High Dose Ifosfamide in Relapsed and Unresectable High-Grade Osteosarcoma Patients: A Retrospective Series

Background: The evidence on high-dose ifosfamide (HD-IFO) use in patients with relapsed osteosarcoma is limited. We performed a retrospective study to analyze HD-IFO activity. Methods: Patients with osteosarcoma relapsed after standard treatment [methotrexate, doxorubicin, cisplatin +/− ifosfamide (MAP+/−I)] with measurable disease according to RECIST1.1 were eligible to ifosfamide (3 g/m2/day) continuous infusion (c.i.) days 1–5 q21d. RECIST1.1 overall response rate (ORR) (complete response (CR) + partial response (PR)), progression-free survival at 6-month (6m-PFS), duration of response (DOR), and 2-year overall survival (2y-OS) were assessed. PARP1 expression and gene mutations were tested by immunohistochemistry and next-generation sequencing. Results: 51 patients were included. ORR was 20% (1 CR + 9 PR). Median DOR was 5 months (95%CI 2–7). Median PFS, 6m-PFS, OS, and 2y-OS were 6 months (95%CI 4–9), 51%, 15 months (10–19), and 30%, respectively. A second surgical complete remission (CR2) was achieved in 26 (51%) patients. After multivariate analysis, previous use of ifosfamide (HR 2.007, p = 0.034) and CR2 (HR 0.126, p < 0.001) showed a significant correlation with PFS and OS, respectively. No significant correlation was found between outcomes and PARP1 or gene mutations. Conclusions: HD-IFO should be considered as the standard first-line treatment option in relapsed osteosarcoma and control arm of future trial in this setting.


Introduction
High-grade osteosarcoma is the most frequent primary bone tumor that usually occurs in children and young adults [1,2]. Since the introduction of multi-agent chemotherapy (cisplatin, adriamycin,  [20] IFO 12 g/m 2 CI d 1-14 30 1/2 10 Harris MB, Med Ped Oncol 1995 [21] %ORR: overall response rate; CR: complete response; PR: partial response; JPHO: Journal of Pediatric Hematology and Oncology; nr: not reported. At present, high dose ifosfamide (HD-IFO) is largely used at the time of recurrence in osteosarcoma. Several studies are available on this regimen with doses up to 12-15 g/m 2 in patients with relapsed osteosarcoma. Interestingly, they mainly focused on response rate whereas clinically relevant information such as PFS rate and variables influencing response and survival are lacking [18][19][20][21] ( Table 1). We performed a retrospective study to assess the clinical activity of HD-IFO in patients with relapsed and unresectable osteosarcoma after standard treatment of their localized disease and explored its correlation with key molecular features.

