Neuroblastoma (NB) is one of the most common malignant tumors in children, which originates from neural-crest cells, and related with about 12% to 15% of pediatric cancer deaths [
Cytology of aspirates and histology of biopsies have been the gold standard to evaluate BMI in NB for many years. However, these methods have limited sensitivity when NB involvement is less than 10% and could seriously underestimate the prevalence of BMI [
18F-FDG PET/CT, as a major functional imaging method, has been widely used in pediatric tumors for diagnosis and staging [
The aim of the present study was to investigate the value of pretreatment 18F-FDG PET/CT bone marrow uptake pattern (BMUP) in detecting BMI in newly diagnosed pediatric NB and compared it with bone marrow biopsy and PHOX2B.
We retrospectively analyzed the medical record of pediatric patients (age <18 years old) from January 2018 to December 2019 in the nuclear medicine department, Beijing Friendship Hospital, Capital Medical University. A total of 188 NB patients were identified from the electronic patient files. This study followed the inclusion criteria: (1) newly diagnosed pediatric NB patients performed 18F-FDG PET/CT scan before surgery; (2) 18F-FDG PET/CT, bone marrow testing, and RT-qPCR test were performed within two weeks. The main exclusion criteria included the following: (1) NB patients receiving chemotherapy, glucocorticoids, or hematopoietic cytokines before 18F-FDG PET/CT; (2) patients with severe active inflammation (Figure
Flow diagrams of study populations.
Clinical data of NB patients were obtained from medical records, including age, gender, weight, height, primary tumor site, and follow-up information. All patients must be followed up for six months or more. The permission to conduct this retrospective study was obtained from the Institutional Review Board of Beijing Friendship Hospital, Capital Medical University, which waived the requirement of informed consent (L-2019-18).
All 18F-FDG PET/CT imaging were acquired from Siemens Biograph 64-slice mCT (Siemens Medical Systems, Germany) according to the guidelines recommended by the European Association of Nuclear Medicine in pediatric oncology [
All images were independently reviewed by two groups of nuclear medicine physicians (two residents and two attending physicians), blinded to laboratory indexes, clinical information, and BMB results. Disagreements were discussed and adjudicated through a third physician.
Based on our routine clinical practice experience, the general form of bone/bone marrow metastasis, and previous researches [
18F-FDG PET/CT bone marrow uptake pattern (BMUP). (a) BMUP1 was defined as direct invasion or being surrounded by adjacent tumor tissue (black arrow). (b) BMUP2 was defined as focal/multifocal uptake (white arrow), which was one or more (less than 30) restricted area(s) of FDG uptake. (c) BMUP3 was defined as diffuse and homogeneously increased FDG uptake within the skeleton. (d) BMUP4 was defined as diffuse and heterogeneously increased FDG uptake within the skeleton.
Bone marrow aspiration and biopsy at unilateral iliac crest were performed routinely at the time of initial diagnosis of NB. BMB would be tested using normal cytology, which was based on standard immunohistochemistry or morphological analysis of bone marrow aspirates. PHOX2B of bone marrow (PHOX2B of BM) and peripheral blood would be tested through RT-qPCR, which were expressed as the fold change in PHOX2B gene expression relative to the healthy controls. All patients were classified into four BMUP, according to the 18F-FDG PET/CT uptake patterns of bone marrow. Since children were unlikely to develop degenerative diseases, those typical focal and heterogeneous FDG uptake patterns were easy to diagnose BMI in pediatric patients, so patients with BMUP1, BMUP2, and BMUP4 were interpreted as having BMI on 18F-FDG PET/CT images. BMI was hard to identify in patients with homogeneous FDG uptake, so further evaluation of BMUP3 was required. If only one of the four test methods were positive, we would review the positive result and check again.
The diffuse and homogeneous 18F-FDG uptake may be caused by tumor infiltration or myeloid hyperplasia reaction. In order to differentiate reactive uptake in BMUP3, semiquantitative 18F-FDG PET/CT parameters would be measured. A circular region of interest was placed in the normal area of left and right lobe of liver, respectively. The average SUVmax value of two regions was considered as the maximum standardized uptake value of liver (SUVmax-Liver). A spherical volume of interest was placed in L1 to L5 vertebral body, and the maximum standardized uptake value of lumbar spine was determined as SUVmax-Bone. Then, the ratio of SUVmax-Bone to SUVmax-Liver (SUVmax-Bone/Liver) was calculated.
The presence of BMI was based on one or more results of the following: (1) directed biopsies guided by PET/CT were confirmed positive; (2) supplementary targeted other diagnostic imaging (including enhanced CT, MRI, and 123I-MIBG SPECT/CT) within two weeks confirmed positive for corresponding site; (3) regression of bone marrow lesions in parallel with other neuroblastoma lesions on clinical follow-up (including the imaging results and relevant laboratory tests).
