An extreme example of focal bone marrow involvement in acute myeloid leukemia


  • Conflict of interest: Nothing to report.

A 59-year-old man presented to the Emergency Department (ED) with a 2 month history of progressive fatigue, dyspnea, and exertional chest pain. For 9 years, he had had chronic, mild fluctuating cytopenias of uncertain etiology. A bone marrow biopsy of the left iliac crest at the onset of the cytopenias demonstrated aplasia, but a follow-up biopsy of the contralateral iliac crest 2 months later revealed a hypercellular marrow with mild erythroid hyperplasia and no dysplasia. Given these discrepant results, an extensive investigation for potential causes of cytopenias was undertaken, but failed to yield a specific diagnosis. Several subsequent bone marrow biopsies from both iliac crests showed only a severely hypocellular marrow with a normal male karyotype, but a sternal aspirate performed 8 years prior to presentation was hypercellular and without morphologic abnormalities. Four months prior to ED presentation, the patient had a normal complete blood count (CBC).

On presentation to the ED, CBC revealed hemoglobin of 7.8 g/dL, hematocrit of 21.7%, white cell count of 0.81 × 109/L (absolute neutrophil count 190/mm3), and platelets of 103 × 109/L. Peripheral blood smear showed evidence of dysplasia, with poikilocytosis, anisocytosis, basophilic stippling, and hypogranular neutrophils, but no circulating blasts. Bone marrow aspirate and biopsy of the left iliac crest was performed, which was again hypocellular and non-diagnostic, but a few suspicious blasts were observed. Due to the patient's unusual history, [18F] fluorodeoxyglucose positron emission tomography (FDG-PET) was performed, with concurrent non-contrast helical computed tomography (CT) to enable anatomic correlation and attenuation correction of the PET images.

FDG-PET/CT revealed scattered, patchy high-intensity FDG uptake in the sternum, part of the right ilium and pelvis, and 8 of 24 vertebral bodies, as seen in the coronal section PET image in Fig. 1A. The remainder of the marrow, including the left iliac crest (Fig. 1B) was markedly photopenic, in keeping with the findings of a hypocellular marrow on previous aspirates. Figure 1C,D demonstrates different FDG uptake in two adjacent vertebrae, showing how the T12 vertebral body (Fig. 1C) is highly FDG-avid, while the L1 vertebral body (Fig. 1D) is not. Normal marrow shows low metabolic activity on FDG-PET and appears nearly black.

Figure 1.

Panels A-D depict FDG uptake in several vertebral bodies and patchy areas of the pelvic bones, including the right iliac crest, on PET/CT scanning of a patient with focal involvement by AML. Other vertebral bodies (panel D) and the left iliac crest are photopenic, and the left iliac crest was hypocellular (panel G) and non-diagnostic upon biopsy. Panels E and F depict a cellular marrow with myeloblasts, obtained from a PET-avid area of the pelvis; panel G is a marrow biopsy from the left pelvis and is hypocellular. Panel H shows G-banded karyotyping with der(6)t(6;12) (arrow). [Color figure can be viewed in the online issue, which is available at]

Bone marrow aspirates targeted to FDG-avid areas of the right ilium and sternum showed 51% and 56% blasts, respectively, with a myeloid phenotype (expressing CD34, dim CD45, HLA-DR, and the myeloid markers CD13, CD33, CD117), consistent with a diagnosis of acute myeloid leukemia (AML) (Fig. 1E,F; Wright–Giemsa ×40 and hematoxylin–eosin ×10, respectively). Erythroid elements exhibited megaloblastoid maturation and frequent (>15%) ring sideroblasts, suggesting that the AML might have arisen on a background of myelodysplastic syndrome. Samples obtained from a non-FDG avid area of the marrow were again hypocellular (Fig. 1G, hematoxylin–eosin ×10). Karyotyping (Fig. 1H) demonstrated an unbalanced rearrangement of chromosomes 6 and 12, resulting in partial deletion of 6q and 12p/q. FLT3 and NPM1 genes were wild-type, and telomere length analysis and chromosome breakage studies were unremarkable.

