A systematic review of early phase studies for children and young people with relapsed and refractory rhabdomyosarcoma: The REFoRMS‐SR project

Rhabdomyosarcoma is the commonest soft tissue sarcoma in children. Around one‐third of children with rhabdomyosarcoma experience relapse or have refractory disease, which is associated with a poor prognosis. This systematic review of early phase studies in pediatric relapsed/refractory rhabdomyosarcoma was conducted to inform future research and provide accurate information to families and clinicians making difficult treatment choices. Nine databases and five trial registries were searched in June 2021. Early phase studies of interventions for disease control in patients under 18 years old with relapsed/refractory rhabdomyosarcoma were eligible. No language/geographic restrictions were applied. Studies conducted after 2000 were included. Survival outcomes, response rates, quality of life and adverse event data were extracted. Screening, data extraction and quality assessment (Downs and Black Checklist) were conducted by two researchers. Owing to heterogeneity in the included studies, narrative synthesis was conducted. Of 16,965 records screened, 129 published studies including over 1100 relapsed/refractory rhabdomyosarcoma patients were eligible. Most studies evaluated systemic therapies. Where reported, 70% of studies reported a median progression‐free survival ≤6 months. Objective response rate was 21.6%. Adverse events were mostly hematological. One‐hundred and seven trial registry records of 99 studies were also eligible, 63 of which report they are currently recruiting. Study quality was limited by poor and inconsistent reporting. Outcomes for children with relapsed/refractory rhabdomyosarcoma who enroll on early phase studies are poor. Improving reporting quality and consistency would facilitate the synthesis of early phase studies in relapsed/refractory rhabdomyosarcoma (PROSPERO registration: CRD42021266254).


| INTRODUCTION
Rhabdomyosarcoma accounts for 4.5 cases/million children/adolescents per year. 1 Overall around two-thirds of patients diagnosed with rhabdomyosarcoma are alive at 5 years after diagnosis, but outcomes vary by risk group.Around one in three children and young people treated for rhabdomyosarcoma experience relapsed or refractory disease. 2,3Outcomes are much poorer in this situation, where historically only 17% of patients survived. 4portantly, the prognosis associated with relapsed and refractory rhabdomyosarcoma varies greatly with the timing and location of the relapse as well as the intensity of prior therapies used; for example, over 40% of children and young people with originally localized disease who relapse in the same location may be cured, but the chances of cure are much lower in those with metastatic relapse. 5With this in mind, it can be difficult for clinicians, parents and patients to decide what treatments should be given for relapsed and refractory rhabdomyosarcoma.
Across Europe, the standard of care treatment for first relapse of rhabdomyosarcoma that has already received an alkylating agent (ifosfamide or cyclophosphamide based induction therapy) is currently the combination of vincristine, irinotecan and temozolomide (VIT) together with appropriate local control measures including surgery and/or radiotherapy wherever feasible. 6Furthermore, the ongoing European pediatric Soft tissue Sarcoma study Group Frontline and Relapse Rhabdomyosarcoma Study (FaR-RMS) is exploring the combination of backbone vincristine and irinotecan chemotherapy with the tyrosine kinase inhibitor regorafenib. 7e options for subsequent lines of treatment are much less clear.Alongside symptom-directed interventions such as pain relief, anticancer treatment options may be considered and can include aggressive treatment with the intention to cure, palliative treatments to reduce treatment burden and early phase studies.
These early phase studies involve investigating new treatments or combinations of treatments, such as, including systemic chemotherapy, novel agents and targeted therapies, radiotherapy, cellular therapy and/or vaccinations.As these treatments are new and experimental, the goal of these early phase studies is primarily to assess the dosing and/or safety of a novel treatment.The findings of effectiveness within these types of studies are often secondary and therefore useful in generating knowledge of potentially effective treatments which need to be synthesized to support further investigations.Previous reviews have shown a low success rate in terms of tumor response and overall survival times in early phase studies, 8 but this response for rhabdomyosarcoma patients specifically, has not been examined and thus warrants review.
Within the REFoRMS-SR project, we conducted a systematic review of early phase studies of interventions for children and young people with relapsed and refractory rhabdomyosarcoma with the aim of synthesizing the current evidence to inform clinicians, parents and patients about the effectiveness of interventions that have been evaluated in this way.This review has been conducted alongside a qualitative study to understand the decision-making process of patients and families with experience of relapsed and refractory rhabdomyosarcoma.Both work-streams will be combined to generate a best practice statement to support healthcare professionals in pediatric oncology services.This article reports the systematic review.

