Molecular analysis of lineage-specific chimerism and minimal residual disease by RT-PCR of p210^BCR-ABL and p190^BCR-ABL after allogeneic bone marrow transplantation for chronic myeloid leukemia: increasing mixed myeloid chimerism and p190^BCR-ABL detection precede cytogenetic relapse

Although allogeneic bone marrow transplantation (BMT) may cure 50% to 60% of patients with chronic myeloid leukemia (CML), disease relapse still represents the major cause of treatment failure.1,2 Patients who relapse after BMT may be reinduced into durable second remission with distinct therapeutic strategies including α-2a-interferon (INF) and/or donor lymphocyte infusion (DLI).3-6 Among prognostic determinants of response to salvage therapy, disease burden appears to represent a significant factor.7,8 Thus, minimal residual disease (MRD) evaluation aimed at early detection of relapse has relevant therapeutic implications in this context.

Monitoring of MRD after BMT for CML has mainly involved a karyotypic search for the Philadelphia (Ph1) chromosome and/or reverse transcriptase–polymerase chain reaction (RT-PCR) amplification of the p210^BCR-ABL hybrid transcript. Detection of the Ph1 chromosome or p210^BCR-ABL messenger RNA (mRNA) after 6 or more months post-BMT has been associated with subsequent hematologic relapse.9-11 Conventional karyotyping has limited sensitivity, whereas RT-PCR enables smaller amounts of MRD to be detected and might permit therapeutic intervention at an earlier stage. However, after BMT a sizable proportion of patients in long-term remission test p210^BCR-ABL positive,12,13 and the issue of whether or not molecular relapse should be treated remains unresolved. Recently quantitative PCR analysis of p210^BCR-ABL has provided more precise prognostic information on post-BMT disease evolution in CML patients, and it is likely that this technology will aid therapeutic decision making in the future.14,15 

As a further approach to monitoring post-BMT outcome, several investigators have employed chimerism analysis using chromosome Y body or highly polymorphic loci detection. These techniques permit the relative proportions of host and donor cells in the post-BMT period to be identified and quantified. Although chimerism analysis cannot directly detect residual leukemia, it permits the definition of specific patterns, such as increasing autologous hemopoiesis (mixed chimerism [MC]), that are known to be associated with disease recurrence.16-18 On the other hand, MC detection in whole blood or marrow may be transient if it is the result of residual host T cells that have survived the conditioning regimen. This drawback could be overcome by analyzing lineage-specific chimerism in highly purified cell fractions (myeloid, B and T lymphocytes, and natural killer [NK] cells).

Recently the p190^BCR-ABL transcript, which is classically associated with Ph1-positive acute lymphocyte leukemia (ALL), has been detected at diagnosis in virtually all patients with CML, where it occurs as a consequence of alternative or missplicing events in the BCR gene.19,20 Because the amount of p190^BCR-ABLhas been correlated with that of p210^BCR-ABL, corresponding to 0.02%-30% of the total BCR-ABL transcripts, one could speculate that a rising tumor burden is needed for p190^BCR-ABL to become detected by RT-PCR during hematologic remission. Therefore we tested the hypothesis that p190^BCR-ABL mRNA detection could be used, in addition to other markers, as an indicator of disease evolution in the post-BMT outcome of CML patients.

We report here a molecular study that was performed on a series of CML patients who underwent a BMT. The study used p210^BCR-ABL and p190^BCR-ABL RT-PCR combined to lineage-specific chimerism analysis in highly purified cell fractions. We found that this strategy closely traces the kinetics of leukemic regrowth after allogeneic BMT. In fact, disease evolution in relapsed patients was consistently characterized by the sequential detection of p210^BCR-ABL transcripts; increasing amounts of myeloid MC; and finally, p190^BCR-ABL positivity preceding impending cytogenetic and hematologic recurrence. This model was further reinforced by an inverse sequence of molecular findings detected after DLI treatment for relapse in 2 patients, both of whom experienced progressive disappearance of p190^BCR-ABL, chimerism pattern conversion to donor type, and p210^BCR-ABLnegativity.

