In our study, the success rate for DNA and RNA analysis of routinely processed bone marrow trephines was 100%. In contrast to these results, Agarwal et al. reported a success rate of only 69% for extraction and amplification of gDNA (63 out of 202 bone marrow trephines failed [6], indicating suboptimal fixation and decalcification conditions and/or inefficient DNA extraction. Greiner et al. [2] reported a success rate similar to our study for the analysis of gDNA extracted form 211 FFPE bone marrow trephines. In a much smaller group of 20 bone marrow trephines, Tan et al. achieved also a success rate of 100% for the analysis of gDNA [7]. The number of analyzed archival bone marrow trephines in several previous studies is too small in order to draw any meaningful conclusions about feasibility, reliability, and sensitivity (Kähler et al. 2007: n = 5 [8]; Schumacher et al. 2008; n = 13 [9]; Kristensen et al. 2011: n = 1 [10]).

Sotlar et al. [11] described already in 2003 a real-time PCR assay for the detection of KIT p.D816V in archival FFPE biopsies. However, in contrast to our study, Sotlar et al. analyzed cutaneous manifestations of mastocytosis. Formalin-fixed, decalcified, and embedded bone marrow trephines were only included as negative controls which were all KIT Codon 816 wildtype. Therefore, nothing can be deduced from this data set about the reliability and sensitivity of KIT p.D816V mutation detection in decalcified bone marrow trephines.

Corless et al. [12] reported already in 2006 on the analysis of bone marrow biopsies for the detection of KIT codon 816 mutations in gDNA. However, from “Materials and Methods” and the results in table 2 in reference [12], it remains unclear whether the eight bone marrow samples designated as either “formalin-fixed” or “embedded” have been decalcified.

Kristensen et al. [13] performed a direct comparison of gDNA-based versus mRNA-based analysis of KIT p.D816V mutation before. However, they did not analyze formalin-fixed specimens. Instead, they compared the results for peripheral blood and bone marrow aspirate, respectively, describing a significant difference in diagnostic yield for gDNA versus mRNA in peripheral blood. The same group also demonstrated analysing various cell line mixtures and fresh patient samples that the potentially very sensitive next-generation sequencing methodology can be false negative if the VAF of the KIT p.D816V mutation is very small [14], an observation confirmed by our results analyzing fixed and decalcified bone marrow trephines.

In comparison digital PCR with gDNA which can only detect KIT mutation c.2447A > T, cDNA-based pyrosequencing can detect more than one activating KIT mutation. This limitation explains some discrepancies in the results especially concerning cases without KIT mutation in the digital PCR assay and KIT mutation c.2446G > C in the pyrosequencing (ID 42, 69, 70, and 77). Eleven patient samples showed the KIT mutation c.2447A > T only in the digital PCR assay (ID 41, 46, 47, 52, 53, 61, 62, 65, 80, 82, and 83) and three patient samples (ID 57, 68, and 96) showed a KIT mutation c.2447A > T only in the pyrosequencing assay with cDNA. In addition to a possible inhomogeneous distribution of neoplastic mast cells in the sections for the DNA and RNA extraction, the determination of threshold values was based on an absolute quantification of FAM labelled signals in the digitalPCR (five positive FAM-signals) whereas the results of the pyrosequencing were only based on relative quantification (4% A > T at position 2447 in the forward assay as cut off).

In conclusion, this study demonstrates that reliable and sensitive analyses of nucleic acids extracted from formalin-fixed, EDTA decalcified, and paraffin-embedded bone marrow trephines is possible under routine conditions and that the analysis of mRNA for mutation detection is a useful complement for the comprehensive diagnostic work-up in hematopathology.

In addition, this approach enables the direct correlation of morphological and immunohistochemical findings with the results of mutational analyses.