Rare coincident NPM1 and RUNX1 mutations in intermediate risk acute myeloid leukemia display similar patterns to single mutated cases

Recently, Mendler et al. reported a low incidence of 4 of 472 (0.85%) acute myeloid leukemia (AML) cases that carried concurrent NPM1 and RUNX1 mutations. Interestingly, they found that RUNX1 mutations in these rare cases with concurrent NPM1 mutations were structurally unusual when compared to RUNX1 mutations observed in NPM1 wild-type cases. All these 4 cases had RUNX1 mutations that were in-frame, located outside the Runx homology (RH) domain and were also present in the germline.

To further investigate these findings in an independent cohort, we screened 2722 adult de novo AML cases with intermediate-risk cytogenetics (1171 females, 1551 males; median age 68.4, range 15.7–100.4 years) for NPM1 and RUNX1 mutations. Patients provided written informed consent and study protocols were in accordance with the Declaration of Helsinki. We found co-existent NPM1 and RUNX1 mutations in a similar rare subset of 0.44% of all cases (11 of 2722 cases) as described by Mendler et al.1 Clinical and molecular characteristics of these patients are shown in Table 1. Three patients were female, 8 patients were male. Median age was 67.7 years (range 42.0–82.0 years). Regarding NPM1 mutations, 9 patients had subtype A, one patient subtype I, and one patient harbored an unusual mutation in NPM1 consisting of a missense mutation (p.Trp290Leu) and a complex frameshift mutation in the 3′-UTR.

In the NPM1 mutated cases, we confirmed a high percentage of RUNX1 missense mutations: n=6 (54.5%) as compared to 37.0% in an independent cohort of RUNX1 mutated/NPM1 wild-type cases.2 However, in 5 cases, other various RUNX1 mutations were detected: n=2 frame-shift; n=2 nonsense; n=1 splice-site mutation. Regarding the localization of the RUNX1 mutations, 4 mutations were localized in the RH domain, 4 mutations in the TAD domain, and only 1 RUNX1 mutation was located downstream the RHD domain. This is in contrast to the report of Mendler et al. who report that all their 4 mutations detected in RUNX1 were located outside the TAD or RHD domain.

In 5 of our cases, follow-up material was available. In 3 cases (Patient ns. 4, 5 and 10), RUNX1 mutations were clearly somatic as they were non-detectable in complete remission material. In 2 cases (Patient ns. 2 and 7), a germline mutation could not be excluded as NPM1 and RUNX1 mutation loads did not decrease during follow up even though complete remission had been achieved. In 6 patients, no follow up or germline material was available.

Regarding cytogenetics, the patients did not differ from single NPM1 - or single RUNX1 mutated cases.2,3 In detail, 8 patients were cytogenetically normal, one patient had trisomy 8, one case showed loss of a sex chromosome, and one patient had a translocation t(5;12)(q33;p13).

Table 1.Clinical and molecular characteristics of primary AML patients with co-existing NPM1 and RUNX1 mutations^#.

Table 2.Clinical and molecular characteristics of relapsed AML patients with co-existing NPM1 and RUNX1 mutations.^#

RUNX1 mutation loads ranged between 3% and 48%. Interestingly, all 3 assured somatic mutations had a very low mutation load of less than 10%. In contrast, the NPM1 mutation load ranged between 30% and 50%. To analyze the disease-causing potential of RUNX1 mutations, all mutations were analyzed by PolyPhen prediction (genetics.bwh.harvard.edu/pph2/) and Mutation Taster (mutationtaster.org) algorithms and were identified as probably damaging to the protein function. We also subjected the detected RUNX1 mutations to the catalog of somatic mutations in cancer (COSMIC; cancer.sanger.ac.uk/cancergenome/projects/cosmic/), an online tool for storage and display of somatic mutation information and related details, also containing information relating to human cancers. Two mutations had an entry in COSMIC (p.Lys83Arg and p.Arg293*), the others had not yet been described. However, as all RUNX1 mutations except one involved functional domains of RUNX1, we suspect them to be disease-associated rather than polymorphisms.

Besides the 11 patients described above, our cohort also contained 2 patients with NPM1 mutated de novo AML who gained a RUNX1 mutation at relapse (Table 2), indicating that RUNX1 mutations can be acquired during disease progression

Taken together, we were able to confirm the rare concomitance of NPM1 and RUNX1 mutations in de novo intermediate risk karyotype AML. However, we could not confirm that RUNX1 mutations are always structurally unusual or germline in NPM1 mutated cases. In fact, in our cohort, most of them were not structurally unusual as had been postulated by Mendler et al.1 In our cohort, the majority of detected RUNX1 mutations in NPM1 mutated cases were located in functional domains of RUNX1, the remaining cases had one mutation located downstream the RHD domain and two splice-site mutations. This pattern does not differ from mutation patterns reported for RUNX1 mutations in NPM1 wild-type cases.