From June 1989 to December 1997, 55 consecutive recipients with Ph1-positive CML presented with both p210^BCR-ABL and p190^BCR-ABL mRNA types at diagnosis and just before transplantation. All patients underwent allogeneic BMT from human leukocyte antigen–identical (HLA-identical) siblings (n = 50) or HLA-matched unrelated volunteers (n = 5) at the Reina Sofia Hospital, Córdoba, Spain, and the Carlos Haya Hospital, Malaga, Spain. Partially T-cell–depleted marrow grafts containing 1 × 10⁶ donor T-CD8+ cells/kg were implanted in 11 patients as previously described,21 whereas 44 patients received unmanipulated BMT. The main patient characteristics at the time of BMT, including disease phase, conditioning regimen, and graft-versus-host disease (GVHD) prophylaxis, are reported in Table1. Results were analyzed from December 31, 1998.

Conventional karyotyping, chimerism analyses, and p210^BCR-ABL RT-PCR studies were prospectively performed in diagnostic material and in sequential post-BMT blood or marrow samples collected at monthly intervals during the first year, at 3-month intervals during the second year, and at 6-month intervals thereafter. Viable cells were also cryopreserved at each time point. Lineage-specific chimerism and p190^BCR-ABL RT-PCR were analyzed in cryopreserved samples collected before September 1996 and prospectively in fresh samples thereafter.

Cell separation of the myeloid and lymphoid subsets was performed using immunomagnetic beads (Dynal, Oslo, Norway) either coupled with the specific antibody (CD15, CD3, and CD19) or coupled with an antimouse immunoglobulin G (IgG) for purification of the NK fraction previously incubated with an anti-CD56 antibody (Becton Dickinson, San Jose, CA). Separation procedures were performed according to the manufacturer's instructions. Purity was assessed using a direct immunofluorescence technique with specific monoclonal antibodies (mAbs) conjugated with fluorescein isothiocyanate (FITC) or R-phycoerythrin (R-PE) (Becton Dickinson). Acquisition and analysis were performed in a cytometer (FACScalibur, Becton Dickinson). Cells were only further processed when purity was at least 98%. DNA of distinct cell populations was isolated using a modified salting-out procedure.22 VNTR analysis was performed as described below.

Total RNA was isolated from mononuclear peripheral blood or bone marrow cells as described by Chomczynski and Sacchi.23Reverse transcription was performed by first heating 1 μg total RNA at 70°C for 5 minutes then using the heated RNA with random hexamers as reaction primers. The reaction was carried out at 42°C for 45 minutes in the presence of 12 units of avian myeloblastosis virus (AMV) reverse transcriptase. The method used to amplify in 2 different “nested” PCRs, respectively, the b2a2/b3a2 types of BCR-ABL hybrid mRNA and the e1a2 type, was performed essentially as described by Saglio et al.20 The second-round PCR products were electrophoresed on a 1.5% agarose gel containing 1 μg/mL ethidium bromide and photographed under ultraviolet (UV) light. Precautions were taken to ensure high PCR quality as recommended by Kwok and Higuchi.24 Using this approach, we could detect a single BCR-ABL positive cell in 10⁵ normal cells.11 

Patients who always had positive BCR-ABL results or who had both BCR-ABL positive and negative time points and a positive last sample were considered as BCR-ABL positive patients. Recipients who presented BCR-ABL negative assays or presented initial positive assays followed by negative results on subsequent analyses were considered as BCR-ABL negative patients.

High molecular weight DNA was extracted from donor and recipient mononuclear cells using a salting-out procedure.25 For PCR amplification, we used specific primers designed to flank the repeat units of the following human minisatellite regions (VNTR-PCR): D1S80, 33.6, 33.1, 33.4, YNZ-22, APO-B, λg3, DXS52, and HVR-3′. Primer sequences and conditions for each reaction have been described elsewhere.26 We defined a VNTR locus as being informative if the analysis of recipient and donor samples prior to BMT showed a unique band for the recipient and another unique band for the donor or if it showed a unique band for the recipient only. The VNTR-PCR assay allows the detection of a minor cell population at the 0.5% to 1.5% level.26 

Patients who exhibited complete donor hematopoiesis with at least 2 markers were defined as having full donor chimerism (DC). Patients who exhibited mixed populations of donor and host cells with at least 2 markers on more than 1 occasion were considered as having MC (Figure1).

