[18F]FDG PET/CT for evaluating early response to neoadjuvant chemotherapy in pediatric patients with sarcoma: a prospective single-center trial

Introduction This is a prospective, single-center trial in pediatric patients with sarcoma aiming to evaluate [18F]FDG PET/CT as a tool for early response assessment to neoadjuvant chemotherapy (neo-CTX). Methods Bone or soft tissue sarcoma patients with (1) baseline [18F]FDG PET/CT within 4 weeks prior to the start of neo-CTX (PET1), (2) early interim [18F]FDG PET/CT (6 weeks after the start of neo-CTX (PET2), (3) evaluation of neo-CTX response by histology or MRI, and (4) definitive therapy after neo-CTX (surgery or radiation) were included. Semiquantitative PET parameters (SUVmax, SUVmean, SUVpeak, MTV and TLG) and their changes from PET1 to PET2 (ΔPET) were obtained. The primary endpoint was to evaluate the predictive value of PET1, PET2 and ΔPET parameters for overall survival (OS) and time to progression (TTP). The secondary outcome was to evaluate if [18F]FDG PET/CT can predict the response to neo-CTX assessed by histopathology or MRI. Primary and secondary outcomes were also evaluated in a subpopulation of patients with bone involvement only. Results Thirty-four consecutive patients were enrolled (10 females; 24 males; median age 15.1 years). 17/34 patients (50%) had osteosarcoma, 13/34 (38%) Ewing's sarcoma, 2/34 (6%) synovial sarcoma and 2/34 (6%) embryonal liver sarcoma. Median follow-up was 39 months (range 16–84). Eight of 34 patients (24%) died, 9/34 (27%) were alive with disease, and 17/34 (50%) had no evidence of residual/recurrent disease. Fifteen of 34 (44%) and 19/34 (56%) were responders and non-responders, respectively. PET2-parameters were associated with longer TTP (p < 0.02). ΔMTV was associated with tissue response to neo-CTX (p = 0.047). None of the PET1, PET2 or ΔPET parameters were associated with OS. Conclusion [18F]FDG PET/CT performed 6 weeks after the start of neo-CTX can serve as an early interim biomarker for TTP and pathologic response but not for OS in pediatric patients with sarcoma.


Introduction
Bone and soft tissue sarcomas are the most common primary bone malignancies in children. Their incidence ranges from 0.2 to 0.3/100,000/year [1,2]. Depending on the histologic sarcoma subtype, patient management may include neo-adjuvant chemotherapy (neo-CTX) and Open Access *Correspondence: jcalais@mednet.ucla.edu † Giulia Polverari and Francesco Ceci contributed equally to this work † Jeremie Calais and Noah Federman contributed equally to this work 1 Ahmanson Translational Theranostics Division, Department of Molecular and Medical Pharmacology, David Geffen School of Medicine, University of California Los Angeles, 200 Medical Plaza, Suite B114-61, Los Angeles, CA 90095, USA Full list of author information is available at the end of the article radical surgery with or without radiation therapy (RT), followed by adjuvant CTX. For soft tissue sarcomas, the therapeutic strategy is based on a risk classification that considers TNM stage, histologic subtype, and primary tumor location. Local recurrence or distant metastases occur in up to 40% of patients who initially receive treatment with curative intent [3]. year survival rate in patients with metastases is 20% compared to 65% for patients with localized disease [4]. Clinical characteristics such as tumor grade, size, presence of distant metastases or skip lesions, surgical margin status and histologic response to neo-CTX have been reported to be predictors of survival in bone and soft tissue sarcomas [5][6][7]. However, these prognostic factors are not highly accurate. Histologic response is determined by examination of resected specimens after the completion of neo-CTX. A noninvasive early interim biomarker that could permit reliable response predictions would be useful to guide changes in treatment of non-responding patients [8].
[ 18 F]FDG PET/CT is used to accurately stage and assess treatment response in almost all cancers, including those in pediatric patients [9][10][11][12]. However, for pediatric bone and soft tissue sarcoma, the role of [ 18 F]FDG PET/ CT is not clearly defined. In this prospective study, we assessed the value of semiquantitative [ 18 F]FDG PET/CT parameters as a potential early intermediate biomarker for response to neo-CTX in pediatric patients with highgrade bone and soft tissue sarcomas.

