Hierarchical Cluster Analysis of Immunophenotype in AML Patients with NPM1 Gene Mutation Reveals Two Distinct Groups with Different Prognosis.

Acute myeloid leukemia (AML) is a group of heterogeneous diseases characterized by increasing immature progenitors in the bone marrow and peripheral blood. Prior reports suggested a relationship between surface antigens expression and prognosis of AML. However, there were some debates in subsequent studies. In this study, we performed cluster analysis of the immunophenotype expression profile in AML patients with NPM1 gene mutation, and correlated the result with clinical characteristics, other gene mutations and prognosis.

Patients and Methods A total of 534 patients diagnosed as having AML at the National Taiwan University Hospital from 1987 to 2007 were recruited in this study. A panel of monoclonal antibodies was used to characterize the phenotypes of the leukemic cells. Expression of surface antigens on the leukemic cells was shown by an indirect immunoalkaline phosphatase method before 1998 and by flow cytometry thereafter. Cytogenetic study was done on bone marrow cells after 1–3 days of unstimulated culture. Analysis of gene mutation of NPM1, FLT3, CEBPA, MLL/ITD, AML1, JAK2, PTPN11, WT1, NRAS and KRAS were performed by polymerase chain reaction and direct sequencing. Abnormal sequencing results were confirmed by at least two repeated analyses. Comparisons were made with the ANOVA and the Chi-square test. Hierarchical cluster analysis was performed with agglomeration schedule, and cluster distance measured with Binary Square Euclidean distance. Dendrogram was plotted with average linkage method. Survival curves were plotted by the Kaplan-Meier method; differences between curves were analyzed by the log-rank test. All statistical analyses were performed by use of SPSS 13.0 for Windows (SPSS, Chicago, IL). Values of P< 0.05 were considered significant.

Result: Clinical characteristics of patients withNPM1gene mutation. There were 309 men and 225 women. 51 were children less than 18 years and 483 were adults. NPM1 gene mutations were detected in 106 (19.8%) of the total AML patients and 37% of those with a normal karyotype, similar to our previous report (Cancer Res. 2006). Most patients with NPM1 mutation were HLADR(−)CD33(+) CD34(−) (p<0.001). The patients with NPM1 mutations were older than the others (56 vs 47 years, p<0.001).

Hierarchical cluster analysis of immunophenotypes in patients withNPM1gene mutations. The cluster analysis was performed based on the expression profile of eight surface markers (HLADR, CD34, CD13, CD33, CD7, CD14, CD15, CD56). The clustering result was displayed by the Treeview dendrogram. The rows represented individual cases and the columns were the results of expression of individual surface markers. The blue color indicates positive expression and yellow color, negative. A total of 89 patients with NPM1 mutation and complete immunophenotype data were enrolled to the hierarchical cluster analysis. The clustering analysis divided the patients with NPM1 mutations into two groups, designated as cluster groups I (n=73, green) and II (n=16, red), based on the expression profiles of immunophenotype (Figure 1). Most patients in group I showed CD34(−)/CD7(−) but with variable expression of other antigens, while most group II patients showed HLA DR(+)/CD34(+)/CD7(+). With a median follow-up time of 56 months, groups II patients showed significantly shorter relapse free survival (RFS, median: 4 vs 22 months, p=0.005) and overall survival (OS, median: 8 vs 94 months, p=0.05) than group I patients. There was no association of immunophenotypic cluster with any other gene mutation, such as FLT3, CEBPA, MLL, AML1, JAK2, PTPN11, WT1, NRAS or KRAS. Multivariate analysis of variables including clinico-laboratory data (age, sex, WBC, Hb, PLT, LDH) and gene mutations revealed that immunophenotype cluster was an independent prognostic factor (RFS, p=0.002; OS, p=0.024) in AML patients with NPM1 mutation. For more practical use, we compared the outcome of patients with positivity for all HLADR, CD34 and CD7, with that of other patients; it was also shown that the former groups had significantly poor prognosis than the latter one (RFS, p=0.0005, OS, p=0.01).

Conclusion: Hierarchical cluster analysis of the immunophenotype profile of patients with NPM1 gene mutation could reveal two distinct groups with different prognosis. The immunophenotypic cluster with HLADR(+) CD34(+) CD7(+) predicts poor prognosis in AML patients with NPM1 mutation, independently from other gene mutations and clinico-laboratory variables.

View largeDownload slide

Figure

View largeDownload slide

Figure

Close modal