Leukaemia cutis: the skin nodules that can be the first sign of infant leukaemia
also called leukemia cutis · skin infiltration · blueberry muffin baby · cutaneous leukaemic infiltrate · aleukaemic leukaemia cutis
Leukaemia cells can settle in the skin and raise firm blue, red or purple lumps, most often in newborns and babies, and sometimes the lumps appear before anything abnormal shows up in the blood. In childhood myeloid leukaemia only about 5% of children have it, but among those who do, nearly two thirds are under two years old and almost half carry a broken KMT2A gene.
The same thing, for a scientist
Leukaemia cutis is a specific dermal or subcutaneous leukaemic infiltrate, distinct from the non-specific rashes and petechiae of marrow failure. It is strongly age- and lesion-associated: in the ELAM02 paediatric AML cohort 24 of 438 (5.5%) presented with leukaemia cutis, the LC-positive group had a median age of 1.2 years versus 8.7 years, 63% (15/24) were under 2 years, KMT2A/MLL rearrangement was present in 46% versus 21% of LC-negative cases (p = 0.008), more than 60% were FAB M5, and overall survival was significantly inferior (p = 0.006). In congenital leukaemia a systematic review of the world literature found specific cutaneous infiltrates in 25-30% of infants (not the 50% frequently quoted), typically firm blue, red or purple nodules in a generalised distribution, and the lesions may precede any other manifestation of leukaemia by up to 4 months ('aleukaemic' leukaemia cutis). In congenital disease, unlike in adults, leukaemia cutis does not alter the natural history.
What it looks like, and what it is not1 deeper
Leukaemia cutis presents as firm, non-blanching nodules or plaques, classically blue, red or purple, in a generalised distribution; in a neonate a widespread nodular eruption of this kind is one cause of the 'blueberry muffin' appearance 2. The lesions are infiltrated rather than haemorrhagic, so they are palpable and do not fade with pressure.
It must be distinguished from three things that are far more common in a leukaemic infant:
- Petechiae and bruising from thrombocytopenia, which are macular, non-palpable and blanch-negative but not infiltrated (
concept-bruising-petechiae-thrombocytopenia). - Non-specific rash, recorded in 35% of children with leukaemia in the presenting-features meta-analysis - a figure that includes drug eruptions, viral exanthems and seborrhoeic-looking rashes and is not leukaemia cutis 3.
- Other blueberry-muffin causes in a newborn: congenital infection (CMV, rubella, toxoplasmosis), neuroblastoma, rhabdomyosarcoma, extramedullary haematopoiesis and vascular lesions.
Diagnosis is by skin biopsy with immunophenotyping, and in an infant it can be the biopsy that establishes the diagnosis before the marrow does.
Aleukaemic leukaemia cutis: skin first, blood later
A striking minority of infants have skin infiltration with a normal blood count and no marrow disease. The world-literature review found that congenital leukaemia cutis may precede other manifestations of leukaemia by as much as 4 months 2, and the case literature includes lesions that spontaneously regress before overt leukaemia declares itself. Clinically this means a firm nodular eruption in a neonate warrants biopsy even when the blood count is normal, and a child with biopsy-proven aleukaemic leukaemia cutis needs surveillance rather than reassurance.
Biologically it is a more interesting observation than it looks: it implies the transformed clone can be present, viable and tissue-resident while the marrow and blood look normal. For a fusion that is thought to arise in utero, that is a natural experiment in latency.
