Prognostic significance of quantitative analysis of WT1 gene transcripts by competitive reverse transcription polymerase chain reaction in acute leukaemia

Authors


John A. Liu Yin, University Department of Haematology, Manchester Royal Infirmary, Oxford Road, Manchester M13 9WL, UK. E-mail: john.yin@cmmc.nhs.uk

Abstract

Summary. We have developed a sensitive, competitive, nested reverse transcription polymerase chain reaction (RT-PCR) titration assay that quantifies the number of Wilm's tumour (WT1) gene transcripts in bone marrow (BM) and peripheral blood (PB), coupled with a competitive RT-PCR protocol for the ABL gene as control. We studied BM/PB samples from 107 acute myeloid leukaemia (AML) patients and 22 acute lymphoblastic leukaemia (ALL) patients at presentation and detected the WT1 gene in > 90% of patients by a qualitative assay. Quantitative analysis of WT1 transcript at presentation in 66 patients (52 AML, 14 ALL) correlated significantly with remission rate, disease-free survival (DFS) and overall survival (OS) (P = 0·003). WT1 levels were normalized to 105ABL transcripts. Within good and standard cytogenetic risk groups, high WT1 levels correlated with poorer outcome. Serial quantification was performed in 35 patients (28 AML, seven ALL); those with less than 103 copies of WT1 after induction and second consolidation chemotherapy had significantly better DFS and OS. Fourteen patients have relapsed with a median complete remission duration of 12 (range 4–49) months. We detected a rise in WT1 levels in nine out of 14 patients, 2–4 months before the onset of haematological relapse, whereas in the remaining five patients, WT1 levels remained persistently high during the disease course. WT1 levels were lower in PB than in BM, but mirrored changes in the BM samples and were equally informative. We suggest that WT1 is a useful molecular target to monitor minimal residual disease in acute leukaemia, especially in cases without a specific fusion gene.

Relapse remains the main cause of treatment failure in acute leukaemia (AL). There is now cumulative evidence that monitoring of minimal residual disease (MRD) is useful in predicting relapse in patients with AL (San Miguel et al, 1997; Cave et al, 1998; Grimwade, 1999; Liu Yin & Tobal, 1999). Detection of MRD is typically based on either molecular or immunological markers, which are present in leukaemic cells but not in normal cells, thus allowing for their specific identification. Polymerase chain reaction (PCR) techniques can detect one leukaemic cell in 105−106 normal cells, but are applicable only to leukaemias that bear specific DNA markers including fused genes such as BCR-ABL, PML-RARA, AML1-MTG8 and CBFβ-MYH11. Hence, they are suitable for only about 30% of cases with acute myeloid leukaemia (AML) where specific translocations such as t(8:21), t(15:17) and inv(16) have been characterized (Liu Yin & Tobal, 1999). Moreover detection of MRD in acute lymphoblastic leukaemia (ALL) by PCR amplification of junctional regions of rearranged immunoglobulin and TCR genes requires sequencing and cloning of the junctional regions in the presentation sample (Cave et al, 1998). There is therefore a critical need to identify alternative gene targets that are expressed in the majority of AL patients. The behaviour or level of these molecular markers needs to be altered in the leukaemic clone even though they may not be specific to a particular type of leukaemia. One such example is the Wilm's tumour gene (WT1).

