Omitting Bone Marrow Sampling in FDG-Avid Rhabdomyosarcoma With 2-[¹⁸F]FDG PET-Negative Marrow: Results From an International Retrospective Study

Abstract

PURPOSE

Detection of bone marrow disease has been a major component of rhabdomyosarcoma (RMS) staging with bilateral bone marrow aspirates/biopsies (BMAB) from the iliac crests considered the diagnostic gold standard. This study's goal was to determine if 2-¹⁸F-labeled fluorodeoxyglucose positron emission tomography/computed tomography ([¹⁸F]FDG PET/CT) or PET/magnetic resonance imaging (MRI) may replace BMAB in RMS staging.

METHODS

Patients were recruited in Europe and North America. Medical records, FDG PET/CT or FDG PET/MRI (PET), and BMAB reports were collected retrospectively. Images were re-reviewed by nuclear medicine physicians at the patients' home institutions if PET reports indicated increased FDG uptake at bone/bone marrow sites or if tumor cells were detected in bone marrow samples.

RESULTS

PET imaging and BMAB were performed before chemotherapy and tumor resection in 301 patients with FDG-avid RMS. Bone marrow metastases were detected by PET and BMAB in 54, by PET only in 24, and by BMAB only in one of 301 patients with FDG-avid primary tumors. Absence of bone marrow metastases was confirmed by PET and BMAB in 222 of 301 patients. These observations translated into 98% sensitivity and 90% specificity for PET in detecting bone marrow metastases when BMAB was considered as gold standard. FDG-avid marrow disease was patchy (1-5 foci) in one third of cases. Patients with marrow disease indicated by PET and BMAB had more FDG-avid bone marrow foci, more metastatic sites, and lower survival than those with marrow disease indicated by PET only, suggesting more advanced disease stages.

CONCLUSION

Bone marrow sampling may be omitted in patients without evidence of FDG-avid bone marrow disease by PET. If there are FDG-avid bone marrow lesions, further investigation by MRI and/or biopsy should be considered.

INTRODUCTION

Rhabdomyosarcomas (RMS) are the most common soft tissue sarcomas in children and adolescents. Clinical behavior is diverse.^1-4 Approximately 6% of all patients with RMS⁵ and 51% of those with metastatic RMS^4,5 present with bone marrow or bone metastases at diagnosis. Their 3-year event-free survival is 14%-17% compared with 34%-37% in patients with metastatic disease but without marrow or bone involvement.^4,5

CONTEXT

• Key Objective

• Are routine bone marrow aspirates/biopsies (BMAB) necessary for detecting bone marrow involvement in rhabdomyosarcoma (RMS) staging?

• Knowledge Generated

• Fluorodeoxyglucose (FDG)-positron emission tomography/computed tomography (PET/CT) or FDG-PET/magnetic resonance imaging demonstrated high sensitivity (98% in FDG-avid tumors) and 90% specificity for detecting bone marrow involvement in RMS. PET indicated morphologically confirmed bone marrow involvement in all but one FDG-avid case, whereas concordant PET-positive/BMAB-positive findings were associated with more advanced disease.

• Relevance (R.G. Maki)

• This study confirms the utility of FDG-PET in detecting metastatic disease in RMS patients non-invasively. Though there remain research considerations for bone marrow biopsy testing in RMS, it appears that in the substantial majority of RMS patients, what you see is what you get.*

• *Relevance section by JCO Associate Editor Robert G. Maki, MD, PhD, FACP, FASCO.

Historically, detection of marrow disease has been an important component of RMS staging. Based on the assumption that RMS cells spread through the marrow and may be detected by focal sampling, bilateral bone marrow aspirates and biopsies (BMAB) obtained from iliac crests have been considered the gold standard.⁶ Yet, heterogeneous marrow involvement has been well-documented and may limit the sensitivity of this approach.⁷ Also, BMAB are invasive procedures associated with stress/pain and in children require anesthesia. Performing BMAB in all patients with RMS has been called into question, as retrospective analyses detected no marrow metastases in patients with embryonal histology without local invasion or lymph node/lung metastases.⁸

In recent years, 2-[¹⁸F]fluoro-2-deoxy-D-glucose (FDG) positron emission tomography coregistered to either computed tomography (FDG PET/CT) or magnetic resonance imaging (FDG PET/MRI) has been established as a sensitive and specific method for detecting marrow involvement in various cancer types, including Hodgkin lymphomas.⁹ In pediatric patients with sarcoma, FDG PET/CT detection of primary tumors, lymph node, and bone metastases reached >90% sensitivity.^10-12 FDG PET/CT appeared to be especially sensitive in detecting increased FDG uptake at skeletal sites.¹³ Cortical destruction and soft tissue masses have emerged as reliable signs of bone metastases. Skeletal FDG uptake without these features indicates marrow disease without bone involvement⁷ (Data Supplement, Fig S1, online only).

The study reported here builds on an international collaboration facilitated by the International Soft Tissue Sarcoma Consortium (INSTRuCT)^14-16 to address the hypothesis that PET/CT and PET/MRI may replace BMAB in RMS staging. The primary objective was to determine the sensitivity/specificity of FDG PET in detecting marrow disease using BMAB as the gold standard. Secondary objectives were to evaluate FDG avidity of primary tumors and explore marrow disease patterns in RMS.