The numbers, and a disagreement worth noticing
| Statement | Frequency | Population | Source |
|---|---|---|---|
| Leukaemia cutis at presentation | 5.5% (24/438) | childhood AML, ELAM02 trial | 1 |
| Of those with leukaemia cutis, proportion under 2 years | 63% (15/24) | childhood AML, ELAM02 | 1 |
| KMT2A/MLL rearrangement | 46% vs 21% in LC-negative (p = 0.008) | childhood AML, ELAM02 | 1 |
| Specific cutaneous infiltrate | 25-30% | congenital leukaemia, world literature | 2 |
| Skin involvement ('cutaneous infiltrates') | 67.8% | published congenital leukaemia cases | 4 |
| Skin involvement | 14.5% under 2 years vs 2.6% over 2 years | childhood leukaemia | 4 |
The 25-30% and the 67.8% figures cannot both be right, and the disagreement is instructive. The dermatology review explicitly set out to test the widely repeated claim that 50% of infants with congenital leukaemia have skin involvement, counted approximately 175 reported congenital leukaemia cases and identified 'at least 41 and possibly 56' with leukaemia cutis, and concluded 25-30% 2. The higher figure in the tertiary source is drawn from a pooled case-report series in which a striking skin eruption is itself part of the reason a case gets published 4. This node takes 25-30% in congenital leukaemia and ~5% in childhood AML overall as the defensible figures, and treats anything above 50% as publication bias.
Why leukaemia cutis is a KMT2A signal1 deeper
In the ELAM02 cohort, KMT2A/MLL rearrangement was more than twice as frequent in children with leukaemia cutis (46%) as in those without (21%, p = 0.008), the LC-positive group was dominated by infants (median age 1.2 years, 63% under 2 years) and by FAB M5 monoblastic morphology (>60%) 1. In other words leukaemia cutis in a child is, probabilistically, a flag for KMT2A-rearranged monoblastic disease of infancy - which makes it a clinical sign that should accelerate the split-signal FISH request (concept-fish-split-signal-kmt2a).
The tissue-tropism question this raises connects to the CNS story. KMT2A-rearranged blasts express NG2/CSPG4, a proteoglycan with a role in melanoma metastasis and brain development, and NG2 is upregulated on blasts infiltrating extramedullary sites (see concept-cns-involvement-at-diagnosis). Whether the same migratory programme underlies skin, CNS and gingival infiltration in KMT2A-rearranged disease is, as far as this node's search found, untested - and it is a directly answerable question.
Does it change the prognosis?
The two sources disagree, and both may be right in their own population. In childhood AML, overall survival was significantly inferior in children with leukaemia cutis (p = 0.006) with relapse in 50% versus 35% 1. In congenital leukaemia, by contrast, the review concluded that 'whereas leukaemia cutis is associated with a poor prognosis in adult leukaemics, the natural history of congenital leukaemia is not altered by leukaemia cutis' 2 - congenital leukaemia is already so aggressive that skin involvement adds no further information. Note also that the ELAM02 association is confounded by age and by KMT2A status, both of which are independently prognostic, so leukaemia cutis may be a marker of the underlying lesion rather than an independent adverse factor.
| sign | leukaemia cutis (specific cutaneous leukaemic infiltrate) |
|---|---|
| appearance | firm, non-blanching blue, red or purple nodules or plaques, generalised distribution; a cause of the 'blueberry muffin' neonate |
| aleukaemic variant | cutaneous infiltrate may precede other manifestations by up to 4 months; biopsy indicated even with a normal blood count |
| oral gingival involvement congenital PCT | 1 |
| associated morphology | FAB M5 monoblastic (>60% of LC+ in ELAM02) |
| prognostic | adverse in childhood AML (OS p=0.006) but not in congenital leukaemia; confounded by age and KMT2A status |
frequencies — the numbers behind the chart
[
{
"finding": "leukaemia cutis at presentation",
"pct": 5.5,
"population": "childhood AML (ELAM02, n=438)"
},
{
"finding": "proportion of leukaemia cutis cases aged under 2 years",
"pct": 63,
"population": "ELAM02 LC+ group (15/24)"
},
{
"finding": "KMT2A/MLL rearrangement in LC+ vs LC-",
"pct": 46,
"comparator_pct": 21,
"p_value": 0.008,
"population": "childhood AML (ELAM02)"
},
{
"finding": "specific cutaneous infiltrate",
"pct_range": [
25,
30
],
"population": "congenital leukaemia, world literature review",
"preferred": true
},
{
"finding": "skin involvement",
"pct": 14.5,
"comparator_pct": 2.6,
"population": "children under 2 years vs over 2 years"
}
]rejected figure
{
"value_pct": 67.8,
"reason": "pooled case-report publication bias; systematic review of the same literature gives 25-30%"
}differential blueberry muffin
[ "congenital CMV", "rubella", "toxoplasmosis", "neuroblastoma", "rhabdomyosarcoma", "extramedullary haematopoiesis", "vascular lesions" ]
- Skin is a large, poorly perfused sanctuary compartment; a systemic agent that clears marrow and blood may leave dermal infiltrate behind, and the resulting relapse will not be detected by marrow MRD monitoring.