WT1 maps to the chromosome 11p13 and encodes a tumour suppressor gene associated with the development of Wilm's tumour, a paediatric kidney neoplasm (Call et al, 1990). WT1 has been shown to be overexpressed in > 90% of leukaemia cells, especially those with myeloid characteristics (Miwa et al, 1992; Miyagi et al, 1993; Inoue et al, 1994). Thus, the use of WT1 as a molecular target for MRD detection is potentially attractive, as it can be applied to the majority of AL patients, particularly those who lack a specific fusion gene. Several preliminary studies in the use of WT1, both as a prognostic marker and for monitoring MRD in AL, have been published (Inoue et al, 1994, 1996; Bergmann et al, 1997). Data on the prognostic significance of WT1 expression at diagnosis and during remission in AL are somewhat conflicting (Inoue et al, 1996; Schmid et al, 1997; Gaiger et al, 1998, 1999). This could be due to the variable levels of sensitivity of the protocols used for quantification of WT1 transcripts, with some of these studies relying on either qualitative or semi-quantitative reverse transcription (RT)-PCR protocols. Moreover, the quantification of WT1 gene transcript appears to be useful in predicting early relapse in AL patients (Cilloni et al, 2002); these results therefore emphasize the need for additional studies using sensitive methods to quantify and examine the kinetics of the WT1 transcript during different phases of the disease.

We have developed a sensitive, quantitative, nested RT-PCR assay to estimate levels of WT1 transcript and, by inference, levels of MRD that is coupled with a competitive RT-PCR for the ABL transcript as a control. We report here our evaluation of this method in the assessment of presentation and serial bone marrow (BM) and peripheral blood (PB) samples from a large group of patients with AL.

Materials and methods

Patients.  Patients with AL who presented between June 1994 and May 2000 were studied and followed until December 2001. The diagnosis of AL was made according to the French–American–British (FAB) classification. Acute leukaemia patients were treated according to United Kingdom (UK) Medical Research Council AML 10/12 or UK ALL XII protocols.

AML 10/12 protocols included remission induction with daunorubicin, cytarabine and 6-thioguanine (DAT 3 + 10) followed by consolidation with DAT 3 + 8, amsacrine, cytarabine and etoposide (MACE), and mitoxantrone and cytarabine (MiDAC). Patients aged < 50 years either had an allogeneic stem cell transplant if they had a matched sibling donor or an autograft, except for patients with good risk cytogenetics (Grimwade et al, 1998).

The UK ALL XII protocol comprises remission induction in two phases with daunorubicin, vincristine, prednisolone, cyclophosphamide, 6-thioguanine, l-asparaginase and cytarabine followed by consolidation with three infusions of high-dose methotrexate. Patients aged < 50 years had an allogeneic stem cell transplant if they had a matched sibling donor; otherwise, they received maintenance chemotherapy for 2 years, which also included two intensification courses.

BM and/or PB samples at presentation from 129 patients (107 AML, 22 ALL) and BM/PBPC from 24 normal donors were analysed qualitatively for the presence of WT1 transcript. BM and PB paired samples were available in 19 patients and seven normal donors. The median age of these patients was 51 years (range 16–78 years). Quantitative analysis of the presentation BM and/or PB samples was performed in 66 patients (52 AML, 14 ALL) and 24 normal samples. Cytogenetic data at presentation were available in 50/52 patients with AML and 11/14 patients with ALL.

Serial samples were analysed at presentation, after induction chemotherapy, second consolidation course, during remission at 3-monthly intervals, after bone marrow transplantation and at relapse in 28 patients with AML. The equivalent time points for patients with ALL (seven patients) were at presentation, after phase 1 of induction chemotherapy, after consolidation with methotrexate, during remission at 3-monthly intervals, after bone marrow transplantation and at relapse.

RNA preparation.  Mononuclear cells (MNC) were isolated from BM and PB using the Ficoll-Hypaque density gradient centrifugation method and stored at −80°C. RNA was extracted from 1–2 × 106 cells of the NB4 cell line and patients' samples according to the guanidinium–phenol–chloroform method of Chomzynski & Sacchi (1987), with minor modifications as described previously. Extracted RNA was quantified by spectrophotometry at 260 and 280 nm.