| Parent and clinical advisory groups
The REFoRMS-SR project was guided by a group of bereaved parents whose children had experienced relapsed and refractory rhabdomyosarcoma, and a clinical advisory group consisting of healthcare and research professionals with expertise in soft tissue sarcoma.The parent group were identified through a combination of open and closed invites to known contacts, and the clinical advisory group by invitation through professional contacts for their specific clinical and/or methodological interests.Both groups were involved continuously and through direct interaction as defined by the ACTIVE framework. 9The parent and clinical advisory groups were involved in influencing and/or controlling the study design (stages 1-4 of ACTIVE framework) and interpretation of findings (stages 10-12 of ACTIVE framework) throughout the REFoRMS-SR project and are co-authors to this article.

| Search strategy and selection criteria
This systematic review followed a protocol registered on the International Prospective Register of Systematic Reviews (PROSPERO) (CRD42021266254 10 ), and was written in accordance with the PRISMA 2020 guidelines. 11It was conducted following standardized systematic review methods as depicted in Studies identified in the searches were screened using the Rayyan Software, 12 based on the following criteria: • Population: Patients with relapsed and/or refractory rhabdomyosarcoma aged 0-17 years inclusive.Patients aged 18 years and above were considered adults and therefore excluded.Studies including patients with other conditions/ages were eligible for inclusion provided that the data relating to the population of interest could be extracted separately, or where 50% or more patients were from the population of interest.Pre-clinical and animal studies of treatments for rhabdomyosarcoma were excluded.
• Intervention: Any treatment given with the intention of disease control, including with palliative or curative intent.Studies which evaluated treatments for symptom management in patients with rhabdomyosarcoma were not eligible.
• Comparator: Studies did not need to have a comparator group but were still eligible if reporting relevant outcomes.
• Outcomes: Survival (progression free survival, overall survival), Radiological response rates by RECIST criteria, Quality of Life (measured by specific assessment tools [eg, PedsQL] and also by experiential or qualitative data), side effects/adverse events, burden of therapy, costs/measures of costeffectiveness.
• Study Design: Early phase studies, including single arms or randomized between two or more options, including, but not limited to: "First in child" studies (traditionally phase 1), Dose finding studies (traditionally phase 1b/2a), Proof of concept/efficacy studies (traditionally phase 2b), Early effectiveness studies (traditionally phase 2b/3).Studies were excluded if enrolment ceased before 2000.With regards to publication type, we included full-text articles, conference abstracts and clinical trial registry records.
Screening was conducted independently and in duplicate by at least two researchers (CE, LB, JEM and GB).Conflicts were resolved by a third reviewer or discussion with the review team.Authors of full-text publications were contacted to clarify whether studies were eligible for inclusion if the information provided was unclear (eg, if the study enrolled participants with rhabdomyosarcoma but the age of these participants was not reported).Authors of clinical trial registrations were contacted if the trial was completed but no corresponding publication could be identified.

| Data extraction
Before data extraction, eligible clinical trial registrations, conference abstracts and full-text publications were linked.For studies where multiple sources of data were available, data were extracted from the source with the most information.
Data extraction was performed by one reviewer (CE, LB and JEM) and checked by a second (CE, LB and JEM).Disagreements were resolved following discussion with the review team.For full-text publications and conference abstracts, patient demographic and disease characteristics were extracted for all patients unless rhabdomyosarcoma specific data were available; adverse event data were extracted for all participants; and data regarding clinical outcomes were extracted for rhabdomyosarcoma patients only (see Data S1 for the full-text data extraction template).

| Quality assessment
Quality assessment was conducted by one reviewer (CE, LB and JEM) and checked by a second (CE, LB and JEM), using a modified version of the Downs and Black Checklist 13 (see Data S1), owing to the absence of any validated quality assessment tool for early phase studies.Two questions regarding the external validity (Questions 11 and 12) were removed as they were not deemed relevant for early phase studies.For single-arm studies, only 15 of the 27 items were applicable.Quality assessment was only conducted for full-text publications and conference abstracts.
Disagreements were resolved by consensus.

