3. Results
3.1. Study selection
The initial search identified 11,110 titles reviewed (PubMed: 8,324; Google Scholar: 1,208; Cochrane library: 1,578). Reviewer 1 read 330 abstracts, from which 97 articles (including 17 articles screened from references) were flagged for Reviewer 2 (Fig. 1). Seventy-five articles were identified for full-text review. Of these, 37 were deemed ineligible for inclusion due to the following reasons: 18 did not report incidence of BM metastasis; 7 did not utilize FDG-PET scans to assess for BM metastasis; 6 had missing data on diagnostic testing used for metastatic disease; 4 were not in English; 2 involved patients with diseases other than ES.
3.2 Quantitative analysis: Incidence of BM metastasis
We identified 10 studies providing quantitative data describing the incidence of BM metastasis in all unselected patients with newly diagnosed ES (Table 1).[21,23-31] These studies ranged in year of publication from 1995 – 2020. Eight studies were completed in a retrospective manner, one was a phase 3 clinical trial and one was a pilot trial. Combined, they comprise 1,663 patients (median 75 patients/study; range 17 – 647 patients).
Of the total of 1,663 patients, 79 had BM metastasis, for an incidence of 4.8% in this unselected group of patients. Among the 450 patients with any metastatic disease, the incidence of BM disease was 17.5% (79/450). Three studies reported the presence or absence of isolated BM metastases, with 1.2% (8/692) of patients with isolated BM metastasis.[24,25,27] Ten additional patients with isolated BM metastatic disease were identified in the literature for a total of 18. Eight of these 18 cases reported the location of the primary tumor: three pelvic primary tumors and five located in the appendicular skeleton.
We identified an additional 8 studies for which inclusion was restricted to newly diagnosed ES patients with metastatic disease only or a specific subset of metastatic ES. Five of these studies reported the incidence of BM metastasis in a selected population of ES patients (Table 2).[2,32-35] Of these, three excluded patients with newly diagnosed localized ES and reported on 934 total patients with newly diagnosed metastatic ES. Of these, 19.0% were reported to have BM metastasis. Two other studies describe further subsets of metastatic ES patients. One excluded patients with lung only metastatic disease and reported an incidence of BM metastasis of 43.8% (123/281).[2] A second study limited its inclusion to patients with metastatic bone ES and reported an incidence of BM metastasis of 23.7% (23/97).[35]
Two additional studies in patients with metastatic disease reported patients with isolated BM metastatic disease without providing an incidence of any BM metastasis. One study reported 7% (8/110) of patients with newly diagnosed metastatic ES to present with isolated BM metastasis,[36] another reported 3.3% (2/60) incidence of isolated BM metastatic disease.[37]
Patients with isolated lung metastasis have superior outcomes compared to patients with other metastasis.[3] Since detecting BM metastasis in patients who might otherwise have a more favorable outcome may provide important prognostic information, we also assessed the frequency of BM metastases in patients with lung metastasis but lacking bone metastases. We evaluated all studies included in Tables 1 2. Six of the 15 studies analyzed reported relevant data, with 8.8% (20/225) of patients with reported lung metastatic disease also having BM metastasis but not bone metastasis.
3.3 Quantitative analysis: FDG-PET scan vs. BMBA
Four studies (Table 3) were identified which utilized FDG-PET scans for staging of BM metastasis in newly diagnosed ES patients.[21,29,38,39] Data from these studies were aggregated for quantitative analysis to define testing characteristics of FDG-PET scans compared to BMBA. The studies ranged in publication year from 2013 - 2019 and all were completed in a retrospective manner. Combined, these studies comprised 142 patients (median 23 patients/study; range 16-80 patients).
Pooled patient level data were used to construct a contingency table to determine sensitivity, specificity, PPV and NPV. Results revealed FDG-PET scans to have a sensitivity of 100% (3/3) and specificity of 96% (96/100) to detect BM metastasis. Compared to BM biopsy, the PPV of a positive FDG-PET scan was found to be 75% (3/4) and the NPV of a negative FDG-PET scan was 100% (106/106). The patient with a negative BMBA but positive FDG-PET scan was identified by Kasalek et al on FDG-PET scan to have BM metastatic lesions involving bilateral posterior iliac crest and extensive BM involvement elsewhere in the axial and appendicular skeleton. [38]
3.4 Qualitative analysis
Several studies were selected for inclusion in our review based upon qualitative analysis or descriptive findings contributing to the understanding of the incidence of BM disease at presentation in ES as well as the advantages and disadvantages to BMBA and FDG-PET scan in assessment of BM metastatic disease.
The most common reported association for patients with BM metastasis was the presence of other metastatic lesions, and specifically bone metastasis. Kopp et al found BMBA to be positive in 41.9% (13/31) of patients with the presence of other metastatic but 0% (0/85) in ES patients with non-metastatic disease.[26] A positive association of the number of bone metastatic lesions and presence of BM metastasis was reported by Inagaki et al. Of 26 patients with newly diagnosed ES, 100% (3/3) of patients with BM metastasis had bone metastases and 4.3% (1/23) patients without BM metastases had bone metastases.[29] Newman et al reported 0 of 91 ES patients with BM metastases in cases without osseous metastatic disease.[21] Khan et al reported that 45.5% (5/11) of patients with BM mets had concurrent bone metastatic disease.[25]
Additionally, BM metastasis was reported to be associated with primary tumors of the pelvis, large tumor size, and presentation with fever and/or elevated lactate dehydrogenase (LDH) level. Two studies reported that pelvic primary site was significantly associated with BM metastatic disease at diagnosis. One reported a risk of BM involvement three times higher in patients with pelvic lesions compared with other primary sites; the second reported BM involvement in 22% (13/59) of patients with pelvic primary disease.[24,27] Tumor diameter (≥80mm) and LDH level at diagnosis (> x ULN) were reported to be factors significantly related to BM involvement by Inagaki et al.[29] Oberlin et al reported fever at diagnosis was significantly associated with the presence of BM metastasis.[27]
One study analyzed the concordance between BM aspirate (BMA) and BM biopsy (BMB). Among 59 patients with bilateral BMA and 62 patients with bilateral BMB, concordance rates between sides were 97% and 98%, respectively. BMB was concluded to be more sensitive to detect metastatic disease than BMA as 100% (4/4) cases of positive BMA were also found to have positive BMB, three positive BMB procedures had negative ipsilateral BMA.[21] In a study including adult patients with various types of sarcoma, discrepancies in findings of metastatic disease between bilateral samples by morphologic evaluation were found in 29% of (27/94) of cases.[40]