To evaluate the possible dynamics of chimerism after BMT, we established a quantitative PCR approach. Briefly, recipient pretransplant DNA and donor genomic DNA were mixed in different percentages and were subsequently amplified. PCR products were stained with ethidium bromide after separation by agarose gel electrophoresis (Figure 2A). Light emissions of the gels were directly digitalized under UV-light stimulation and subsequently analyzed densitometrically (Bio-1D software, Vilber Lourmat, Marne-La-Vallee, France). The peak areas of donor and recipient bands directly correlated to DNA concentrations. After determining the percentage of host signal intensities, standard curves for each individual patient were generated. Signal intensities of individual samples were related to the patient's standard curve (Figure 2B). The degree of MC was expressed as the percentage of host DNA. We prepared standardized chimeric samples by mixing pretransplant recipient and donor DNA in dilution experiments (75% to 0.39%) in relation to host DNA. Each sample was amplified with the informative primer. The intensity of each allele was measured densitometrically. The proportion of host DNA was expressed according to the quotient, where peak area recipient/peak area donor equals quotient host/donor DNA.

The quotient host DNA was correlated to the percentage of host DNA given to the chimeric samples. The mixing experiments were carried out in at least duplicate, and standard curves were then generated from mean values. The analysis of each prior posttransplant sample was repeated when an actual blood sample was received. In general, measurements were completed 5-12 times, and mean values were calculated from repeated analyses.

For cytogenetic analyses, cells from bone marrow were cultured in McCoy's medium supplemented with 20% fetal calf serum and antibiotics for 48 hours. Chromosome preparations were stained with 5% Giemsa solution according to standard procedures. A minimum of 25 metaphases were analyzed per sample. This allowed the detection of a minor clone at the 5% level. Cytogenetic relapse was considered to have occurred if one or more Ph1 positive metaphases were detected at any time following BMT without evidence of hematological relapse.

Cumulative actuarial probabilities (plus or minus standard errors) were calculated by the product-limit method.27 Differences between time-to-event distribution functions were compared by the log-rank test. Comparisons between continuous covariates were performed by the 2-tailed Wilcoxon signed rank test across strata. Differences between frequencies were compared by the 2-tailed Fisher exact test. Stepwise proportional hazards general linear model analysis was used to evaluated the effect of different covariates on the analytical endpoints of cytogenetic relapse and p190^BCR-ABLexpression.28 

During the post-BMT period, 41 patients remained in cytogenetic and hematologic remission at a median duration of 34 months (range: 12.5-107 months), and 14 patients (including 6 T-cell– depleted patients) experienced cytogenetic relapse at a median duration of 10.5 months (range: 5-80 months). Cytogenetic relapse was followed in all cases by hematologic relapse at a median duration of 2 months (range: 1-11 months). Of the 14 patients with cytogenetic relapse, 10 patients were in the chronic phase, 2 in the accelerated phase, and 2 in the blastic phase. The 2 patients with blastic phase relapse were given chemotherapy, and the 12 patients with chronic/accelerated phase relapse received α-INF at the time of hematologic recurrence. Of these latter 12 patients, 3 patients (unique patient numbers [UPNs] 235, 251, and 2056) were also treated with donor leukocyte infusion (DLI) between 7-36 months post-BMT, and 1 patient (UPN 199) underwent a second BMT.

Of the 41 patients in this group, 31 patients tested p210^BCR-ABL negative, 28 had signs or symptoms of persistent full DC, and 3 had a transient MC pattern at low levels (less than 30%) in whole blood, which lasted 4-5 months. Chimerism studies on purified cell fractions in these latter cases showed persistent recipient-type CD3+ cells and full DC in CD15+, CD19+, and CD56+ cell fractions. The chimerism pattern of the T-cell fraction converted to the donor type in subsequent evaluations of these 3 patients.

