An institutional audit of the use of novel drugs in pediatric oncology

Abstract Background Significant challenges persist in treating children with rare, relapsed, or refractory malignancies. Novel molecularly targeted drugs promise improved outcomes for these children with reduced toxicity. However, there is often limited evidence to substantiate their clinical efficacy and guide their use. This raises issues for clinical decision‐making, ethical concerns surrounding equity of access to these often‐expensive agents, and the management of families' expectations for cure. This audit evaluated the off‐label use of novel drugs and associated clinical outcomes in order to guide the development of future clinical and ethical guidelines. Aim To evaluate the patterns in the off‐label use of novel drugs for treating childhood cancer and the associated clinical outcomes to guide prospective studies and inform ethical and clinical governance protocols for the use of these agents. Methods A retrospective audit was performed for all patients who received novel drugs off‐label as treatment for their malignancy at an Australian pediatric oncology center between 2010 and 2019. Results One hundred patients with 32 unique diagnoses received 133 novel drugs across 124 regimens. Eighty‐four patients received these drugs at the second line of treatment or greater. Novel drug median cost was $15 521 AUD (Range: $6.53 AUD to $258 339 AUD) and was primarily funded by the hospital (N = 60/133, 45.1%) or compassionate access from pharmaceutical companies (N = 52/133, 39.1%). Decision‐making related to novel drugs was inconsistently documented. Ninety‐one of 124 treatment regimens commenced between 2010 and 2019 resulted in objective responses (73.4%), but only 35 were still ongoing upon review in June 2020 (38.5%). Median response duration was 12.6 months (Range: 0‐93.2 months). Conclusions While novel drugs were largely unable to definitively cure patients, most achieved objective responses. Prospective trials and more rigorous documentation are needed to fully inform the future use of these agents given the heterogeneity of their applications.


| INTRODUCTION
Significant challenges persist in improving treatment outcomes for children with rare, relapsed, or refractory malignancies. 1 Intensifying traditional chemotherapeutic regimens achieves little benefit at the cost of significant toxicity, 1 highlighting the need for new treatment approaches. Over the past decade, technological advancements have greatly facilitated the expansion of precision medicine, a novel treatment paradigm integrating advanced molecular analytical techniques into the diagnosis and personalized treatment of malignancies. 2 By targeting cancer-specific biomarkers and genomic changes, precision medicine promises greater clinical benefit without increase in toxicity.
Despite this theoretical potential, there is currently limited evidence to substantiate the efficacy of novel drugs and guide their optimal use in clinical practice. 3 Few clinical trials have been undertaken to evaluate these agents. Owing to the rarity of pediatric cancers and few eligible patients, 3 clinicians have often relied on off-label or compassionate-use programs to access these drugs. 4,5 However, the significant costs of these novel agents have raised ethical issues surrounding their widespread implementation and accessibility to all patients. 6 The limited evidence for the clinical efficacy also makes it challenging to justify funding drugs through government mechanisms, 6 and to manage families' expectations for cure. 7 The off-label use of novel drugs is undertaken in our pediatric oncology center, particularly for relapsed and refractory diseases, despite a paucity of data around their use. In this context, it remains unclear whether these novel drugs provide enough benefit that their off-label use should be routine in pediatric oncology or if they are best restricted to a controlled clinical trial setting. The aim of this study therefore was to evaluate current practices of prescribing novel drugs off-label to pediatric patients at a large tertiary pediatric oncology center, including the associated clinical outcomes, in order to inform future studies and provide clinical and ethical recommendations for the use of these medications.

| Study design and research ethics
We conducted a retrospective audit at a large tertiary pediatric hospital utilizing patient data extracted from the hospital electronic medical record (EMR). Ethics approval was gained from the RCH Human Research Ethics Committee (HREC) to access the required data from the EMR records (QA/60416/RCHM-2019). For retrospective audits, it is not a requirement of the HREC to obtain informed consent from patients or guardians where the audit involves the use of existing de-identified clinical data with no foreseeable risk to the participants. In this study, the HREC deemed this to be the case.

| Study population
All patients who received a novel drug off-label for the treatment of their malignancy between 2010 and 2019 were included in the study.
Novel drugs were defined as medications, which were considered above the "standard-of-care" for their prescribed indication by a senior clinical oncologist. Patients who accessed novel drugs through clinical trials were not considered eligible for this study, given the stringent governance and reporting procedures in the clinical trial setting.