Continuous data were expressed as mean ± standard deviation (mean ± SD), and categorical data were expressed as numbers (percentages). Clinical characteristic factors and diagnostic accuracy of different BMUP groups were compared by using Student’s
Ninety-eight pediatric patients (48 boys and 50 girls) with newly diagnosed NB were finally enrolled in our study. In order to evaluate different observer agreements for BMUP, we divided the observers into two groups (two residents and two attending physicians). Kappa coefficients of two residents and two attending physicians were 0.857 (95%CI: 0.793–0.957) and 0.891 (95%CI: 0.815–0.968), respectively. The kappa coefficient between resident group and attending physician group was 0.845 (95%CI: 0.756–0.935). All groups had excellent kappa coefficients, indicating that BMUP was a simple and easy method for diagnosis BMI in pediatric NB. Results are summarized in Table
18F-FDG PET/CT bone marrow uptake pattern diagnostic consistency in different physicians.
BMUP1 | BMUP2 | BMUP3 | BMUP4 | |||
---|---|---|---|---|---|---|
Residents | One | 7 | 12 | 34 | 45 | 0.857 (0.793–0.957) |
Two | 6 | 13 | 37 | 42 | ||
Attending physicians | One | 8 | 13 | 34 | 43 | 0.891 (0.815–0.968) |
Two | 9 | 10 | 35 | 44 | ||
Between group | Residents | 7 | 12 | 36 | 43 | 0.845 (0.756–0.935) |
Attending physicians | 9 | 12 | 35 | 42 |
BMUP: bone marrow uptake pattern, BMUP1: direct invasion, BMUP2: focal/multifocal uptake, BMUP3: diffuse and homogeneous uptake, BMUP4: diffuse and heterogeneously uptake, and CI: confidence interval.
According to the classification of attending physician group, there were 9, 12, 35, and 42 patients in BMUP1, BMUP2, BMUP3, and BMUP4, respectively. The mean age of patients was 3.53 ± 2.17 years old, and about 79.6% (78/98) of pediatric patients were younger than five years old. Most patients (68/98, 69.4%) had advanced stage disease (stage M), and 11 patients (11/98, 11.2%) were early-stage (stage L1). The tumors of 83.6% (82/98) of the patients were located in the abdomen, while only three patients had the tumor in the pelvic cavity. In this study, despite significant differences in age, primary sites, INRGSS, and International Neuroblastoma Risk Group classification system (INRGCS) among different BMUP, the patient’s gender did not show significant difference to the pattern of uptake. Clinical data of these patients are shown in Table
Clinical characteristics of different18F-FDG PET/CT bone marrow uptake patterns.
Total | BMUP1 | BMUP2 | BMUP3 | BMUP4 | ||
---|---|---|---|---|---|---|
Age | 3.53 ± 2.17 | 2.97 ± 3.08 | 3.98 ± 2.57 | 2.42 ± 1.31 | 4.45 ± 2.02 | <0.001 |
Female | 50 (51.0%) | 7 (77.8%) | 5 (41.7%) | 19 (54.3%) | 19 (45.2%) | 0.316 |
Male | 48 (49.0%) | 2 (22.2%) | 7 (58.3%) | 16 (45.7%) | 23 (54.8%) | |
Chest | 13 (13.3%) | 6 (66.7%) | 1 (8.3%) | 2 (5.7%) | 4 (9.5%) | <0.001 |
Abdomen | 82 (83.6%) | 3 (33.3%) | 11 (91.7%) | 30 (85.7%) | 38 (90.5%) | |
Pelvic | 3 (3.1%) | 0 (0.0%) | 0 (0.0%) | 3 (8.6%) | 0 (0.0%) | |
L1 | 11 (11.2%) | 0 (0.0%) | 0 (0.0%) | 11 (31.4%) | 0 (0.0%) | <0.001 |
L2 | 17 (17.3%) | 5 (55.6%) | 0 (0.0%) | 12 (34.3%) | 0 (0.0%) | |
M | 68 (69.4%) | 4 (44.4%) | 12 (100.0%) | 10 (28.6%) | 42 (100.0%) | |
c | 2 (2.1%) | 0 (0.0%) | 0 (0.0%) | 2 (5.7%) | 0 (0.0%) | |
Extremely low risk | 12 (13.2%) | 0 (0.0%) | 0 (0.0%) | 12 (37.5%) | 0 (0.0%) | <0.001 |
Low risk | 6 (6.6%) | 4 (57.1%) | 0 (0.0%) | 2 (6.3%) | 0 (0.0%) | |
Medium risk | 11 (12.1%) | 2 (28.6%) | 1 (9.1%) | 7 (21.9%) | 1 (2.4%) | |
High risk | 62 (68.1%) | 1 (14.3%) | 10 (90.9%) | 11 (34.3%) | 40 (97.6%) |
INRGSS: International Neuroblastoma Risk Group Staging System; INRGCS: International Neuroblastoma Risk Group Classification System; BMUP: bone marrow uptake pattern; BMUP1: direct invasion; BMUP2: focal/multifocal uptake; BMUP3: diffuse and homogeneous uptake; BMUP4: diffuse and heterogeneously uptake;
Semiquantitative parameters of 18F-FDG PET/CT were used to assess BMI in BMUP3 patients. The optimal cut-off value of SUVmax-Bone/Liver for distinguishing metastasis from reactive uptake in BMUP3 patients was 2.08, and the area under curve (AUC) was 0.873 (95% CI: 0.714–0.962,
Receiver operating characteristic curve of SUVmax-Bone/Liver for BMUP3 patients.