Patients and Methods
Patients with high-grade osteosarcoma who received chemotherapy with HD-IFO when relapsing after standard treatment were selected from the database of the Rizzoli Institute. Only patients with measurable disease according to RECIST 1.1, deemed not suitable for upfront surgery after Cells 2020, 9, 2389 3 of 12 multidisciplinary team (MDT) evaluation (including thoracic surgeons, orthopedics, radiologists, radiation therapists and medical oncologists), who received at least two courses of HD-IFO, and with available demographic, clinical, imaging and follow-up data were included in the present analysis. This study was approved by the local ethical committee. Demographic and clinical data (age, sex, ECOG scale of performance status, pattern of relapse, type, number and prior treatments (metastasectomy before HD-IFO and histologic response on primary tumor after induction chemotherapy), number of HD-IFO courses, radiologic response according to RECIST 1.1, toxicity, date of progression, type of treatment after HD-IFO, date of last follow-up or death) were collected from patient chart. Radiological images were reviewed (D.V., P.P. and E.P.) for the purpose of this study.
Treatment consisted of ifosfamide 3 gr/m 2 /day combined with MESNA at same dose, given as continuous infusion from day 1 to 5, every 21 days [22], until progression or unacceptable toxicity. Prophylactic use of G-CSF was recommended from day 7 of each cycle.
According to institutional guidelines, response to HD-IFO was assessed by means of CT scan every two cycles according to RECIST criteria version 1.1. The achievement of a complete surgical excision of all metastases (second complete surgical remission = CR-2) was the mainstay of the treatment and surgery was offered to selected patients in case of partial response or stable disease, after multidisciplinary team discussion. Archival formalin-fixed paraffin-embedded (FFPE) tumor tissue were analyzed by standard immunohistochemistry procedure using DAKO Autostainer (Agilent, Santa Clara, CA, United States) and recombinant Anti-PARP1 antibody (E102, ab32138, Abcam, Cambridge, MA, United States). An expert pathologist evaluated PARP1 expression in tumor nuclei and scored as positive (+) tumor samples with more than 50% expressing tumor nuclei in three different optical fields. Visible images were acquired with a DM1000 microscope (Leica, Wetzlar, Germany) equipped with a color 3.1 M PixelCMOS digital camera. Tumor areas were microdissected from 10 µM slices and nucleic acids were extracted by Maxwell RSC DNA FFPE Kit and Maxwell RSC RNA FFPE Kit using a Maxwell RSC Instrument (Promega, Madison, WI, United States), according to manufacturer's instructions. DNA and RNA were quantified using a Qubit ® 3.0 Fluorometer (Thermo Fisher Scientific, Life Technology Italia, Monza, Italy) and Qubit dsDNA HS (high sensitivity) assay kit (Thermo Fisher Scientific, Life Technology Italia). DNA and RNA quality were assessed by gel electrophoresis and 2100 Bioanalyzer High Sensitivity DNA and RNA assay Kits (Agilent Technologies, Agilent Technologies, Inc., Santa Clara, CA, USA). Nucleic acids purity was evaluate using NanoDropTM (Thermo Fisher Scientific). Ten ng of QC-passing nucleic acid samples were loaded on IonChefTM System for library preparation according to Oncomine Comprehensive Cancer panel assay v3 TM (Thermo Fisher Scientific). In the presence of IonCodeTM barcodes, eight 100 pM prepared DNA and RNA libraries were diluted to 50 pM, pooled and the single stranded template libraries and loaded on the Ion 540™ chip by IonChefTM System. Sequencing was achieved using the Ion GeneStudioTM S5 Plus System. The panel results were optimized for a total of 8 to 10 million reads with a 500× median coverage. The reads were aligned to assembly hg19 of the human reference genome by the Torrent SuiteTM (v 5.8). To standardize the analyses and to reduce the impact of sequencing artifacts derived from the formalin fixation, we set the allele frequency limit of detection at 10% for all the samples.
Toxicity data were collected both from clinical chart and from a "toxicity data form" filled by patients or their guardians. Toxicity data were analyzed and graded according to the Common Toxicity Criteria for Adverse Events (CTCAE) version 4.03.
Primary endpoint was overall response rate (ORR) [defined as complete response (CR) + partial response (PR)], according to RECIST 1.1. Progression-Free survival (PFS), PFS at 6-month (6m-PFS) defined as the ratio between patients in CR, PR or stable disease (SD) and those ones progressing (PD) after six months from study entry. Overall survival (OS), OS at 2-year and duration of response and toxicity were secondary end-points of the study. Patients who underwent radical surgery after HDIFO were defined as CR2.
PFS was calculated from the date of the first cycle of HD-IFO to the date of tumor progression or last follow-up. Duration of response was calculated from first day of chemotherapy (HD-IFO) to the date of tumor progression. Overall survival (OS) was calculated from the date of the first day of HDIFO chemotherapy to the date of death or last follow-up (post-HDIFO survival). Patients were censored at the date of last follow up in the absence of death or progression.
Statistical analyses were performed using SPSS v26.0 (IBM, Armonk, NY, United States). We used descriptive statistics for baseline patients' characteristics. Qualitative variables were compared using the χ 2 and Fisher's exact tests and/or the Mantel-Haenszel odds ratio (OR) estimates. Survival endpoints were computed by Kaplan-Meier method. Log-rank test and hazard ratio (HR) estimates calculated by Cox regression were used for comparisons. Multivariate analysis was performed using the Cox proportional hazards model including covariates with p-value ≤ 0.05 in the univariate analysis. Whenever indicated, tests were performed two-sided and results were reported with 95% confidence intervals (95%CI). We considered a p-value ≤ 0.05 as statistically significant.

Overall Survival (OS) Post HDIFO
The median post HD-IFO survival was 14.5 months (95%CI 10.1-18.9; range 2-260 months). 1-year and 2-year OS were 62% (95%CI 49-75) and 30% (95%CI 19-45), respectively ( Figure 2). Patients reaching CR/PR and SD had a 2-year OS of 44% and 38%, respectively, whereas none was alive at 2 years in case of PD (HR for OS in patients achieving at least a disease stabilization vs. progression patients was 0.223, 95%CI 0.106-0.470 p < 0.001; Table 5). Twenty-three responding patients (including patients with tumor shrinkage not reaching the RECIST 1.1 threshold for defining a partial response) underwent metastasectomy, achieving a CR2 status. Twenty-two patients had lung metastases (with concomitant local recurrence in three patients). One had a synchronous bone lesion. Patients achieving a CR2 had a significantly longer survival compared to others with a median OS of 66 (95%CI 2-130) vs. 10 months (95%CI 7-13) and an HR for death of 0.100 (95%CI 0.045-0.224, p < 0.001). Two-year OS in the group of patients achieving a CR2 was 61% vs. 4% in patients who did not (Figure 2). Prior treatment with standard-dose ifosfamide was associated with a poorer prognosis (median OS

Toxicity
All the 51 patients were evaluable for safety analysis. The median number of cycles administered was four (range two to seven). Nine (18%) patients experienced a febrile neutropenia. Grade 3-4 anemia in two (4%) patients and thrombocytopenia in three (6%) cases was reported, with one patient interrupting treatment due to thrombocytopenia. Grade 3-4 neurological toxicity was described in two (4%) of the patients. In both cases neuropathy resolved after chemotherapy interruption, hydration, diuretics and methylene blue. Two patients (4%) experienced a grade 1 persistent kidney injury. Subsequent dose reduction was required in all patients with non-hematological toxicity.

