2b (PHOX2B), and tyrosine hydroxylase (TH; NB5 assay)] correlated with results from standard clinical evaluations and provided independent information for prognostication. CHGA, DCX, DDC, PHOX2B, and TH genes are rarely and only weakly expressed by bone marrow or blood mononuclear cells from normal adults but are strongly expressed by neuroblastomas in vivo and by cell lines that are MYCN amplified or nonamplified and by multidrug sensitive or resistant neuroblastoma cell lines. The combined signature of all five genes is more sensitive than two genes (TH and PHOX2B) in detecting neuroblastoma mRNA (5) and is able to detect one neuroblastoma cell among 106 normal cells.

Relapsed/refractory neuroblastoma patients were studied because they generally have detectable disease and because evaluations of changes in disease burden could be enhanced by a sensitive and quantitative assay. This is the first prospective study to quantify expression of five neuroblastoma-associated genes in bone marrow and blood of patients with relapsed/refractory neuroblastoma and to correlate expression with concurrent CT/MRI and MIBG imaging and morphologic bone marrow evaluations. The NB5 assay improves definition of disease burden and provides prognostic information that is independent from that derived from clinical disease and MYCN gene assessments.

Patients

The study was designed and conducted in accordance with the U.S. Common Rule, the Declaration of Helsinki, and local regulations including the US Code of Federal Regulations Title 21. The study was performed after approval of the protocol and any informed consent documents by an institutional review board at each site. Investigators obtained written consent from all subjects prior to study enrollment.

Patients were treated at a New Approaches to Neuroblastoma Therapy (NANT) Consortium institution and were enrolled in the NANT Biology Study (NANT 2004-05). All patients had high-risk neuroblastoma as defined by the (INRG; ref. 1) and for the purposes of this study, only patients with relapsed/progressive disease (at any time point prior to enrollment) or no response (refractory disease) per International neuroblastoma response criteria (INRC; ref. 24) were included. Therapy at the time of clinical and NB5 assay assessments varied, and included NANT therapeutic protocols and other therapies for relapsed/refractory disease (Supplementary Table S1). Once enrolled in NANT 2004-05, bone marrow and blood samples were obtained prospectively for the NB5 assay. All patients submitted at least one bone marrow and/or blood sample between October 11, 2011, and April 24, 2014, for quantification of neuroblastoma mRNA with the NB5 assay at the time of standard disease evaluation that included bone marrow morphology (bilateral aspirates and biopsies) and MIBG and CT/MRI imaging. All imaging, bone marrow aspirates/biopsies, and samples for NB5 assay were obtained within one month of each other (median, 3 days) with exception of those for two patients that spanned 38 and 53 days.

Sample processing and NB5 assay

NB5 assays were performed on mononuclear cells from heparinized blood and bone marrow (pooled bilateral aspirates) isolated by density separation with Ficoll–Hypaque (25). The bilateral aspirates were pooled in order to have a sample more reflective of overall bone marrow disease burden detectable by this assay. The NB5 assay quantified expression of neuroblastoma-associated genes CHGA, DCX, DDC, PHOX2B, and TH and of housekeeping genes B2M, GAPDH, HPRT1, and SDHA with predesigned and preoptimized probe and primer sets (Supplementary Table S2), 2,500 ng of cDNA, and standard cycling conditions using the 7900HT fast real-time PCR system (Applied Biosystems). The cycle threshold (Ct) value for each gene was the cycle number where the amplification signal reached a threshold of 0.5 over baseline, and Ct 40 was assigned when this threshold was not reached by the 40th cycle. A summary ΔCt for the five detection genes was computed as the geometric mean (GM) of the Ct values for the five detection genes minus the GM of the Ct values of the four housekeeping genes (NB5 assay ΔCt). Lower ΔCt values indicate higher NB-mRNA. NB-mRNA was “undetectable” when none of the five neuroblastoma genes had a Ct < 40. Experiments in which neuroblastoma cell line RNA was seeded into peripheral blood mononuclear cell RNA showed that the NB5 assay can detect one neuroblastoma cell among 106 PBMCs (5). ΔCt was chosen instead of Ct alone as it accounts for RNA quality of the samples obtained. Note that ΔCt values may be equivalent when one specimen may have no detectable neuroblastoma mRNA and other may be mildly positive. This is due to variability of the geometric mean of housekeeping genes. Undetectable versus detectable samples are distinguished in all analyses. Details are provided in the Supplemental Patients and Methods section.

Disease evaluation

For all patients included in this analysis, disease evaluations and central review were performed as follows (including patients treated on NANT therapeutic trials or non-NANT therapies). CT/MRI images were reviewed by one radiologist (F. Goodarzian) using RECIST 1.0 criteria for presence and size of target and nontarget lesions (26), and the sum of the longest diameter (LD) of all target lesions was determined. MIBG scans were reviewed by one radiologist (H.A. Lai), who performed Curie scoring (27, 28). Histopathology of bilateral bone marrow biopsies was reviewed by one pathologist (H. Shimada), and the maximum percentage of tumor cells from either side was assigned as the percent of neuroblastoma cells (29).

Response was graded at each disease evaluation time point for CT/MRI, MIBG, and bone marrow assessments, and these were combined into an overall response per the NANT Response Criteria (v1.0; ref. 30). The overall response assigned was CR, PR, mixed response (MR), SD, or PD. For patients with multiple disease evaluation time points, each disease progression was considered a new baseline for subsequent response evaluation. Time to progression was calculated from the first baseline and after each subsequent progression. Details are provided in the Supplementary Patients and Methods section.

Statistical analysis

Where appropriate, standard descriptive and analytic statistical methods such as t test, ANOVA, ordinary least squares regression, and contingency table analyses were used (31). Reported P values were two sided, with P < 0.05 considered significant. Statistical computations were performed with Stata 11 (Stata Statistical Software: Release 11). Individual NB5 assay or disease evaluations were the analytic units and therefore a single patient may contribute more than one time point with paired NB5 and disease assessment. As the majority of patients had only one or two of each assessment type, formally accounting for repeated measures had negligible effect on the results in most analysis or was technically not feasible in selected analyses. Therefore, for consistency we report the analyses that do not account for repeated measures. For the ANOVA analysis of differences in average ΔCt between groups and for regression of ΔCt on clinical evaluations, assays classified as undetectable represented right censored data, and they were analyzed using normal theory maximum likelihood interval regression analysis as implemented in Stata module “intreg.”

