DISCUSSION
Our study on osteonecrosis was carried out on one of the largest cohort
of children and adolescents followed for leukemia, and it was the first
radiological study conducted in pediatrics on such a population. We
demonstrated that osteonecrosis in the follow-up of childhood acute
leukemia was a rare complication (2.5%) except for children over 10
years at the time of diagnosis (10.6%). Early access to MRI allowed
characterization of the lesions which were most often multifocal at
diagnosis. Moreover, the occurrence of osteonecrosis had a global and
rapid negative impact on quality of life which lasted in adulthood.
The incidence of symptomatic ON remained similar (2.8% in 2019 vs
2.5%) 12 in comparison with the L.E.A. 2013 study but
the current size of the cohort made the exploration of the risk factors
of occurrence and consequences on QoL more relevant.
We confirmed that being a female older than 10 years and suffering from
ALL increased the risk of ON 8,12,27,28. The absence
of ON in AML patients who did not undergo SCT could be explained by
shorter treatment not involving corticosteroid therapy. Whereas an
hormonal role on the occurrence of ON has been described14 in literature, the lack of significance of the
gender in the transplanted population could be explained by two factors:
the major role of overall metabolic disturbances induced by the
transplant and the possible lack of vigilance for post-transplant
hormonal substitution in female for the oldest patients. We also
confirmed in our large cohort the impact of chronic GVHD12. We did not evaluate in our current cohort the
cumulative steroid dose because its role in ON occurrence was well
described in various studies but we can speculate that the negative
impact of an history of relapse could be in part due to an higher
cumulative steroid dose 5,9,10.
Furthermore our study showed for the first time on an homogeneous
population that children who received treatment against leukemia,
presented with multifocal joint involvement (mainly weight-bearing
joints) at ON diagnosis while in other systemic pathologies, such as
sickle cell anemia, also known for the frequent occurrence of
osteonecrosis, involvement is predominantly unifocal (80% of cases)29. This is probably due to corticosteroid therapy,
since Kuhlen et al has shown it in adult population30 and Krez et al also showed that a high
cumulative dose of glucocorticoids was associated with a risk of
multifocal disease in patients with different underlying conditions like
systemic lupus erythematosus 31.
We also described the continuing negative impact of the occurrence of
osteonecrosis on physical, social, and psychological scores. The absence
of changes in QoL between the onset of ON and the last assessment
suggests that QoL was permanently impaired as soon as ON occurred. The
weak therapeutic arsenal, represented mainly by chirurgical joint
prosthesis 32, could explain the lack of possible
improvement in the quality of life in these patients.
Finally, our study gave new insights about radiological presentation of
ON in the context of leukemia. The widespread access to MRI since 2009
has allowed a representative description of more than 70% of patients
with osteonecrosis. We chose to exclude the re-reading of the images of
the patients who benefitted from MRI with a delay longer than 6 months
(10%) to avoid the risk of lesion modification over time. Because a
minor stage change for the same articulation on the follow-up MRI was
found in only 20% of patients, we could probably include them in a
further study. The centralized blinded proofreading by two radiological
experts was a strength of our study. They could correct the diagnosis of
osteonecrosis for 3 patients whose MRI lesions were osteochondritis, an
isolated diaphyseal infarction and an inflammatory arthritis. The rare
disagreements during the re-reading of the MRI have always been linked
to images that were difficult to interpret due to a lack of sequences,
when the measurement of the affected joint surface was very close to
30%.
We found mainly severe multifocal radiological damages probably because
we included only symptomatic ON. We questioned here the interest to
perform systematically a whole joint MRI for searching other damaged but
asymptomatic joints in the aim to detect milder lesions and preserve the
joint destruction. The radiological diagnosis of these asymptomatic
joints could then result in a need for restrictive care like discharge
by crutch or wheelchair, balneotherapy and physiotherapy sessions while
the patient is not in pain. The benefit of this early diagnostic imaging
remains to be demonstrated by a novel study. Our study also raised the
issue to perform systematically whole weight bearing joint MRI for
children over 10 years old, especially for females with multiple
sequelae. The radiological follow-up of these asymptomatic patients
should be analyzed in a prospective study to find out whether they
progress to regression of the MRI stage. Indeed, Inaba showed recently
that MRI at the end of induction of ALL treatment could identify
susceptible patients who could benefit of an extensive follow-up imaging
and early medical intervention 33 but this study
should be confirmed.
Our retrospective study had two main limits: the frequent absence of
available MRI for the ON diagnosed before 2009 which limited the size of
the study n°2 and the lack in the L.E.A registry of data of accurate
information on medical intervention, the use of physiotherapy or
discharge for the ON. We were not able to conclude that medical
intervention has an impact by the retrospective nature of our study.
Prospective data about treatment are essential to establish management
recommendations in the future as well as biological insights. Since not
all patients exposed to the same treatment will develop ON, we can
assume that pathophysiological mechanisms are underpinned by genetic
polymorphisms. Some studies go in this direction, with an impact of the
nucleotide polymorphisms of the plasminogen activator inhibitor 1-gene,
or glutamate receptor GRIN3A 34,35. The nested case
control French GENLEA01 study (Genome-wide association studies GWAS
identification of genetic factors affecting the occurrence of late side
effects in childhood leukemia survivors from the L.E.A cohort) could
help to solve these questions in the future.