- Leukaemia cutis is over-reported in the case-report literature, so any prior built from published cases will overestimate how often a targeted agent needs to penetrate skin.
- Because leukaemia cutis flags infant KMT2A-rearranged monoblastic disease, a trial that enriches for it will simultaneously enrich for the youngest and highest-risk patients - improving the biological match but worsening the apparent outcome of the experimental arm.
- Is the skin tropism of KMT2A-rearranged blasts driven by the same NG2/CSPG4-dependent migratory programme implicated in CNS infiltration, and would NG2 blockade reduce cutaneous as well as meningeal disease?
- In aleukaemic congenital leukaemia cutis the clone is tissue-resident while marrow and blood are normal; what keeps it there, and does that latency state share a transcriptional programme with the pre-leukaemic clone detectable at birth in later-onset infant ALL?
- No modern cohort reports leukaemia cutis frequency in KMT2A-rearranged infant ALL specifically, as opposed to infant AML - the Interfant and COG databases could answer this in a single query.
- Does spontaneous regression of aleukaemic leukaemia cutis reflect immune clearance, and if so, what antigen is being recognised?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In the ELAM02 childhood AML study (n = 438), 24 patients (5.5%) presented with leukaemia cutis; their median age was 1.2 years versus 8.7 years for those without, 63% (15/24) were under 2 years, MLL/KMT2A gene rearrangements were significantly more frequent in the leukaemia cutis group (46% vs 21%, p = 0.008), more than 60% were FAB M5, initial WBC count did not differ, relapse occurred in 50% versus 35%, and overall survival was significantly inferior in patients with leukaemia cutis (p = 0.006).
Leukemia Cutis in Childhood Acute Myeloid Leukemia: Epidemiological, Clinical, Biological, and Prognostic Characteristics of Patients Included in the ELAM02 Study · Gouache E, Greze V, Strullu M, et al. · HemaSphere 2018What this evidence cannot show
The leukaemia cutis group is 24 patients, so subgroup percentages carry wide uncertainty and the survival difference (p = 0.006) is not adjusted for the age and KMT2A imbalance that defines the group - leukaemia cutis may be a marker for infant KMT2A-rearranged monoblastic AML rather than an independent adverse factor. This is an AML cohort; it says nothing directly about leukaemia cutis frequency in infant ALL.
Evidence that argues against this (1)
- In congenital leukaemia the same sign appears not to be prognostic at all, so its adverse association in childhood AML may not generalise to the youngest patients. source 1993
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review
A review of the world literature on congenital leukaemia identified approximately 175 reported cases, of which at least 41 and possibly 56 had leukaemia cutis, giving specific cutaneous leukaemic infiltrates in 25-30% of infants with congenital leukaemia - not the 50% frequently cited; the lesions usually appear as firm blue, red or purple nodules in a generalised distribution; only 1% of congenital leukaemia patients have oral involvement, in contrast to 50% gingival/oral infiltration in adult monocytic leukaemia; congenital leukaemia cutis may precede other manifestations of leukaemia by as much as 4 months; and the natural history of congenital leukaemia is not altered by leukaemia cutis.