| Data collection
In the Australian healthcare system, drugs are traditionally dispensed by the hospital at little to no cost to the patients through the Pharmaceutical Benefits Scheme (PBS). For drugs for which approved governmental funding is not available, the hospital Drug Usage Committee (DUC) determines patient eligibility for hospital-funded access to medication on a case-by-case basis. As such, eligible patients were identified using the pharmacy dispensary database and the EMR.
Data, including patient and parents' demographics, were extracted from the clinical notes stored on the EMR. Each patient was assigned a unique study ID and data were stored using REDCap electronic data capture tools on a secure server. 8 Disease data collected included diagnosis, tumor classification, and date of treatment allocation as well as disease state (relapsed, refractory or other) and line of treatment at the time of novel drug prescription. If a novel drug was added to an existing treatment regimen, this was considered another line of treatment. Data extracted to analyze the patterns of prescription for these novel drugs included date of prescription, drug name, billing type, and the overall cost of the medication in Australian dollars (AUD). Outcome data were recorded, including response type, defined according to the documented interpretation of radiographic results including any criteria cited to describe clinical response. Clinical response descriptions were used only where no imaging results were available. Response durations were defined as the time between the first documentation of response to the first documentation of disease progression, recurrence, and patient death or, in the case of ongoing responses, at the time of review, June 2020.
To further characterize the clinical practice of prescribing novel drugs, additional details were audited for example, prescribing consultant, whether there was external consultation and/or multidisciplinary meeting discussion. The DUC meeting minutes from 2010 to 2019 were also systematically reviewed to assess factors, which influenced drug allocation as well as the quality of documentation surrounding each decision. Following data collection, deidentified patient data were exported from the REDCap database for descriptive statistical analysis.

| Patient demographics
One-hundred and twenty-one patients were identified for study inclusion. Twenty-one patients were excluded owing to insufficient documentation prior to the introduction of an EMR in 2016. The remaining 100 patients were prescribed a total of 133 drugs across 124 treating regimens, where nine treatment regimens combined two novel drugs.
Twenty-eight patients received more than one novel drug in total: 23 were prescribed two, and five were prescribed three. Patient demographic details are summarized in Table 1.
There were 32 unique diagnoses across the patients included in this study (Table 1). Over half were diagnosed with a brain tumor (N = 52, 52%); with solid tumors the next most common diagnostic group (N = 31, 29%). Eighty percent of patients were experiencing refractory or relapsed disease, and 84% received a targeted drug at their second line of treatment or greater, indicating a relatively pre-treated population.

| Patterns of treatment allocation
The majority of novel drug regimens were prescribed with curative LGG encompassed diffuse astrocytoma, disseminated glioneuronal tumor, ganglioglioma, low-grade glioma, oligoastrocytoma, pilocytic astrocytoma and pleomorphic xanthroastrocytoma. b HGG encompassed glioblastoma multiforme, anaplastic astrocytoma and diffuse intrinsic pontine glioma. marrow transplantation. Two treatment regimens did not fit into any of these categories: One where treatment intent was undefined, and another where the novel drug was used to enhance standard therapy.
The primary reason cited explicitly by oncologists in their clinical notes for their choice to use a novel drug was genomic results identified through molecular diagnostics or inferred based on clinical diagnosis in the case of NF1 in neurofibromatosis (N = 44, 33.1%) ( Table S1). The next most frequent category was "reason not clearly documented" (N = 23, 17.3%). For seven patients, evidence from the literature or unpublished clinical trials was the primary justification for the allocation of a novel drug, while for 18 patients, novel treatments were trialed due to conventional treatment failure, but no specific evidence was cited (13.5%) ( Table 2). For specific indications, such as radiation necrosis, treatment for a bleeding syrinx, or to avoid resistance to targeted agent monotherapy, there was no explicit documented reference to the evidence for the use of the selected drug in that context. Overall, documentation for the specific rationale behind the choice to use novel drugs varied in detail between oncologists and also between documentation prior to and following the introduction of the EMR.
Frequency of prescription also varied between oncologists, with Oncologists 1-4 collectively accounting for 85 novel drug prescriptions (63.9%) ( Table 2). Oncologist 1 alone prescribed 38 (28.6%) of the novel drugs in the study period. The majority of novel drugs were prescribed following discussion at a multidisciplinary meeting (90.2%).
A significantly lower proportion (32.3%) was chosen following consultation with external experts, which ranged from discussions with colleagues at other local hospitals, to national meetings and consultations with international experts. Documentation for these consultations was irregular, particularly prior to the introduction of the EMR. They were only occasionally referenced in clinical notes or letters written by the primary oncologist, so it was unclear if these data accurately represented the prevalence of external opinion in treatment decision-making processes.