| Synthesis
Meta-analyses were planned but were not performed due to significant heterogeneity in the included interventions.A narrative synthesis was performed.Results are presented in order of importance to the parent advisory group.

| Study selection
From 16,965 studies identified from the database searches, 584 were deemed eligible at title and abstract screening, including 203 clinical trial registry records, 99 conference abstracts and 282 full-text publications.
An additional 83 studies and clinical trial registry records were identified by additional searches, 32 of which were eligible for inclusion (Supporting Information Refs.S1-S32).Of the 75 authors contacted for further information, 32 replied (43% response rate) and four studies (Supporting Information Refs.S13, S24, S33, S34) were eventually included.Excluded studies information is provided in the Data S1.
Overall, 122 studies from 124 full-text articles (Supporting Information Refs.S1-S27, S33-S129) alongside seven studies from conference abstracts (Supporting Information Refs.S28, S130-S135), were included in the synthesis of published studies (n = 129).Three of these studies (Supporting Information Refs.S63, S92, S94) included seven noncomparative arms which have been extracted separately, resulting in a total of 133 individual cohorts being included in the synthesis.Where applicable, the data has been explicitly reported as either the number of cohorts or number of studies.An additional 107 clinical trial registry records of 99 trials were included in the synthesis of clinical trial data.
Further details of the study selection process are provided in Figure 2.

| Quality assessment
One hundred and twenty single-arm studies (Supporting Information Refs.S1, S3-S27, S33-S47, S49-S62, S64-S89, S93, S95-S105, S108-S111, S113-S135) and three non-comparative, multiarm studies (Supporting Information Refs.S63, S92, S94) were assessed using a 17-item modified Downs and Black checklist. 13In general, studies reported the methods and results well, although several studies did not report study selection criteria.Similarly, almost 20% did not provide random variation of the data, and almost 20% did not report adverse events appropriately.Internal validity was deemed to be at low risk of bias across the studies included.
Six multiarm, comparative studies (Supporting Information Refs.S2, S28, S90, S91, S106, S112) were assessed using the 27-item Downs and Black Checklist, 13 with the majority of studies providing comprehensive reporting of their trial.The internal validity across the studies was mixed; although subjects were randomized in five of the six studies, randomization was only concealed in one of those studies.Only one study blinded participants to the intervention, and two blinded assessors to the intervention.Studies did use appropriate statistical tests and outcome measures.
For all studies, external validity was difficult to determine.By their very nature, early phase studies often investigate novel drugs only available in highly specialized centers and have stringent eligibility criteria.Risk of bias assessments are summarized in Figure 3, with further details provided in Data S1.

| Response rates
Overall, 59 of 1151 children and young people showed a complete response (CR), and 190 experienced a partial response (PR).
Therefore, the objective response rate (CR + PR) across all interventions for children and young people with relapsed and refractory rhabdomyosarcoma was 21.6%.Cohorts reporting more than 10 children and young people with relapsed and refractory rhabdomyosarcoma, where the objective response rate is greater than 30% have been identified in blue fill within Table 2.Ten cohorts reported a 100% response rate amongst children and young people with rhabdomyosarcoma, but these studies all had fewer than five participants, so results should be interpreted with caution (Supporting Information Refs.S8, S10, S14, S33, S62, S72, S96, S105, S128, S131).Where data was reported separately for children and young people with a first-relapse, the overall response rate was 33.7% (29/86 children and young people, from seven cohorts [Refs.S12, S42, S53, S74, S92, S108, S112]).No studies assessed differential efficacy by ethnicity or sex.Other planned subgroup analyses were unable to be performed due to availability of data.

| Adverse events
Data on adverse events of interventions included in this systematic review were available for over 4500 children and young

| Deaths
Nineteen studies (15%) explicitly reported deaths (Supporting Information Refs.S1, S3, S5, S10, S11, S19, S22, S24, S65-S67, S76, S82, S91, S103, S112, S115, S128, S134).From these studies 69 deaths were reported out of a total of 1011 patients.Nine deaths were deemed to be related to the study treatment, while 32 were due to progressive disease.Children and young people progressed both early within a study (either before the intervention was administered or within the first cycle of the intervention) and within 30 days of treatment administration.