Because individual patients could have multiple disease evaluations performed during follow-up, and the results of these evaluation could change, implying that the patient is now potentially at higher or lower risk of relapse or progression, time-dependent covariate (TDC) Cox regression analysis was used to examine the influence of ΔCt and other variables of interest (e.g., CT/MRI, bone marrow morphology) on time to progression (32, 33). Individual patients could contribute multiple follow-up periods in this analysis, with time 0 reset at study entry or at occurrence of disease progression, and where values of ΔCt, CT/MRI, Curie score, maximum percent bone marrow involvement by morphology were all time varying, that is could change at each evaluation during the follow-up period thus shifting individuals to different risk groups at the times of these changes. P values were based on the likelihood ratio test. Product limit (Kaplan–Meier) curves using these intervals were constructed to visualize the magnitude of the NB5 effect represented in a univariate TDC analysis of NB5. As above, time 0 is reset at study entry or at progression, and patients whose NB5 status changes during follow-up are shifted to the curve representing that new status at the time of the change. Although any one curve may not reflect the outcome of a particular well-defined group of patients with a known NB5 assessment history, the difference between curves effectively conveys the magnitude of the difference in outcome resulting from different NB5 assay results.

Patients and disease status evaluations

One hundred one patients who submitted at least one specimen for TLDA analysis with relapsed (n = 81) or refractory (n = 20) neuroblastoma are included in this study (Table 1). A total of 305 standard disease evaluations were performed, and concurrent NB5 assay data were obtained for 259 with a median of 3 days between the disease evaluation and NB5 assay (Table 1 and Supplementary Table A3). RNA integrity numbers (RINs) of the specimens for NB5 assays were the following: mean RINs for bone marrow 9.4 (range, 6.1–10) and blood 9.3 (range, 8.1–10). Patients were allowed to submit samples at the time of each disease evaluation, generally every 2 and 3 months. The median time from the first to last assessment, which included CT/MRI and MIBG scans, bone marrow morphology, or NB5 analysis, was 24 weeks (range, 5–127 weeks). The number of evaluation time points was 1 (n = 33 patients), 2 (n = 28), 3 (n = 11), and 4+ (n = 29). Review of disease evaluations performed demonstrates the following: MIBG scans were most frequently positive (77%) whereas CT/MRI and bone marrow evaluations were positive in 53% and 45% of assessments. The percentage of morphologically identified neuroblastoma cells in bone marrow correlated with the MIBG Curie score (Supplementary Fig. S1; Spearman r = 0.4; P < 0.0001) but not with CT/MRI-defined tumor LD. All bone marrow samples with >30% neuroblastoma cells on routine morphology were associated with positive MIBG scans, whereas morphologically negative bone marrow samples were associated with a range of Curie scores.

Table 1.

Patient and tumor characteristics, clinical disease status evaluations, and NB5 TLDA assays

VariableCategory descriptionSummary/count
All patients  101 
Sex Male 66 
 Female 35 
Age at diagnosis Median (range) in months 50 (3–236)a 
INSS stage at diagnosis 
 
 
 93 
MYCN gene status Amplified 34 
 Non-amplified 60 
 Unknown/not done 
Histopathology Favorable 
 Unfavorable 70 
 Unknown/not done 25 
Response to frontline therapy Refractory, no progression 20 
 Relapsed, progression 81 
CT/MRI # Positive/# totalb (%) 147/279 (52.7%) 
 Median (range) LD in cm for positives 3.7 (1.1–36) 
MIBG # Positive/# totalb (%) 204/264 (77.3%) 
 Median (range) Curie score for positives 7 (1–27)c 
Bone marrow morphology # Positive/# totalb (%) 110/243 (45.3%) 
 Median (range) % neuroblastoma cells for positives 5 (0.5–95) 
Bone marrow NB5 TLDA # Detectable/# totalb (%) 185/223 (83%) 
 Median (range) ΔCt for detectable 15.3 (0.02–21.2) 
Blood NB5 TLDA # Detectable/# totalb (%) 89/142 (63%) 
 Median (range) ΔCt for detectable 18.2 (6.8–21.8) 
VariableCategory descriptionSummary/count
All patients  101 
Sex Male 66 
 Female 35 
Age at diagnosis Median (range) in months 50 (3–236)a 
INSS stage at diagnosis 
 
 
 93 
MYCN gene status Amplified 34 
 Non-amplified 60 
 Unknown/not done 
Histopathology Favorable 
 Unfavorable 70 
 Unknown/not done 25 
Response to frontline therapy Refractory, no progression 20 
 Relapsed, progression 81 
CT/MRI # Positive/# totalb (%) 147/279 (52.7%) 
 Median (range) LD in cm for positives 3.7 (1.1–36) 
MIBG # Positive/# totalb (%) 204/264 (77.3%) 
 Median (range) Curie score for positives 7 (1–27)c 
Bone marrow morphology # Positive/# totalb (%) 110/243 (45.3%) 
 Median (range) % neuroblastoma cells for positives 5 (0.5–95) 
Bone marrow NB5 TLDA # Detectable/# totalb (%) 185/223 (83%) 
 Median (range) ΔCt for detectable 15.3 (0.02–21.2) 
Blood NB5 TLDA # Detectable/# totalb (%) 89/142 (63%) 
 Median (range) ΔCt for detectable 18.2 (6.8–21.8) 

aN = 99 patients.

bNumber of assessments.

cn = 145 positive evaluations with Curie score; 59 assessments were positive but Curie score unknown.

NB5 assay ΔCt in bone marrow and blood specimens

Neuroblastoma mRNA for at least one of the five genes was detected in 185/223 (83%) of bone marrow and 89/142 (63%) of blood specimens using the NB5 assay (Table 1). Eighty-three of 101 (82%) bone marrow and blood specimens were obtained on the same day while the remaining 18 pairs were obtained a median of 2 days apart (range, 1–6 days). Analysis of ΔCt at these 106 time points showed high correlation between bone marrow and blood (Spearman r = 0.67, P < 0.0001), but also showed that the level of neuroblastoma mRNA is less in blood than in bone marrow (Fig. 1). In 19 strongly positive bone marrow specimens (ΔCt ≤ 8), bone marrow gave a signal that was 7.9 ± 0.46 Ct stronger than blood. Bone marrow was negative in only 8% of instances (5/62) when blood was positive, but blood was negative in 35% of instances (31/88) when bone marrow was positive (Fig. 1).