There were 10 patients (all of whom had received unmanipulated BMT) who tested p210^BCR-ABL positive at various time points and in the last follow-up control. All 10 patients displayed persistent full DC in whole blood as well as in purified cell populations. All 41 patients who remained in cytogenetic and molecular remission tested persistently negative for the p190^BCR-ABL mRNA by RT-PCR.

The results of molecular monitoring in this group are shown in Table2. All 14 patients persisted (n = 6) or converted (n = 8) to a positive p210^BCR-ABL at 1-62 months following BMT. Results of sequential chimerism analyses in whole blood and purified cell fractions were as follows.

MC was detected in whole blood at 1-68 months in 13 cases; this preceded p210^BCR-ABL mRNA detection in 1 patient (UPN 322, Table 2), was observed concurrently with p210^BCR-ABL in 3 cases (UPNs 245, 305, and 2072), and followed p210^BCR-ABLdetection by 1-6 months in 9 cases. Chimerism analysis in purified cell fractions revealed myeloid MC (detected in CD15+ cells) simultaneously with whole blood MC in 12 patients. In 1 patient (UPN 322), myeloid MC was detected 6 months after MC was detected in the whole blood. Longitudinal myeloid MC quantitative evaluation showed a progressive increase of autologous DNA in these 13 patients (Table 2). Increasing myeloid MC preceded cytogenetic and hematologic relapse by 2-12 months and 3-16 months, respectively. Chimerism analysis of the CD3+ fraction showed only donor-type T cells in 8 patients at the first detection of MC. Interestingly, this molecular profile with full-donor origin in T-cell fraction persisted in 4 of these 8 patients at the time of cytogenetic relapse (UPNs 235, 251, 267, and 2056).

Chimerism evolution in CD19+ and CD56+ cell fractions was not too different from that observed in the CD3+ subset, although an earlier reappearance of recipient cells in patients who presented T-cell MC was detected. One patient (UPN 286) had a persistent full DC pattern in whole blood and fractionated cell populations. This patient underwent an unusual pattern of hematologic relapse in donor cells.

All 14 patients (including UPN 286) had a positive PCR conversion to p190^BCR-ABL at 4-80 months post-BMT. The detection of p190^BCR-ABL followed p210^BCR-ABL detection in all but 1 case (UPN 267), and in 12 of 14 cases MC was detected. RT-PCR positivity for p190^BCR-ABL preceded cytogenetic relapse by 1-6 months in 11 patients and was detected simultaneously with cytogenetic recurrence in 3 patients.

No major cytogenetic and molecular changes were observed after treatment of relapse in patients receiving INF. The complete disappearance of p210^BCR-ABL and p190^BCR-ABLtranscripts and chimerism conversion to donor type were documented in one patient (UPN 199) after the second BMT. The 3 patients who received DLI following INF were also prospectively monitored. In 1 patient (UPN 235) who received a single dose of 1 × 10⁷ CD3 cells/kg, the positive RT-PCR for both p210^BCR-ABLand p190^BCR-ABL transcripts persisted, and there was increasing autologous hemopoiesis. This patient refused further treatment and remains currently in the second CML chronic phase.

Complete hematologic and cytogenetic response following DLI was noted in 2 patients. One patient (UPN 251) received 2 escalating T-cell doses (1 × 10⁷ and 1 × 10⁸ CD3 cells/kg) and an infusion of 1.5 × 10⁶ and 4.5 × 10⁶ CD34 cells/kg in an attempt to abrogate post-DLI aplasia. A transient response was observed after the first dose, with total disappearance of recipient T cells and a 2-fold decrease of recipient granulocytes. After the second infusion, sequential molecular analyses showed the inverse pattern observed prior to relapse, ie, progressive negativity of p190^BCR-ABL RT-PCR (the day of infusion plus 15 days [day +15] following the second DLI), myeloid cell chimerism conversion to donor type (day +90), and finally disappearance of p210^BCR-ABL transcripts at day +120 (Figure3). One patient (UPN 2056) did not experience a hematologic response following the first DLI dose of 1 × 10⁷ CD3 cells/kg, and a second dose of 1 × 10⁸ CD3 cells/kg was infused. Despite the concomitant administration of 1.3 × 10⁶ CD34 cells/kg, this patient developed a severe pancytopenia, which required an additional dose of 2 × 10⁶ CD34 cells/kg. After hemopoietic recovery, myeloid chimerism conversion to donor type was observed at day +50 following the second DLI, whereas p190^BCR-ABL and p210^BCR-ABL RT-PCR converted to negative at +90 and +120 days, respectively.