The optimal cut-off value of SUVmax-Bone/Liver was 2.08, and areas under curve (AUC) was 0.873 (95% CI: 0.714–0.962,
There were 70 positive patients and 28 negative patients in 18F-FDG PET/CT, 41 positive patients and 57 negative patients in BMB, 48 positive patients and 50 negative patients in PHOX2B of blood, and 62 positive patients and 36 negative patients in PHOX2B of BM. Positive predictive value (PPV), negative predictive value (NPV), sensitivity, and specificity, were 92.9%, 92.9%, 97.0%, and 83.9% for 18F-FDG PET/CT and 96.7%, 80.6%, 89.6%, and 93.5% for PHOX2B of BM, respectively (Table
BMI positive patients, BMI negative patients, PPV, NPV, sensitivity and specificity of four tests.
BMI positive | BMI negative | PPV (%) | NPV (%) | Sensitivity (%) | Specificity (%) | |
---|---|---|---|---|---|---|
PET/CT | 70 | 28 | 92.9 | 92.9 | 97.0 | 83.9 |
BMB | 41 | 57 | 100.0 | 54.4 | 61.2 | 100.0 |
PHOX2B of blood | 48 | 50 | 95.8 | 58.0 | 68.7 | 93.5 |
PHOX2B of BM | 62 | 36 | 96.7 | 80.6 | 89.6 | 93.5 |
BMI: bone marrow involvement, BMB: bone marrow biopsy, PHOX2B of blood: paired-like homeobox 2b of blood, PHOX2B of BM: paired-like homeobox 2b of bone marrow, PPV: positive predictive value, and NPV: negative predictive value.
The diagnostic efficiency of different tests and compared with PET/CT.
AUC | 95%CI | ||
---|---|---|---|
PET/CT | 0.904 | 0.828–0.955 | — |
BMB | 0.806 | 0.714–0.879 | 0.028 |
PHOX2B of blood | 0.811 | 0.719–0.883 | 0.078 |
PHOX2B of BM | 0.916 | 0.842–0.962 | 0.817 |
BMB: bone marrow biopsy, PHOX2B of blood: paired-like homeobox 2b of blood, PHOX2B of BM: paired-like homeobox 2b of bone marrow, AUC: area under curve, and CI: confidence interval;
Diagnostic efficiency between 18F-FDG PET/CT with other three tests in all pediatric NB patients.
Areas under curve (AUC) were 0.904 (95% CI: 0.828–0.955) for 18F-FDG PET/CT, 0.916 (95% CI: 0.842–0.962) for PHOX2B of BM, 0.811 (95% CI: 0.719–0.883) for PHOX2B of blood, and 0.806 (95% CI: 0.714–0.879) for BMB.
We further investigated the diagnostic accuracy of 18F-FDG PET/CT, BMB, PHOX2B of blood, and PHOX2B of BM in the different BMUP. According to the BUMP grouping, the diagnostic accuracy of BMI in BMUP1, BMUP2, BMUP3, and BMUP4 of FDG PET were 66.7% (6/9), 100.0% (12/12), 88.6% (31/35), and 100.0% (42/42), in BMB were 33.3% (3/9), 16.7% (2/12), 80.0% (28/35), and 92.8% (39/42), in PHOX2B of blood were 33.3% (3/9), 33.3% (4/12), 82.9% (29/35), and 92.8% (39/42), and in PHOX2B of BM were 77.8% (7/9), 83.3% (10/12), 85.7% (30/35), and 100.0% (42/42). 18F-FDG PET/CT was better than BMB and PHOX2B of blood in diagnosing BMI in BMUP2 patients with significant difference, and the results are shown in Table
The diagnostic accuracy of different tests based on the BMUP grouping and compared with PET/CT.