Objectives
The primary aim of this prospective study was to evaluate whether semiquantitative [ 18 F]FDG PET/CT parameters acquired at baseline (PET1) and during therapy (PET2) are predictive of time to progression (TTP) and overall survival (OS) in children with high-grade bone or soft tissue sarcomas.
The secondary objective was to determine whether [ 18 F]FDG PET/CT was able to predict the response to neo-CTX defined by percent tumor necrosis in the resected tumor or by MRI performed after the completion of neo-CTX. Primary and secondary aims were also evaluated in the subpopulation of patients with bone sarcomas.

Study design and participants
This was a prospective, open-label, observational, singlearm, single-center study approved by the local ethics committee (UCLA-IRB#10-000246) in pediatric patients with high-grade bone or soft tissue sarcomas. All patients with histologically or cytologically confirmed bone or soft tissue sarcoma who were evaluated for management of disease prior to neo-CTX before definitive therapy (surgery, radiation therapy (RT)) were eligible. Written informed consent was obtained from all participants or guardians at enrollment along with signed participant assent, when applicable.
Enrolled patients underwent [ 18 F]FDG PET/CT at 2 time points: within 4 weeks before the start of neo-CTX (baseline, PET1) and at 6 weeks after the start of CTX (early interim, PET2).
[ 18 F]FDG PET/CT findings were confirmed by pathology when available or by follow-up [ 18 F]FDG PET/CT and/or standard clinical follow-up.
Excised tumors were examined for extent of necrosis, and ≥ 90% necrosis (< 10% viable tumor cells) was considered a complete histopathological response to neo-CTX [13]. In patients undergoing definitive RT, response was assessed by MRI at the end of neo-CTX [14,15]. Patients with a complete disappearance of the soft tissue component of the tumor on MRI were considered responders. Pathology and MRI clinical reports were used to obtain the pathological response.

[ 18 F]FDG PET/CT image acquisition
Patients were instructed to fast for at least 6 h before the scan, and blood glucose levels were measured before injection of [ 18 F]FDG. All patients had serum glucose levels of < 150 mg/dl prior to the scan. None of the patients had a history of diabetes.
[ 18 F]FDG was administered by intravenous injection at the activity of 0.1 mCi/Kg and up to a total maximum of 10 mCi. After 60 min of uptake time, images were acquired using a 64-detector PET/CT scanner (2007 Biograph 64 Truepoint or 2010 Biograph mCT 64; Siemens). A low-dose CT for attenuation correction (132 kVp, 35 mAs (CareDose protocol), 0.5-s tube rotation, 5-mm slice collimation, bed speed 8 mm/s) was performed after administration of intravenous contrast (115 mL of iohexol [Omnipaque 350; GE Healthcare]) unless contraindicated. CT images were acquired along the same length of the patient's body as the PET (full-body PET/ CT, from vertex to toes). The time per bed position was 2 min. Iterative methods were used to reconstruct the PET images with a slice thickness of 2 mm. All PET images were reconstructed using attenuation, deadtime, random-event and scatter corrections. PET images were reconstructed with an iterative algorithm (orderedsubset expectation maximization) in a 200 × 200 matrix (3-dimensional, 2 iterations, 24 subsets, Gaussian filter 5.0).

Visual analysis
PET/CT images were retrospectively analyzed on an Osi-riX workstation by two UCLA investigators (GP, FC), with more than 5 years of experience in reading oncologic PET images. The readers had access to all patient medical information. Images were interpreted by consensus. Any focal non-physiologic [ 18 F]FDG uptake above surrounding background activity was considered consistent with malignancy. Metastatic sites were classified as regional lymph nodes (LNs), lung, or other distant metastases (other skeletal segments and/or other distant sites).