METHODS

Study Design

This was a multicenter retrospective study facilitated by INSTRuCT.^14-16 Eligibility criteria were histologically proven RMS (≤40 years), availability of FDG PET/CT or FDG PET/MRI, and direct marrow assessment at diagnosis. All patients underwent BMAB before starting chemotherapy.

Marrow Disease Detected by PET

Marrow involvement was defined by increased FDG uptake at skeletal sites without cortical bone destruction/soft tissue masses. FDG-avid lesions showing cortical disruption/soft tissue involvement, but no marrow uptake, were classified as bone metastases. FDG-avid foci with cortical destruction/soft tissue masses and marrow uptake were considered as combined bone marrow and bone disease. Uniform, homogenous FDG uptake throughout the marrow space, without focal abnormalities, was considered physiologic or reactive and not consistent with malignant infiltration. Marrow lesions solely defined by increased FDG uptake were not confirmed by biopsy.

Study Patients

Children and adolescents were recruited at 18 institutions across Europe and North America (Data Supplement, Table S1). All institutions were required to obtain ethics approval. Written consent from patients/guardians was obtained from the institution or as part of a prior clinical trial. The Data Supplement (Table S2) lists the data elements that were recorded. A total of 287 patients were recruited directly at partner institutions. Ninety-one patients were included in the SoTiSaR registry of the Cooperative Weichteilsarkom Studiengruppe (CWS).¹⁷ Forty-five patients were registered in the European pediatric Soft tissue sarcoma Study Group (EpSSG) clinical trial MTS-2008.^18,19 Marrow data from patients registered in MTS-2008 were previously reported by Mercolini et al.¹¹ Data were collected from 423 patients in total. Ultimately, 403 patients were considered in the analyses (Fig 1). One PET scan was considered per patient.

FIG 1.

Flow diagram. A total of 620 patients were screened, including 213 SoTiSaR patients, 120 MTS-2008 patients, and 287 patients identified from COG-, EpSSG, and CWS-affiliated partner institutions. Among SoTiSaR patients, 71 had BMAB only, 13 had PET only, and 38 had neither; these were excluded. It was not possible to obtain scan reviews for 75 MTS-2008 patients; these were also excluded. Twenty patients were excluded, because their PET was performed >30 days after chemotherapy start (n = 4). Their BMAB was obtained after chemotherapy start (n = 14) or both (n = 2). Consequently, 403 patients who underwent BMAB before chemotherapy and PET within 30 days of chemotherapy initiation were considered for analyses. The main cohort comprised 313 patients who had BMAB before chemotherapy and PET before both chemotherapy and tumor resection (n = 272 with data on progression; n = 307 with data on overall survival; one patient died at diagnosis). Among the remaining 90 patients, PET was performed before chemotherapy but after tumor removal (n = 51) 1-30 days after chemotherapy start and before tumor removal (n = 38), or 1-30 days after chemotherapy start and after tumor removal (n = 1). In the main cohort, 301/313 patients had FDG-avid primary tumors. BMAB, bone marrow aspirates/biopsies; COG, Children's Oncology Group; CWS, Cooperative Weichteilsarkom Studiengruppe; EpSSG, European pediatric Soft tissue sarcoma Study Group; FDG, fluorodeoxyglucose; PET, positron emission tomography.

FDG PET/CT Findings

The original PET reports available in the medical records were reviewed for all patients. Non-English language reports were reviewed by native speakers or translated using an artificial intelligence tool. FDG PET/CT or FDG PET/MRI images were re-reviewed by board-certified nuclear medicine/radiology physicians according to predefined criteria at the patients' home institutions if PET reports indicated increased FDG uptake at bone/bone marrow sites or if tumor cells were detected in marrow samples. PET images were not re-reviewed if neither the original PET nor the original marrow reports indicated bone/bone marrow metastases. Illustrative PET/CT images were obtained from seven patients with newly diagnosed RMS.

Bone Marrow Aspirates/Biopsies

For patients recruited at individual institutions, detailed information on BMAB quality and results were obtained from the original BMAB reports. For EpSSG MTS-2008 patients, data were obtained from the MTS-2008 database. Illustrative images of marrow smears were obtained from seven patients with newly diagnosed RMS.

Statistics

Patient demographics were described by median/range for continuous and by absolute/relative frequencies for categorical variables. Disease characteristics, PET, and marrow data were compared between patients with marrow disease indicated by PET only compared with those with marrow disease indicated by PET and confirmed by BMAB using Cochran-Armitage trend tests for ordinal and Fisher exact tests for nominal variables. Overall survival (OS) was calculated as the date of diagnosis to the date of death; patients who did not die during the follow-up period were censored at their last follow-up. Progression-free survival (PFS) was calculated from the date of diagnosis to the date of the first event, defined as tumor progression, relapse, second malignancy, or death as a result of any cause. Cox regression was applied for group comparison and estimation of hazard ratios with corresponding 95% CIs. All statistical tests were performed two-sided on a significance level of 5%.

RESULTS

Patient Characteristics

Within the main cohort (Fig 1), the median patient age was 8.9 years (range, 0-39.8 years). 34.8% of tumors were FOXO1-rearranged. Frequent primary sites were the extremities (n = 85, 27.2%), parameningeal (n = 54, 17.3%), genitourinary (n = 53, 16.9%), and orbit and nonparameningeal head/neck (n = 50, 16%). 46.6% of patients had metastatic disease (n = 146). Common sites of metastases were bone and/or bone marrow (n = 89, 28.4%), distant lymph nodes (n = 48, 15.3%), and lungs (n = 45, 14.3%; Table 1, Data Supplement, Table S3).