Figure 1.

Correlation of NB5 assay ΔCt between bone marrow (BM) and blood. NB5 analysis was performed on bone marrow and blood specimens that were obtained concurrently from 106 time points where both assays were performed. Spearman r = 0.67, P < 0.0001. Values are denoted “undetectable” when mRNA for none of the five neuroblastoma (NB) genes had a Ct < 40 (see Patients and Methods).

Figure 1.

Correlation of NB5 assay ΔCt between bone marrow (BM) and blood. NB5 analysis was performed on bone marrow and blood specimens that were obtained concurrently from 106 time points where both assays were performed. Spearman r = 0.67, P < 0.0001. Values are denoted “undetectable” when mRNA for none of the five neuroblastoma (NB) genes had a Ct < 40 (see Patients and Methods).

Close modal

NB5 assay ΔCt and clinical disease status

In univariate analysis, bone marrow and blood NB5 ΔCt correlated with both the percentage of neuroblastoma cells in bone marrow by morphology and with the MIBG Curie score (Fig. 2A–D; P < 0.0001). 81/177 (69%) of bone marrow with negative morphology had detectable neuroblastoma mRNA with the NB5 assay. Nondetectable neuroblastoma mRNA in bone marrow in all but one case was associated with morphologically nondetectable neuroblastoma cells. The bone marrow from this patient had large areas of pink neuropil on one side but no neuroblastoma cells. Two patients with a high Curie score of 10 and 20 and nondetectable neuroblastoma mRNA in bone marrow had large soft tissue tumors 7 years from diagnosis (12 and 35 cm LD on anatomical imaging), extensive bone disease and no morphologically detectable neuroblastoma cells in bone marrow. NB5 ΔCt in blood, but not in bone marrow, correlated with the CT/MRI defined LD of soft tissue tumors (Fig. 2E and F).

Figure 2.

Correlation of NB5 ΔCt in bone marrow (BM) and blood with concurrently performed standard disease evaluations. A and B, Correlation of bone marrow and blood NB5 ΔCt and the percent of neuroblastoma (NB) cells in bone marrow defined morphologically. C and D, Correlation of bone marrow and blood NB5 ΔCt and MIBG Curie score. E and F, Correlation of bone marrow and blood NB5 ΔCt and the CT/MRI defined tumor LD. Specimens in which NB mRNA was detectable or nondetectable with the NB5 assay are coded blue or red, respectively. Inset tables provide summary data for each comparison.

Figure 2.

Correlation of NB5 ΔCt in bone marrow (BM) and blood with concurrently performed standard disease evaluations. A and B, Correlation of bone marrow and blood NB5 ΔCt and the percent of neuroblastoma (NB) cells in bone marrow defined morphologically. C and D, Correlation of bone marrow and blood NB5 ΔCt and MIBG Curie score. E and F, Correlation of bone marrow and blood NB5 ΔCt and the CT/MRI defined tumor LD. Specimens in which NB mRNA was detectable or nondetectable with the NB5 assay are coded blue or red, respectively. Inset tables provide summary data for each comparison.

Close modal

Bone marrow and blood NB5 ΔCt values associated with bone marrow, MIBG, and CT/MRI findings on clinical evaluations as shown in Table 2. Note, only evaluations with no missing evaluations of either bone marrow, MIBG, and CT/MRI were included in this analysis. Notably, 11/20 (55%) of patients with no clinically detectable disease had neuroblastoma mRNA detectable in bone marrow and/or blood. Analysis of variance assessed whether tumor burden (nondetectable or detectable disease) contributed independently to the average NB5 ΔCt (Table 2). Analysis of variance of the association between NB5 ΔCt and tumor burden assessed by bone marrow morphology, MIBG, or CT/MRI (nondetectable vs. detectable disease) showed that the presence of disease by bone marrow morphology and MIBG imaging were independently associated with a stronger NB5 ΔCt signal in bone marrow and in blood (Table 2, bottom). In contrast, presence of soft tissue disease by CT/MRI was not associated with strengthening of the NB5 assay signal in either bone marrow or blood independently of bone marrow morphology and MIBG imaging.

Table 2.