At a median follow-up of 34 months (range: 12.5-107 months), 14 of the 55 patients had developed cytogenetic and clinical relapse. To evaluate interactions between different potential influential factors of relapse, we used an analysis that included clinical and molecular parameters as time-dependent covariates (Table3).

We detected no significant association of cytogenetic relapse with either patient age, the time interval between diagnosis and BMT, the clinical phase at BMT, or the type of immunosuppressive treatment. Acute and chronic GVHD and T-cell depletion seem to interfere with disease recurrence. In addition, a significant association between cytogenetic relapse and BCR-ABL expression or myeloid and T-cell MC was observed by univariate analysis (Figure 4). However, multivariate analysis confirmed a positive p190^BCR-ABL as the only independent variable (P = .002, Table 3). After adjustment for the influence of a positive p190^BCR-ABL, no other factor had a significant influence on the occurrence of relapse. Detection of MC in myeloid fraction only presented a trend toward significance (P = .082). This later event could be justified by the exceptional relapse observed in donor cells (as in patient UPN 286).

Interestingly, neither T-cell MC nor p210^BCR-ABL expression was an independent predictor of relapse. Of the 24 patients who showed p210^BCR-ABL positivity, 10 cases did not develop either cytogenetic relapse or p190^BCR-ABL expression. The influence of clinical characteristics on relapse in these patients is summarized in Table 4. Development of chronic GVHD was independently associated with p190^BCR-ABL negativity (P = .034) and also with a better leukemia-free survival in this group of p210^BCR-ABL positive patients.

We report in this study a consistent sequence of molecular findings that closely parallels disease evolution after allogeneic BMT for CML. Of the several markers used here for monitoring patient outcome, chimerism assessment in myeloid cells and p190^BCR-ABL mRNA detection were the best prognostic indicators. Besides allowing a novel strategy for early identification of CML relapse after BMT, our study provides new insights into the kinetics of leukemia recurrence and response to adoptive immunotherapy after relapse in this setting.

Longitudinal monitoring studies of CML patients following BMT were completed using p210^BCR-ABL RT-PCR and chimerism assessment. Although p210^BCR-ABL mRNA positivity, particularly if detected 6 months post-BMT, has been associated with subsequent relapse,10 this correlation is not absolute; in several studies11,29 patients have remained positive in sustained remission. In the present series, only 14 of 24 patients who tested positive for p210^BCR-ABL after 6 months post-BMT have undergone relapse. While recently developed quantitative PCR assays for p210^BCR-ABL seem to provide a more accurate prediction of disease outcome,13,14,30,31 these techniques are quite laborious and are not suitable for routine clinical work in most transplantation centers.

In contrast to p210^BCR-ABL mRNA, p190^BCR-ABLemerges from our study as a novel marker of CML evolution after BMT. In fact, p190^BCR-ABL positivity by nonquantitative RT-PCR was associated with impending cytogenetic relapse in the majority of patients. Moreover, p190^BCR-ABL mRNA was not detected in any patient as a reversible finding nor was it ever found in long-term survivors. Recently, Lichty et al32 reported that p190^BCR-ABL mRNA is more likely coexpressed, together with the p210^BCR-ABL transcript, in CML patients with high white blood cell and blast counts as well as in advanced disease. In keeping with these findings, Carlo-Stella et al33 found that p190^BCR-ABL expression is consistently detectable in the CML CD34+ cell compartment and progressively decreased in differentiating elements, and mature cells probably reflect a differentiation-linked reduced transcription. Finally, the p190^BCR-ABL protein is known to be more strongly transforming than the p210^BCR-ABL protein.34Therefore p190^BCR-ABL detection after transplant could be not only a marker of total tumor burden but also a more aggressive feature of the CML clone.