The patient was treated with standard daunorubicin/cytarabine induction chemotherapy and achieved remission [1]. He then proceeded to hematopoietic stem cell transplantation (HSCT) and successfully engrafted.


It is well recognized that AML can present or relapse at extramedullary sites (e.g., as a “granulocytic sarcoma” or “chloroma) [2]. However, localized or patchy intramedullary involvement at initial presentation of myelodysplastic syndrome (MDS) or AML with discrepant results from sampling bilateral iliac crests has been assumed to be rare, in contrast to the relatively frequent pattern of focal marrow involvement by lymphoplasmacytic neoplasms or solid tumors [3-6]. In AML, the marrow is usually diffusely hypercellular throughout the axial skeleton due to massive expansion of clonal myeloid blasts [7], and this is reflected in results of imaging studies that visualize the red marrow, including magnetic resonance imaging (MRI) [8, 9] (MRI was not performed in this case). The localized proliferation of blasts demonstrated in this case by FDG-PET, resulting in an extremely hypocellular marrow aspirate and biopsy on repeated good-quality sampling of the customary trephining sites in the iliac crests, is highly unusual.

The mechanism behind focal involvement by AML is unclear. When localized relapse follows HSCT, failure of the graft-versus-leukemia effect in specific extramedullary or marrow “sanctuary” sites has been postulated [10]. Similarly, heterogeneity in the marrow microenvironment or hematopoietic stem cell niche may explain differential ability of myeloid blasts to proliferate in one site over another. Given our patient's widely variable bone marrow cellularity documented on serial bone marrow biopsies 9 years prior to the AML diagnosis, the anatomic pattern of leukemic involvement could also reflect intrinsically patchy distribution of hematopoietic elements. Patterns of expression of cell–surface adhesion molecules by the clonal cells could explain a particular anatomical distribution.

This patient's leukemia was characterized by a novel karyotype, t(6;12)(q27;q13), of uncertain molecular consequence. A reciprocal translocation with similar breakpoints was reported in a childhood acute lymphoblastic leukemia (ALL) case, but may not be molecularly equivalent [11].

The relatively long prodrome of fluctuating cytopenias prior to AML diagnosis and extent of cellular dysplasia at the time of diagnosis here is consistent with an MDS-like prophase, but the lack of cellular atypia on prior specimens and normal karyotype meant that a specific diagnosis could not be made. MDS may be morphologically occult until the time of leukemic transformation; persistent unexplained cytopenias that could represent MDS but defy diagnosis despite extensive investigation have been termed “idiopathic cytopenias of undetermined significance” (ICUS) [12, 13].

In our patient, FDG-PET revealed increased cellular metabolism in blast-filled marrow in certain skeletal sites compared to low metabolic activity in biopsy-proven hypocellular marrow at other sites. FDG-PET is well-established in the clinical diagnosis and staging of many tumors, particularly breast, colorectal, esophageal, head and neck, non-small cell lung cancers, melanoma, and lymphoma [14], but FDG-PET scans are rarely used or required in leukemia. A new injectable agent, FLT (3′-deoxy-3′-fluorothymidine), may increase the sensitivity and applicability of PET for assessment of hematological disease. FLT is retained intracellularly secondary to phosphorylation by thymidine kinase 1, enabling direct evaluation of tissue/tumor nucleic acid proliferation, rather than using glucose metabolic activity as a surrogate marker [15]. Initial validation studies for the use of FLT-PET in AML have shown preferential FLT uptake in disease-infiltrated extramedullary sites but the clinical utility of such imaging studies is as yet unclear [16].

The case reported here demonstrates that marrow involvement by AML may be localized at diagnosis, and that even repeated bone marrow biopsy in patients with cytopenias may result in sampling error and diagnostic failure. Performing a PET/CT scan can help reveal sites of marrow activity, when warranted by clinical suspicion.


Authors thank the patient for giving permission to report details of his case history.