Clinical disease status and NB5 assay ΔCt

Bone marrow NB5 assay ΔCtBlood NB5 assay ΔCt
NB5 assay and clinical disease assessmentsN positive/N total (%)Mean ΔCt for total (SE)aMean ΔCt for positive (SE)N positive/N total (%)Mean ΔCt for total (SE)aMean ΔCt for positive (SE)
All NB5 assays 185/223 (83) 15.7 (0.46) 13.9 (0.42) 89/142 (63) 19.2 (0.46) 16.4 (0.39) 
All NB5 assays with concurrent clinical disease assessments 158/190 (83) 15.6 (0.50) 13.7 (0.46) 64/104 (62) 19.2 (0.54) 16.3 (0.45) 
No disease found 11/20 (55) 20.0 (1.40) 16.0 (1.46) 4/9 (44) 22.3 (1.72) 19.6 (0.66) 
Soft tissue only 2/7 (29) 24.5 (2.79) 18.8 (1.59) 3/6 (50) 21.7 (2.02) 18.9 (0.37) 
MIBG only 28/34 (82) 17.6 (1.00) 16.2 (0.72) 12/20 (60) 20.6 (1.07) 18.4 (0.53) 
Bone marrow only 9/9 (100) 15.3 (1.91) 15.3 (1.29) No sample No sample No sample 
MIBG and bone marrow 32/33 (97) 11.2 (1.00) 10.8 (1.02) 12/19 (63) 17.5 (1.06) 14.3 (1.09) 
MIBG and soft tissue 27/38 (71) 18.7 (0.97) 16.3 (1.02) 10/19 (53) 20.3 (1.12) 17.3 (0.86) 
Soft tissue and BM 8/8 (100) 12.8 (2.02) 12.8 (1.82) 2/5 (40) 20.8 (2.24) 15.9 (4.24) 
Soft tissue, bone marrow, and MIBG 41/41 (100) 11.7 (0.89) 11.7 (1.00) 21/26 (81) 16.4 (0.88) 14.9 (0.86) 
Main effects ANOVAb BM NB5 ΔCt, 190 time pointsc  Blood NB5 ΔCt, 104 time pointsc  
Clinical disease comparison ΔCt difference (SE) P  ΔCt difference (SE) P  
Bone marrow (negative vs. positive patients) 6.8 (0.78) <0.0001  3.2 (0.96) <0.001  
MIBG (negative vs. positive patients) 2.9 (1.07) 0.007  2.4 (1.3) 0.06  
CT/MRI (negative vs. positive patients) −0.77 (0.87) 0.88  −0.59 (0.94) 0.53  
Bone marrow NB5 assay ΔCtBlood NB5 assay ΔCt
NB5 assay and clinical disease assessmentsN positive/N total (%)Mean ΔCt for total (SE)aMean ΔCt for positive (SE)N positive/N total (%)Mean ΔCt for total (SE)aMean ΔCt for positive (SE)
All NB5 assays 185/223 (83) 15.7 (0.46) 13.9 (0.42) 89/142 (63) 19.2 (0.46) 16.4 (0.39) 
All NB5 assays with concurrent clinical disease assessments 158/190 (83) 15.6 (0.50) 13.7 (0.46) 64/104 (62) 19.2 (0.54) 16.3 (0.45) 
No disease found 11/20 (55) 20.0 (1.40) 16.0 (1.46) 4/9 (44) 22.3 (1.72) 19.6 (0.66) 
Soft tissue only 2/7 (29) 24.5 (2.79) 18.8 (1.59) 3/6 (50) 21.7 (2.02) 18.9 (0.37) 
MIBG only 28/34 (82) 17.6 (1.00) 16.2 (0.72) 12/20 (60) 20.6 (1.07) 18.4 (0.53) 
Bone marrow only 9/9 (100) 15.3 (1.91) 15.3 (1.29) No sample No sample No sample 
MIBG and bone marrow 32/33 (97) 11.2 (1.00) 10.8 (1.02) 12/19 (63) 17.5 (1.06) 14.3 (1.09) 
MIBG and soft tissue 27/38 (71) 18.7 (0.97) 16.3 (1.02) 10/19 (53) 20.3 (1.12) 17.3 (0.86) 
Soft tissue and BM 8/8 (100) 12.8 (2.02) 12.8 (1.82) 2/5 (40) 20.8 (2.24) 15.9 (4.24) 
Soft tissue, bone marrow, and MIBG 41/41 (100) 11.7 (0.89) 11.7 (1.00) 21/26 (81) 16.4 (0.88) 14.9 (0.86) 
Main effects ANOVAb BM NB5 ΔCt, 190 time pointsc  Blood NB5 ΔCt, 104 time pointsc  
Clinical disease comparison ΔCt difference (SE) P  ΔCt difference (SE) P  
Bone marrow (negative vs. positive patients) 6.8 (0.78) <0.0001  3.2 (0.96) <0.001  
MIBG (negative vs. positive patients) 2.9 (1.07) 0.007  2.4 (1.3) 0.06  
CT/MRI (negative vs. positive patients) −0.77 (0.87) 0.88  −0.59 (0.94) 0.53  

aMeans and SEs of NB5 ΔCt were obtained from an analysis that accounts for right censoring due to mRNA levels below the level of detectability of the NB5 assay (see Patients and Methods). These represent the best estimate of average NB5 ΔCt in the different disease status groups regardless of their detection status. This analysis was performed for patients who had CT/MIBG/bone marrow clinical evaluations and NB5 assay testing.

bFor the main effects analysis of variance, values represent the difference in the mean NB5 ΔCt in patients negative for a clinical assessment compared to positive patients, controlling for the status of the remaining clinical assessments. Interaction terms were not significant and were excluded from this model.

cAll NB5 assays with concurrent clinical disease assessments.

Change in NB5 ΔCt was assessed between pairs of time points where both NB5 ΔCt and clinical disease assessments were performed (Supplementary Fig. S2). Change in NB5 ΔCt in these sequential bone marrow samples was correlated with change in the percentage of bone marrow neuroblastoma cells (Spearman r = −0.34, P < 0.0001), the MIBG Curie score (r = −0.48, P < 0.0001), and the CT/MRI LD (r = −0.21, P < 0.001). For bone marrow, the change in NB5 ΔCt that was associated with progressive disease (ΔCt for PD specimen − ΔCt for previous specimen) was −2.7 ± 0.83 and with nonprogressive disease was 1.9 ± 0.27 (P < 0.0001, two-sample t test). For blood, these values were −2.0 ± 0.66 and 0.52 ± 0.25, respectively (P < 0.0001).

NB5 assay ΔCt and progression-free survival

Time-dependent covariate analysis of ΔCt and progression-free survival (PFS) was performed. NB5 assay results were classified as undetectable, ΔCt > 15 and ΔCt ≤ 15, where the cut-off point of 15 is approximately the median of ΔCt detectable values. All NB5 assays were included in this analysis if they met the criterion of having one subsequent clinical disease assessment. On univariate analysis, both bone marrow and blood NB5 ΔCt were significant predictors of subsequent progression for all patients including those without morphologically detectable neuroblastoma cells in bone marrow (Figs. 3A–D; Table 3). When comparing the baseline characteristics of the undetectable versus detectable patients by NB5 assay, there were no statistically significant differences in MYCN status, age at diagnosis, or history of prior relapse.

Figure 3.

Correlation of NB5 assay ΔCt in bone marrow (BM) and blood with PFS. The probability of progression for all patients (A and B), for patients with negative bone marrow by morphology (C and D), and for patients with positive bone marrow by morphology (E and F) in relationship to NB5 ΔCt in bone marrow (A, C, E) and blood (B, D, F) is shown. P values in the figures correspond to the categorical univariate likelihood ratio test from the time-dependent covariate analysis shown in Table 3 (see Patients and Methods). ΔCt of 15 chosen as the median of positive ΔCt values. N is the number of unique patients/number of samples.

Figure 3.

Correlation of NB5 assay ΔCt in bone marrow (BM) and blood with PFS. The probability of progression for all patients (A and B), for patients with negative bone marrow by morphology (C and D), and for patients with positive bone marrow by morphology (E and F) in relationship to NB5 ΔCt in bone marrow (A, C, E) and blood (B, D, F) is shown. P values in the figures correspond to the categorical univariate likelihood ratio test from the time-dependent covariate analysis shown in Table 3 (see Patients and Methods). ΔCt of 15 chosen as the median of positive ΔCt values. N is the number of unique patients/number of samples.