On the other hand, chimerism studies to identify donor versus recipient hemopoiesis following BMT have been hampered by the use of whole blood instead of lineage-specific hemopoiesis. This latter issue is particularly relevant in CML patients after allogeneic BMT. In fact, the disease that is predominantly expressed in the myeloid compartment and T lymphocytes rarely belongs to the leukemic clone. Moreover, T cells frequently survive the conditioning regimen and, therefore, could affect interpretation of the chimerism findings concerning their prognostic impact. Besides, analysis of separated cell populations, especially of antileukemic effector cells (T and NK cells), has not been studied thoroughly. Although by using quite sophisticated techniques, such as chromosome-specific fluorescent in situ hybridization with simultaneous immunophenotyping of interphase cells (FICTION),35 close detection of tumoral kinetic after DLI has recently been reported, the issue of lineage-specific chimerism evolution and its relationship with MRD and relapse using a simpler molecular technique remains a matter of controversy.

Mackinnon et al,36 studying a group of T-cell–depleted patients who had BMTs, found a high incidence of T-cell MC and subsequent relapse, which suggests that T-cell MC could induce immune host and/or donor tolerance and thereby abrogate the graft-versus-leukemia (GVL) effect. However, we and others37 have detected full DC in T lymphocytes in patients who relapse, meaning that they do not mediate an efficient immune effect toward leukemic cells. This immune escape might be due to marked alterations in expression of surface costimulator molecules and might lead to anergy or to a lack of alloreactivity in donor lymphocytes grown in a donor environment.38,39 

An important finding of our series is that the presence of T-cell MC fails to clearly identify subsequent relapses. All 3 patients who exhibited a transient MC in this subset remain in cytogenetic remission, and 4 patients had full DC T lymphocytes at the moment of relapse and thereafter. In contrast, when myeloid cells were studied, we found that all MC patients who expressed recipient type CD15+ cells did not convert to donor origin, and all relapsed. This makes biological sense when considering the predominant expression of CML on myeloid series and its susceptibility to be destroyed with the myeloablative conditioning regimen. This is also in agreement with van Leeuwen et al,40 who suggest that the identification of persistent host cells only within the original leukemia lineage can be associated with leukemia relapse. The regrowth of a tumoral burden, as a consequence of an inefficient immune surveillance, will be easily detected and found first in the myeloid compartment. In our experience, when this pattern is established, it signifies a point of no return, and progressively, B lymphocytes, NK cells, and even T cells can evolve to recipient origin. We suggest that the evolution of relapse after BMT is not too different from the development at the onset of CML, where only myeloid and erythroid lineages are invariably derived from the leukemic clone.

In our study the consistent kinetic pattern observed in 2 patients was inversely reproduced after successful DLI was given for relapse. The disappearance of recipient type CD15+ cells and the negativity of p190^BCR-ABL were the first indications of this change. Bearing in mind that the expression of p190^BCR-ABL can be interpreted as an indication of the expansion capacity of the Ph1-positive stem cell compartment,33 these initial molecular events could be in agreement with the recent description of Ph1-positive CD34+ cells as the target of the alloimmune response.41 Moreover, this latter hypothesis also seems to be sustained by the disappearance of p190^BCR-ABL, which occurred in one patient (UPN 251) when 60% of the host-type granulocytes were still present. Both patients ultimately converted to RT-PCR negative for the p210^BCR-ABL transcript and remained in hematological remission.

In conclusion, we provide new insights into the biology of the CML evolution after BMT and suggest a molecular monitoring strategy suitable for routine work, which may be used to better address clinical decisions in the post-BMT outcome. For example, patients in hematologic remission who convert to a positive p210^BCR-ABL might be periodically tested for whole blood chimerism and, whenever increasing amounts of MC are observed, further tested for myeloid-specific chimerism. Detection of MC and particularly p190^BCR-ABL mRNA in granulocytes infers impending cytogenetic relapse and might allow early administration of salvage treatment.

We are particularly grateful to Dr F. Lo Coco for his interest in our work and his valuable suggestions. We also thank Dr G. Cimino, Dr G. Saglio, and Dr W. Arcese for helpful discussions and for the critical reading of this manuscript.