Results:
One hundred and five trials were included: 93 from the COG website (COG/CCG/POG) and 12 protocols recruiting patients in European countries—ten from European collaborative groups and two from national groups (DCOG in the Netherlands and GPOH in Germany). The trials included are listed in Supplemental Table S5 . Most of the protocols involved concurrent chemoradiation with chemotherapeutic agents including vincristine, cisplatin, carboplatin, cyclophosphamide, ifosfamide, and etoposide. Anthracyclines were frequently used but generally not during the radiation therapy course.
Thirty-eight unique OARs were found. The number of different constraints per OAR within the protocols varied widely, ranging from a single constraint for the hypothalamus from COG ACNS 0222 to 29 unique kidney constraints within 68 protocols. Other OARs with at least 10 unique constraints included the liver, lungs, spinal cord, optic chiasm, optic nerves, heart, brainstem, and brain at 26, 24, 20, 16, 14, 13, 11, and 11, respectively. An example of the variability of dose constraints in a critical organ can be seen with the spinal cord constraints ranging from a Dmax of 40 Gy up to V57 Gy < 10% (less than 10% of the volume receiving 57 Gy). For several OARs, different protocols chose the same dose metric but assigned a wide range of volumetric limits. For example, the heart V30 Gy limit ranged from 40-100%. For kidney, V12 Gy was limited by some protocols to 20% but for others, up to 100% and the V14.4 Gy limit ranged from 33-100%. For kidney D50%, constraints ranged from 8-24 Gy. For cochlea, the allowed dose for 50% volume ranged from 20 Gy to 40 Gy. Further dose-volume metrics are listed inTable 1 . Example diagrams illustrating the range of constraints, pediatric group, and number of trials associated with each constraint can be seen in Figure 1A-C .
When comparing the high dose-volume constraints (Dmax or dose to a volume ≤ 20%) between pediatric US constraints and European constraints, 13 of the 38 OARs had at least two constraints with either a Dmax or a volumetric parameter of ≤ 20%. US and European constraints matched in five of these OARs (brain, cornea-lacrimal gland, optic chiasm/nerves, and small-large bowel). European constraints had higher dose allowances in one OAR (brainstem), while US constraints were higher in seven OARs (spinal cord, bladder, heart, kidneys, lungs, liver, and mandible), Table 2 .
The conventionally fractionated constraints for all trials organized by the OARs are listed in Supplemental Table S1 with available Pediatric Normal Tissue Effects in the Clinic (PENTEC) data listed for comparison 15–19. PENTEC is an ongoing systematic effort to summarize and, where feasible, suggest OAR-based constraints for children and adolescents based on published evidence15. Four protocols addressed proton constraints (an additional 3 trials that are currently active were not included) and three trials included SRS constraints (an additional 2 trials currently active were not included) listed in Supplemental Table S2 and S3 . Additionally, the interactive web application URL is also available in Supplemental Data S4 . To account for changes in constraints over time, we included within the web application the option to filter by start date of the protocol. Although much has changed technologically over the past 30 years, radiotherapy dose constraints generally did not show any consistent pattern of change over time for any of the OARs including the spinal cord, brainstem, optic apparatus, lungs, heart, and kidneys.