| Institutional approval
Twenty-seven applications to DUC for off-label novel drugs were identified for 19 patients. Seventeen patients were among the 100 included in this study and the remaining two were excluded, as meeting entries could not be reconciled with patient records. Of these, 18 applications were accepted. Two were rejected for insufficient evidence provided in the application, and imminent transition to seven who specialize in brain and solid tumors and three who specialize in hematological malignancies.
adult care for a 19-year-old patient, respectively. Four applications were withdrawn due to patient disease progression, compassionate access obtained elsewhere, and an unspecified reason (n = 2). In three applications excluded from the aforementioned counts, the decision of the DUC was not explicitly documented but was inferred from clinical notes or subsequent meeting minutes: two further approvals, and one additional rejection. In all instances, the precise clinical indication for the novel drug was not documented, and clinical context was limited to descriptors such as "urgent" or "for end-of-life care." It was unclear if DUC possessed additional information in attachments, which were not included in the minutes.
Fourteen out of the 18 accepted applications were approved for a set timeframe (doses, cycles, months, or weeks), ranging from a single dose to up to 12 months. Extensions were requested on six occasions for three patients: one had four extensions approved, another had one extension approved, and one outcome was unknown following escalation to the hospital executive. The criteria used by the committee to discuss the extensions were not explicitly documented beyond commentary that the treatment appeared to be working.
While progress was discussed for these patients, outcomes were not documented for patients, who received DUC approval but for whom extensions, were not requested. Overall, it was not possible to draw conclusions regarding the decision-making process of the DUC.

| Treatment outcomes
Ninety-one of 124 treatment regimens with novel drugs (73.4%) achieved an objective response of complete remission, partial remission, stable disease, or mixed response. Response rates were based on documentation by the respective clinical oncologist and were not reported using specified criteria such as the Response Assessment for Neuro-Oncology (RANO), 9

or Response Evaluation Criteria in Solid
Tumors (RECIST). 10 Of these 91 responses, 35 were still ongoing at the time of review in June 2020, 55 had ceased and one was lost to follow-up (     (Table 4).
Five-year overall survival (OS) rates were also calculated for the three most prevalent diagnoses in this patient cohort: low-grade glioma (LGG), high-grade glioma (HGG), and precursor B-cell acute lymphoblastic leukemia (Pre-B ALL) (Figure 1). Only 14 LGG, 11 HGG, and three deceased Pre-B ALL patients had sufficient data for analysis. No surviving Pre-B ALL patients could be evaluated as none had a sufficient follow-up duration.

| Key findings
Novel drugs were primarily utilized for patients experiencing refractory or relapsed disease (80%) and were prescribed at the second line of treatment or greater (84%, Table 1). The majority were funded by the hospital and compassionate access schemes, with only five regimens funded by patients. These novel drugs were mostly prescribed with the intention to cure (54.8%) or to palliate (24.2%), and almost three quarters of these regimens achieved an objective response. However, more than half of the objective responses (55/91, 60.4%) did not continue past an average of 9 months. The vast majority of novel drug regimens still eventually resulted in progression of disease or no response at all, implying that these novel drugs did not definitively cure these patients.
More specifically, while the majority of patients prescribed novel drugs with curative intent experienced objective responses less than half of these responses were still ongoing at the time of the audit (Table 3).
Further, since follow-up durations varied greatly, these latter patients may yet experience progression of disease as more time passes.
A recent study identified patient survival as a primary consideration for patients and their families in approaching treatment decision-making. 7 Estimating the benefit of these novel drugs on patient survival was limited in this study by the small sample size, even for the three most prevalent diagnoses of LGG, HGG, and Pre-B ALL. Five-year OS following standard-of-care treatment has been reported at 90% for LGG, 11 15% for HGG, 12 and 15% to 51% for early and late relapse in Pre-B ALL, respectively. 13 Comparing these rates with those observed in this study (Figure 1)

| Study limitations and future directions
As it is retrospective, this study is limited by lack of control for key variables such as follow-up duration, prior lines of treatment, patient diagnoses, tumor genomic profile, and the precise indications, which prompted the use of novel drugs. Promisingly, large, prospective clinical trials seeking to assign children with novel drugs based on genomic sequencing such as iCat2, 16

| CONCLUSION
Novel drugs present a promising solution to improving clinical outcomes for children with rare, relapsed, and refractory malignancies. The results of this study indicate that while using novel drugs off-label has provided patients with some benefit, more rigorous and standardized approaches to documentation are required to fully characterize the best use of novel drugs, and to advance the ethical and clinical discussions surrounding their complex implications for patients and their families.

ACKNOWLEDGMENTS
This study was supported by the Victorian Government Department of Human Services and Operational Infrastructure Support Program.
MM was funded by a Victorian Cancer Agency grant (#HSR15034).

JH was funded by a Royal Children's Hospital Foundation Grant
(2015-500).