F I G U R E 3 Risk of bias assessment for single arm and multiarm non-comparative studies (A) and multiarm comparative studies (B).
T A B L E 1 Summary of study characteristics and demographics of included children and young people from the included studies.Note: Detailed demographic information for each study can be found in the project's full report. 24bbreviations: HSCT, hematopoietic stem cell transplant; NOS, not otherwise specified; RMS, rhabdomyosarcoma; UK, the United Kingdom; USA, the United States.
T A B L E 2 Disease response and survival outcomes for the included, published studies.Fischer, 2021 (S3)   ensured this project will have significant impact within the community.
As in much evidence synthesis work, the main challenges related to the poor reporting of data within included studies.In particular, data relat-

Figure 1 .
Figure 1.Searches were developed by an information specialist (HF); the full search strategies are provided in Data S1.MEDLINE, Embase, Cochrane Central Register of Controlled Trials (CENTRAL), Cochrane Database of Systematic Reviews (CDSR), Science Citation Index-Expanded, Database of Abstracts of Reviews of Effects (DARE), the International HTA database, PROSPERO and Conference Proceedings Citation Index-Science were searched to identify published articles.ClinicalTrials.gov,European Union Clinical Trials Register, WHO International Clinical Trials Registry Platform (WHO ICTRP), International Standard Randomised Controlled Trial Number (ISRCTN) and ANZCHOG Children's Cancer Clinical Trials Repository (ACCCTR) were searched to identify additional unpublished, ongoing or completed studies.No language or geographical limitations were applied, but studies were only included if they were published from 2000 onwards.All databases were searched on 30 June 2021 and were deduplicated in EndNote 20.Reference lists of relevant systematic reviews and included articles were searched on 11 April 2022.
people (not rhabdomyosarcoma-specific).Although the majority of studies used a standardized tool (including the Common Terminology Criteria for Adverse Events [CTCAE], and the World Health Organisation [WHO] classification), the reporting of adverse events varied across studies making it difficult to synthesize the data.Hematological adverse events were most common.Laboratory test abnormalities were also common, although the impact of these on children and young people's symptoms was unclear.
but assumed PD.One patient in each stratum (where stratified by previous treatment) T A B L E 2 (Continued) evaluable with no objective response and either SD/PD. 2 patients with SD had STS but unclear if these had PD or SD, and one unclear if PD/SD or nonfor pt with CR and 2 with PR; 3 course for other pt with PR.No responses observed had partial response after 4 courses and was alive with SD at the end of study.The other RMS patient had SD after 6 courses but died from reported but not CR/PR/SD.Of the two RMS patients with SD, ing to outcomes of children and young people with rhabdomyosarcoma was frequently not separable from other tumor types; 35 studies on 31 therapies including almost 80 potentially eligible children and young people were excluded for this reason.Trials including multiple tumor types are essential in oncology; nonetheless we encourage reporting of patient demographics and outcomes by tumor type to improve the transparency and clinical utility of these data.We selected a search strategy focused on soft tissue sarcoma.This facilitated screening more broadly than a pure rhabdomyosarcoma search, but may potentially have missed a small number of studies which included "all relapsed/refractory pediatric malignancies."Testing of strategies in advance, including screening samples of broader searches, suggests this number is likely minimal and is unlikely to have included data which would substantially impact on the review conclusions.Furthermore, additional strategies within the study identification process (including linking clinical trial registries and conference abstracts to published studies, and reference list searching) will have helped to mitigate any potential deficiencies in the database searches.In relapsed and refractory rhabdomyosarcoma, one of the greatest future research challenges is the speed at which early phase studies are conducted, and thus the risk of any evidence synthesis becoming rapidly out of date.Children, young people, families and clinicians require innovative solutions to provide high quality data syntheses in a form that is continually updated.To address this, the REFoRMS-SR will now become the first living systematic review in childhood cancer: Living-REFoRMS.The Living-REFoRMS team will perform regular updates of the evidence synthesis, whilst also working on the methodological challenges of living reviews, including evaluating different methods for searching, screening, quality assessment and synthesis.The first update review is in progress and an interactive and user-friendly online resource is being developed to facilitate access to the Living-REFoRMS data for children, young people, families, clinicians and researchers.
a Mean, (SE).b Evaluable, RMS patients.c Plus italicized indicates studies where exact number of evaluable RMS patients is unknown but is definitively >1.d Calculated from provided information.