Although most CML patients treated with tyrosine kinase inhibitors (TKI's) achieve durable responses, some develop drug resistance that is usually due to a heterogeneous array of acquired mutations in the BCR-ABL kinase domain (KD). While many of these mutations confer resistance to imatinib, most (but not all) of these mutations will respond to a second-generation TKI (dasatinib or nilotinib). The qualitative detection of KD mutations, typically by direct DNA sequencing, is thus required for the optimal management of suspected drug resistance. Once a specific mutation is identified, however, a laboratory method to quantitatively monitor the mutation's subsequent response to the new therapy would be desirable. Toward that goal, we have developed and validated 2 different laboratory assays for the quantitative analysis of BCR-ABL KD mutations – pyrosequencing, and allele-specific PCR – and report their performance in the long-term serial monitoring of drug resistant CML patients.
For pyrosequencing, sequencing primers were designed 1–28 nucleotides adjacent to the polymorphic sites of common KD variants T315I, M351T, Y253H/F, E255K, F359V, M244V, Q252H, and G250E, and quantitation of mutant allele burdens was accomplished with the SNP-AQ function of the PyroMark ID instrument (Qiagen). For allele-specific PCR, real-time PCR primers were designed that preferentially amplified the mutant allele of common variants T315I, M351T, Y253H, E255K, and F359V. The 17 patients included in this retrospective study were all of those from our institution with a known KD mutation at any of the 5 loci targeted by our allele-specific PCR assays and with at least 5 available archival samples (from each patient) with known Sanger sequence information.
Of the 17 patients (65% male, average age=51), 16 had CML (1 had Ph+-ALL), and all were treated with imatinib as the initial TKI. 11 of the 17 patients achieved a major molecular response on imatinib. The total follow-up duration, from the time of imatinib initiation, was 6.9 years [median (IQR 4.0–8.8)], during which samples for BCR-ABL RQ-PCR were drawn every 3.0 months [median (IQR 1.9–4.2)]. The second-generation TKI was dasatinib in 9 patients, nilotinib in 1 patient, and AP24534 in one patient. The spectrum of KD mutations included T315I (8 pts), M351T (3 pts), Y253H/F (4 pts), E255K (4 pts), F359V (4 pts), Q252H (2 pts), and G250E (2 pts). Eight patients had 2 different KD mutations, and one patient had 3 different mutations. The first detectable KD mutation was found after 1.7 years of TKI therapy [median (IQR 1.0–2.0)]. From these 17 patients, 269 archival samples were available for quantitation of the mutation burden by pyrosequencing and allele-specific PCR [median 17 samples per patient (IQR 8–27)]. For allele-specific PCR (AS-PCR), the lower limit of detection was 100 copies of mutant DNA per PCR reaction. For pyrosequencing (Pyro), the lowest BCR-ABL transcript level that reliably yielded a signal above background was ∼0.03% on the international scale, and a mutant allele burden below 5% could not be reliably detected. For Sanger sequencing, a mutant allele burden below ∼20% could not be reliably detected. Of the 217 samples for which readable data could be generated by both Pyro & AS-PCR, AS-PCR was slightly more sensitive for the detection of a KD mutation - yielding positive results in 84 samples, as compared to 79 mutations detectable with Pyro. In contrast, Sanger sequencing detected slightly fewer mutations than either Pyro or AS-PCR, consistent with its presumed lower detection sensitivity. In 12 patients, there were a total of 48 samples that had a KD mutation detectable by both Pyro and allele-specific PCR in both the analyzed sample and an immediately prior sample, allowing a “delta allele burden” value to be calculated. The change in mutant allele burden between consecutively drawn sample pairs was no different for allele burdens quantitated by pyrosequencing as compared to those quantitated by allele-specific PCR (average 0.05 log difference; P>0.8).
Quantitative monitoring of the BCR-ABL kinase domain mutation allele burden can be accurately accomplished with either pyrosequencing or allele-specific PCR.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.