Classimed logo

Outcome Prediction for High Risk AML Patients

G.Valet1), R.Repp2)

1) Max-Planck-Institut für Biochemie, Martinsried
2) Medizinische Klinik III der Universität Erlangen, Germany

1. Background: AML patients are frequently stratified by prognostic parameters into therapeutic subgroups (*external link L2). Stratification is useful for therapy improvement in patient subgroups but does not provide indivualized outcome predictions.

2. Goal: Classification of the SHG-96 multicenter AML trial database (*external link L2) to identify high risk AML-patients prior to therapy (L1.).

3. Results: Algorithmic classification of the available immunophenotype, cytogenetic and clinical parameters (fig.1) provides predictive values of 100% for 5-year and of 88.6% for 2-year nonsurvival (fig.2) of individual patients.
Discriminatory data patterns (disease classification masks) contain 7 parameters for 5-year prediction and 12 parameters for 2-year classifications (fig.3) . Patient age and %CD4, %CD45 positive AML blasts are equally selected in both classifications.
The reclassification of the learning set indicates correct patient classifications with mask coincidence factors between 0.57-1.00 (fig.4) despite only partial coincidence of patient classification masks with the two disease classification masks >5-year survivors and <5-year nonsurvivors. Individually predictive (fig.3) and group oriented prognostic (fig.5) disease classification masks contain different parameters.

4. Conclusion and Outlook: Immunophenotype parameter pattern analysis identifies early on individual high risk patients with high predictive values. Cytogenetic parameters were not selected, probably because they occur only in about half of the patients as opposed to CD antigens expressed on all cells.
It seems promising to routinely evaluate multiparameter CD antigen expression intensity, antigen ratios and antigen spread of cell populations to enrich the percent cell frequency determination of the present study with additionally relevant information. This might further increase predictive values beyond the 95% level, aiming at individualized outcome predictions for all considered AML patients.

Literature References:
L1. Valet G, Repp R, Link H, Ehninger G, Gramatzki M and SHG-AML study group. Pretherapeutic identification of high risk acute myeloid leukemia (AML) patients from immunophenotype, cytogenetic and clinical parameters. Cytometry 53B:4-10, (2003)
*external link L2. Repp R, Schaekel U, Helm G, Thiede C, Soucek S, Pascheberg U, Wandt H, Aulitzky W, Bodenstein H, Kuse R, Link H, Ehninger G, Gramatzki M and AML-SHG Study Ehninger G, Gramatzki M and SHG-AML study group. Immunophenotyping is an independent factor for risk stratification in AML. Cytometry 53B:11-19, (2003)

© 2022 G.Valet
last update: Jun 03,2022
first display: Mar 31, 2003