Reverse transcription (RT).  Total RNA (1 µg) was denatured at 72°C for 5 min and snap cooled on ice. RT was performed by adding RT reaction mixture (final concentration in a 20-µl reaction volume; 1× first-strand buffer, 10 mmol/l dithiothreitol, 0·25 µg of pd(N)6, 0·5 mmol/l dNTP, 200 U of Moloney murine leukaemia virus reverse transcriptase, 40 U of RNAsin) to the RNA and incubation at room temperature for 10 min. RT reaction was performed at 37°C for 60 min, then at 45°C for 30 min and at 72°C for 5 min.

ABL RT-PCR (qualitative).  The ABL gene transcript was chosen as a control gene to assess the quality and quantity of amplifiable mRNA used in the RT-PCR assays. PCR amplification was performed in a 25-µl reaction containing 1× PCR buffer, 0·5% W-1, 1·5 mmol/l MgCl2, 0·25 mmol/l dNTP, 15 pmol of primers (A2 and CA3), 1 U of Taq (Gibco) with 2 µl of cDNA. ABL PCR was performed at 97°C for 1 min 30 s, 64°C for 50 s, 72°C for 1 min (one cycle); 97°C for 30 s, 64°C for 50 s, 72°C for 1 min (40 cycles); 72°C for 5 min (one cycle). PCR products were electrophoresed on a 2% agarose gel. Any sample demonstrating the presence of genomic DNA was subjected to repeat RNA extraction. However, contamination with genomic DNA was encountered in only a minority of samples. Expected band size for ABL transcript was 276 bp.

WT1 RT-PCR (qualitative).  Two microlitres of cDNA was subjected to two rounds of PCR amplification for the WT1 transcript. First-round PCR was performed in a 50-µl reaction containing 1× PCR buffer, 0·5% W-1, 1·5 mmol/l MgCl2, 0·25 mmol/l dNTP, 15 pmol of primers (WT1 and WT4) and 1 U of Taq (Gibco). The first-round PCR was performed in a 50-µl reaction containing primers WT1 and WT4 at 93°C for 2 min (one cycle); 93°C for 30 s, 55°C for 50 s, 72°C for 1 min (40 cycles); 72°C for 5 min (one cycle). Two microlitres of first-round products were used in a 50-µl second-round PCR containing nested primers WT3 and WT11 under the same PCR conditions. PCR products were electrophoresed on a 2% agarose gel. The expected band size for WT1 transcript is 412 bp in the first round and 242 bp in the second round.

ABL competitor construction (Fig 1).  The ABL competitor was constructed using splicing by the overlap extension technique as described before (Clarkson et al, 1991; Tobal & Liu Yin, 1996). The size of the constructed ABL competitor was 285 bp.

Figure 1.

Diagrammatic description of the preparation of ABL gene competitor using the ‘splicing by overlap extension’ technique.

ABL competitive RT-PCR.  This is based on co-amplification of two fragments (gene and competitor) with the same set of primers, but produces two different-sized products. If a PCR contains the same amount of both transcript and competitor, then equal amounts of product will be seen on the agarose gel (same intensity of the bands) (Cross et al, 1993). Competitive RT-PCR was performed using 2 µl of sample cDNA and 2 µl of competitor DNA in a 25-µl reaction subjected to the same PCR conditions as described previously (Tobal & Liu Yin, 1996). The expected band size for the ABL transcript was 276 bp and ABL competitor was 232 bp. We estimated that there were 105 copies of ABL transcript in 1 µg of RNA.

WT1 competitor construction (Fig 2).  The WT1 competitor was prepared by PCR amplification of WT1 transcript from NB4 cDNA using primers WT5 and WT12. Primer WT12 is the reverse primer and has a tail that corresponds to reverse primers WT4 and WT11 that were used in two rounds of PCR amplification. The expected band size of the WT competitor was 323 bp.

Figure 2.

Diagrammatic description of the preparation of WT1 gene competitor.

