Pallor and fatigue: the anaemia of a marrow taken over by leukaemia
also called pallor · fatigue · anaemia at presentation · tiredness · pale baby
As leukaemia cells crowd out the marrow, the body stops making enough red blood cells, so the child becomes pale and tired; pallor is found in about 54% of children with leukaemia and fatigue in about 46%. In a baby this looks like poor feeding, being floppy, sleeping more and breathing faster rather than like a complaint of tiredness.
The same thing, for a scientist
Anaemia at presentation reflects displacement of erythropoiesis by leukaemic infiltration rather than haemolysis or blood loss, and is typically normocytic with a low reticulocyte count. Pallor (54%) and fatigue (46%) rank third and seventh among presenting features in a meta-analysis of 33 childhood leukaemia cohorts (n = 3084). In the congenital leukaemia literature anaemia was recorded in 35.7% of cases, which is lower than in older children and consistent with a very acute presentation dominated by proliferation rather than by prolonged marrow failure. In infants the clinical expression is non-verbal and easily attributed to something else: reduced feeding volume, irritability, hypotonia, tachypnoea and tachycardia. Correction requires care when the white count is very high, because raising the haematocrit raises whole-blood viscosity and can precipitate leukostasis.
Why the baby is pale, and why nobody notices for weeks1 deeper
Erythropoiesis is the most volume-hungry activity in the marrow and it is the first to be squeezed out as the leukaemic clone expands. The resulting anaemia is normocytic and hypoproliferative: haemoglobin falls, reticulocytes do not rise to compensate, and there is no jaundice or dark urine to suggest haemolysis.
The diagnostic problem is that an infant cannot report tiredness. The presentation is a list of things parents notice but doctors discount:
- taking smaller feeds, or falling asleep during feeds
- less interested in play, more content to be held
- pale conjunctivae, palms and mucosae rather than pale skin, which is unreliable
- tachypnoea and tachycardia at rest, which is the body compensating
- irritability
Because each of these is common and benign in isolation, the pre-diagnostic interval is measured in weeks, and 6% of children with leukaemia are asymptomatic when the diagnosis is made incidentally 1.
The numbers in context
| Feature | Frequency | Population |
|---|---|---|
| Pallor | 54% | childhood leukaemia, 33 cohorts, n = 3084 1 |
| Fatigue | 46% | same 1 |
| Anorexia, weight loss, abdominal pain, abdominal distension | reported as common | same 1 |
| Asymptomatic at diagnosis | 6% | same 1 |
| Anaemia | 35.7% | published congenital leukaemia cases 2 |
That congenital figure is the counterintuitive one: anaemia is less often recorded in congenital leukaemia than pallor is in childhood leukaemia generally. The most likely explanation is tempo. Neonatal disease presents within days of a rapidly proliferating clone, dominated by hyperleukocytosis (62.7%) and organomegaly, before the weeks of suppressed erythropoiesis that produce a profoundly anaemic older child have had time to accumulate 2.
Transfusing an infant with a white count of 200
The instinct to correct a haemoglobin of 45 g/L quickly is wrong in a child with hyperleukocytosis. Whole-blood viscosity depends on total cell fraction, so raising the haematocrit in a patient whose blood is already loaded with large, stiff blasts can convert borderline flow into frank pulmonary or cerebral leukostasis - and intracranial haemorrhage, not renal failure, is the main cause of early death in paediatric hyperleukocytosis 3. Standard practice is therefore to transfuse slowly, to a modest target, and preferentially after cytoreduction has begun; platelets by contrast are given freely because they do not materially change viscosity and because bleeding is the acute threat. This is covered further in concept-hyperleukocytosis-emergency.
Why this matters to a targeted-therapy programme
Anaemia is the clinical face of the differentiation block that the fusion enforces (cell-state-differentiation-block). It is also the most direct measure of what a patient gains from restoring normal haematopoiesis rather than merely killing leukaemia cells. A menin inhibitor acts by releasing the differentiation block, so recovery of erythropoiesis and haemoglobin is a mechanistically meaningful pharmacodynamic signal, and one that can be sampled from a few hundred microlitres of blood rather than a marrow aspirate - which matters when the patient weighs 5 kg. The flip side is that the same mechanism produces differentiation syndrome, so a rapid haematological shift is not automatically good news.
| sign | pallor and fatigue secondary to anaemia |
|---|---|
| mechanism | displacement of erythropoiesis by leukaemic marrow infiltration; normocytic, hypoproliferative, low reticulocyte count |
| management caveat | transfuse slowly to a modest target in hyperleukocytosis; raising haematocrit raises whole-blood viscosity and can precipitate leukostasis |
| pharmacodynamic relevance | recovery of erythropoiesis is a mechanistically direct readout of release of the differentiation block |
frequencies — the numbers behind the chart
[
{
"finding": "pallor",
"pct": 54,
"population": "childhood leukaemia 0-18 y, n=3084"
},
{
"finding": "fatigue",
"pct": 46,
"population": "childhood leukaemia 0-18 y, n=3084"
},
{
"finding": "anaemia",
"pct": 35.7,
"population": "published congenital leukaemia cases",
"reliability": "low"
},
{
"finding": "asymptomatic at diagnosis",
"pct": 6,
"population": "childhood leukaemia 0-18 y, n=3084"
}
]infant specific signs
[ "reduced feed volume", "falling asleep during feeds", "pale conjunctivae/palms/mucosae", "resting tachypnoea and tachycardia", "irritability", "hypotonia" ]
- Rapid correction of anaemia in a hyperleukocytotic infant can precipitate the leukostasis it was meant to relieve, so supportive care itself carries an early-death risk.