Close modal
Table 3.

Univariate and multivariate time-dependent covariate analysis of association of NB5 assay ΔCt with PFS

Bone marrow NB5 assayBlood NB5 assay
Type of analysisPatient groupNCategoricalaContinuousaNCategoricalContinuous
Univariate All 216 <0.001 0.005 134 0.001 <0.001 
Univariate Bone marrow morph. negative 113 0.001 0.008 62 0.014 0.017 
Univariate Bone marrow morph. positive 98 0.33 0.65 50 0.72 0.19 
Multivariateb All 178 <0.001 0.025 99 0.055 0.006 
Multivariateb Bone marrow morph. negative 92 0.004 0.066 51 0.0013 0.009 
Multivariateb Bone marrow morph. positive 86 0.098 0.62 48 0.87 0.48 
Bone marrow NB5 assayBlood NB5 assay
Type of analysisPatient groupNCategoricalaContinuousaNCategoricalContinuous
Univariate All 216 <0.001 0.005 134 0.001 <0.001 
Univariate Bone marrow morph. negative 113 0.001 0.008 62 0.014 0.017 
Univariate Bone marrow morph. positive 98 0.33 0.65 50 0.72 0.19 
Multivariateb All 178 <0.001 0.025 99 0.055 0.006 
Multivariateb Bone marrow morph. negative 92 0.004 0.066 51 0.0013 0.009 
Multivariateb Bone marrow morph. positive 86 0.098 0.62 48 0.87 0.48 

aFor the categorical analysis, P values are based on two degree of freedom Cox likelihood ratio test using three NB5 ΔCt categories as shown in Fig. 3, and for the continuous analysis they are based on a two degree of freedom likelihood ratio for linear and quadratic terms for NB5 ΔCt.

bMultivariate analysis includes only time points where the NB5 assay and CT/MRI, MIBG, and bone marrow morphology assessments of disease status were performed. The multivariate analysis simultaneously adjusts for positive/negative status by CT/MRI, MIBG, and bone marrow morphology and includes NB MYCN gene status (amplified or nonamplified) for the 94 patients with available data.

The NB5 assay using all five genes was compared in this dataset with other analyses using other signatures (PHOX2B + TH + DCX; PHOX2B + TH + DDC; PHOX2B + TH; and TH + DCX) based on prior publications (refs. 17, 19, 34, 35; Supplementary Fig. S3). The correlation between these signatures and the NB5 was very high, but there were a larger number of undetectable results in the signatures with fewer than five genes. Dividing the sample in Supplementary Fig. S3 into two groups at various cut-off points and comparing 1-year PFS, the NB5 signature resulted in a wider separation between groups than the other signatures. ROC analysis confirmed that NB5 signature resulted in a higher AUC than the other signatures for predicting PFS. In addition, individual genes were removed from the NB5 assay and change in AUC was calculated (Supplementary Table S4). Based on this analysis, no one gene dominates all the others but the singly most influential gene in this context appears to be CHGA whose removal results in 4.8% drop in AUC.

To determine whether NB5 ΔCt was associated with disease progression independently of clinically-defined disease and MYCN gene status, multivariate time-dependent covariate analysis was performed using time points where all three clinical assessments (CT/MRI, MIBG, and bone marrow morphology) and MYCN data were available and classified as either positive or negative. These analyses confirmed that NB5 ΔCt is significantly associated with PFS independently of clinical disease and MYCN gene status (Table 3). Analyses using ΔCt as a continuous rather than a categorical variable yielded similar results.

Disease evaluation is important for guiding therapeutic decisions and assessing prognosis. This is the first study of patients with high-risk neuroblastoma to quantify expression of five neuroblastoma-associated genes in bone marrow and blood concurrently with standard disease evaluations at several sequential times. We show a high correlation between the quantity of neuroblastoma-associated mRNA in bone marrow and blood and disease status defined by percent of neuroblastoma cells in bone marrow biopsy and MIBG Curie score but also show that neuroblastoma mRNA is detectable when these standard evaluations are negative. Importantly, quantifying neuroblastoma mRNA in both bone marrow and blood provides an independent predictor of PFS. Thus, the NB5 assay provides a useful new molecular biomarker that improves both assessment of disease and prognostication.

Neuroblastoma mRNA in bone marrow and blood, as quantified by the NB5 assay, were strongly correlated, but the amount in bone marrow was significantly greater. Although it is less invasive to perform this assay on blood samples, our data suggest that bone marrow samples provide more sensitive detection. Indeed, when bone marrow was positive, 35% of blood specimens did not have detectable neuroblastoma mRNA, which indicates that testing blood cannot be a sensitive surrogate for bone marrow disease. NB5 assay ΔCt values of both bone marrow and blood correlated with the percentage of neuroblastoma cells in bone marrow and with the MIBG Curie score. Furthermore, even when controlling for the status of the other clinical disease assessments, stronger bone marrow and blood NB5 ΔCt values correlated with neuroblastoma cells in bone marrow (present or absent) and with MIBG uptake (present or absent). Correlation of blood ΔCt with both bone marrow and MIBG defined disease implies that both can contribute to circulating neuroblastoma cells. Blood ΔCt (and not bone marrow) also correlated with soft tissue tumor size, which suggests that soft tissue disease also can contribute circulating tumor cells. The absence of correlation between bone marrow NB5 ΔCt and tumor size suggests that neuroblastoma cell growth/survival in the two sites may be independent.

In many instances, quantification of neuroblastoma-associated mRNA in bone marrow and blood using the NB5 assay detected disease when CT/MRI and MIBG scans and morphologic assessment of bone marrow were negative. For example, the NB5 assay detected neuroblastoma mRNA in bone marrow of 11 of 20 patients and in blood of four of nine patients who were in complete clinical remission. Although not significant by logrank test, 5 of 11 patients with NB5 detectable versus 1 of 9 with NB5 nondetectable disease in bone marrow subsequently developed progressive disease. Improved sensitivity of disease quantification in bone marrow and blood will likely improve assessment of response to treatment and provide an early surrogate for outcome. Although others have reported that quantitative RT-PCR is more sensitive than morphology for detecting bone marrow disease (14, 16, 36, 37), our study is the first to comprehensively compare quantification of mRNA (for five neuroblastoma-associated genes) in bone marrow and blood to current standard imaging and bone marrow evaluations. Our data shows that the NB5 assay for mRNA in bone marrow and blood improves assessment of neuroblastoma disease status.

