Final Report Summary - EPOC (European Paediatric Oncology Off-Patent Medicines Consortium)
Background of consortium: Research in paediatric oncology is of necessity a collaborative enterprise. The consortium is built on existing networks of clinical centres, together with networks established through groups such as the UK CCLG, BFM in Germany and the PAMM group of the EORTC.
Following the award of funding for the doxorubicin study, the consortium held a number of teleconferences and meetings aimed at establishing the management of the group, writing the protocol and addressing the various regulatory steps necessary for the clinical trial. There were some initial adjustments to the consortium, due to a change in role at ULeic, a transfer of a beneficiary lead (from ICR to UCL) and an institutional reconfiguration (at UnivMed). A further significant change in the function of the consortium came with the transfer of some trial and data management functions from PTL to WWU.
The trial was set up with the University Hospital Münster as the sponsor. 21 clinical centres were identified for patient recruitment. One centre was eventually excluded due to contractual delays. The major obstacle to completion of the study on time turned out to be the length of time it took to set up contractual agreements with each of the centres, with some initial delay due to ethical and insurance issues. These delays necessitated an extension of the project by 9 months.
The majority of the laboratory work associated with the study was to be performed at WWU, with additional pharmacogenetic work at UNEW. CRUK commissioned and reviewed an audit of laboratory procedures and facilities at WWU prior to the start of the study.
Patient recruitment proceeded at the projected rate, aided by the website set up by PTL for patient registration and distribution of study documents. Samples were collected at the clinical centres, aggregated by the NSMs in each country and forwarded to WWU for analysis. This arrangement permitted the resolution by the NSM of any queries relating to samples or documentation.
A planned interim analysis was performed after the analysis of data from 30 patients. This was associated with a physical meeting of the consortium. The interim analysis confirmed that the study design was appropriate and that recruitment was proceeding towards targets, especially in the target cohorts of children less than 3 years and less than 1 year.
After the completion of patient recruitment, the focus shifted towards analysis of samples and collation of study data. These tasks were completed in good time, and allowed a discussion of the preliminary results of the study at a scientific symposium. The meeting was attended by all members of the consortium, invitees from tumour specific groups and clinical centres which had contributed to the study, and by representatives from the Data Safety Monitoring Committee, the EMA and EC. In addition a reviewer appointed by the EC also attended the meeting.
The primary aims of the study were achieved, in that the age-dependency of doxorubicin pharmacokinetics was determined. The influence of age and other factors will be further established by the application of population pharmacokinetic models to the data.
The consortium have presented their data at a number of international meetings, with the final results presented at the ECCO meeting in Amsterdam in September 2013. Further dissemination and publication is planned, with 7 publications to be written from the data. UNEW will maintain a website hosting all of the relevant documents and publicizing the activities of the consortium, which will continue to contribute to the activities of Enpr-EMA and other groups. The consortium will continue dialogue with the EMA and with the tumour specific groups to establish the optimal method for incorporation of the study findings in SMPCs and treatment guidelines.
Project Context and Objectives:
The treatment of cancer in children has made significant advances over the last 30 years. Up to 80% of patients now achieve long-term cures. However, for several tumour types, the cure-rate is less optimistic. For instance, patients with neuroblastoma or Ewings sarcoma have a survival rate of approximately 60%, with an even worse prognosis for particular sub-groups of patients. Conversely, other tumours may have better survival rates, but at the expense of potentially severe long-term toxicities associated with chemo- and radiotherapy.
Chemotherapy has made a major contribution to improvements in the treatment of childhood cancer. However, progress has, in many cases, been achieved in the absence of a systematic pharmacological evaluation of the active agents. Often the dosing of these drugs in children has been based on extrapolation from studies in adult patients (McLeod 1992). In many cases, there are no pharmacological data on which to base current dosing regimens.
In choosing the appropriate dose of an anticancer drug in paediatric oncology, the initial starting point will be the corresponding adult dose, usually in mg/m2. Some downward adjustment of this dose may be applied to give a margin of safety and to allow for combination with other drugs in a multi-agent regimen. Subsequently, doses may be adjusted based on observed toxicities. Once established, this maximum tolerated dose will be applied in other therapeutic scenarios. Even at this dose, a proportion of patients will experience clinically-significant toxicity. Also, given the narrow therapeutic window of these drugs, it is likely that a number of patients will be given less than the optimal dose.
For many drugs it is established that such variation in clinical outcome depends on variability in the pharmacokinetics and metabolism of the drug in individual patients. It has been known for some time that the clearance of carboplatin is determined by renal function (Newell 1993), and that the area under the plasma time curve (AUC) is closely linked to both toxicity and antitumour effect (Jodrell 1992). The inverse relationship between clearance and AUC is exploited to guide dosing of carboplatin based on renal function in both adult and paediatric patients (Thomas 2000). This pharmacological approach to dosing is particularly important in high-dose carboplatin regimens, where the therapeutic window is particularly narrow (Veal 2007). Dosing to a target AUC has also been applied in anephric patients (Veal 2004).