Leukemia cutis in congenital leukemia. Analysis and review of the world literature with report of an additional case · Resnik KS, Brod BB · Archives of Dermatology 1993What this evidence cannot show
A 1993 literature review assembled from published case reports, so its denominator (~175 congenital leukaemia cases) is itself a biased sample of reported rather than occurring disease - the 25-30% is a ratio of two biased counts and is best read as 'substantially less than half'. It predates molecular KMT2A diagnostics entirely, so it cannot attribute the cutaneous phenotype to a specific lesion. 'At least 41 and possibly 56' reflects genuine ambiguity in the original case descriptions.
Evidence that argues against this (1)
- The congenital-leukaemia literature is a collection of individually reported cases, not a cohort. Roughly 175 cases assembled from the world literature over decades carries publication bias toward the visually striking and the unusual, so every frequency derived from that denominator is a frequency among CASES SOMEONE CHOSE TO PUBLISH. Leukaemia cutis is the most visible presentation there is, so it is exactly the finding a case-report denominator over-represents. The entry is right to correct 50% down to 25-30% and the corrected figure inherits the same bias. Blood Cells, Molecules and Diseases 2010
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In a meta-analysis of 33 cohort studies of childhood leukaemia (n = 3084, ages 0-18), rash was recorded as a presenting feature in 35% of children, bruising in 52% and bruising/petechiae in 42%, and 6% of children were asymptomatic at diagnosis.
Clinical presentation of childhood leukaemia: a systematic review and meta-analysis · Clarke RT, Van den Bruel A, Bankhead C, et al. · Archives of Disease in Childhood 2016What this evidence cannot show
'Rash' in this meta-analysis is a heterogeneous category that certainly includes non-specific eruptions, drug reactions and petechial rashes as well as true leukaemic infiltrates, so the 35% figure must not be read as a leukaemia cutis frequency. Included studies had no control groups and spanned ages 0-18 without an infant-specific breakdown.
Evidence that argues against this (1)
- This meta-analysis covers ages 0 to 18, and infants are a small minority of its 3,084 children. Infant KMT2A-rearranged leukaemia presents differently from the disease that dominates the pooled figure — higher white cell counts, more CNS and extramedullary involvement, a pro-B rather than common-ALL phenotype — so a frequency computed across the whole paediatric range is not the frequency in the patients this graph is about. The signs are also chart-recorded retrospectively, which systematically under-records soft findings and over-records the ones that prompted referral. "Rash" in 35% of children with leukaemia is also a non-specific term covering viral exanthem, drug reaction and petechiae, which is a different thing from the specific cutaneous leukaemic infiltrate this entry is about. Statistics in Medicine 2006
- E4weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.book
A tertiary source reports cutaneous infiltrates in about two thirds of congenital leukaemia patients (67.8%), alongside hepatosplenomegaly 47.5%, CSF blasts 25.4%, hyperleukocytosis 62.7% and spontaneous remission 11.9%, and reports skin involvement in 14.5% of children under two years versus 2.6% of older children.
Infant Leukemia (StatPearls) · StatPearls Publishing · NCBI Bookshelf 2024What this evidence cannot show
Cited here specifically as the figure this node rejects. The 67.8% is more than double the 25-30% arrived at by a systematic review of the same literature, and the discrepancy is best explained by publication bias in pooled case reports. The 14.5%-versus-2.6% age contrast is directionally consistent with the ELAM02 cohort and is the more usable part of this source.
Evidence that argues against this (1)
- A systematic review of the world literature that explicitly set out to test the commonly quoted high frequency concluded 25-30%, and this node adopts that figure. source 1993
age gradient within infancy for skin infiltration
presenting sign: cutaneous infiltration, disproportionately common in infant KMT2A-r disease
- ← supportsSkin first, blood later: a three-day-old referred for purple spots, with KMT2A::MLLT3 neonatal ALL
aleukaemic skin-first presentation: skin biopsy diagnostic on day 7 while marrow was not
diffuse non-blanching bluish-red macular rash with palpable areas; post-mortem showed diffuse dermal leukaemic infiltration