The most striking result of this study is the association of neuroblastoma mRNA quantity in bone marrow and blood with PFS of patients with refractory/relapsed neuroblastoma. This was highly significant for all patients and even for those whose bone marrow samples were negative morphologically. In fact, the absence of NB5 detectable mRNA in bone marrow, which is almost always associated with a negative bone marrow morphologically, predicts a high probability of PFS. By contrast, any level of neuroblastoma mRNA in bone marrow, even when neuroblastoma cells were not detected morphologically, predicted a poor PFS. Importantly, multivariate analysis demonstrated that the quantity of neuroblastoma mRNA in bone marrow and blood defined with our NB5 assay predicts PFS for all patients and for those with morphologically negative bone marrow samples independently from disease status defined with standard clinical evaluations and from MYCN gene status. Although time to first relapse, age, stage of disease, and MYCN gene copy number have been reported to be independently predictive of postrelapse survival (6), our study is the first to demonstrate that subgroups of patients with relapsed/refractory neuroblastoma can be identified who have different likelihoods of PFS when treated in early phase clinical trials. Our data suggest that quantifying neuroblastoma mRNA for multiple neuroblastoma-associated genes in bone marrow and blood will enable risk stratification of patients with relapsed/refractory neuroblastoma before and during their treatment in early phase clinical trials.

A criticism of an assay using five genes would be that there could be increased sensitivity but with addition of false positivity. To answer this, a comparison of the 5-gene signature to other published gene combination signatures was performed (Supplementary Fig. S3). This showed that the NB5 assay is detectible more often and also better able to predict PFS at different cut-off points. This argues that the positive predictive value is enhanced by quantifying expression of all five genes.

Predicting PFS by quantifying neuroblastoma mRNA with a multigene assay could be affected by clinical variables. First, aspirating a single site may not provide bone marrow that is representative of multiple sites with sufficient RNA quality (36, 38), although one study that utilized immunocytology to identify neuroblastoma cells indicated that a sensitive assay can overcome this potential problem (39). To circumvent this problem, we obtained specimens from both iliac crests and pooled them for testing, and the RNA quality for this multi-institution study was excellent. Second, some neuroblastoma cells may be tightly adherent in the bone marrow microenvironment and not readily aspirated. However, we used biopsy specimens to define neuroblastoma cells morphologically, and there was only one specimen that had neuropil in the biopsy but that did not have detectable neuroblastoma mRNA, which suggests that the sensitivity of the NB5 assay renders this an unlikely limitation. Although testing blood could potentially overcome sampling concerns, there is less neuroblastoma mRNA in blood than bone marrow, and some blood specimens did not have detectable neuroblastoma mRNA even though disease was present by standard evaluations. The type of therapy could affect test results, but our finding that NB5 ΔCt predicts PFS for patients enrolled in a variety of early phase clinical trials suggests independence from treatment. However, this needs to be further studied in trials in which patients are uniformly treated and have complete disease assessments along with NB5 assay testing of bone marrow and blood.

Other studies have shown that quantifying mRNA for three and five neuroblastoma-associated genes [TH, PHOX2B, and DCX; (17) PHOX2B, TH, DDC, GAP43, and CHRNA3; (21)] in bone marrow provides prognostic information for patients with stage 4, high-risk neuroblastoma at diagnosis and during and at the end of induction therapy (17, 21). We previously reported that subgroups of high-risk neuroblastoma patients with different outcomes can be identified by quantifying neuroblastoma mRNA in peripheral blood stem cells obtained during induction therapy with the NB5 assay used in this study (5). Quantification of mRNA for four neuroblastoma-associated genes (B4GALNT1, PHOX2B, CCND1, and ISL1) has been reported to predict PFS and overall survival for patients in first or second remission or with refractory disease and treated with the anti-GD2 antibody 3F8 (19). Our study demonstrates for the first time an improvement in the definition of disease status by quantifying mRNA for five neuroblastoma-associated genes (CHGA, DCX, DDC, PHOX2B, and TH) in bone marrow and blood and confirms the correlation with PFS in relapsed or refractory neuroblastoma. Analysis of bone marrow and blood with this new molecular biomarker assay predicts PFS independently of standard clinical evaluations of disease and of MYCN gene status. Further studies that utilize this assay in this and other contexts will aim to confirm its utility in managing patients with neuroblastoma. To this end, the NB5 assay has been incorporated into all NANT consortium therapeutic studies in order to validate the assay and prospectively analyze disease burden assessment and to provide standard clinical disease response testing in the context of uniform therapy in large numbers of patients.

H.A. Lai reports receiving commercial research grants from Siemens Medical. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A. Marachelian, J.G. Villablanca, H.A. Lai, K.K. Matthay, S. Groshen, S. Asgharzadeh, R.C. Seeger

Development of methodology: A. Marachelian, C.W. Liu, S. Young, S. Asgharzadeh, R.C. Seeger

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Marachelian, H. Shimada, H.C. Tran, R. Gallego, S. Young, S. Czarnecki, B.D. Weiss, K. Goldsmith, M. Granger, S. Asgharzadeh, R.C. Seeger

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Marachelian, J.G. Villablanca, C.W. Liu, F. Goodarzian, H.C. Tran, S. Groshen, R. Sposto, R.C. Seeger

Writing, review, and/or revision of the manuscript: A. Marachelian, J.G. Villablanca, H.A. Lai, H. Shimada, H.C. Tran, B.D. Weiss, K. Goldsmith, M. Granger, K.K. Matthay, S. Asgharzadeh, R. Sposto, R.C. Seeger

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Marachelian, C.W. Liu, B. Liu, R. Gallego, N. Bedrossian, S. Young, S. Czarnecki, R. Kennedy, R.C. Seeger

Study supervision: A. Marachelian, K.K. Matthay, R.C. Seeger

Other (acquisition of data via laboratory assays, provision of data for statistical analysis, reporting and organizing data, database entry, etc.): J.A. Parra

Leader of NANT; oversee all studies: K.K. Matthay

This work was supported in part by grants from the National Cancer Institute [5 P01 CA81403 (to R. Seeger); 1 R33 CA152809 (to R. Seeger); 1 R01 CA182633 (to S. Asgharzadeh and R. Seeger), and 5 P30 CA014089 (to A. Marachelian)]; from the Alex's Lemonade Stand Foundation (to R. Seeger); and from the St. Baldrick's Foundation (to R. Seeger).