TABLE 1.

Characteristics of Patients Whose FDG PET/CTs and FDG PET/MRIs Took Place Before the Start of Chemotherapy and Removal of the Tumor Compared With Those Whose PETs Were Performed After the Start of Chemotherapy or After the Removal of the Tumor

Characteristic	PET Before Surgery and Chemo (n = 313), No. (%)	PET After Chemo Starta and Before Tumor Surgery (n = 38), No. (%)	P	PET Before Chemo and After Tumor Surgery (n = 51), No. (%)	P
Age at diagnosis			.15d		.23d
<12 months	7 (2.2)	1 (2.6)		1 (2.0)
≥12 months, ≤10 years	158 (50.5)	24 (63.2)		21 (41.2)
≥10 years	148 (47.3)	13 (34.2)		29 (56.9)
FOXO1 fusion status			.54c		<.0001c
FOXO1-rearranged	109 (34.82)	10 (26.3)		3 (5.9)
Negative	127 (40.6)	16 (42.1)		34 (66.7)
Unknownb	77 (24.6)	12 (31.6)		14 (27.5)
Primary tumor site			.67c		<.0001c
Extremity	85 (27.2)	9 (23.7)		2 (3.9)
Head/neck (non-pm)	41 (13.1)	2 (5.3)		9 (17.6)
Orbit	9 (2.9)	2 (5.3)		2 (3.9)
Parameningeal	54 (17.3)	5 (13.2)		2 (3.9)
GU, bladder/prostate	30 (9.6)	4 (10.5)		1 (2.0)
GU, nonbladder/prostate	23 (7.4)	3 (7.9)		31 (60.8)
Nonspecifiedb	71 (22.7)	13 (34.2)		4 (7.8)
Oberlin factors			.16d		.91d
0	7 (2.2)	3 (7.9)		0 (0)
1	40 (12.8)	5 (13.2)		3 (5.9)
2	36 (11.5)	7 (10.5)		1 (2.0)
3	47 (15.0)	3 (7.9)		4 (7.8)
4	16 (5.1)	2 (5.3)		0 (0)
NA (nonmetastatic)b	167 (53.4)	18 (47.4)		43 (84.3)
Head-to-toe scan			1.0c		.06c
Yes	222 (70.9)	25 (65.8)		30 (58.8)
No	75 (24)	8 (21.1)		19 (37.3)
Unknownb	16 (5.1)	5 (13.2)		2 (3.9)
FDG-avidity of tumor			.60c		NA
FDG-avid	301 (96.2)	36 (94.7)		0 (0)
Not FDG-avid	7 (2.2)	1 (2.6)		0 (0)
No data availableb	5 (1.6)	1 (2.6)		51 (100)
FDG uptake in bone/bone marrow			.62c		.01c
Bone only	11 (3.5)	1 (2.6)		0 (0)
Bone marrow only	51 (16.3)	3 (7.9)		2 (3.9)
Bone and bone marrow	27 (8.6)	4 (10.5)		3 (5.9)
Normal	205 (65.5)	24 (63.2)		46 (90.2)
Not availableb	19 (6.1)	6 (15.8)		0 (0)
Number FDG-avid bone marrow foci			.69d		.04d
None	231 (73.8)	30 (78.9)		46 (90.2)
One focus	12 (3.8)	0 (0)		1. (2.0)
2-5 foci	15 (4.8)	2 (5.3)		0 (0)
>5 foci	46 (14.7)	5 (13.2)		4 (7.8)
Unknownb	9 (2.9)	1 (2.6)		0 (0)
Bone marrow assessment			.07c		.72c
Aspirate and biopsy	253 (80.8)	25 (65.8)		40 (78.4)
Aspirate only	57 (18.2)	13 (34.2)		11 (21.6)
Biopsy only	3 (1)	0 (0)		0 (0)
Tumor cells			1.0c		.16c
Detected	57 (18.2)	7 (18.4)		5 (9.8)
Not detected	256 (81.8)	31 (81.6)		46 (90.2)

NOTE. All patients underwent BMAB before the start of chemotherapy.

Abbreviations: BMAB, bone marrow aspirates/biopsies; CT, computed tomography; FDG, fluorodeoxyglucose; MRI, magnetic resonance imaging; NA, not available; PET, positron emission tomography.

^a PET performed within 1-30 days after starting chemotherapy.

^b Not included in statistical comparisons.

^c Statistical significance was tested by Fisher exact test.

^d Statistical significance was tested by Cochran Armitage Trend test.

The disease characteristics of 38 patients whose PETs were performed after chemotherapy and before tumor removal (Fig 1) were similar to those in the main cohort (Table 1, Data Supplement, Tables S3 and S6). By contrast, those with PET scans conducted before chemotherapy but after tumor removal (n = 51, Fig 1) included more patients with embryonal histology tumors (P < .0001), tumors without FOXO1-rearrangements (P < .0001), tumors arising at genitourinary, nonbladder/prostate sites (P < .0001), and without metastatic disease (P < .001; Table 1, Data Supplement, Tables S3 and S6).

PET Findings

Scan reviews were performed for 193 patients. The re-review agreed with the original report for 182 scans (94%). For eight patients, the original report described unspecific marrow uptake. Upon re-review, FDG-avid foci were confirmed in five patients; three patients were considered free of FDG-avid marrow lesions. For one patient, the PET report indicated FDG-avid marrow disease; the re-review was consistent with homogenous marrow activation. For two patients, the report described no FDG-avid marrow foci, but the re-review indicated FDG-avid marrow disease.