Semiquantitative analysis
Standardized uptake value (SUV) was defined as activity concentration (Bq/mL) divided by injected activity (Bq) normalized to body weight. The highest voxel value (SUVmax) was obtained in a volume of interest (VOI) covering the entire tumor as defined by the investigator (GP). SUVpeak and SUVmean were also calculated within the same VOI. The metabolic tumor volume (MTV) was determined with a threshold of 40% of the SUVmax. When normal tissues with high [ 18 F] FDG uptake were included in the VOIs or the VOIs excluded obvious tumor tissue, manual adjustment was applied. Total lesion glycolysis (TLG) was defined as the product of SUVmean and MTV. Reduction in SUV parameters was defined as ΔSUV = [(SUV2 − SUV1)/ SUV1]. Change in MTV and TLG was calculated as follows: Finally, response to neo-CTX was evaluated applying PET EORTC criteria [16].

Statistical analysis
Median and interquartile range (IQR) were used as descriptive statistics for continuous variables, while absolute and relative frequencies for categorical ones. For time-to-event data, the endpoints were: OS defined as the time interval from the start of neo-CTX to the date of last follow-up or the date of death from any cause. TTP was defined as the time interval from the start of neo-CTX to the date of the first event or the date of last follow-up for patients who had no events (recurrent or progressive disease and death from any cause).
A univariate Cox proportional hazards regression model was used to assess the association between TTP, OS and the following covariates of interest: PET1-SUVmax, PET1-SUVmean, PET1-SUVpeak, PET1-MTV, PET1-TLG, PET2-SUVmax, PET2-SUVmean, PET2-SUVpeak, PET2-MTV, PET2-TLG and changes therein. Changes between PET1 and PET2 parameters (ΔPET parameters) were expressed in percentage of reduction. Kaplan-Meier curves and log-rank test were used to summarize and compare the survival experience between PET1, PET2 and ΔPET parameters on cutoff values (median, upper quantile and lower quantile). Mann-Whitney test was used to test the association between PET1, PET2 and ΔPET parameters to the neo-CTX response status (yes/no). Data were analyzed by R 3.6.1 (R Foundation for Statistical Computing, Vienna-A, https ://www.R-proje ct.org).

Population characteristics
The study flowchart is provided in Fig. 1, and patients characteristics are listed in Table 1. Thirty-four consecutive patients (

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's sarcoma patients (n = 13) were treated as per COG AEWS1031 protocol composed of 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 Fig. 2. The treatment schema for patients with Ewing's sarcoma is not included as the data from that study are 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 were 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).

[ 18 F]FDG PET parameters for prediction of pathological response
No significant association between PET1 and PET2parameters and the response to neo-CTX was observed (Table 4).Despite the 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 Additional file 1: Table S3.

Bone sarcoma only subpopulation
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's sarcoma = 9). In this bone sarcoma subpopulation, all PET2 parameters showed a statistically significant association with TTP, confirming the results also observed in the full study population (Additional file 1: Table S4).