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.

1.
Cohn
SL
,
Pearson
AD
,
London
WB
,
Monclair
T
,
Ambros
PF
,
Brodeur
GM
, et al
The International Neuroblastoma Risk Group (INRG) classification system: an INRG Task Force report
.
J Clin Oncol
2009
;
27
:
289
97
.
2.
Stram
DO
,
Matthay
KK
,
O'Leary
M
,
Reynolds
CP
,
Haase
GM
,
Atkinson
JB
, et al
Consolidation chemoradiotherapy and autologous bone marrow transplantation versus continued chemotherapy for metastatic neuroblastoma: a report of two concurrent Children's Cancer Group studies
.
J Clin Oncol
1996
;
14
:
2417
26
.
3.
Matthay
KK
,
Villablanca
JG
,
Seeger
RC
,
Stram
DO
,
Harris
RE
,
Ramsay
NK
, 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
1999
;
341
:
1165
73
.
4.
Yu
AL
,
Gilman
AL
,
Ozkaynak
MF
,
London
WB
,
Kreissman
SG
,
Chen
HX
, et al
Anti-GD2 antibody with GM-CSF, interleukin-2, and isotretinoin for neuroblastoma
.
N Engl J Med
2010
;
363
:
1324
34
.
5.
Kreissman
SG
,
Seeger
RC
,
Matthay
KK
,
London
WB
,
Sposto
R
,
Grupp
SA
, et al
Purged versus non-purged peripheral blood stem-cell transplantation for high-risk neuroblastoma (COG A3973): a randomised phase 3 trial
.
Lancet Oncol
2013
;
14
:
999
1008
.
6.
London
WB
,
Castel
V
,
Monclair
T
,
Ambros
PF
,
Pearson
AD
,
Cohn
SL
, et al
Clinical and biologic features predictive of survival after relapse of neuroblastoma: a report from the International Neuroblastoma Risk Group project
.
J Clin Oncol
2011
;
29
:
3286
92
.
7.
London
WB
,
Frantz
CN
,
Campbell
LA
,
Seeger
RC
,
Brumback
BA
,
Cohn
SL
, et al
Phase II randomized comparison of topotecan plus cyclophosphamide versus topotecan alone in children with recurrent or refractory neuroblastoma: a Children's Oncology Group study
.
J Clin Oncol
. 
2010
;
28
:
3808
15
.
8.
Pugh
TJ
,
Morozova
O
,
Attiyeh
EF
,
Asgharzadeh
S
,
Wei
JS
,
Auclair
D
, et al
The genetic landscape of high-risk neuroblastoma
.
Nat Genet
2013
;
45
:
279
84
.
9.
Bergfeld
SA
,
Blavier
L
,
DeClerck
YA
: 
Bone marrow-derived mesenchymal stromal cells promote survival and drug resistance in tumor cells
.
Mol Cancer Ther
2014
;
13
:
962
75
.
10.
Fang
H
,
Declerck
YA
: 
Targeting the tumor microenvironment: from understanding pathways to effective clinical trials
.
Cancer Res
2013
;
73
:
4965
77
.
11.
Asgharzadeh
S
,
Salo
JA
,
Ji
L
,
Oberthuer
A
,
Fischer
M
,
Berthold
F
, et al
Clinical significance of tumor-associated inflammatory cells in metastatic neuroblastoma
.
J Clin Oncol
2012
;
30
:
3525
32
.
12.
Yanik
GA
,
Parisi
MT
,
Shulkin
BL
,
Naranjo
A
,
Kreissman
SG
,
London
WB
, et al
Semiquantitative mIBG scoring as a prognostic indicator in patients with stage 4 neuroblastoma: a report from the Children's oncology group
.
J Nucl Med
2013
;
54
:
541
8
.
13.
Brisse
HJ
,
McCarville
MB
,
Granata
C
,
Krug
KB
,
Wootton-Gorges
SL
,
Kanegawa
K
, et al
Guidelines for imaging and staging of neuroblastic tumors: consensus report from the International Neuroblastoma Risk Group Project
.
Radiology
2011
;
261
:
243
57
.
14.
Beiske
K
,
Ambros
PF
,
Burchill
SA
,
Cheung
IY
,
Swerts
K
. 
Detecting minimal residual disease in neuroblastoma patients-the present state of the art
.
Cancer Lett
2005
;
228
:
229
40
.
15.
Viprey
VF
,
Lastowska
MA
,
Corrias
MV
,
Swerts
K
,
Jackson
MS
,
Burchill
SA
. 
Minimal disease monitoring by QRT-PCR: guidelines for identification and systematic validation of molecular markers prior to evaluation in prospective clinical trials
.
J Pathol
2008
;
216
:
245
52
.
16.
Beiske
K
,
Burchill
SA
,
Cheung
IY
,
Hiyama
E
,
Seeger
RC
,
Cohn
SL
, et al
Consensus criteria for sensitive detection of minimal neuroblastoma cells in bone marrow, blood and stem cell preparations by immunocytology and QRT-PCR: recommendations by the International Neuroblastoma Risk Group Task Force
.
Br J Cancer
2009
;
100
:
1627
37
.
17.
Viprey
VF
,
Gregory
WM
,
Corrias
MV
,
Tchirkov
A
,
Swerts
K
,
Vicha
A
, et al
Neuroblastoma mRNAs predict outcome in children with stage 4 neuroblastoma: a European HR-NBL1/SIOPEN study
.
J Clin Oncol
2014
;
32
:
1074
83
.
18.
Cheung
IY
,
Lo Piccolo
MS
,
Kushner
BH
,
Cheung
N-KV
. 
Early molecular response of marrow disease to biologic therapy is highly prognostic in neuroblastoma
. J Clin Oncol
2003
;
21
:
3853
3858
.
19.
Cheung
NV
,
Ostrovnaya
I
,
Kuk
D
,
Cheung
IY
. 
Bone Marrow minimal residual disease was an early response marker and a consistent independent predictor of survival after anti-GD2 immunotherapy
.
J Clin Oncol
2015
;
33
:
755
63
.
20.
Stutterheim
J
,
Gerritsen
A
,
Zappeij-Kannegieter
L
,
Yalcin
B
,
Dee
R
,
van Noesel
MM
, et al
Detecting minimal residual disease in neuroblastoma: the superiority of a panel of real-time quantitative PCR markers