For the main cohort, 295 (94.3%) of 313 patients underwent PET/CT, and 18 of 313 (5.8%) patients underwent PET/MRI. For 301 of 313 (96.2%) patients, the scans depicted an FDG-avid primary tumor. Seven tumors (2.2%) were not FDG avid. For five tumors, data were not available. Seventy-five (24%) of 313 PETs did not cover patients from head to toe (including one FDG-negative parameningeal tumor). The body coverage of 16 (5%) of 313 patients was not known (including one FDG-negative orbital tumor). Abnormal FDG uptake was noted in bone only in 11 (3.5%), in bone marrow only in 51 (16.3%), and in both bone and bone marrow in 27 (8.6%) patients, respectively (Table 1, Data Supplement, Table S4).

FDG-avid marrow disease rates were similar (P = .62) in patients whose PET was performed after chemotherapy and before tumor removal (n = 38, Table 1, Data Supplement, Tables S4 and S6). By contrast, there were fewer patients with FDG-avid marrow foci (P = .01) among those whose PET was conducted before chemotherapy and after tumor removal (n = 51, all tumors FDG-unknown, Table 1, Data Supplement, Tables S4 and S6).

Bone Marrow Findings

Aspirates and biopsies were obtained from 253 (80.8%), aspirates only from 57 (18.2%), and biopsies only from three (1%) of 313 patients (Table 1). The median length of left-sided core biopsies was 1.1 cm (range, 0.1-3.1 cm); the median length of right-sided core biopsies was 1 cm (range, 0.1-2.5 cm). Bilateral cores ≥0.5 cm were obtained in 151 of 313 (48.2%) patients. Of 310 aspirates, 247 were obtained bilaterally and 15 unilaterally (no data available for 48 aspirates). Of 256 biopsies, 187 were obtained bilaterally and 34 unilaterally (no data available for 35 biopsies). Tumor cells were detected in marrow samples from 57 patients (18.2%). Marrow disease was detected by aspirates and biopsies in 36 (63.2%), by aspirates only in 10 (17.5%) and by biopsies only in 11 (19.3%) of 57 patients. Marrow sampling data were similar among patients whose PET was performed after chemotherapy or after tumor removal (Table 1, Data Supplement, Tables S5 and S6).

Diagnostic Performance of BMAB versus FDG PET/CT or FDG PET/MRI

Historically, bilateral BMAB from iliac crests has been considered the gold standard for marrow examination in patients with RMS (Fig 2). Marrow metastases were detected by PET and BMAB in 54, by PET only in 25, and by BMAB only in 3 of 313 patients whose PET was performed before treatment start. Absence of marrow metastases was confirmed by PET and BMAB in 231 patients. Based on tumor detection by BMAB as gold standard, PET indicated marrow disease with 95% sensitivity (95% CI, 85 to 99), 99% negative predictive value (95% CI, 96 to 100), 90% specificity (95% CI, 86 to 94), and 68% positive predictive value (95% CI, 57 to 78) (Fig 2A, Data Supplement, Table S7).

FIG 2.

Diagnostic performance of BMAB versus FDG PET/CT and FDG PET/MRI in the detection of bone marrow disease in patients with newly diagnosed RMS. All patients underwent BMAB before the start of chemotherapy. FDG PET/CTs or PET/MRIs took place before the start of chemotherapy and before the removal of the primary tumors. BMAB results were considered as gold standard for the detection of bone marrow disease. (A) All 313 patients in the main group were considered, including 301 patients with FDG-avid RMS, seven patients with FDG-negative RMS, and five patients with FDG-unknown RMS. (B) Only patients with FDG-avid primary tumors (n = 301) were considered. BMAB, bone marrow aspirates/biopsies; CT, computed tomography; FDG, fluorodeoxyglucose; MRI, magnetic resonance imaging; NPV, negative predictive value; PET, positron emission tomography; PPV, positive predictive value.

If only patients with FDG-avid primary tumors were considered, marrow metastases were detected by PET and BMAB in 54, by PET only in 24, and by BMAB only in one of 301 patients. Absence of marrow metastases was confirmed by PET and BMAB in 222 patients. Sensitivity and negative predictive value reached 98% (95% CI, 90 to 100) and 100% (95% CI, 98 to 100), respectively (Fig 2B, Data Supplement, Table S7). Tumor cells were detected in marrow samples from only one patient with FDG-avid RMS and no FDG-avid marrow foci. Re-review confirmed the presence of scarce, myogenin-expressing tumor cells in this patient.

For 38 patients whose PET was performed after chemotherapy and before tumor removal, the sensitivity of detecting marrow disease by PET dropped to 86% (95% CI, 42 to 100, Data Supplement, Table S7). For those whose PET was obtained before chemotherapy and after tumor removal (n = 51), sensitivity dropped to 60% (95% CI, 15 to 95, Data Supplement, Table S7).

As diagnostic performance measures depend on the underlying gold standard, we recalculated these parameters using any indication of marrow disease by BMAB or PET as reference, which translated into 99% sensitivity (95% CI, 93 to 100) and 100% negative predictive value (95% CI, 98 to 100) of PET for patients with FDG-avid primary tumors (Data Supplement, Table S7). Finally, we chose to recalculate the sensitivity and negative predictive value of BMAB using any indication of marrow disease by BMAB or PET as reference, resulting in inferior sensitivity (70%, 95% CI, 58 to 80) and negative predictive of BMAB (90%, 95% CI, 86 to 94, Data Supplement, Table S7).