- Because anaemia and fatigue are non-specific, they cannot be used to detect relapse early; a child who becomes pale again on maintenance therapy is far more often cytopenic from chemotherapy than relapsing.
- If haemoglobin recovery is adopted as a pharmacodynamic surrogate, a menin inhibitor causing differentiation syndrome could produce an apparently favourable haematological trend at the same time as a clinically dangerous one.
- What is the distribution of haemoglobin at presentation in KMT2A-rearranged infant ALL specifically, and does it differ from KMT2A-germline infant ALL in a way that reflects the cell of origin?
- Does the speed of haemoglobin recovery under menin inhibition track depth of response better than blast clearance does, and could it serve as a low-volume pharmacodynamic biomarker in infants?
- How long is the pre-diagnostic interval in infant leukaemia, and would a primary-care rule combining pallor, poor feeding and abdominal distension shorten it enough to lower presenting white-cell counts?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In a systematic review and meta-analysis of 33 cohort studies of childhood leukaemia (n = 3084, ages 0-18), pallor was present in 54% and fever in 53% of children, fatigue in 46%, limb pain in 43% and lymphadenopathy in 41%; abdominal symptoms including anorexia, weight loss, abdominal pain and abdominal distension were common; 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
All included studies were cohort studies without control groups, so these are frequencies among diagnosed children and cannot be used as positive predictive values in primary care. Pallor and fatigue are subjective, clinician- or parent-reported, and systematically harder to elicit in a pre-verbal infant, so the true infant frequencies may be higher than recorded while the recorded ones are less reliable. Ages 0-18 pooled, with no infant-specific stratum.
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. Pallor and fatigue are additionally the hardest signs to elicit in an infant, who cannot report tiredness and whose skin tone makes pallor an unreliable observation, so the pooled 54% and 46% are least applicable to exactly this population. Statistics in Medicine 2006
- E2weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.book
In reported congenital leukaemia, anaemia was recorded in 35.7% of cases while hyperleukocytosis was present in 62.7% with a mean WBC of 68.5 x 10^9/L, hepatosplenomegaly in 47.5% and cutaneous infiltrates in 67.8%.
Infant Leukemia (StatPearls) · StatPearls Publishing · NCBI Bookshelf 2024What this evidence cannot show
Tertiary source pooling published congenital leukaemia case reports, with heavy publication bias toward visually striking features; anaemia is exactly the sort of unremarkable finding likely to go unrecorded in a case report, so 35.7% is probably an underestimate rather than a real difference from older children. Neonatal reference ranges for haemoglobin also differ substantially from those of older infants, so 'anaemia' is not a single threshold across these cases.
Evidence that argues against this (1)
- Pallor is recorded in 54% of children with leukaemia in a much larger and more systematically assembled dataset, so the lower congenital figure should not be taken as evidence that neonates are less anaemic. source 2016
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
In a paediatric cohort with hyperleukocytosis, intracranial haemorrhage rather than renal failure was the main cause of early death, cerebral leukostasis or coagulopathy-related death within 14 days occurred in 4.2% of ALL and 7.7% of AML patients, and all such patients had a white cell count of at least 400 x 10^9/L; 21.3% had pulmonary leukostasis on admission.
Management of Hyperleukocytosis in Childhood Acute Leukemia Without Leukapheresis and Rasburicase Prophylaxis · Aylan Gelen S, Sarper N, Zengin E, et al. · Journal of Pediatric Hematology/Oncology 2022What this evidence cannot show
Retrospective single-centre series of 61 children, not infants specifically, in a setting without leukapheresis. The paper does not directly test the effect of red-cell transfusion on leukostasis; the transfusion caution is an inference from viscosity physiology plus the observed dominance of haemorrhagic and leukostatic early death, not a demonstrated causal link.
Evidence that argues against this (1)
- This is a mixed ALL and AML cohort and the mechanism differs between them: myeloblasts are larger and less deformable than lymphoblasts, so leukostasis is substantially more a myeloid phenomenon at any given count. Quoting 4.2% for ALL and 7.7% for AML in one sentence, with a shared threshold of 400 x 10^9/L, obscures that the same count means different things in the two diseases. Transfusion 2020
no infant-specific pallor or anaemia frequency exists; the 54% is an age 0-18 pooled figure
fatigue is a reported symptom with no pre-verbal equivalent
- co-occurs with →Hepatosplenomegaly: the swollen abdomen that is often the first thing a parent notices
both follow from the same process: blast expansion displacing normal haematopoiesis and seeding organs
haemoglobin 49 g/L at presentation
asthenia and pallor among the first five presenting features
progressive anaemia and worsening fatigue with haemoglobin 6.7 g/dL and weekly transfusion as the presenting change