Discussion
In this prospective single-center study, a significant association between early interim [ 18 F]FDG PET2 parameters and TTP was observed. Patients with higher residual disease metabolic activity at 6 weeks after initiation of neo-CTX study had worse outcomes compared to those with no or only mild residual [ 18 F]FDG activity. Additionally, patients showing a lower MTV reduction from PET1 to PET2 had a lower pathological response rate. However, none of the PET parameters were predictive of OS.
[ 18 F]FDG PET/CT is important for staging and therapy response assessment of patients with high-grade bone and soft tissue sarcomas [10,20]. However, limited data are currently available regarding the potential role of early interim PET/CT performed 6 weeks after neo-CTX initiation as a predictor of patient outcome, especially in the pediatric population. Conflicting published results are probably due to heterogeneous patient populations, different therapy regimen and different time points of PET/CT evaluations. Costelloe et al. reported in a mixed population of pediatric and adult bone sarcoma patients that SUVmax and TLG values measured before and after neo-CTX provided predictive information about treatment response [21]. In a small study of bone sarcoma patients, changes in [ 18 F]FDG SUVmax at the end of neo-CTX predicted histopathologic responders and non-responders [22]. However, as both studies measured glucose metabolic parameters after completion of neo-CTX, the impact on managing these patients appears limited. In the present study, glucose metabolism responses were measured early during neo-CTX. Earlier identification of non-responders to neo-CTX could lead to meaningful treatment changes. In the current study, a significant association between the early interim [ 18 F] FDG PET parameters and patient outcome was observed. However, no significant association between the baseline [ 18 F]FDG PET parameters and TTP was observed. This contrasts with other studies reporting PET1-SUVmax as a prognostic biomarker [23,24].
None of PET1, PET2 and ΔPET parameters were associated with OS. This is probably due to the relatively small sample size and the limited number of events in the current study population.
Histologic response to neo-CTX is known to be a prognostic indicator in bone and soft tissue sarcoma, especially in osteosarcoma. Patients with > 90% tumor necrosis in response to treatment have improved outcomes [13]. In contrast to other studies [24][25][26][27], we did not observe a significant association between any baseline or early interim [ 18 F]FDG PET parameters and the response to treatment assessed by the percentage of tumor necrosis (n = 29) or by MRI evaluation (n = 5). However, we observed a significant association of ΔMTV with the pathological response to neo-CTX. Additionally, patients with higher PET2-SUVmax were less likely to be responders to neo-CTX although without statistical significance. These results suggest that tumor metabolic activity changes under neo-CTX as assessed on early interim [ 18 F]FDG PET2 can be integrated into the clinical risk prognostic assessment.
[ 18 F]FDG PET/CT is a whole-body imaging modality and can detect distant metastatic lesions. In our cohort, patients with metastatic disease had worst outcome. Of note, all metastatic lesions showed partial or complete response on PET2. It is considered standard of care to treat distant metastases with local control methods (surgery and/or radiation). Whether or not the prognosis is altered with this aggressive approach is controversial.
The main limitation of the study is its small sample size and the heterogeneity of the included sarcoma sub-types. However, each tumor sub-type was treated under the same therapy protocol. Results of the study were comparable in the subpopulation of patients with bone sarcoma only (osteo-and Ewing's sarcoma). Of note, patients with rhabdomyosarcoma were not included in this study as they were enrolled on competing COG therapeutic clinical trials, which included PET imaging as an experimental aim. Further sub-analyses considering the different tumor subtypes were not feasible considering the small sample size and the limited number of events. Larger and more homogeneous cohorts will be required to determine whether early interim [ 18 F]FDG PET/CT imaging can be useful for early treatment response predictions and prognostic information. However, such studies are difficult to conduct as bone and soft tissue sarcomas are rare neoplasms, especially when only considering pediatrics patients only. Another limitation is the lack of control for tumor necrosis. The data reported were obtained using the pathology and MRI clinical reports considered as reference in the treatment management of the patient. Finally, the comparison with the RECIST criteria for assessing the response to neo-CTX was not performed because of heterogeneous conventional imaging follow-up (modality, time points).

Conclusion
In this prospective single center study of 34 pediatric patients with sarcoma, the intensity of residual metabolic tumor activity on early interim [ 18 F]FDG PET/CT studies performed 6 weeks after the start of neo-CTX was associated with earlier TTP but not OS. Additionally, MTV reduction after neo-CTX was associated with tumor pathological response. [ 18 F]FDG PET/CT may serve as a useful early prognostic marker in pediatric patients with high-grade bone and soft tissue sarcoma.