Disease Characteristics and Survival Among Patients With FDG-Avid Marrow Disease

OS and PFS were analyzed in the main cohort of 313 patients. Data on OS and PFS were available for 307 and 272 patients, respectively.

Compared with those without marrow disease, PFS was lower for patients with marrow metastases indicated by PET only (n = 23; hazard ratio [HR], 1.90 [95% CI, 1.08 to 3.36]) and by PET and BMAB (n = 49; HR, 4.51 [95% CI, 3.06 to 6.65]; Fig 3A). Similarly, OS was lower for patients with marrow metastases indicated by PET only (n = 25; HR, 2.71 [95% CI, 1.51 to 4.85]) and by PET and BMAB (n = 53; HR, 6.88 [95% CI, 4.57 to 10.36]; Fig 3B). Survival estimates were uncertain for patients with marrow disease indicated by BMAB only (n = 3 patients, including two patients with FDG-negative and FDG-unknown primary tumors, respectively).

FIG 3.

(A) PFS and (B) OS of patients without any evidence for bone marrow disease, with bone marrow disease by PET, bone marrow disease by BMAB, and bone marrow disease by PET and BMAB. Within the group of patients whose BMAB were obtained before chemotherapy start and whose PETs were performed before chemotherapy start and tumor removal (n = 313), data on disease progression and OS were available for 272 and 307 patients, respectively. One patient died shortly after diagnosis (time point 0 months). Patients were censored at their last known encounter because of loss to follow-up or end of data collection. BMAB, bone marrow aspirates/biopsies; HR, hazard ratio; OS, overall survival; PET, positron emission tomography; PFS, progression-free survival

Interestingly, OS and PFS of patients with marrow disease indicated by PET and BMAB appeared to be substantially worse than OS and PFS of those with marrow metastases indicated by PET only. Disease characteristics, PET, and marrow data—including Oberlin scores—were mostly similar (Table 2, Data Supplement, Tables S8-S10). However, a trend toward a higher number of metastatic sites (P = .02), FDG-avid marrow foci (P = .02), and marrow aspirates without megakaryocytes (P = .01) was noted in those with marrow metastases indicated by PET and BMAB (Table 2).

TABLE 2.

Comparison of Patients With Bone Marrow Disease Indicated by BMAB Only, by BMAB and by PET, and by PET Only

Characteristic	Bone Marrow Disease Indicated by BMAB Only (n = 3), No. (%)	Bone Marrow Disease Indicated by BMAB and PET (n = 54), No. (%)	Bone Marrow Disease Indicated by PET Only (n = 25), No. (%)	P (BMAB and PET v PET only)
Age at diagnosis				.50b
<12 months	0 (0)	0 (0)	0 (0)
≥12 months, ≤10 years	2 (66.7)	17 (31.5)	6 (24.0)
>10 years	1 (33.3)	37 (68.5)	19 (76.0)
FOXO1 fusion status				.74c
FOXO1-rearranged	0 (0)	33 (61.1)	15 (60)
Negative	1 (33.3)	8 (14.8)	5 (20)
Unknowna	2 (66.7)	13 (24.1)	5 (20)
No. of metastatic sites				.02b
0	0 (0)	0 (0)	2 (8.0)
1	3 (100)	8 (14.8)	5 (20.0)
2-4	0 (0)	40 (74.1)	18 (72.0)
≥5	0 (0)	6 (11.1)	0 (0)
Lung metastases				.6c
Yes	0 (0)	11 (20.4)	7 (28.0)
No	3 (100)	43 (79.6)	18 (72.0)
Oberlin factors				.8b
1	2 (66.7)	4 (7.4)	3 (12.0)
2	0 (0)	12 (22.2)	2 (8.0)
3	1 (33.3)	28 (51.9)	13 (52.0)
4	0 (0)	10 (18.5)	5 (20.0)
NA (non-metastatic)a	0 (0)	0 (0)	2 (8.0)
No. of FDG-avid bone marrow foci				.02b
None	3 (100)	0 (0)	0 (0)
One focus	0 (0)	5 (9.3)	7 (28.0)
2-5 foci	0(0)	9 (16.7)	6 (24.0)
>5 foci	0 (0)	35 (64.8)	11 (44.0)
Unknowna	0(0)	5 (9.3)	1 (4.0)
Bone marrow assessment				.19b
Aspirate and biopsy	2 (66.7)	46 (85.2)	18 (72.0)
Aspirate only	1 (33.3)	7 (13)	7 (28.0)
Biopsy only	0 (0)	1 (1.9)	0 (0)
Megakaryocytes in aspirates	1 (33.3)	29 (53.7)	21 (84.0)	.01c
Present	0 (0)	10 (18.5)	0 (0)
Absent—unknowna	2 (66.7)	14 (25.9)	4 (16.0)
No aspirates performeda	0 (0)	1 (1.9)	0 (0)

NOTE. All patients underwent BMAB before the start of chemotherapy. FDG PET/CT or PET/MRI imaging took place before the start of chemotherapy and before the removal of the primary tumors.

Abbreviations: BMAB, bone marrow aspirates/biopsies; CT, computed tomography; FDG, fluorodeoxyglucose; MRI, magnetic resonance imaging; PET, positron emission tomography.