WT1 competitive RT-PCR.  cDNA (2 µl) positive for WT1 and 2 µl of competitor DNA were mixed together in a 50-µl reaction and subjected to PCR amplification as described above. The expected band size of the competitor (WTC) produced was 275 bp in first round and 185 bp in the second round. Each sample was quantified at every order of magnitude and then at every half order of magnitude. The point of equivalence was assessed by gel densitometry. Because of the size difference between the WT1 transcript (242 bp) and the competitor band (185 bp), the number of competitor molecules at the point of equivalence was multiplied by 0·76 (the ratio of the size of competitor and the transcript). The level of WT1 transcript was normalized to 105 copies of ABL transcript present, thus eliminating errors resulting from sample variation and handling.

Sequences of PCR primers. Outer forward primer WT1 5′-GGCATCTGAGACCAGTGAGAA-3′

Inner forward primer WT3 5′-GCTGTCCCACTTACAGATGCA-3′

Inner reverse primer WT11 5′-GACAGCTGAAGGGCTTTTCA-3′

Outer reverse primer WT4 5′-TCAAAGCGCCAGCTGGAGTTT-3′

Forward WT5 5′-TCAGGATGTGCGACGTGTGC-3′

Reverse WT12 5′-TCAAAGCGCCAGCTGGAGTTTGACAGCTGAAGGGCTTTTCATTTCGCTGACAAGTTTTA-3′

CA3 5′-TGTTGACTGGCGTGATGTAGTTGCTTGG-3′

A2 5′-TTCAGCGGCCAGTAGCATCTGACTT-3′

ABL and WT1 degradation rates.  Rates of degradation studies were carried out for both transcripts at 24 h and 48 h using the NB4 cell line. The NB4 cell line sample was divided into three equal parts. One part was subjected to Ficoll-Hypaque density gradient centrifugation as described above, then frozen directly at −80°C; the other two parts were incubated at room temperature for 24 and 48 h, respectively, before centrifugation. The levels of ABL and WT1 transcripts were estimated in all three parts to determine the rates of degradation.

Accuracy and reproducibility of assays.  Positive and negative controls were used in each assay. The NB4 cell line was used as positive control. Negative controls included sterile H2O as a replacement for RNA or cDNA. Contamination was avoided through the use of an ultraviolet flow cabinet, designated PCR pipettes and filtered tips. Each assay was repeated at least twice to confirm the results.

Statistical analysis.  Overall survival (OS) was calculated from presentation until death, and the disease-free survival (DFS) from the achievement of complete remission (CR) until relapse. Kaplan and Meier life tables were constructed for the survival data, i.e. OS and DFS, and were compared by means of the log rank test.

Results

Degradation rates for ABL and WT1 transcripts

The degradation rates of the ABL and WT1 transcripts were found to be equal. After incubation of samples at room temperature, the levels of both transcripts decreased equally, by 0·5 log after 24 h and 1 log after 48 h. These results indicated that ABL is a suitable control transcript for quantification of WT1 transcript. RNA and cDNA samples stored at −80°C for up to 2 months showed no degradation of either transcript.

Linearity and sensitivity analysis (Fig 3)

Figure 3.

Linearity and sensitivity studies of the competitive nested RT-PCR titration assay for WT1. The figure above each track indicates the number of WT1 competitor molecules added. Dilution of the NB4 cell line in sterile H2O at factors of 100×, 1000× and 10 000× resulted in a reduction in the detectable number of competitor molecules 10-, 100- and 1000-fold respectively.

Serial dilutions of NB4 cells were made in sterile water. The RT-PCR method was found to be linear over a wide range of WT1 transcript levels as shown. The method has a sensitivity level of one NB4 cell in 104 dilution and, in some experiments, we were able to detect 1 in105. With this method, we were able to detect as few as eight copies of WT1 transcripts per 105ABL transcripts.

Qualitative assay (Table I)

Table I.  Disease characteristics and qualitative WT1 expression.
Disease
(FAB type)
No. of
patients
WT1 expression
positive
Percentage
  1. RAEBt, refractory anaemia with excess blasts in transformation.

