Population Characteristics
The study flowchart is provided in Figure 1 and patients characteristics listed in Table 1. Thirty-four consecutive patients (10 females and 24 males; median age of 15.1 years; 7.4-19.7 years) were enrolled from July 2010 to November 2016: 17/34 (50%) had Osteosarcoma, 13/34 (38%) had Ewing Sarcoma (4/13 (31%) soft tissue and 9/13 (69%) bone), 2/34 (6%) had synovial sarcoma and 2/34 (6%) had embryonal sarcoma of the liver. According to the revised American Joint Committee on Cancer staging system (AJCC) (15), 16/34 (47%) patients were classified as stage IIb, 11/34 (32%) as stage IIa and 7/34 (21%) had skip or distant metastasis at diagnosis (stage III, IV; 1 skip lesion, 1 distant bone metastasis and 5 lung metastases).
Therapy protocols
All 17 patients with osteosarcoma were treated according to the Children’s Oncology Group (COG) Protocol with 10 weeks of neo-CTX (high-dose methotrexate, doxorubicin and cisplatin) before surgery (16 resections with reconstruction/replacement, 1 amputation) followed by 18 weeks of adjuvant-CTX (high-dose methotrexate, doxorubicin and cisplatin) for low-risk patients or 29 weeks of adjuvant-CTX with addition of ifosfamide and etoposide in high-risk patients (17,18). Ewing sarcoma patients (n=13) were treated as per COG AEWS1031 protocol composed by an initial 12 week course of interval compression CTX (Vincristine, Doxorubicin, Cytoxan, alternating with Ifosfamide and Etoposide) followed either by surgical removal of the tumor (8/13 (62%) or definitive RT (5/13 (38%)). Following definitive therapy, CTX was continued for approximately 6 months (consolidation). Patients with synovial sarcoma (n=2) and embryonal sarcoma of the liver (n=2) were treated with doxorubicin and ifosfamide according to COG ARST0332 combined with surgery and radiation (19). The treatment schemas for patients with osteosarcoma, synovial sarcoma, and embryonal sarcoma of the liver are shown in Figure 2. The treatment schema for patients with Ewing Sarcoma is not included as the data from that study is not yet published.
Follow-up and therapy response assessment
Median follow-up was 39 months (range 16-84). 8/34 patients (24%) died from cancer-related causes while 9/34 (27%) were alive with disease and 17/34 (50%) had no evidence of residual/recurrent disease at the last follow-up. The median time between primary therapy and the disease relapse was 14.4 months (4.2 – 53.4 months) and the median time between primary therapy and cancer-related death was 33.6 (19.7 – 73.5 months). The median OS was 71 months while the median TTP was 33.5 months. The shortest follow-up in patient who did not show disease progression was 16 months. In 7/34 patients with metastatic disease (21%) the median OS and TTP was 35 and 13.2 months, respectively (3/7 died (43%), 4/7 were alive with disease (57%) at last follow-up).
Tumor tissue response to neo-CTX was evaluated in all patients (necrosis> 90% at histopathology of excised tumors in 29 and by MRI in 5 patients). Fifteen of 34 patients (44.1%) were classified as responders (15/15 by histopathology evaluation) while 19/34 patients (55.9%) were considered non-responders (14/34 by histopathology and 5/34 by MRI evaluation). The average percentage of CTX induced tumor necrosis was 68%, ranging from 5% to 99%. Six deaths were reported among the non-responders (n=14, median OS=72 months), while no events were observed in the responders group (n=15, median OS=not reached).
[18F]FDG PET/CT findings
Primary tumors were identified on [18F]FDG PET/CT in all patients. Twenty-seven /34 patients (79%) had localized disease while 7/34 patients (21%) had metastatic disease (Figure 3). These included 2 patients with Ewing sarcoma (left femur with left iliac bone metastasis; 1 patient with Ewing sarcoma of the chest wall (11th rib) with vertebral body metastasis). Five patients (3 osteosarcomas, 1 synovial and 1 liver embryonal sarcoma) showed sub-centimeter bilateral lung nodules with faint [18F]FDG uptake. All metastatic lesions seen at PET1 showed partial or complete metabolic response at PET2. No new metastatic lesion at PET2 was observed.
All primary tumors had increased [18F]FDG uptake. Baseline SUVmax, SUVmean, SUVpeak, MTV and TLG median values were 7.9 (5.3-10.6 IQR), 3.8 (3.0-4.9 IQR), 6.0 (4.0-7.4 IQR), 161.0 (85.2-262.4 IQR) and 104.7 (54.8-259.9 IQR), respectively. In PET2 the same parameters decreased to 3.1 (2.2-4.0 IQR), 2.0 (1.4-2.7 IQR), 2.4 (1.8-3.4 IQR), 72.0 (34.8-131.8 IQR) and 45.4 (19.4-105.5 IQR), respectively. PET1, PET2-parameters and their changes are listed in Table 2. Figure 4 and figure 5 show two examples of patients responding and non-responding to neo-CTX, respectively.
[18F]FDG PET parameters for prediction of survival
None of PET1-parameters or the ΔPET were significantly associated with TTP in the univariate Cox regression model (Table 3). All PET2-parameters were significantly associated with TTP (Table 3, p< 0.02): patients with higher SUVmax, SUVmean, SUVpeak, MTV and TLG on PET2 had earlier disease progression than those patients presenting with lower values. TTP was shorter in patients with PET2-SUVmax >3.1 (median PET2-SUVmax) (Figure 6, p=0.016) and with PET2-SUVpeak >2.4 (median value) (p=0.02), PET2-SUVmean >1.9 (median value) (p=0.007), PET2-MTV >131.8 (upper quartile value) (p=0.01) and PET2-TLG >105.5 (upper quartile) (p=0.002). All these data are summarized in Supplemental Figure 1. An association was observed for ΔSUVmean (p=0.017), ΔMTV (p=0.028) and ΔTLG (p=0.031) with the relapse status at last follow-up (Supplemental Table 1). None of the PET1, PET2 and ΔPET parameters were associated with OS (Supplemental Table 2). Due to the low number of events (8/34 patients died while 17/34 had a recurrent/progressive disease), multivariate Cox analysis was not performed. Finally, response to neo-CTX has been evaluated using PET parameters, applying EORTC criteria (18). One patient was considered as complete responder according to PET EORTC criteria, 23/34 as partial responder, 4/34 as stable disease and 3/34 as progressive disease. No statistically significant association has been observed between PET EORTC response to therapy criteria and TTP (p=0.59) or OS (p=0.94).
[18F]FDG PET parameters for prediction of pathological response
No significant association between PET1 and PET2-parameters and the response to neo-CTX was observed (Table 4). Despite lack of statistical significance, the response rate was higher in patients with PET2-SUVmax <3.1 (median value) (60% vs 21%; p>0.05). An association between ΔMTV (p=0.037) and response to neo-CTX was observed while other ΔPET parameters did not show significant associations (Table 4). No statistically significant associations have been observed between PET EORTC response to therapy criteria and the pathological response to neo-CTX. These results are summarized in Supplemental Table 3.
Bone Sarcoma Only Sub-population
Considering the presence of multiple tumor types in our pediatric population, a post-hoc sub-analysis was performed in patients with bone sarcoma only (n=26; osteosarcoma=17; Ewing sarcoma=9). In this bone sarcoma sub-population, all PET2 parameters showed a statistically significant association with TTP, confirming the results also observed in the full study population (Supplemental Table 4).