^a Not included in statistical comparisons.

^b Statistical significance was tested by Cochran Armitage Trend test.

^c Statistical significance was tested by Fisher exact test.

If FDG-avid marrow foci had been considered, the Oberlin scores of three patients would have changed. In a 12-year-old child with embryonal RMS of the head/neck and a 16-year-old child with FOXO1-rearranged RMS of the extremities, no other metastases were identified. Consideration of FDG-avid marrow foci would have resulted in Oberlin scores of two and three, respectively. Another 6-year-old child with FOXO1-rearranged RMS of the extremities had been assigned an Oberlin score of one; consideration of FDG-avid marrow foci would have increased the score to two. One of these three patients died of disease (corrected Oberlin score 3).

Patterns of Marrow Disease in Rhabdomyosarcoma

PET allows for an assessment of the extent of FDG-avid marrow disease. Among 79 patients with FDG-avid marrow disease, 12 (15.2%) had one FDG-avid marrow focus, 15 (19%) had 2-5 FDG-avid foci, and 46 (58.2%) had >5 FDG-avid foci (including three patients with diffuse FDG-avid marrow disease). For six patients, no information on marrow disease pattern was available.

Figure 4 illustrates patterns of findings in seven children and adolescents with newly diagnosed RMS, including five patients with FDG-avid marrow foci detected by PET (Figs 4A-4E). RMS cells were detected in BMAB in four patients; all had FDG-avid marrow lesions in their iliac crests (Figs 4A-4D). BMAB were free of tumor cells in three patients without FDG-avid marrow foci in their iliac crests (Figs 4E-4G), including one patient with FDG-avid lesions in several other bones (Fig 4E). The latter case was described in detail by Kroening et al.⁷

FIG 4.

Detection of bone marrow disease by BMAB or PET in patients newly diagnosed with RMS. (A) An 8-year-old female patient with FOXO1-rearranged RMS. FDG-avid bone marrow lesions were noted in both iliac bones. Tumor cells were present in bilateral bone marrow aspirates and biopsies. (B) An 8-year-old male patient with FOXO1-rearranged RMS. FDG-avid bone marrow lesions were noted in both iliac bones. Tumor cells were present in unilateral, left-sided bone marrow aspirates and biopsies. (C) An 11-year-old female patient with FOXO1-rearranged RMS. FDG-avid bone marrow lesions were noted in both iliac bones. Tumor cells were present in bilateral bone marrow aspirates; no bone marrow biopsies were performed. (D) A 5-year-old female patient with FOXO1-rearranged RMS. FDG-avid bone marrow lesions were noted in both iliac bones. Tumor cells were present in bilateral bone marrow aspirates and a unilateral, right-sided bone marrow biopsy. (E) A 6-year-old male patient with FOXO1 fusion-negative RMS. Multifocal FDG-avid bone marrow lesions were noted, but there were no FDG-avid bone marrow lesions in iliac bones. No tumor cells were present in bilateral bone marrow aspirates; no biopsies were performed. (F) An 18-year-old female patient with FOXO1-rearranged RMS. No FDG-avid bone marrow lesions were noted. No tumor cells were seen in bilateral bone marrow aspirates and biopsies. (G) A 1-year-old male patient with FOXO1 fusion-negative RMS. No FDG-avid bone marrow lesions were noted. No tumor cells were present in unilateral bone marrow aspirates and biopsies. BMAB, bone marrow aspirates/biopsies; FDG, fluorodeoxyglucose; RMS, rhabdomyosarcoma.

DISCUSSION

Findings from this study show that FDG-PET/CT or FDG-PET/MRI indicated marrow disease in patients with RMS with 98% sensitivity and 90% specificity, if PET imaging was conducted before treatment started and if tumors were FDG avid (Fig 2B), supporting omission of BMAB in up-front staging. This would be practice-changing, as the current paradigm requires BMAB in most patients. Yet, marrow sampling may result in anesthesia complications, infection, or bleeding. Scheduling, performing, and interpreting BMAB may delay therapy. Above all, the stress and pain associated with BMAB may be avoided.

PET sensitivity in this study was substantially higher than the 92% sensitivity reported in a previous retrospective assessment of 114 patients registered in MTS-2008.¹¹ We note that 45 of these patients were included in our study. However, no scan reviews were performed in the previous assessment. FDG avidity of tumors and PET timing were not considered. Four of 114 patients had PET-negative/biopsy-positive marrow metastases.¹¹ The lower PET sensitivity in the previous study aligns with the drop in sensitivity observed when FDG-negative cases or patients whose PET was obtained after treatment start were considered in our analysis (Data Supplement, Table S7).

When PET was performed after starting chemotherapy or tumor resection, the sensitivity of detecting marrow disease by PET dropped to 86% and 60%, respectively. This may reflect rapid tumor cell elimination by chemotherapy and the inability to assess FDG avidity after tumor resection. PET should be performed before starting treatment according to European Association of Nuclear Medicine guidelines,²⁰ cover the entire body, and verify primary tumor FDG avidity. Twenty-two (7%) of 313 primary tumors in the main cohort were located at the level/below the knees and would not have been detected by skull base-to-thigh coverage. Moreover, 2.2% of primary RMS tumors were not FDG avid (Table 1), which is consistent with lack of FDG avidity in 2% of adult soft tissue sarcomas.²¹ In such cases, PET allows no conclusions regarding metastases. If PET scans were not performed before treatment, if the primary tumor was FDG-negative, and if the PET did not cover the entire body, bilateral BMAB should be performed.

