Therapy-related acute myeloid leukaemia with mutated NPM1: treatment induced or de novo in origin?

Acute myeloid leukaemia (AML) carrying NPM1 mutations that cause aberrant cytoplasmic expression of nucleophosmin,^{1, 2} accounts for about one-third of de novo adult AML (50–60% of AML with normal karyotype), and shows distinctive biological and clinical features.³ Until now, little information was available on the incidence of NPM1 mutations in therapy-related AML (t-AML) or on the molecular and clinical characteristics of t-AML with mutated NPM1.^{1, 4, 5 and 6} Thus, the first comprehensive study of t-MDS (t-myelodysplasia)/t-AML with mutated NPM1 reported by Andersen et al.⁷ in this issue of the Journal is timely and remarkable.

Andersen et al.⁷ observed that NPM1 mutations occurred at lower frequency in t-AML than in de novo AML and were not related to any specific type of therapy. Notably, about 70% of the NPM1-positive t-MDS/t-AML had normal karyotype (an unusual finding in these cases, which generally show an abnormal karyotype) and none of them carried 7q−/−7, the most common abnormality in t-AML.⁸ Finally, about half of t-MDS/t-AML with mutated NPM1 expressed the FLT3 internal tandem duplication.

These findings are of great relevance since they indicate that t-AML with mutated NPM1 markedly differs cytogenetically and molecularly from other t-AML subtypes. Interestingly, normal karyotype and a high incidence of FLT3 internal tandem duplication are both distinctive features of NPM1-mutated AML with de novo origin.¹ The finding by Andersen et al.⁷ that two t-AML with mutated NPM1 carried a ring chromosome 10 and a trisomy 8, respectively, is in keeping with the observation that about 14% of de novo AML with mutated NPM1 harbour chromosomal aberrations, which are thought to be secondary.¹

De novo AML with mutated NPM1 shows a distinct immunophenotype¹ and gene expression profile,^{9, 10 and 11} which is characterized by CD34 downregulation (at RNA and protein levels) and upregulation of most HOX genes. Unfortunately, Andersen et al.⁷ were not able to provide any information on the HOX gene signature and CD34 protein expression in their cases. However, the few NPM1-mutated t-AML with normal karyotype so far investigated were usually negative for the CD34 molecule,⁴ suggesting they may have the same immunophenotype (and possibly the same molecular signature) as NPM1-mutated AML arising de novo.¹ Future efforts are warranted to clarify this issue.

Thus, although some information is still missing, the evidence strongly suggests that de novo and therapy-related NPM1-mutated AML share common biological features. This finding is hardly surprising since t-AML with other recurrent genetic abnormalities, such as t(15;17) inv(16) or t(8;21), usually show the same biological and clinical findings as de novo AML with the corresponding karyotypes.¹² Rather, it further supports the view that NPM1 mutation is a founder genetic alteration, which defines a distinct leukaemia entity.³

The results from Andersen et al.⁷ also raise the intriguing question of whether their t-AML cases with mutated NPM1 are truly secondary leukaemias induced by previous treatment or whether they represent de novo NPM1-mutated AML arising in patients with a history of therapy. If NPM1-mutated t-AML is related to previous treatment, the close association with normal karyotype clearly indicates that the leukaemogenic mechanisms must differ from those that operate in t-AML carrying an abnormal karyotype. In t-AML with t(8;21) or t(15;17), DNA double breaks caused by drugs, especially topoisomerase II inhibitors, are thought to favour the development of balanced translocations.^{13, 14} However, the role of NPM1 mutants in the development of de novo AML still remains to be elucidated,³ and, therefore, no causal relationship with previous treatment can be established at this time.

Conversely, like t-AML carrying internal-tandem duplication of the MLL gene,¹⁵ it is conceivable that a significant fraction of the t-AML reported by Andersen et al.⁷ are de novo leukaemias occurring incidentally in patients with a history of therapy. This is supported by the finding that these cases often received local radiotherapy or methotrexate plus prednisone, which are treatments whose leukaemogenic potential remains controversial. Moreover, two patients developed t-AML after unusually short or long latent periods. Another argument against these cases being secondary to therapy is that AML with mutated NPM1 has been constantly associated with a de novo origin. Indeed, NPM1 gene mutations are not usually observed in AML secondary to myelodysplasia or chronic myeloproliferative disorders,^{1, 3} and in-depth analysis of the only reported case of NPM1-mutated AML in this setting revealed a possible de novo origin.¹⁶

Therapy-related AML appears to be molecularly and clinically heterogeneous and its prognosis varies with the underlying genetic lesion. Cases harbouring 7q−/−7 genetic alterations are usually associated with poor prognosis, while t-AML carrying the t(15;17), t(8;21) or inv(16) translocations generally have a good outcome. Thus, whatever the nature of NPM1-mutated t-AML (therapy-related or arising de novo), future clinical studies should address the issue of whether t-AML with the NPM1-mutated/FLT3 internal tandem duplication-negative genotype shows the same favourable prognosis as the corresponding de novo cases.