In order to apply the principles of pharmacologically-guided dosing, it is first necessary to characterize the pharmacokinetics, and variability therein, within the relevant patient population. There are few studies in paediatric oncology that have generated data in a sufficiently large patient population such that pharmacologically-guided dosing could be applied systematically. Furthermore, there are subpopulations of patients, such as those less than 3 years old, for whom almost no data exist. Yet another consideration is the quality of the data from existing studies, and compliance with ICH-GCP. Such compliance would be essential if these data were to be used to guide therapy or in a submission to the appropriate regulatory authorities.
While an improved regulatory framework for drugs in paediatrics has been the major motivating force behind the EMEA priority lists, the selection of the individual drugs and the requirement for specific investigations detailed in these lists has been based on clinical need. Of the several oncology drugs on the EMEA list, doxorubicin in particular is used very widely in the treatment of a variety of paediatric tumours. Moreover, there is a clear need for an improvement in our understanding of the pharmacology of this agent, especially in very young patients.
The collaborators on this project represent groups who have successfully performed many pharmacological and biological studies in paediatric cancer patients. However, these studies have been limited to their respective home countries, which in turn has limited the number of patients available. Patient numbers are particularly limited in very young patients. Such restrictions have placed these groups at a disadvantage, particularly compared to investigators in countries with larger patient populations. It was, therefore, essential that a successful collaboration was established across the European member states of those investigators with the appropriate experience. Without such a collaboration, the aims of the EMEA priority list would not be easily achieved in paediatric oncology. The overall aim of the consortium was to establish a network of collaboration to perform pharmacological studies in paediatric cancer patients, with a study of doxorubicin as an initial, proof-of-concept study.
The proposed research of the EPOC project was based on a clinical pharmacology study protocol for doxorubicin. Patients receiving treatment with doxorubicin were eligible for enrolment, provided that they were also registered on a Europe-wide clinical study protocol. Collaboration with the chief investigators of the latter studies allowed access to the relevant clinical data for each patient, without the need for parallel, duplicate investigations.
The study was to consist of 5 active phases. These were:
1. Establishing the group, developing the trial protocol and addressing regulatory issues.
2. Patient recruitment and sample collection
3. Analysis of samples and generation of pharmacokinetic and pharmacodynamic data
4. Population pharmacokinetic analysis and integration of clinical data and pharmacogenetics
5. Reporting and dissemination of data, together with IP and regulatory management
The second of these phases presented the largest challenge. To optimize successful patient recruitment, lead centres in the four participating countries were designated National Study Managers (NSMs). Each has links to clinical investigators within each country, and in turn these clinicians are leading investigators for clinical trials in the major tumour types for which doxorubicin is most important. This matrix approach was coordinated from the Northern Institute for Cancer Research (NICR) at UNEW and mirrors that applied successfully in paediatric pharmacology studies in the UK. These existing UK studies are funded by Cancer Research UK. Trial management was provided by WWU, who also coordinated the population pharmacokinetic analyses of data. Cancer Research UK provided support for the regulatory aspects of the proposed research. The International Confederation of Childhood Cancer Parent Organisations (ICCCPO) provided input on ethical issues, a link to patient and parent groups and oversaw the dissemination of the results of the research, together with UNEW.
Pharmacology studies in paediatric cancer patients in Europe have proceeded largely within individual countries, with few examples of genuine international collaboration. Notable exceptions to this have been the previous collaboration between CCLG in the UK and SFOP in France, and the ITCC group funded under FP6. In addition, a range of pharmacological investigations in patients with high-risk neuroblastoma have been coordinated by UNEW. These initial collaborations formed the basis for the proposed network for future studies.
The aim of the research proposed was to provide high-quality data on doxorubicin. These data were to serve three main purposes:
a. To provide a rationale for the selection of optimal doses in paediatrics, particularly in patients less than 3 years.
b. To provide high quality data on the pharmacology of doxorubicin.
c. To establish a European network for future clinical pharmacology studies in children with cancer.
Although there are existing reports of the pharmacokinetics of doxorubicin in children, these have largely been single centre studies, with relatively small patient numbers. Moreover, the quality of these data has not been validated in accordance with ICH-GCP and there has been little attempt to apply population pharmacokinetic approaches or to incorporate pharmacogenetic and pharmacodynamic endpoints. The proposed advance to be provided by the consortium was to bring together a collaborative group with the prospect of recruiting sufficient patient numbers to fulfill these aims.
