Skin first, blood later: a three-day-old referred for purple spots, with KMT2A::MLLT3 neonatal ALL

With the skin, and with nothing else. The report describes an apparently healthy newborn referred at 3 days of life for multiple purple-to-violaceous, non-blanching macules a few millimetres across on his face, upper back and right shoulder; only a few lesions were slightly raised 1. The pregnancy had been uneventful and the family history unremarkable for skin and blood disease.

Everything else was normal. The blood count was normal for age. The smear showed 30% neutrophils, 57% lymphocytes, 9% monocytes, a few pro- and metamyelocytes - and no blasts. Liver and spleen were normal in size on ultrasound, fundoscopy showed no chorioretinitis or haematopoietic foci, parvovirus B19 and CMV PCR were negative, and so were the Coombs test and red-cell membrane and enzyme assays. He was negative for neonatal sepsis 1.

So at the moment of referral there was no anaemia, no organomegaly, no bleeding and no leukaemic blood picture: only the skin. This is the aleukaemic presentation that concept-leukemia-cutis-infant describes, seen from the cot side.

Skin biopsy on day 7 made the diagnosis. Marrow on day 7 did not. The bone marrow examination on day 7 was non-diagnostic; it took a peripheral smear on day 14 and a marrow aspirate on day 16 to confirm neonatal ALL 1.

Flow cytometry on peripheral blood showed a monomorphic lymphoid blast population positive for CD19, CD45, CD22 and CD79a and negative for CD10, CD20, CD34 and TdT. The authors call this an aberrant phenotype: CD19 and CD79a fix the lineage, and the absence of CD10 is the CD10-negative pro-B pattern of concept-flow-cytometry-immunophenotype-kmt2a-r, but the absence of CD34 and TdT is not what most KMT2A-rearranged infant ALL looks like. A discrepancy is recorded and left unexplained: marrow blasts were TdT-negative in two independent laboratories, while blasts in the skin biopsy were TdT-positive 1.

Molecular analysis confirmed the KMT2A::MLLT3 (MLL-AF9) fusion transcript arising from t(9;11), and excluded a GATA1 mutation - the test that separates true neonatal leukaemia from transient abnormal myelopoiesis (concept-molecular-confirmation-fusion-transcript).

CSF contained fewer than 5 leucocytes/uL, but the cytospin showed leukaemic blasts, giving a CNS status of CNS2a - low-level CNS involvement detectable only because someone spun the fluid (concept-cns-involvement-at-diagnosis).

The skin lesions regressed spontaneously over several days, then reappeared on day 13 and spread over the entire integument. The leucocyte count climbed steadily, peaking at 125,700/uL on day 24 - crossing into hyperleukocytosis (concept-hyperleukocytosis-emergency) roughly three weeks after a normal count 1.

Treatment was deliberately deferred. An interdisciplinary neonatology and paediatric-oncology team started the prednisone pre-phase of the ALL-BFM protocol on day 24, explicitly to avoid neonatal toxicity. Response was poor: a poor prednisone responder at day 8, more than 10% blasts in the marrow at day 15, morphological remission at day 33 with MRD still detectable.

Subsequent therapy was steered by serial PCR-MRD with the aim of reaching MRD below 10^-4 before transplant. He received an allograft from an unrelated donor matched 10/10 at high resolution and permissive at HLA-DP, conditioned with fludarabine, thiotepa, melphalan and ATG because of his age - a non-total-body-irradiation regimen (concept-hsct-in-kmt2a-r-leukemia). Post-transplant surveillance was PCR-MRD on marrow every 6-8 weeks, so that a rising MRD could be met with a bispecific or conjugated antibody before frank relapse.

Outcome as reported: MRD-negative for 15 months after transplant, alive and in excellent clinical condition at 2 years of age. The report does not state the outcome beyond that point.

concept-leukemia-cutis-infant says that cutaneous infiltrate may precede other manifestations by up to four months and that a firm nodular eruption in a neonate warrants biopsy even with a normal blood count. This case is what that sentence costs and buys.

Four things a category entry cannot convey:

1. The marrow can be negative while the skin is diagnostic. Day 7 marrow: non-diagnostic. Day 7 skin biopsy: leukaemic blasts. If the biopsy had not been done, the diagnosis would have waited for day 14-16.
2. Spontaneous regression is real and is not remission. The lesions faded, then came back disseminated, and the white count went to 125,700/uL.
3. The clone was tissue-resident while marrow and blood looked normal. For a fusion thought to arise in utero, this is a natural experiment in latency, and it happened in a child who was four days old.
4. Treating a neonate is a decision, not a reflex. Therapy was withheld until day 24 to avoid neonatal toxicity, and that delay was a judgement made while the counts climbed.

It is also the graph's one long-ish survivor of neonatal disease. A reader who has read only the outcome statistics should see that the 2-year-old exists.