Acute leukaemia12912093
Acute myeloid leukaemia107100  93·5
 M022100
 M11414100
 M2252496
 M31515100
 M46583
 M55120
 M655100
 M733100
 Not otherwise specified141393
 RAEBt1818100
Acute lymphoblastic leukaemia222091
 Common ALL1212100
 T ALL33100
 B ALL11100
 Biphenotypic11100
 Not otherwise specified5360
Normal controls24416

We performed this assay in 94 BM and 32 PB samples in 107 patients with AML. Twenty-one BM and eight PB samples were analysed in 22 patients with ALL. Altogether, 120/129 patients with AL (> 90%) had detectable levels of WT1 at presentation. It was not detected in five BM and two PB samples in seven patients with AML (patient FAB types were: one M2, one M4Eo, four M5, one not specified), and 2/22 patients with ALL. In the normal donors, WT1 transcript was not detected in 20 BM/PBSC samples, whereas a positive result was obtained in four. A positive qualitative ABL RT-PCR was sufficient to confirm the quality of the extracted RNA and, hence, the samples that were negative for WT1 transcript were not subjected to quantitative RT-PCR for ABL.

Levels at presentation (Table II)

Table II. WT1 expression level at presentation, disease relapse and in long-term remission.
Sample
type
No. of patients WT1
positive/total number
Normalized levels of
WT1 transcripts in BM
At presentation66/66> 103 copies/105 copies of ABL
Normal controls  4/24< 2 × 102 copies/105 copies of ABL
Long-term remission  7/21< 103 copies/105 copies of ABL
Follow-up > 36 months
At relapse18/18> 104 copies/105 copies of ABL

Quantitative analysis was performed in 66 patients (52 AML, 14 ALL). The median level of WT1 in AML patients was 8·9 × 104 (range 2 × 103−3 × 106) and 3·2 × 104 (range 8 × 102−8 × 105) copies/105ABL for BM and PB respectively. Patients with ALL had a median level of 6 × 104 (range 8 × 103−8 × 104) and 2·4 × 104 (range 8 × 102−8 × 104) copies/105ABL for BM and PB respectively. Eleven PBSC and 13 BM were analysed from normal donors. The gene was not detected in 9/11 PBSC and 11/13 BM, whereas in the remaining four normal samples, the median level of WT1 transcript was 80 copies/105 ABL (range 8–2 × 102).

Critical presentation level (Fig 4)

Figure 4.

Kaplan–Meier plot showing the correlation between the disease-free survival and WT1 level in BM at presentation in patients with acute leukaemia.

Patients in our cohort could be divided into three groups according to the levels of WT1 in the BM samples at presentation, namely < 104, 104−105 and > 105 copies. Sixteen patients (10 AML, six ALL) had between 103 and 104 copies/105ABL, and 100% achieved CR with induction chemotherapy, whereas in 32 patients (24 AML, eight ALL) with levels between 104 and 105 copies at presentation, the CR rate was 68%. Eighteen patients (all AML) had more than 105 copies, and only 50% achieved CR. The difference in the CR rates in the three groups was highly significant (P = 0·001).

Probability of survival at 3 years was 81%, 37% and 22% for patients with levels < 104, between 104 and 105 and > 105 copies respectively. Patients with WT1 levels < 104 had significantly better OS (P = 0·0003) and DFS (P = 0·0004) compared with patients with levels between 104 and 105 or > 105 copies (Fig 4). We did not find a significant difference in DFS (P = 0·1577) and OS (P = 0·2394) when patients with levels between 104 and 105 and > 105 copies were compared. Similar survival results were obtained when patients with AML were analysed separately.

Correlation between cytogenetics and WT1 level in the BM at presentation

Acute myeloid leukaemia.  Patients were classified according to their karyotypic abnormalities (Grimwade et al, 1998).

Good risk cytogenetics (13 patients).  A total of 8/13 had WT1 levels > 104, only two patients have remained in CR, whereas six have succumbed to the disease. Five out of 13 had < 104 copies; all remained in stable remission with a median follow-up of 46 months (range 16–73 months).