Limitations of this study include its retrospective approach, skewing toward patients with metastatic and/or FOXO1-rearranged disease and lack of MRI data for most patients. There was one patient with FDG-avid RMS, who had marrow disease indicated by BMAB, but not PET. This patient did not have other metastases and died of disease. Without BMAB, marrow metastases would have gone undetected. However, any recommendation in favor of universal marrow sampling based on a single case must be weighed not only against the burden of stress and pain but also the fact that BMAB were frequently inadequate as per published standards,⁶ with issues including unilateral sampling and/or short biopsy cores (Data Supplement, Table S5). These observations are consistent with published rates of inadequate marrow biopsies.²² If BMAB are adequate, absence of tumor cells does not rule out marrow metastases at other sites, as illustrated by lower sensitivity and negative predictive value of BMAB compared with PET. When BMAB rates were screened in 213 CWS patients treated at seven German centers, we noted that 51 (24%; range, 4%-52% for individual centers) did not undergo marrow sampling (Fig 2). These observations were unexpected, given that a recent international survey suggested that most providers routinely perform bilateral BMAB in patients with RMS (Just et al, article in preparation). Yet, in clinical practice, marrow sampling may be limited by patients' critical illness, restricted anesthesia resources, high patient volumes, and individual physicians' preferences.

Detection of marrow metastases by PET was associated with suboptimal specificity of 90% (Figs 2A and 2B). FDG-avid marrow lesions may not exclusively represent tumor infiltration. Injuries, inflammation, or marrow activation may provide alternative explanations.²³ FDG-avid marrow sites were not biopsied, and relative timing of BMAB and PET was not evaluated for all patients. Nevertheless, it is unlikely that single FDG-avid marrow foci were caused by preceding BMAB in 12 patients with FDG-avid tumors, as marrow disease was confirmed by BMAB in five, and PET was performed before BMAB in the remaining seven patients. Systematic correlation of FDG-avid marrow foci in iliac crests with tumor detection in BMAB was not possible. Yet, FDG-avid marrow disease was patchy (1-5 foci) in 34% of patients, and the illustrative cases depicted in Figure 4 suggest that marrow sampling from iliac crests is less likely to yield positive results, if there are no FDG-avid marrow foci in iliac crests. If FDG-avid marrow lesions are verified, biopsies should be obtained from sites of abnormal FDG uptake.

As diagnostic performance measures depend on the chosen gold standard, we recalculated these parameters using either BMAB or PET as the reference. This analysis demonstrated inferior sensitivity/negative predictive value of BMAB versus PET. Therefore, we propose that tumor detection by BMAB should no longer be considered the gold standard for diagnosing marrow disease in RMS.

Marrow disease in RMS has been long recognized as a predictor of poor outcomes. Yet, the survival differences between patients with marrow disease indicated by PET only as opposed to those with marrow metastases indicated by PET and BMAB were unexpected. Higher number of metastatic sites and FDG-avid marrow foci hint toward more extensive disease states in the latter group. Alternatively, underlying biologic differences may be responsible. We recommend to consider focal sites of increased FDG uptake at marrow sites in patients with RMS as true metastases. This will reshape the definition of marrow disease and influence risk stratification.

High sensitivity and negative predictive value of PET support omission of BMAB in patients without evidence of FDG-avid marrow disease. If FDG-avid marrow lesions are indicated by PET/CT or PET/MRI, further investigation should be considered. This may include focused MRI or BMAB if FDG-avid marrow sites are accessible without undue risk.

APPENDIX. CONTRIBUTOR INFORMATION

TABLE A1.

