We thank the editor for giving us the opportunity to reply to Dr Dupont's letter, in which an assay for the detection of the MLL/AF4, MLL/AF6, MLL/AF9, MLL/ENL, and MLL/ELL fusion genes is described. In the reported assay, a total of five primer mixes are used. One mix (mix 1) is a ‘multiplex mix’ containing a common 5′ MLL primer together with a set of five different 3′ primers that are specific for each fusion partner gene. Three mixes contain a single primer pair designed to detect MLL/AF4, MLL/AF6 and MLL/AF9, respectively. The fifth mix (mix 19bis) contains a single 5′ MLL primer together with two 3′ primers located in the ENL and ELL genes, respectively. All reaction mixes are run in parallel at initial testing, and a final hybridization step with a 5′ MLL-specific biotinylated probe is performed to increase the specificity. Dr Dupont concludes that the described one-step multiplex assay allows a rapid and accurate detection of the mentioned fusion genes in ‘only one assay, while the method by Anderson et al needs a second step to specify the fusion gene’. Given this wording, the described protocol may at first sight seem more appropriate than the one reported by us, and we would therefore like to discuss some aspects of relevance in this context.
Apart from the necessity of an additional hybridization step, the inability to detect an MLL/AF10 fusion gene, and different primer locations (see below), the strategy by Dr. Dupont differs from ours in that we use two ‘multiplex mixes’ (out- and in-mix) that are run in two parallel single-step PCR reactions. Hence, with few exceptions, only if a sample is positive in both reactions, a split-out reaction with single primer pairs for each fusion gene is performed. In our view, the advantage, or disadvantage, of the two different strategies depends on the clinical setting in which the tests are carried out. If the expected frequency of an MLL fusion gene in a given sample is low, as in older children and adults with acute leukemia (5–10%), fewer reactions have to be prepared using our protocol. Our assay was in part developed to allow the detection of a clinically relevant MLL fusion gene upon initial diagnosis and bone marrow sampling when it still may be unclear if the patient suffers from acute myeloid or lymphocytic leukemia (ALL). If 10 such samples are tested consecutively, one sample harboring one of the mentioned fusion genes would mean setting up a total of 180 PCR reactions (excluding analysis of MLL/AF10, our assay requires eight reactions for the out- and in-mix, respectively, including appropriate positive and negative controls; a positive sample is verified by a split-out reaction requiring 20 reactions; no hybridization step necessary). Using the protocol described by Dr Dupont in a similar setting, a total of 250 PCR reactions would be necessary (each assay would require 25 reactions, including appropriate controls; no split-out reaction necessary, but subsequent hyridization needed). Only if the expected frequency of the detectable fusion genes exceeds 45% does it become more ‘economical’ and less labor intensive to use a similar strategy as proposed by Dr Dupont, provided that the protocol is optimized to obviate the need for hybridization. Naturally, if more clinical information is available before testing, eg a child with ALL, it may be reasonable to include a directed split-out reaction with MLL/AF4-specific primers already at the first PCR step.
This is a preview of subscription content, access via your institution