Representative simulations visualising the impact of current DOX treatment schedules and age-dependent PK differences on systemic therapy intensity are illustrated. As outlined by fig. 2 exemplarily for three treatment regimens, substantial differences in drug exposure and peak concentrations have to be expected among children aged 0 – 18. In very young children who are subject to dose adaptations these may lead to particularly sharp steps in therapy intensity. For instance, simulated typical AUC and cmax are lowest in neonates (AUC = 507 μg·L-1·h, cmax = 56 µg·L-1) and increase towards a maximum in children slightly above one year of age (AUC = 1002 μg·L-1·h, cmax = 138 µg·L-1) when treated according to the CWS-guidance (dose: 20 mg·m-2, infusion time: 3 h). Here, the standard BSA-based dose is reduced in children < 1 year or weighing < 10 kg to 67 % (< 6 months) or 100 % (≥ 6 months) of the body weight-based dose (additional file 1). Due to age-dependent differences in PK, therapy intensity can be expected to decrease with growing age. Taking the CWS-guidance as an example, simulated typical AUC decreases from its maximum 1002 μg·L-1·h to 688 μg·L-1·h at the age of 18. Similarly, however less pronounced, typical cmax decreases from 138 µg·L-1 to 117 µg·L-1.
Irrespective of schedule- and age-dependent variations a substantially broad distribution of individual AUC and cmax has to be considered due to the high variability in PK that cannot be sufficiently explained by age and BSA (fig. 3 A, B). As a result of conversion rules from BSA-based dosing to body weight-based dosing, therapy intensity will differ among infants of the same age but heterogeneous in body composition depending on the specific regimen-defined boundaries. For example, simulated median doxorubicin AUC and cmax differ more than 30 % between a one-year old child on the 95th percentile of body weight who already receives the full BSA-based dose according to the CWS-guidance and a child on the 5th or 50th percentile of body weight who still receives the body-weight-based dose (fig. 3 C, D). A comparison of simulated AUC and cmax following a single drug administration between selected treatment regimens is displayed in fig. 3 (E, F). Besides the dose, the duration of infusion determines peak concentrations leading to large differences between treatment regimens. Though the dose is lower in the NB 2016 N4 regimen (15 mg·m-2) compared to the CWS-guidance (20 mg·m-2) median peak concentrations simulated for a 2-year-old child are more than 3 times higher due to the difference in infusion time (30 min vs. 3 h).
Exemplarily, we assessed the impact of the dose adaptation that has been described by Völler et al.  on drug exposure for 94 children of the EPOC patient population (fig. 4). Application of this dosing algorithm allows to achieve a defined target AUC without relevant bias (-2.5 %, 95 % confidence interval -8 – 3 %), however, variability in drug exposure is still substantial underlined by the small decrease in precision between observed (21 %, 95 % confidence interval 18 – 23 %) and hypothetical, dose-adjusted AUC values (17 %, 95 % confidence interval 13 – 19 %) (p < 0.05). The percentage of AUC attaining the range of 80 – 125 % around the target AUC was 58.5 % for the observed AUC and 69.1 % for dose-adjusted AUC values. This difference was not statistically noticeable.
Delphi consensus procedure
The Delphi consensus procedure was conducted between September 2017 and April 2018. Overall, 28 experts were invited to participate in the Delphi procedure of whom 11 agreed to participate, one expert refused and 16 did not respond. Though 11 experts initially agreed to participate only 8 completed the first pilot phase questionnaire. The 2nd and 3rd round questionnaires were each completed by 11 experts. Both clinical centres (7 experts) and relevant paediatric study groups (4 experts) were represented in the final Delphi panel (fig. 1). In general, most questions were answered with high agreement among the panel members. For some questions, however, consent was marginally below or above the pre-specified threshold of 67 % with levels of agreement after round three of 64 % (7/11 ratings) to 73 % (8/11 ratings). Comparing outcomes from round two and three, trends in responses observed during round two were generally confirmed during the 3rd round. The median score, range of scores and the level of agreement for the 2nd and 3rd Delphi round are summarised in tables 1 – 3.
Dose adaptations for infants/young children
Panel members generally acknowledged the clinical relevance of individual differences in systemic therapy intensity which arise from regimen-specific dose modifications as well as individual differences in PK. Nevertheless, dose adaptations are still considered necessary to reduce the risk of cardiac injury in the very young. To improve the current practice, experts agreed on a standardised a priori dose adaptation that takes into account patient’s age and BSA thus allows compensating for age-dependent differences in PK. The pharmacological goal of targeting equal AUC levels across the age range was deemed appropriate. Further, a reduction of peak concentrations in younger children was favoured, albeit only 8 out of 11 experts agreed. This could be achieved by a prolongation of infusion time in these children. As current conversion rules from BSA- to body weight-based dosing can be expected to cause arbitrary differences in therapy intensity among children of the same age this should be avoided by future dose adaptations. Any reduction of the cardiotoxic risk in the high-risk population of very young children at the expense of a potentially lower tumour efficacy was not accepted by the majority of the panel members (table 1).
Standardised dosing targets
Chemotherapy regimens for childhood malignancies vary substantially in dose and infusion time with a large impact on systemic therapy intensity (fig. 3 E, F). Given the unclear role of PK characteristics such as AUC and cmax we asked whether it might be desirable to target certain minimum and maximum therapy intensity thresholds, meaning in practice to constrain the range of currently applied doses and infusion times. The participating experts considered AUC, cmax as well as the time of exposure to be important and favoured the establishment of both minimum and maximum threshold levels to balance the risk of cardiac side effects and tumour efficacy. We further asked whether such target ranges should be uniformly defined across different tumour entities. Here, outcome is not unequivocal as levels of agreement ranged from 64 % (7/11 ratings) to 73 % (8/11 ratings). Apart from a priori dose adaptations, all experts agreed that defined patient populations might additionally benefit from therapeutic drug monitoring approaches for DOX (table 2).
In their answers to the open-ended questions of the 1st Delphi round panel members raised additional aspects which were addressed in a separate section of the 2nd and 3rd round questionnaires. Adjusting DOX administration to the particular clinical needs of special patient populations was found to be relevant, which where infants/children with good prognosis disease, patients with tumour predisposition syndromes, Down syndrome patients with AML/ALL, and syndromes associated with higher toxicity of chemotherapy (e.g. Fanconi anaemia). Mediastinal/lung radiotherapy, pharmacogenetic analysis and use of liposomal DOX were considered as potentially relevant factors for DOX administration. Other co-medication and the use of a cardioprotectant were not regarded relevant (table 3).