Project Results:
Addressing pre-analytical issues
Paediatric oncology patients, almost without exception, have an in situ central venous access. This may be a single or double lumen line, and in some cases may feed into a sub-cutaneous reservoir (Portocath). This IV access is convenient for the administration of chemotherapy, and also provides a means of obtaining blood samples for biochemistry and haematology. While the CVC is also suitable for obtaining blood samples for pharmacokinetic studies, care must be taken to avoid contamination with the administered infusion fluid, and to obtain a sample that is representative of the circulating blood, without dilution due to the deadspace of the CVC. This issues have been considered in the design of the EPOC study, and appropriate guidelines for sampling have been established. The results of these evaluations have been published in a peer-reviewed journal .
Study design
Although doxorubicin has been used in paediatric oncology for many years, there are relatively few studies that provide a sound pharmacological basis for the dosage regimens employed. Indeed, one of the key logistical problems for the EPOC study was to design pharmacokinetic sampling schedules with sufficient flexibility to accommodate the different doses, infusion durations and schedules employed for the administration of doxorubicin in different treatment regimens for different tumours. Using raw data generously provided from studies by Thompson in children and by Callies in adults, it was possible to simulate plasma concentration time profiles associated with each of the different treatment regimens. Thus, it was possible to identify the optimum sampling times for the estimation of the parameters of the pharmacokinetic model, minimizing the number of samples required especially in very young patients. The results of this investigation were published and formed the starting point for the analysis of the data generated in the EPOC study. Although the dataset provided by Thompson was from a paediatric study, there were no data available from children less than 3 years of age.
PK data.
Based on the available pre-existing data, and confirmed by the planned interim analysis, a 3 compartment model was the starting point for the population pharmacokinetic analysis of doxorubicin. The model comprised a central compartment (volume V1), with elimination characterized by a clearance term Cl. There were two peripheral compartments of volumes V2 and V3, with transfer between V1 and each of the other two compartments governed by clearance terms Q2 and Q3 respectively. The analysis proceeded in a stepwise fashion:
1. Comparison of different error models
2. Consideration of inter-individual variability (IIV) in all PK parameters, where possible
3. Test of covariance effects between IIVs.
4. Test of inter-occasion variability (IOV) on Cl, V1 and both of these primary parameters together.
5. Comparison of different estimation methods within NONMEM.
6. Test of covariate effects, including pharmacogenetic variants.
For doxorubicinol, a fourth compartment was introduced into the model, with parameters for the clearance of doxorubicin to doxorubicinol (Q4) and for the clearance of doxorubicinol from this fourth compartment (ClM). As only limited data were available on the fate of the metabolite, Q4 was fixed to a value estimated from the literature.
The pharmacokinetics of doxorubicin and doxorubicinol was to be studied in 100 patients, with a focus on patients less than 3 years (minimum 20 patients) and on those less than 1 year (minimum 5 patients). Any patient enrolled in a European protocol which included more than one cycle of administration of doxorubicin was eligible. Patients were to be studied on two cycles to allow the investigation of between occasion or intra-individual variability. Patients less than 3 years who were treated according to a European protocol could be studied regardless of whether or not they were treated as part of an ongoing therapeutic study.
Pharmacokinetic data were available from 94 patients, with data from at least one sampling period. The second sampling period was missing for 4 patients, either because of disease progression or relapse, or because patients or parents withdrew their consent. There were particular problems with obtaining samples during the infusion, particularly in countries where peripheral sampling (either by venipuncture or be needle stick capillary sampling) was not available. Of the samples obtained, 2 had insufficient volume for determination of drug or metabolite, and 41 (6.2%) had insufficient volume for determination of the metabolite. Two samples were lost as blood rather than plasma was shipped to the lab, and 35 samples (5.3%) were mis-directed during shipping and had thawed. A risk assessment based on the known stability of the analytes in plasma resulted in the exclusion of these samples from the analysis.
Analysis of samples from 181 sampling periods in 94 patients was completed and data reported for audit by CRUK. These data formed the basis of the population pharmacokinetic analysis described below and a report describing these data is uploaded as a deliverable.
PD data
The aim was to collect plasma samples for the measurement of potential biomarkers of cardiotoxicity, together with data from measures of cardiac function. Overall, 13.7% of samples provided incomplete data, either because there was insufficient plasma for the measurement of all markers, the sample was affected by haemolysis or the labeling was unclear. These PD data are available in a final pharmacodynamic report, uploaded as a deliverable.
PG data
Samples for determination of genotype were transferred to the NICR, Newcastle. Of the 101 patients recruited, DNA was available from all but 5. One patient withdrew consent, 2 had clotted blood samples from which DNA could not be extracted and in one case plasma was provided instead of whole blood. Determination genetic variability was performed by standard methods, with the results delivered to WWU for inclusion in the population PK analysis. The output from the pharmacogenetic analysis are the subject of a separate report uploaded as a deliverable of the project.
