METHODS

We analyzed all the patients included on June 11, 2019, in the L.E.A program (clinicaltrials.gov identifier: NCT 01756599). This is a national cohort initiated in 2003 to assess the long-term health of patients who were treated in childhood or adolescence for AL, from 1980. The precise functioning of the cohort has already been defined previously 18. Briefly participants are summoned to a follow-up clinic at predefined dates, starting one year after SCT or after completion of chemotherapy. These visits continue every two years until the age of 20 and at least 10 years of complete remission (CR), and every four years thereafter. Late effects were detected by physicians through regular visits to reference centers, comprising a medical examination and adequate additional tests. Thus, fourteen late effects were prospectively assessed in the entire cohort in which ON was diagnosed.
All patients (or their parents) provided written informed consent. This study was approved by the Sud Méditerranée V ethics committee (opinion n° 2012-A00984-39), in compliance with the General Data Protection Regulation.
We divided our study in two parts. The first study was conducted on all patients with a symptomatic ON identified in the L.E.A. program, to identify the risk factors for the occurrence of this complication, as well as those for having multifocal involvement at diagnosis, and to evaluate the impact of this side effect on long term QoL
The second one focused on radiological assessment for patients who met the following criteria: presence of symptomatic ON with a MRI performed at diagnosis and available for centralized double blinded review at time of the study by 2 experienced pediatric radiologists. The proofreading was performed independently by these radiologists.
The positive diagnosis of ON was retained in the presence of a serpiginous borderline in T1 hypointense delimiting the necrosis area, joining the subchondral bone lamina for epiphyseal damage.19 The parameters concerning the location and extension of the necrosis were collected in order to classify each location of symptomatic ON according to a grade of severity in accordance with Niinimäki’s classification 16 (see Appendix 1) : site of necrosis (epiphysis, metaphysis or diaphysis), bearing character of the joint, affected articular or epiphyseal surface, presence of joint deformation. Were also listed parameters which are known to potentially modify the prognosis 20: presence of bone edema associated with the periphery of the necrosis area, presence of intra-articular effusion, other ON locations or distant bone marrow edema in the field of exploration, growth cartilages fusion and signal in T1 of the necrosis range (variable depending on the course, initially fatty then progressing to fibrosis).
In the event of disagreement on the severity grade, a joint review was carried out and a consensus was reached. Radiological severity was defined by the presence of V grade, i.e., joint deformity. In the event of multifocal involvement, we have chosen to consider the most severe impairment.
Quality of life of children and adolescents was assessed using the parents’ version of the Vécu et Santé Perçue de l’Adolescent et l’enfant (VSP-Ap) which is completed by the parents of the children and adolescents 21–23 . The questionnaire is comprised of nine dimensions and a summary score. All scores range between 0 and 100, with higher scores indicating a better Health Related Quality of Life (HRQoL). Reference values are available for the general French population > 7 years of age for sex- and age-matched comparison purposes 22,24. Adult patients were asked to complete SF-36 questionnaires which consists of 36 items. Results are divided into eight sub scales and two calculated composite scores (physical and mental composite score). The reliability of this scale both in survivors of childhood cancers and in its French version has already been validated 25

Statistical methods

The data collected is presented in the form of counts and percentages for the qualitative variables and in the form of means and standard deviations for the quantitative variables. The normality of the distribution of quantitative variables was verified. Chi-square tests or Fisher’s exact tests made it possible to compare the qualitative variables according to the different groups of interest. For the quantitative variables, these comparisons were made by Student’s tests.
Logistic regressions were carried out, making it possible to model: the risk factors for the onset of ON, multifocal and severe involvement. First univariate analyzes were performed. Variables whose p-value was less than 0.05 in the univariate analysis were retained for the multivariate analysis; a descending step-by-step selection was then carried out. Only the significant variables below the significance level (p-value < 0.05) were maintained in the final multivariate model. All analyses were carried out with a significance level of 0.05. Analyses were performed using IBM SPSS Statistics Version 20 software.
HRQoL scores of our patients were compared with the French reference scores for age and sex using paired Student’s t-tests. The calculation of inter-observer reproducibility was determined by kappa statistics between raters, and the results was interpreted according to guidelines adapted from Landis and Koch. 26