BMUP1 | BMUP2 | BMUP3 | BMUP4 | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Correct | Incorrect | Correct | Incorrect | Correct | Incorrect | Correct | Incorrect | |||||
PET/CT | 6 | 3 | — | 12 | 0 | — | 31 | 4 | — | 42 | 0 | — |
BMB | 3 | 6 | 0.347 | 2 | 10 | <0.001 | 28 | 7 | 0.324 | 39 | 3 | 0.240 |
PHOX2B of blood | 3 | 6 | 0.347 | 4 | 8 | 0.001 | 29 | 6 | 0.495 | 39 | 3 | 0.240 |
PHOX2B of BM | 7 | 2 | <1.000 | 10 | 2 | 0.478 | 30 | 5 | <1.000 | 42 | 0 | 1.000 |
BMB: bone marrow biopsy, PHOX2B of blood: paired-like homeobox 2b of blood, PHOX2B of BM: paired-like homeobox 2b of bone marrow, BMUP: bone marrow uptake pattern, BMUP1: direct invasion, BMUP2: focal/multifocal uptake, BMUP3: diffuse and homogeneous uptake, and BMUP4: diffuse and heterogeneously uptake;
We proposed 18F-FDG PET/CT BMUP and evaluated the value of BMUP in detecting BMI compared with PHOX2B of blood and PHOX2B of BM in newly diagnosed NB. Pretreatment 18F-FDG PET/CT BMUP was a simple method for diagnosing BMI in pediatric NB patients, which had excellent coefficients between different physicians. We demonstrated that the optimal cut-off value was 2.08 for SUVmax-Bone/Liver of BMUP3 to diagnose BMI. The PHOX2B of BM had the highest AUC in diagnosing BMI, but there was no significant difference with 18F-FDG PET/CT. 18F-FDG PET/CT was better than normal BMB and PHOX2B of blood in detecting BMI, especially in BMUP2.
NB is a common neuroectodermal tumor, which originates from the neural crest, while the bone marrow was the most metastatic site of NB [
There was direct evidence of BMI in PET/CT images, for BMUP1, BMUP2, and BMUP4. However, for diffuse and homogeneous FDG uptake (BMUP3), there was still lack of standardized methods to assess BMI in PET/CT images. For children especially, the diffuse and homogeneous uptake of bone marrow may be caused by paraneoplastic bone marrow activation, cancerous infiltration, and inflammatory reaction or as a result of hematopoietic growth factors [
BMI was an important prognostic factor for NB; however, there was still much controversy about detecting BMI [
There were some limitations in the present study. First, this was a retrospective single-center study with a small sample. Second, we only took samples from the unilateral iliac crest for RT-qPCR testing. For pediatric NB patients, it was difficult for us to perform multiple bone marrow biopsies as recommended. Third, there was no acceptable way to confirm every lesion showing in PET/CT, especially in our pediatric NB patients. Then, in order to avoid the impact of bone marrow puncture of the ilium on SUVmax-Bone in PET/CT, we adopted the method of drawing ROI on the lumbar spine and abandoned the method of drawing ROI on the iliac bone, which might better reflect bone marrow pathology. Finally, the SUVmax-Bone/Liver >2.08 for predicting BMI in BMUP3 is only obtained from limited data in our single institute, which may lack universal representation, and the final diagnosis of BMI still has some shortcomings, which might need further refinement. A large prospective multi-center PET/CT multiparameter study should be taken to validate our findings in the future.
18F-FDG PET/CT BMUP is a simple and practical method to evaluate BMI and has a relatively high diagnostic efficiency. 18F-FDG PET/CT could play a complementary role in accurately diagnosing BMI and decrease unnecessary invasive inspections for some pediatric NB patients.
The data used in this study are available upon reasonable request from the corresponding author.
All authors declare that there are no conflicts of interest.
Jun Liu, Xu Yang, Xia Lu, Mingyu Zhang, Wei Wang, Ying Kan, and Jigang Yang have made substantial contributions to the conception or design of the work; the acquisition, analysis, or interpretation of data; the creation of new software used in the work. Jun Liu drafted the work or revised it critically for important intellectual content. Jigang Yang approved the version to be published and agree to be accountable for all aspects of the work by ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Jigang Yang was supported by Capital’s Funds for Health Improvement and Research (no. 2020-2-2025), National Natural Science Foundation of China (no. 81971642), and National Key Research and Development Plan (no. 2020YFC0122000).
Supplementary Table 1: multiple comparisons of clinical characteristics. Supplementary Table 2: sensitivity, specificity, and accuracy of BMUP3. Supplementary Table 3: the sensitivity and specificity of different cut-off values in BMUP3. .