Standard risk cytogenetics (23 patients).  A total of 19/23 patients had a WT1 level > 104, 15 of whom have relapsed after a median CR duration of 14 months (range 6–42 months), two died during induction, one had refractory leukaemia and only one patient is in CR for 19 months. All four of the 23 patients with < 104 copies of WT1 at presentation remain in first CR at a median of 21 months (range 18–29 months).

Poor risk cytogenetics (14 patients).  Thirteen patients had WT1 levels > 104, and all either had refractory disease or have relapsed, whereas the only patient with < 104 copies remained in CR at 30 months.

Acute lymphoblastic leukaemia.  Among six patients with t(9:22) at diagnosis, three patients with WT1 levels > 104 at presentation were refractory to induction chemotherapy or subsequently relapsed at a median duration of 8 months (range 4–15 months), whereas three patients who had < 104 copies of WT1 achieved remission with a median duration of 30 months (range 27–34 months).

Critical remission level (Fig 5)

Figure 5.

Kaplan–Meier plot showing the correlation between the disease-free survival and WT1 level in BM after induction chemotherapy in patients with acute leukaemia.

WT1 levels were assessed at specific time points during the disease course, namely after induction chemotherapy and second consolidation chemotherapy in 28 patients with AML. The equivalent time points for patients with ALL (seven patients) were after phase 1 of induction chemotherapy and after consolidation with high-dose methotrexate. Patients were divided into two groups according to the level of MRD detected by WT1 assay in the BM. Patients who were negative or had < 103 copies of WT1 transcript after induction (Fig 5) and second consolidation chemotherapy (Figure not shown) had significantly better DFS and OS (P < 0·0001).

Levels during remission (Table II and Fig 6)

Figure 6.

Serial quantification of WT1 transcripts in four patients with AML in long-term CR (> 3 years). Levels of 1·00E−01 (= 1·0 × 10−1) are RT-PCR negative.

We studied serial BM and PB samples in 21 patients who continued to be in stable remission for a median follow-up of 51 months (range 15–91 months). WT1 has not been detected in 14 patients (10 AML, four ALL) since post-induction chemotherapy, whereas seven patients (six AML, one ALL) expressed WT1 at very low levels (< 103). Eleven patients have remained in remission for more than 4 years.

Prerelapse levels (Figs 7 and 8)

Figure 7.

Serial quantification of WT1 transcripts in three patients with AML and one patient with ALL who had persistently high levels of WT1 during the disease course. R, relapse. Levels of 1·00E−01 are RT-PCR negative.

Figure 8.

Levels of WT1 transcripts in AL at different phases of disease. A, relapsed patients (n = 14); B, remission patients (n = 21). Levels of 1·00E−01 are RT-PCR negative. Post Con, post consolidation.

We studied serial samples in 14 patients (12 AML, two ALL) who subsequently relapsed at the median CR duration of 12 months (range 4–49 months). In nine patients (eight AML, one ALL), relapse could be predicted 2–4 months before haematological relapse because of rising levels of WT1 (1–5 log rise). The median level of WT1 transcript in these samples was 6 × 104 (range 8 × 103−2 × 105) and 8·9 × 103 (range: 2 × 103−3 × 104) copies/105ABL in BM and PB respectively.

Five patients (three AML, two ALL) had persistently high levels, i.e. > 103 copies of WT1 throughout the treatment course. These patients relapsed within 6 months of diagnosis and succumbed to their disease (Fig 7).

Levels at relapse

Relapsed disease was studied in 14 BM and 11 PB samples from 18 patients. The median level of WT1 transcript was 8·9 × 104 (range 2 × 104−3 × 106) and 8·3 × 104 (range: 8 × 103−3 × 105) copies/105ABL in the BM and PB respectively. Six patients (six BM and four PB samples) had > 105WT1 copies at relapse.

