DFCI 95‐01.

Study characteristics
Methods	Computer‐generated randomisation was performed centrally at the Quality Assurance Office for Clinical Trials (permuted block design with institutional balancing to ensure that a treatment imbalance within an institution was no greater than 3 participants).
Participants	206 children (age for all randomised participants nm: see notes; 120 boys and 86 girls) with high‐risk acute lymphoblastic leukaemia (ALL), treated with multiagent chemotherapy (including doxorubicin: see notes) and CNS irradiation. No prior anthracycline therapy. No prior cardiac radiotherapy. Some of the participants were diagnosed with prior cardiac dysfunction (by either echocardiography or the cardiac marker troponin T; definition used nm), but the exact number of participants was nm.
Interventions	Dexrazoxane (10:1 ratio of dexrazoxane to doxorubicin; IV bolus up to 15 minutes immediately before doxorubicin) (N = 105) versus no cardioprotective intervention (N = 101).
Outcomes	Heart failure (clinical heart failure defined as congestive heart failure or other symptomatic cardiac disease) Tumour response rate (no definition of complete remission provided) Adverse effects (no definition provided)
Notes	Median length of follow‐up: 2.7 years Median age: 7.5 years in intervention group and 7.3 years in control group According to protocol, children in both treatment groups should have received a cumulative doxorubicin dose of 300 mg/m2 (peak dose (i.e. maximal dose received in 1 week) 30 mg/m2; infusion duration nm). Long‐term follow‐up data of this study have been published (DFCI 95‐01; Barry 2008 and Lipshultz 2010 references). Both articles included 205 randomised participants (105 in the dexrazoxane group and 100 in the control group) as opposed to the original publication, which included 206 randomised participants. Lipshultz 2010 provided long‐term follow‐up data (median follow up in the dexrazoxane group 6.2 years; range 3 to 7.7 years and in the control group 5.7 years; range 2.8 to 7.6 years) on clinical heart failure for 134 of the 205 randomised participants, i.e. 68 of the 105 participants in the dexrazoxane group and 66 of the 100 participants in the control group. These were participants for which data were available after treatment completion. It was stated that children leaving the study did not differ in any clinical characteristic from those who stayed in the study. The median cumulative anthracycline dose in the dexrazoxane group was 300 mg/m2 (range 300 mg/m2 to 300 mg/m2) and in the control group it was also 300 mg/m2 (range 288 mg/m2 to 300 mg/m2) with an infusion duration up to 15 minutes (push or bolus). Barry 2008 provided long‐term follow‐up data (median follow up 6.2 years) on secondary malignant neoplasms. Gender: 64 males (61%) males and 41 (39%) females in dexrazoxane group and 56 (55%) males and 45 females (45%) in the control group (DFCI 95‐01 primary reference); 27 (40%) males and 41 (60%) females in the dexrazoxane group and 30 (45%) males and 36 (55%) females in the control group (Lipshultz 2010); in Barry 2008 nm. Stage of disease per treatment group: in both treatment groups all high‐risk ALL. Funding sources: grants from the National Institutes of Health (CA 68484, CA 79060, CA 55576, CA 06516, HL 59837, HR96041, HL 53392, and HL 72705), Pfizer, and Roche Diagnostics (DFCI 95‐01 primary reference); grants from the US National Institutes of Health (HL072705, HL078522, HL053392, CA127642, CA068484, HD052104, AI50274, CA068484, HD052102, HL087708, HL079233, HL004537, HL087000, HL007188, HL094100, HL095127, and HD80002), Children’s Cardiomyopathy Foundation, University of Miami Women’s Cancer Association, Lance Armstrong Foundation, Roche Diagnostics, Pfizer, and Novartis (Lipshultz 2010); grant from the National Institutes of Health (CA 68484) (Barry 2008). Declaration of interests: one of the authors has received investigator‐initiated research grant support from Pfizer, the manufacturer of dexrazoxane (Zinecard), and Roche Diagnostics, the manufacturer of the troponin T assay used in this study. Neither company had any active involvement in the study. This author also reports having received an honorarium as a consultant for Chiron, which manufactures a product related to dexrazoxane (Cardioxane) (DFCI 95‐01 primary reference); one of the authors received investigator‐initiated grants from Pfizer, Novartis and Roche Diagnostics to help support this study. Other authors have received payment for consultancy work (+/‐ stock or stock options) from Enzon Pharmaceuticals, ELISA Pharmaceuticals and/or Genzyme Corporation. The funding sources had no role in the study design, data collection, data analysis, data interpretation, or writing of the report (Lipshultz 2010); some authors received compensation for consultant or advisory roles from Chiron and Enzon Pharmaceuticals, honoraria from Enzon Pharmaceuticals, research funding from Pfizer, Novartis, Chiron and Enzon Pharmaceuticals (Barry 2008). The Vrooman 2011 publication provided long‐term follow‐up data of this study, but provided only information of dexrazoxane participants and results could thus not be used for this review. The Lipshultz 2012 publication also provided long‐term follow‐up data of this study, but a shorter follow‐up than in the Lipshultz 2010 publication and no new information was provided; thus we did not use information from this publication. The Moghrabi 2007 publication again provided long‐term follow‐up data, but no new information was provided and thus results were not included in the review.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated randomisation was performed centrally
Allocation concealment (selection bias)	Low risk	Computer‐generated randomisation was performed centrally.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel were not blinded to the received intervention (either dexrazoxane or no cardioprotective intervention).
Blinding of outcome assessors (detection bias) ‐ clinical heart failure	Low risk	The outcome assessors of clinical heart failure (at long‐term follow‐up) were blinded to treatment.
Blinding of outcome assessors (detection bias) ‐ tumour response rate	Low risk	Central investigators providing summary study results remained blinded throughout the study.
Blinding of outcome assessors (detection bias) ‐ toxicities other than cardiac damage (not diagnosed by laboratory tests)	Unclear risk	No information on blinding of outcome assessors provided for toxicities other than cardiac damage
Incomplete outcome data (attrition bias) ‐ clinical heart failure	High risk	Clinical heart failure at long‐term follow‐up evaluated in 65% of participants in the dexrazoxane group and 66% of the control group (68/105 of the dexrazoxane group and 66/100 of the control group)
Incomplete outcome data (attrition bias) ‐ tumour response rate	Unclear risk	Unclear in how many participants tumour response rate was evaluated
Incomplete outcome data (attrition bias) ‐ SMN	Low risk	SMN evaluated in 95% or 96% of participants in the dexrazoxane group and 95% or 94% of the control group (for 1 participant it was not clear from which treatment group he/she was missing).
Selective reporting (reporting bias)	High risk	Not all expected outcomes were reported (e.g. overall survival was missing)
Other bias	Unclear risk	Block randomisation in unblinded trials: unclear (information on blinding of outcome assessors for some of the outcomes not provided; block randomisation was used). Baseline imbalance between treatment groups related to outcome (prior cardiotoxic treatment (anthracyclines and cardiac irradiation), age, gender, stage of disease and prior cardiac dysfunction): unclear (no prior cardiotoxic treatment, but unclear how many in each treatment group had prior cardiac dysfunction; all other factors comparable) Difference in length of follow‐up between treatment groups: unclear (length of follow‐up not mentioned for DFCI 95‐01 primary reference and Barry 2008 reference; for the Lipshultz 2010 reference, there was a difference in length of follow‐up between treatment groups, but relevance unclear)