Chomel and colleagues1 recently reported that study of blood or marrow cells from patients in complete cytogenetic remission following treatment with interferon-α or allogeneic bone marrow transplantation (BMT) for chronic myeloid leukemia (CML) by fluorescence in situ hybridization (FISH) shows appreciable numbers of BCR-ABL–positive cells. They examined 10 patients treated by BMT using RT-PCR and BCR-ABL D-FISH for evidence of the fusion gene at mRNA and genomic levels, respectively. In all cases theBCR-ABL transcripts were undetectable (BCR-ABL/ABLratio less than 0.002%), whereas FISH studies of interphase nuclei were positive at levels of 3%-11%. The authors concluded that these patients must harbor appreciable numbers of leukemic cells that carried a BCR-ABL fusion gene but did not express BCR-ABL transcripts. In 2 patients, the donor was of opposite sex and the level of genomic rearrangement was confirmed by XY FISH, and in 2 patients the level was confirmed by Southern hybridization. These results are surprising and are at odds with our own experience of minimal residual disease in CML after BMT.
We have found BCR-ABL FISH to be technically demanding. Although in experienced hands with good quality preparations the technique can give a very low false positive rate of 0.1%-0.2%,2,3 in practice we prefer to use cutoff rate of 1% because accuracy is seriously compromised by several factors, including slide quality and operator experience. A similar point has recently been made in a multicenter study of BCR-ABL FISH in clinical practice.4 In this study, slides scored with the knowledge that they were normal were given a meanBCR-ABL score of 0.2% (maximum 0.8%), whereas normal slides scored blind with other mixed positive slides were given a mean score of 1.0% (maximum 8.8%). In practice, therefore, the expectations of the operator can substantially influence the number of cells scored as positive.
To circumvent the problems associated with interpreting BCR-ABLFISH data, we decided to study CML patients who had previously received allogeneic stem cell transplants from opposite sex donors. We made the assumption that residual leukemic cells in a patient's blood after allografting would necessarily have the patient's sex chromosome complement. If so, the proportion of cells with aBCR-ABL fusion gene could not be greater than the proportion with a host-type sex pattern. We used a commercial kit (Vysis, Downer's Grove, IL) for the detection of X and Y chromosomes in interphase cells. This probe has a false positive rate less than 0.1% and is less subject to observer bias than areBCR-ABL probes.5
We have examined 11 consecutive CML patients previously in remission after BMT with a sex-mismatched donor (Table).
Residual host cells were detected using the commercial XY FISH kit, and BCR-ABL mRNA was detected using RT-PCR as previously described.6 FISH slides were scored blind with no knowledge of the RT-PCR result. Nine patients were RT-PCR–negative, and 8 were also negative by FISH; one patient showed a single poor-quality cell of apparently host sex, which we interpreted as a false positive cell. Two patients were RT-PCR–positive and showed very low levels of host cells by FISH. Using these techniques, we therefore found no evidence of residual leukemic cells in RT-PCR–negative patients. Three patients were also examined byBCR-ABL FISH using a commercial kit (BCR-ABL ES probe; Vysis) with no evidence of BCR-ABL–positive cells in excess of our false positive rate of 1%.
A previous study by our group has shown similar results using a different method. Patient-specific PCR primers were designed to amplify genomic BCR-ABL rearrangements in CML patients at a sensitivity of one leukemic cell in 105 cells. Ten patients underwent treatment with BMT, and serial samples were analyzed for levels of genomic and mRNA BCR-ABL using DNA-PCR with patient-specific primers and RT-PCR, respectively. Eight patients became RT-PCR–negative after BMT. All these patients also became DNA-PCR–negative either concurrently or, in one case, shortly after. Again, this study was not consistent with the hypothesis of common, persistent genomic BCR-ABL rearrangement without mRNA expression.
In conclusion, we believe that our data demonstrate concordance ofBCR-ABL mRNA and DNA levels after BMT for CML. But the results reported by Chomel et al deserve further attention because they would, if confirmed by others, require radical revision of our concepts of “minimal residual disease” after allografting for CML.