Comparison of PB and BM analysis (Fig 9)

Figure 9.

Serial quantification of WT1 transcripts in four cases of relapsed AML and comparison of WT1 levels in BM and PB in two cases: FR (PB, BM) and LL (PB, BM). R, relapse. Levels of 1·00E−01 are RT-PCR negative.

BM and PB paired analysis was undertaken in 12 patients at presentation. Sequential paired samples were tested for WT1 levels in 12 patients who have relapsed and 10 patients who continued to be in CR for a median duration of 40 months (range 16–85 months). The levels in PB were lower than in BM by 1–2 log but mirrored changes in BM samples and were therefore equally useful in predicting early relapse in AL patients.

Discussion

The WT1 gene has been shown to be overexpressed in immature leukaemia cells, especially those with myeloid characteristics (Miwa et al, 1992; Miyagi et al, 1993; Inoue et al, 1994). We confirmed that > 90% of AL cases express the WT1 gene as reported previously (Inoue et al, 1994; Menssen et al, 1995). It was not expressed in 80% of acute monocytic and 33% of myelomonocytic (M4, M4Eo) leukaemia cases (Table I). Moreover, it was expressed at a lower level in more differentiated leukaemias such as acute promyelocytic leukaemia. However, unlike previously reported figures (Miwa et al, 1992; Menssen et al, 1995; Patmasiriwat et al, 1999), we found WT1 to be expressed in a higher proportion of ALL (91%). This could be due to the relatively high sensitivity of the nested PCR technique we have used to detect WT1 transcripts. These findings suggest that WT1 gene expression is closely associated with acute leukaemias of both myeloid and lymphoid lineages.

Data regarding the expression of WT1 on normal haemopoietic progenitors are somewhat conflicting (Baird & Simmons, 1997; Inoue et al, 1997; Maurer et al, 1997; Patmasiriwat et al, 1999). This discrepancy appears to result from the variable sensitivities of the different methodologies used to detect the WT1 transcript. Routine qualitative or semi-quantitative RT-PCR assays have not been able to detect the transcript in unsorted normal BM samples (Brieger et al, 1994; Menssen et al, 1995), whereas a sensitive quantitative method has shown WT1 to be expressed in normal CD34-selected progenitors (Inoue et al, 1994, 1996; Baird & Simmons, 1997; Maurer et al, 1997). Using a very sensitive quantitative method, we were able to detect WT1 in four out of 24 normal BM or PBPC donors. The level in these samples was appreciably lower (8–2 × 102 copies) than the level we detected in leukaemic patients at presentation (> 2 × 103). Thus, by quantitative analysis, we have shown WT1 expression to be at least 10 times lower in some normal progenitors compared with leukaemic cells, similar to previously reported results. (Inoue et al, 1996, 1997).

Several studies (Brieger et al, 1994; Schmid et al, 1997) using a qualitative assay for WT1 in AML found no correlation between WT1 expression at diagnosis and achievement of CR, whereas Bergmann et al (1997) have shown a tendency towards a higher CR rate in patients with low-level expression of WT1 mRNA using a semi-quantitative technique. However, Inoue et al (1996) showed a clear inverse correlation between high WT1 levels in AL and prognosis in terms of CR, DFS and OS. Quantitative analysis, using either competitive RT-PCR (Bergmann et al, 1997) or visual densitometry (Inoue et al, 1996), does not take into account the variation resulting from sample handling and quality of RNA and cDNA, whereas by performing double quantification for both WT1 and ABL transcripts and adjusting the WT1 value to 105 copies of ABL transcripts, we attempted to eliminate such potential errors. In this study, we were able to divide the patients into three groups according to the levels of WT1 transcripts at presentation. We showed that the CR rate was 100%, 66% and 50% in patients with WT1 levels of < 104, 104−105 and > 105 copies respectively. Furthermore, the actuarial survival at 3 years was 81%, 37% and 22%, respectively, in these three groups. Patients with < 104 copies in the BM at presentation had significantly better DFS and OS than patients with > 104 copies. Subgroup analysis showed similar results for AML patients. We confirm that WT1 levels at presentation in acute leukaemia, as well as in AML, are prognostically important for CR achievement, DFS and OS.