Discussion
In this series a treatment with HD-IFO (15 g/m 2 ) in recurrent/progressive osteosarcoma after standard treatment showed an ORR of 20% and up to nearly 30% in pediatric patients. The 6m-PFS was 53% and was significantly higher for whom were ifosfamide-naïve and treated in 1st line. The 2-year OS was 52% and markedly influenced by the high proportion of patients who achieved a second complete surgical remission that also in our series was confirmed as the most relevant factor influencing survival.
Interestingly, in our study ORR in pediatric patients was 29% vs. 13% in the adult patients (p = 0.186), with no difference both in PFS and in OS figures in the two different populations, confirming that age per-se does not represent a prognostic factor after relapse in osteosarcoma [8].
Beyond the possibility of prolonging disease control, a major goal of chemotherapy in osteosarcoma metastatic setting is to increase the chance of obtaining a radical surgery and/or delay further progression. In this series, in 23/51 non-progressing patients (including patients with CR/PR and those ones experiencing a tumor shrinkage < 30%) a CR2 status was achieved. Consistently with large previous experiences [7,8], the CR2 status was confirmed also in our series as a key factor associated with an improved survival. Due to the retrospective nature of our study, it is not possible to definitely determine the role of HD-IFO in CR2 achievement. Nonetheless, all the 23 patients who achieved a CR2 were considered not eligible to surgical resection at time of HD-IFO start by an experienced MDT.
A dose-response mechanism is well-known for ifosfamide when used in soft tissue sarcoma [28]. Importantly, HD-IFO activity was observed also in patients previously treated with standard dose ifosfamide (from 6 to 10 g/m 2 ), suggesting that higher doses could overcome at least some of the resistance mechanisms. Similarly, we observed some degree of activity of HD-IFO also in patients who received previous ifosfamide, but the PFS of these patients was significantly worse compared with those ifosfamide-naïve. This result could at least in part be related to both selection bias of poor prognostic features (in most of the of the studies adjuvant ifosfamide was given in case of poor histologic response) and chemo-resistance itself. This is in contrast with lack of association with PFS and histologic response to former induction chemotherapy. Taken together these findings, in view of EURAMOS1 clinical trial results [6], discourage use of adjuvant ifosfamide, unless within a clinical trial.
Of course, HD-IFO regimen is more toxic compared to ifosfamide given at lower doses (9-10 g/m 2 ) in terms of both kidney and central neurological adverse events. Nevertheless, in our experience it was manageable, especially in the pediatric population, and no toxic death was reported in our series. Grade 3 or 4 of hematological side effects were reported in 14% of the patients overall; this rate might reflect under-reporting bias due to the retrospective design of this study. Febrile neutropenia occurred in nine patients (18%) in our study. Consistently, Verspor et al., described febrile neutropenia in 7 (19%) cases in a similar series with ifosfamide given from 6 gr/m 2 to 9 gr/m 2 [18]. Our series confirms that grade 3-4 neurological toxicity should be expected in at least 4% of the patients. Full recovery was achieved in both patients with hydration, diuretics and methylene blue, as described [29].
Molecular markers enabling clinical practices with predictive or prognostic valuable tools are warranted to better define patients who might really benefit from this treatment. PARP1 had shown predictive and prognostic values in bone and soft tissue sarcomas [30][31][32]. Moreover, peculiar gene defects showed implications in response to chemotherapy [33,34]. Unfortunately, in our retrospective case series paucity of available samples impinged on predictive and prognostic evaluation of PARP-1 and genetic abnormalities found on response to HD-IFO. Further prospective evaluation of this markers is needed to define their role, if any, in predicting prognosis and/or response to chemotherapy.

Conclusions
In conclusion, there are few new active regimens for patients with relapsed osteosarcoma following multimodality therapy. To our knowledge, this is the largest study on high-dose ifosfamide in pre-treated relapsed high-grade classic osteosarcoma, showing a relatively high ORR especially in pediatric patients. Our data support the use of this regimen as a first option for the treatment of metastatic disease, especially in oligometastatic cases. In this setting any tumor shrinkage making patients eligible to a second surgical complete remission raises chance of cure to about 25%. Finally, this study might be used as a benchmark for phase II studies in the setting of relapsed osteosarcoma.