Identification of New Markers for the Detection of Minimal Residual Disease in AML by Gene Expression Profiling and Real Time PCR.

Monitoring of minimal residual disease (MRD) has become one of the strongest diagnostic tools in the treatment of acute leukemia. MRD data after induction therapy can be used for risk group stratification. Long-term monitoring of MRD may allow predicting pending relapses. Suitable MRD markers are available for almost all ALL patients. In AML, many patients can not be monitored because there are no specific markers available that distinguish between leukemic cells and healthy blood (hB) or healthy bone marrow (hBM).

It was recently shown that MRD in AML can be monitored by measuring the expression of WT1 (Wilms tumor gene) or PRAME (Preferentially expressed antigen in melanoma). Both genes are strongly overexpressed in the leukemic cells of many AML patients. The aim of the present study was to identify additional genes which are strongly overexpressed in AML.

In a first step, five microarrays (Affymetrix U133 chip) were performed. 1^st: pooled RNA from 10 samples of healthy CD34+ stem cells (hSC); 2^nd: pooled RNA from 10 samples of hBM; 3^rd: pooled RNA from 10 children with AML M1 or M2; 4^th: pooled RNA from 10 children with AML M4; 5^th: pooled RNA from 10 children with AML M5. Twenty-eight genes were found to be absent in hBM and hSC but present in all three subtypes of AML. About 200 genes were absent in hBM and hSC but present in at least two subtypes of AML.

In a second step, 35 genes were selected from these lists and analyzed by real time PCR in the pools of RNA from hBM and from AML patients. Using real time PCR none of the genes was found completely negative in hBM. However, for eight genes the expression in the AML pools was more than 100 times higher than in the normal control (PRAME, WT1, TRH, CTGF, POU4F1, CCL23, ST18, and CCNA1).

These eight genes were analyzed by real time PCR in 51 samples from children with previously untreated AML, 10 samples of hBM and 10 samples of hB.

The most promising candidates for MRD markers were PRAME and ST18. The expression of both genes was up to 10000 times higher in AML patients than the highest levels that were found in hBM or hB. WT1 was completely negative in all hB samples but positive in some samples of hBM. The expression in the patient samples was up to 100 times higher than the highest expression in hBM. This is less than previously described for adult patients. TRH and CTGF could be very useful markers for the analysis of MRD in peripheral blood but the expression in hBM is probably too high to allow a reliable detection of MRD. CCL23 and POU4F1 could be useful marker for the detection of MRD in blood and bone marrow but very high levels were only found in a few patients. CCNA1 was expressed at relatively high levels in some hBM and some hB samples and therefore does not seem to be a useful MRD marker.

It is not yet clear what level of expression is needed to use a gene as MRD marker. It was recently shown that WT1 can be used to predict pending relapses in some children with AML. Our results suggest that PRAME and ST18 are at least as useful. We are currently analyzing the expression of these genes retrospectively in follow up samples of 35 children with AML who were treated in our hospital. We hope to show that the combined analysis of these markers allows a more specific and sensitive monitoring of MRD.