Leukaemic Involvement of Bone

Telling a child with leukaemia from a child with juvenile idiopathic arthritis is the version of this problem that reaches a clinic rather than an emergency department, and it has a quantified answer. A multicentre case-control study of 277 children first seen in paediatric rheumatology clinics - 71 who turned out to have ALL and 206 who turned out to have JIA - tested which findings at that first visit actually predicted leukaemia:

Three-quarters of the children with ALL had no blasts on the peripheral film at that visit. That single figure is why "the film was normal" cannot be allowed to close the question, and why the marrow, not the film, is the test that settles it.

The platelet row inverts the usual instinct, and is worth reading twice. Frank thrombocytopenia and a two-line cytopenia are the late and easy findings; by then the diagnosis is rarely missed. The child who gets missed has a blood count that a quick glance calls normal, and the discriminating observation is a relative one - platelets that ought to be reactively high in a child who is systemically unwell, and are instead sitting quietly in the lower half of the reference range.

Three actions taken in good faith before the marrow is examined each cause specific, avoidable harm.

Why a trial of steroids is the worst of the three. Glucocorticoids are themselves lymphotoxic - they are a component of ALL induction. Given empirically they will partially treat the leukaemia, so the child improves, which is read as confirmation that the diagnosis was inflammatory. Meanwhile the blasts are cleared from the marrow and the peripheral blood, so a marrow examination performed afterwards can be non-diagnostic or falsely reassuring, delaying diagnosis further and making accurate immunophenotyping and cytogenetic risk stratification - which determine the intensity of the treatment the child will receive - unreliable. Rapid cytoreduction in an undiagnosed child also risks tumour lysis without the prophylaxis that would have been given had anyone known. Take the blood count and film, and if there is any doubt the marrow, before the first dose - not after it.

Neuroblastoma deserves particular attention because it is the one other malignancy that reproduces almost every feature relied on above. It occurs in a younger child, typically under five; it metastasises to bone with periosteal reaction and permeative lysis; and because it infiltrates marrow it can produce exactly the same cytopenias, so a low count does not distinguish the two. The features that do are extra-skeletal: periorbital ecchymosis with proptosis from orbital deposits, an abdominal mass crossing the midline, hypertension, and in some children opsoclonus-myoclonus. Urinary vanillylmandelic and homovanillic acid are the screening test. The practical point is that the marrow examination arranged for suspected leukaemia will often reveal it, which is another reason the marrow - not a bone biopsy - is the right first invasive investigation in a child with multifocal bone lesions and abnormal counts.

How much does age matter? More than almost any other risk factor in orthopaedic oncology. In 1,095 children treated on three consecutive Japanese protocols, symptomatic osteonecrosis on the dexamethasone-era protocol occurred in 0.42% of those treated under the age of 10 and 15.6% of those aged 10 or over - a roughly thirty-fold difference, and the reason surveillance and counselling are pitched at adolescents rather than at all children on treatment. Girls fared worse than boys on the same protocol (6.7% versus 1.3%).

The choice of steroid matters too. Across the same three protocols, osteonecrosis occurred in 0.76% and 0.35% when prednisolone was the only steroid, and rose to 3.6% once dexamethasone partly replaced it. A separate prospective Dutch cohort of 466 children aged 4-18 on a dexamethasone-based protocol found symptomatic osteonecrosis in 6.4%. Dexamethasone gives better central nervous system penetration and better leukaemia control; the osteonecrosis is the price, which is why the trade-off is a live one in protocol design rather than a settled question.

That same Dutch cohort answers a question the exam likes: bone density before treatment did not predict who developed osteonecrosis, so a baseline DXA does not identify the at-risk child. Bone density then fell faster in those who developed osteonecrosis than in those who did not - the osteonecrosis drives the bone loss, partly through the enforced restriction of weight-bearing, rather than low bone density causing the osteonecrosis. The practical consequence is that protected weight-bearing, which the joint needs, works against the skeleton, and bone-health measures should be intensified at the moment osteonecrosis is diagnosed rather than left until a fragility fracture appears.