.
Clin Chem
2009
;
55
:
1316
26
.
21.
Stutterheim
J
,
Zappeij-Kannegieter
L
,
Versteeg
R
,
Caron
HN
,
van der Schoot
CE
,
Tytgat
GA
. 
The prognostic value of fast molecular response of marrow disease in patients aged over 1 year with stage 4 neuroblastoma
.
Eur J Cancer
2011
;
47
:
1193
202
.
22.
Park
JR
,
Bagatell
R
,
Cohn
SL
,
Pearson
AD
,
Villablanca
JG
,
Berthold
F
, et al
Revisions to the international neuroblastoma response criteria: a consensus statement from the NCI-clinical trials planning meeting
.
J Clin Oncol
. 
2017 May 4
.
[Epub ahead of print]
.
23.
Matthay
KK
,
Shulkin
B
,
Ladenstein
R
,
Michon
J
,
Giammarile
F
,
Lewington
V
, et al
Criteria for evaluation of disease extent by (123)I-metaiodobenzylguanidine scans in neuroblastoma: a report for the International Neuroblastoma Risk Group (INRG) Task Force
.
Br J Cancer
2010
;
102
:
1319
26
.
24.
Brodeur
GM
,
Pritchard
J
,
Berthold
F
,
Carlsen
NL
,
Castel
V
,
Castelberry
RP
, et al
Revisions of the international criteria for neuroblastoma diagnosis, staging, and response to treatment
.
J Clin Oncol
1993
;
11
:
1466
77
.
25.
Seeger
RC
,
Reynolds
CP
,
Gallego
R
,
Stram
DO
,
Gerbing
RB
,
Matthay
KK
. 
Quantitative tumor cell content of bone marrow and blood as a predictor of outcome in stage IV neuroblastoma: a Children's Cancer Group Study
.
J Clin Oncol
2000
;
18
:
4067
4076
.
26.
Therasse
P
,
Arbuck
SG
,
Eisenhauer
EA
,
Wanders
J
,
Kaplan
RS
,
Rubinstein
L
, et al
New guidelines to evaluate the response to treatment in solid tumors. European organization for research and treatment of cancer, national cancer institute of the United States, National Cancer Institute of Canada
.
J Natl Cancer Inst
2000
;
92
:
205
16
.
27.
Ady
N
,
Zucker
JM
,
Asselain
B
,
Edeline
V
,
Bonnin
F
,
Michon
J
, et al
A new 123I-MIBG whole body scan scoring method–application to the prediction of the response of metastases to induction chemotherapy in stage IV neuroblastoma
.
Eur J Cancer
1995
;
31A
:
256
61
.
28.
Matthay
KK
,
Edeline
V
,
Lumbroso
J
,
Tanguy
ML
,
Asselain
B
,
Zucker
JM
, et al
Correlation of early metastatic response by 123I-metaiodobenzylguanidine scintigraphy with overall response and event-free survival in stage IV neuroblastoma
.
J Clin Oncol
2003
;
21
:
2486
91
.
29.
Burchill
SA
,
Beiske
K
,
Shimada
H
,
Ambros
PF
,
Seeger
R
,
Tytgat
GA
, et al
, 
Recommendations for the standardization of bone marrow disease assessment and reporting in children with neuroblastoma; on behalf of the International Neuroblastoma Response Criteria Bone Marrow Working Group
.
Cancer
2016
;
123
:
1095
105
30.
Yanik
G
,
Villablanca
J
,
Maris
J
,
Weiss
B
,
Groshen
S
,
Marachelian
A
, et al
131 I-Metaiodobenzylguanidine with Intensive chemotherapy and autologous stem cell transplant for high risk neuroblastoma. A new approaches to neuroblastoma therapy (NANT) phase II study
.
Biol Blood Marrow Transplant
2015
;
21
:
673
81
.
31.
Dixon
WJ
,
Massey
FJ
:
Introduction to statistical analysis
. 3rd ed.
New York, NY
:
McGraw-Hill
; 
1968.
32.
Cox
DR
,
Oakes
D
: 
Analysis of survival data
.
New York, NY
:
Chapman and Hall
; 
1984.
33.
Klein
JP
,
Rizzo
JD
,
Zhang
MJ
,
Keiding
N
. 
Statistical methods for the analysis and presentation of the results of bone marrow transplants. Part 2: Regression modeling
.
Bone Marrow Transplant
2001
;
28
:
1001
11
.
34.
van Wezel
EM
,
Decarolis
B
,
Stutterheim
J
,
Zappeij-Kannegieter
L
,
Berthold
F
,
Schumacher-Kuckelkorn
R
, et al
Neuroblastoma messenger RNA is frequently detected in bone marrow at diagnosis of localised neuroblastoma patients
.
Eur J Cancer
2016
;
54
:
149
58
.
35.
Yanez
Y
,
Hervas
D
,
Grau
E
,
Oltra
S
,
Pérez
G
,
Palanca
S
, et al
TH and DCX mRNAs in peripheral blood and bone marrow predict outcome in metastatic neuroblastoma patients
.
J Cancer Res Clin Oncol
2016
;
142
:
573
80
.
36.
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
1998
;
4
:
2801
5
.
37.
Bozzi
F
,
Luksch
R
,
Collini
P
,
Gambirasio
F
,
Barzanò
E
,
Polastri
D
, et al
Molecular detection of dopamine decarboxylase expression by means of reverse transcriptase and polymerase chain reaction in bone marrow and peripheral blood: utility as a tumor marker for neuroblastoma
.
Diagn Mol Pathol
2004
;
13
:
135
43
.
38.
Cheung
NK
,
Heller
G
,
Kushner
BH
,
Liu
C
,
Cheung
IY
. 
Detection of metastatic neuroblastoma in bone marrow: when is routine marrow histology insensitive?
J Clin Oncol
1997
;
15
:
2807
17
.
39.
Cheung
NK
,
Heller
G
,
Kushner
BH
,
Kramer
K
. 
Detection of neuroblastoma in bone marrow by immunocytology: is a single marrow aspirate adequate?
Med Pediatr Oncol
1999
;
32
:
84
7
.