"Lineage switch" is the term used to describe the phenomenon of acute leukemia that meets standard French-American-British (FAB) criteria for a specific lineage (lymphoid or myeloid) at initial diagnosis, but converts to the opposite lineage upon relapse (1). This situation rarely occurs, and prognosis is variable. Although different hypotheses have been proposed to explain the occurrence of lineage switch, the specific causative factors have not yet to be identified. Here, we present the clinical and laboratory features of four cases in which the cell lineage switched from acute lymphoblastic leukemia (ALL) to acute myeloid leukemia (AML).

All the cases reported here were initially diagnosed as ALL, but relapsed as AML. In the first three cases, the original karyotype had been replaced by an entirely different abnormal karyotype, while in case 4, lineage switch represented a relapse of the same leukemic clone.

In case 1, the patient was initially diagnosed as ALL L1 with normal karyotype, which switched to AML M0 with t(9;11) at relapse with complete immunophenotypic changes. In chemotherapy-associated secondary AML, which has a mean latency period of 2 yr, monocytic subtypes are more common and associated with translocations involving the MLL gene on 11q23 (2). Considering the long duration between diagnosis and relapse and the multiple antineoplastic drugs administered to the patient before relapse, we cannot conclude the effect of that chemotherapy for the relapse with clonal change.

In case 2, ALL L1 switched to AML M4, with loss of CD10 and TdT. In terms of gain or loss of phenotypic markers at relapse of acute leukemia, one of the most common changes is loss of CD10 and/or TdT activity (3-5). Pui et al. (3) previously suggested that loss of CD10 might be related to malignant transformation of pluripotent stem cells after eradication of the original stem cell line with chemotherapy, but the precise significance of this finding remains unknown. AML-t(8;16), diagnosed at relapse, is characterized by monocytic differentiation of AML M4/M5 and poor prognosis. The t(8;16) gene fuses the MOZ gene encoding histone acetytransferase located on 8p11 with the CBP gene, which also encodes histone acetyltransferase located on 16p13. Recent reports have indicated that the chimeric transcription of MOZ-CBP is essential for leukemogenesis (5). The majority of the documented AML-t(8;16) cases are de novo, and, to the best of our knowledge, 12 cases of secondary t(8;16) leukemia (including our case 2) have been reported to date (6-8). However, the relationship between histone acetylase and secondary leukemia is yet to be determined.

In case 3, the original diagnosis was B-ALL with aberrant expression of the myeloid marker, CD33. At first relapse, the patient showed immature T-lymphoid phenotype of CD7+, CD4-, CD8-, CD1-, retaining CD 33 expression, in addition to the emergence of a new clone, trisomy 13. Finally, the patient switched to AML retaining aberrant T-lymphoid cell marker CD7. Trisomy 13, observed in our patient at relapse, is associated with undifferentiated or biphenotypic acute leukemia, and is rarely described in lymphocytic disorders (9). The first transformation of B-ALL was T-lineage ALL coexpressing myelocytic and undifferentiated markers, such as CD33 and CD34. A subset of hematopoietic precursors that lack mature T-cell and myeloid antigens retains the potential to differentiate into T-lymphoid and myeloid cells. Hurshfield and co-workers (10) described the in vivo conversion of CD7+, CD4-, CD8- lymphoid cells to the myeloid lineage, while Kurtzberg et al. (11) suggested that CD7+, CD4-, CD8-leukemias arise from immature hematopoietic stem cells with lineage infidelity. Moreover, several authors have proposed that NOTCH1 mutations occur in leukemic stem cells that precede both myeloid and T-lineage commitment (12). Considering the rapid lineage switch between the first and second relapse and early immature T-lymphoid phenotype in case 3, we suggest that lineage switch may arise from very immature cells belonging to a common T-cell/myeloid progenitor.

Lineage switch in infant leukemia has been previously described in the literatures (13, 14). The most common chromosomal abnormality reported in infant leukemia is t(4;11) (15). Rearrangement of the MLL gene is known to play an important role in leukemogenesis (16). Some authors have suggested that the MLL gene rearrangement occurs in B lymphocyte-monocyte bipotential stem cells (13). Although the prognosis of MLL-infant leukemia is poor (15), our patient 4 remained in CR after allo-HSCT.

Lineage switch may be regarded as a variation of mixed-lineage leukemia in which the lineage at onset is converted to another at a later time (17). However, the relationship between lineage switch and mixed-lineage leukemia is yet to be determined. In some cases, lineage switch may be part of the biologic spectrum of mixed-lineage leukemia. An example of this is the patients 3 and 4, who met the morphologic and cytochemical criteria for diagnosis of ALL, but displayed positivity for CD33 with the expanded immunophenotype panel.

In conclusion, although our current findings are insufficient to confirm the mechanism of lineage switch, each case indicates different possibilities. Sequential phenotypic and cytogenetic studies may yield valuable insights into the mechanisms of leukemic recurrence, with possible implications for individualized treatment selection.