We read with great interest the recent reports of recurrent mutations in the putative tumor suppressor gene ASXL1 that were published in this journal, as well as others. Several groups have reported the identification of missense, nonsense and frameshift mutations in ASXL1 in a wide spectrum of myeloid malignancies, including 11% of myelodysplastic syndrome, 17% acute myelogenous leukemia, 8% of myeloproliferative neoplasm, 43% of chronic myelomonocytic leukemia and 15% of chronic myelogenous leukemia patients. 1–5 In each of these publications, the most commonly reported mutation is an insertion of a guanine nucleotide at nucleotide position 1934 that results in a frameshift and resultant premature stop codon 12 codons downstream of the insertion (referred to as ‘c. 1934dupG; pGly646TrpfsX12’). Of note, this alteration accounted for> 50% of the mutations found in ASXL1 in a series of myelodysplastic syndrome and acute myelogenous leukemia patient samples. 2 Close inspection of this variant in ASXL1, as displayed in a unidirectional trace without paired normal tissue in Figure 1 in the publication of Carbuccia et al., reveals that this variant occurs within an 8 base-pair mononucleotide guanine (G) nucleotide repeat (Figure 1a). 4 Although the mononucleotide tract therein may represent a ‘mutational hotspot’, we became concerned that this frequently reported mutation might represent an artifact of PCR and/or sequencing, and not an actual somatic disease allele that contributes to myeloid transformation. It is noteworthy that the majority of reported mutations in ASXL1, including this allele, have not been assessed in paired normal DNA from the same individuals to ascertain if these mutations are acquired somatic mutations. 1–5

As part of our efforts to delineate the frequency of somatic ASXL1 mutations in myeloid malignancies, we have sequenced the entire open reading frame of the ASXL1 gene in paired tumor DNA and DNA extracted from buccal tissue from individuals with a variety of myeloid malignancies and from patients without hematologic disease. The c. 1934dupG; pGly646TrpfsX12 variant is readily apparent in paired tumor and normal DNA (Figure 1a). This finding in combination with the fact that this candidate frameshift alteration occurs within