INSTRuCT Imaging/Pathology/Biology Network

Name	Institution
Giorgio Baldari, MD	Nuclear Medicine Division, Azienda Ospedaliero-Universitaria di Parma, Via Gramsci 14, 43126, Parma, Italy
Isabel Borrego-Dorado, MD	Unidad de Diagnóstico por Imagen (UDIM), Servicio de Medicina Nuclear, Hospital Universitario Virgen del Rocío, Sevilla, España
Nicodemo Baffa, MD	Department of PET-CT and Radiological and Laboratory Imaging Hospital Santa Maria della Misericordia Perugia Italy
Carmelo Caldarella, MD	Nuclear Medicine Unit, Dipartimento di Diagnostica per Immagini e Radioterapia Oncologica, Fondazione Policlinico Universitario Agostino Gemelli IRCCS, Rome, Italy
Edmund Cheesman, MD	Royal Manchester Children's Hospital, Manchester, United Kingdom
Carmen Degitz, MD	Pediatrics I, Martin Luther University Halle-Wittenberg, Halle, Germany
Michael DiGianvittorio, MD	MaineHealth, Maine Medical Center, Portland, ME
Stefano Fanti, MD	Nuclear Medicine, IRCCS, Azienda Ospedaliero-Universitaria Di Bologna, Bologna, Italy
Alexandra Fischer	Department of Pediatric hematology and Oncology, University Medical Center Freiburg, University of Freiburg, Freiburg, Germany
Josef J. Fox, MD	Memorial Sloan Kettering Cancer Center, New York, NY
Giuseppe Arturo Fuso, MD	Nuclear Medicine, IRCCS, Azienda Ospedaliero-Universitaria Di Bologna, Bologna, Italy
Laura Hiemcke-Jiwa, MD	Princess Máxima Center, Utrecht, The Netherlands
Alexander Heinzel, MD	Department for Nuclear Medicine, Martin Luther University Halle-Wittenberg, Halle (Saale), Germany
Sebastian Jacob, MD	Pediatrics I, Martin Luther University Halle-Wittenberg, Halle, Germany
Martha Kleinwechter, cand. med	Pediatrics I, Martin Luther University Halle-Wittenberg, Halle, Germany
Pia Kröning, MD	Department of Pediatric and Adolescent Medicine, University Medical Center Freiburg, University of Freiburg, Freiburg, Germany
Christian Lohrmann, MD	Department of Nuclear Medicine, Klinikum Rechts der Isar, Technical University of Munich, Munich, Germany
Ajaykumar C. Morani, MD	Abdominal Imaging Department, The University of Texas MD Anderson Cancer Center, Houston, TX
Helen Nadel, MD	Department of Radiology, Lucile Packard Children's Hospital at Stanford University, Stanford University, Stanford, CA
Marguerite T. Parisi, MD	Seattle Children's Hospital and University of Washington, Seattle, WA
Katia Perruccio, MD	Pediatric Oncology Hematology, Santa Maria della Misericordia Hospital, Perugia, Italy
Martina Pugliano, MD	Pediatric Oncology Department, Regina Margherita Children's Hospital, Turin, Italy
Carlos Ramos Font, MD	Nuclear Medicine Service, Hospital Universitario Virgen de las Nieves Granada, Granada, Spain
Maria Rosaria Cuomo, MD	Nuclear Medicine, IRCCS Foundation, National Institute of Tumor, Milan, Italy
Livia Ruffini, MD	Nuclear Medicine Division, Azienda Ospedaliero-Universitaria di Parma, Via Gramsci 14, 43126, Parma, Italy
Raya Saab, MD	Department of Pediatrics, Stanford University, Stanford, CA
Maura Scarlattei, MD	Nuclear Medicine Division, Azienda Ospedaliero-Universitaria di Parma, Via Gramsci 14, 43126, Parma, Italy
Jürgen Schäfer, MD	Division of Pediatric Radiology, Department of Radiology, Universitätsklinikum Tübingen, Tübingen, Germany
Kristina Schachtrup, PhD	Clinical Trials Unit, Medical Center University of Freiburg, Faculty of Medicine, University of Freiburg, Germany
Michele Sternberg, JD	Medical Center University of Freiburg, Faculty of Medicine, University of Freiburg, Germany
Markus Uhl, MD	Division of Pediatric Radiology, Department of Radiology, University Medical Center Freiburg, University of Freiburg, Freiburg, Germany
Hein J. Verberne, MD	Department of Radiology and Nuclear Medicine, Amsterdam University Medical Centers, Location AMC, University of Amsterdam, Amsterdam, the Netherlands
Pietro Zucchetta, MD	Unit of Nuclear Medicine, Department of Medicine, Università University Hospital of Padova, 35128 Padua, Italy

AUTHOR CONTRIBUTIONS

Conception and design: Federico Mercolini, Barry L. Shulkin, Douglas Harrison, Maria C Affinita, Sonja Chen, Christine Mauz-Körholz, Suzi Birz, Gianni Bisogno, Martin Ebinger, Aaron R. Weiss, Corinne M. Linardic, Reineke A. Schoot, Simone Hettmer

Administrative support: Suzi Birz, Simone Hettmer

Provision of study materials or patients: Marissa Just, Natalie B. Collins, Jacquelyn Crane, Douglas Harrison, Alessandra Alessi, Sebastian D. Asaftei, Hedieh Khalatbari, Irene von Luettichau, Christine Mauz-Körholz, Claudia Rossig, Stefan Rutkowski, Gianni Bisogno, Martin Ebinger, Johannes HM Merks, Monika Sparber-Sauer, Simone Hettmer

Collection and assembly of data: Federico Mercolini, Marissa Just, Natalie B. Collins, Jacquelyn Crane, Douglas Harrison, Maria C. Affinita, Wendy Allen-Rhoades, Alessandra Alessi, Sebastian D. Asaftei, Michael A. Arnold, Stefano Chiaravalli, Forrest FitzGerald, Martin T. Freitag, Jessica L. Iftner, Bart de Keizer, Irene von Luettichau, Christine Mauz-Körholz, Claudia Rossig, Guido Rovera, Stefan Rutkowski, Stephan D. Voss, Ayami Yoshimi, Gianni Bisogno, Martin Ebinger, Monika Sparber-Sauer, Aaron R. Weiss, Corinne M. Linardic, Reineke A. Schoot, Simone Hettmer

Data analysis and interpretation: Federico Mercolini, Marissa Just, Barry L. Shulkin, Natalie B. Collins, Jacquelyn Crane, Douglas Harrison, Wendy Allen-Rhoades, Michael A. Arnold, Sonja Chen, Martin T. Freitag, Hedieh Khalatbari, Christine Mauz-Körholz, Stefan Rutkowski, Justin B. Sims, Stephan D. Voss, Ayami Yoshimi, Gianni Bisogno, Johannes H.M. Merks, Bernhard Haller, Aaron R. Weiss, Corinne M. Linardic, Simone Hettmer

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Omitting Bone Marrow Sampling in FDG-Avid Rhabdomyosarcoma With 2-[¹⁸F]FDG PET-Negative Marrow: Results From an International Retrospective Study

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).