Population PK model
The major activity of the consortium was in running the clinical study was to provide data for a population pharmacokinetic model. The major question to be addressed in applying this analysis was whether or not there is a significant effect of age on the pharmacokinetics of doxorubicin. Secondary questions were to investigate the pharmacokinetics of the major metabolite (doxorubicinol) and to investigate the potential impact of pharmacogenetic variants. The parameters of the population PK analysis were to be further employed in investigations of the links between pharmacokinetics and the pharmacodynamic observations on markers of cardiotoxicity. The results of the population pharmacokinetic analysis are included in a report which has been uploaded as a deliverable of the project.
Recommendations for infant dosing?
The analysis identified that patients less than 3 years of age had a lower value of clearance than that in older children, even after accounting for the difference in body size. Given that most regimens for drug administration use body size as the main means of determining the appropriate dose, this finding indicates that children less than 3 years are exposed to higher concentrations of doxorubicin for a given dose than are older children. The significance of this is potentially two-fold. Firstly, younger patients may be at greater risk of toxicity associated with their treatment with doxorubicin, including an increased risk of long-term cardiotoxicity. Secondly, since drug administration is assumed to have a uniform impact across all ages in a clinical trial, there may be some bias towards better treatment outcome in younger patients, who have been exposed to higher concentrations of the therapeutic agent. Conversely, this may be offset by an inability to deliver the target regimen in these patients if unacceptable toxicity intervenes.
Potential Impact:
The activities of the consortium have had a number of impacts, some directly as a result of the project, others with slightly wider implications in terms of the EMA and paediatric oncology. There have also been impacts within the consortium as the relative roles of beneficiaries have shifted during the project.
Early on in the project, the UNEW and WWU engaged in discussion with the EMA regarding the regulatory status of our study. Although the call and the application for FP7 funding featured a future application for a Paediatric Usage Marketing Authorisation as a future goal of this project, it was not clear if or how this could be achieved. Although the project met the demands of the EMA priority list, it was not possible to generate interest from a commercial partner and the commercial application or exploitation of the data to be generated could not be envisaged. Using the Scientific Advice service of the EMA, members of the consortium discussed the aims of the project with the EMA, seeking to determine if a PUMA was indeed not possible, and if this was the case, how the data generated in the study could be made available to clinicians to inform dosing of doxorubicin. Given that doxorubicin was already an authorized product, the likelihood of any commercial exploitation was small and the outcome of these discussions was a statement by the EMA that the project would not result in a PUMA. Future discussions would be held to determine the optimal method for dissemination of the results to the clinical community and other interested parties. These discussions have led to further reconsideration of the interplay between the EMA priority list for off-patent medicines and the funding streams made available by the EC.
From the regulatory point of view the project offers an impact on a standard regulatory problem. In order to avoid undue delays in making new drugs to available to patients, labeling for children is often based on a small number of studies with only limited patient numbers. The concept of multicenter international therapeutic drug monitoring study with population-pharmacokinetic analysis study aligned with the standard use of the drug could be a model to address the lack of data on pharmacokinetics in children. Trials with such a design might be optimal for conditional approvals of drugs in children. The EPOC is a model of this approach.
Within the practice of paediatric oncology, and specifically studies of the pharmacology of drugs in children with cancer, there have been a number of impacts resulting from the conduct of the study. Best practice for nurses and clinical staff has been shared amongst the 20 clinics participating in the study, with standardized procedures for drug administration and for blood sampling. Some cultural differences have been highlighted. The major one of these is that peripheral blood sampling is not common in clinical centres in the UK, whereas it is common practice in the other 3 MS.
In terms of designing the study, the approach of giving a precise definition to the nature of relevant SAEs and the duration for which any events (SAEs and SUSARs) were pertinent to this specific study was important and sets a standard for future studies.
From an early stage, the consortium has sought to engage with patient groups, through beneficiary 15 (ICCCPO). The design of the clinical protocol was extensively discussed with the relevant clinical tumour groups and with the EMA. These initial discussions formed the foundation for our ongoing dissemination activities during and beyond the lifetime of the project. The work of the group has been presented at a number of national and international meetings, including the AACR, ECCO and NCRI. The consortium has also published papers based on the study design issues described above and reported as deliverables of the project.
A major dissemination activity of the group was the organization of a symposium towards the end of the project. This was attended by the beneficiaries, clinical collaborators, research nurses, representatives of the clinical tumour groups, EMA and EC, and a reviewer appointed by the EC.
Dissemination beyond the lifetime of the funded project will continue. The website: http://research.ncl.ac.uk/epoc/ will host details of the group’s activities, reports and publications. The outputs from the research will be the subject of up to 7 publications, focusing on the scientific findings, clinical implications and regulatory issues relating to the EMA/FP7 interaction. Further results will also be presented at international meetings.
List of Websites:
http://research.ncl.ac.uk/epoc/