We also attempted to correlate WT1 levels at presentation with cytogenetic risk groups in AML. There appeared to be a trend for a higher proportion of patients with high WT1 levels (> 104 copies) in the less favourable cytogenetic groups (60% of good risk, 82% of standard risk and 93% of poor risk patients had > 104 copies). Within both good and standard cytogenetic risk groups, high WT1 levels correlated with a poorer outcome. Thus, WT1 level at diagnosis may be another useful pretreatment characteristic, similar to Flt3 ligand mutation (Kottaridis et al, 2001), and may enable the identification of high-risk patients in the cytogenetically defined good and standard risk subsets of AML. However, this important observation remains to be confirmed in a larger group of patients.

We carried out quantification of MRD in 35 patients (28 AML, seven ALL) at two specific time points, i.e. after induction and after second consolidation chemotherapy, and serially during remission. We found that patients who were negative or had < 103 copies at these time points had significantly better DFS and OS than those with > 103 copies of WT1. Similarly, during long-term remission, patients had either undetectable or < 103 copies of WT1. Furthermore, we were able to predict relapse in nine patients, in whom WT1 levels rose significantly, 2–4 months before the onset of haematological relapse. In five patients, WT1 levels remained persistently high during morphological and cytogenetic remission, and they all relapsed within 6 months. Thus, our results confirm and extend the observations of Inoue et al (1996), who used semi-quantitative assays for the WT1 transcript. Using a sensitive, quantitative method, we have also shown that PB is a reliable source of MRD in WT1-positive patients and could also be used successfully to predict early relapse. Clearly, a qualitative assay for WT1, because of its lack of sensitivity, is of no value in MRD monitoring in AL. We also postulate that it may be possible to establish critical MRD levels, based on WT1, above and below which patients are likely to relapse or remain in remission. These observations are in accordance with those reported for quantitative MRD monitoring using AML1-ETO transcripts in t(8:21)-positive patients (Tobal et al, 2000). More recent data also suggest that quantification of CBFβ-MYH11 transcripts by real-time PCR in AML patients with inv(16) abnormality may allow the establishment of MRD thresholds for determining relapse risk (Buonamici et al, 2002; Guerrasio et al, 2002). Real-time quantitative (RQ)-PCR has now been applied to quantify the WT1 transcript, and the results of early studies in AML also appear promising (Kreuzer et al, 2001; Cilloni et al, 2002). WT1 may thus be a valuable molecular target, and the results of our study provide the basis for future MRD monitoring by RQ-PCR analysis in AML patients who lack a specific fusion gene and possibly also offer an alternative target in ALL patients. Furthermore, RQ-PCR can produce a result in a few hours and lends itself to greater standardization and quality control.

In conclusion, we have shown that accurate quantification of WT1 transcripts in both AML and ALL patients at presentation is an important prognostic pretreatment characteristic and is also a useful predictive marker of leukaemia relapse. We also showed that, in at least 76% of patients in this series (presentation WT1 level > 104 copies/105ABL, i.e. at least 2 log higher than normal), WT1 is a suitable marker for MRD monitoring. Our data suggest that it may now be possible to offer MRD monitoring for most patients with AML. This information on MRD may in future provide the basis for therapeutic intervention, for example pre-emptive treatment in cases of molecular relapse, including stem cell transplantation or additional therapy for high-risk patients, thus allowing specific treatment to be tailored to individual patients. However, the clinical utility of WT1 as a marker of MRD in AML will need to be evaluated in prospective studies involving large numbers of patients.

Acknowledgment

Dr M. Garg was a Leukaemia Research Fund Clinical Training Fellow.

Ancillary