A Comparative Analysis of Molecular Genetic and Conventional Cytogenetic Detection of Diagnostically Important Translocations in More than 400 Cases of Acute Leukemia, Highlighting the Frequency of False-negative Conventional Cytogenetics

Rebecca L. King, MD; Mojdeh Naghashpour, MD, PhD; Christopher D. Watt, MD, PhD; Jennifer J.D. Morrissette, PhD; Adam Bagg, MD


Am J Clin Pathol. 2011;135(6):921-928. 

In This Article


Identification of genetic aberrations is required in the contemporary diagnosis of AL. With the publication of the WHO 2008 classification,[1] there are now 7 recognized, genetically defined subgroups of AML and 5 subgroups of B lymphoblastic leukemia. In fact, the presence of t(8;21), inv(16)/t(16;16), and t(15;17) can be considered diagnostic of AML, irrespective of the blast count.[1] All 3 of these entities confer a favorable prognosis among patients with AML. Cases with inv(16)/t(16;16) and t(8;21), both of which affect the core-binding factor transcription factor complex, are associated with good response to chemotherapy and relatively long complete remissions.[10,11] AML with t(15;17)/PML-RARA (also referred to as APL) is a prototype of a genetic aberration dictating specific therapy, given the key role of all-trans retinoic acid (ATRA).[12]

Treatment decision making (including remission induction chemotherapy, alone or with stem cell transplantation, and timing of stem cell transplantation) depends on multiple parameters; however, genetic data are paramount.[10,11,34] Failing to detect genetic abnormalities, including the aforementioned fusion transcripts, could potentially lead to inappropriate therapy, for example, treating with transplantation an apparently cytogenetically normal (CN) patient with AML with an unrecognized favorable prognosis, such as a patient with inv(16), or not exposing a patient with a cryptic t(15;17) to the benefits of ATRA.[12]

In our study, there were 5 cases in which CC revealed a normal karyotype, in which we found 4 balanced translocations by RT-PCR that could be targets for specific therapy, 1 with PML-RARA and 3 with BCR-ABL1 (Table 3). However, in the patient with APL, there was a suboptimal yield of analyzable metaphases (6 instead of 20). In this case, the t(15;17) detected by MRP was later confirmed with FISH. The morphologic picture was that of APL, and flow cytometry revealed the leukemia cells to be negative for HLA-DR and CD34. All 3 cases of B-cell ALL with a t(9;22) translocation missed by CC were positive for CD33. In addition, in the case with the cryptic inv(16) (Table 4), CC revealed a gain of chromosome 22 (which is often associated with this inversion), while the morphologic picture was myelomonoblastic with abnormal eosinophils, and the blasts expressed CD2. In the case of the t(4;11) missed by CC, the karyotype revealed abnormalities of chromosomes 4 and 11, albeit in no more than 6 cells in this composite karyotype, but did not demonstrate the presence of the t(4;11) translocation (Table 4). Morphologic features in the case of the cryptic t(8;21) were consistent with an AML with maturation. In addition, the blasts were noted to have large salmon-pink granules, a morphologic feature often associated with the t(8;21). Thus, the MRP results in these cases were not entirely unexpected, based on the additional findings.

Of patients with evaluable RT-PCR results, CC analysis was attempted in 362 patient samples. CC analysis failed in 32 samples (8.8%) owing to poor quality and/or lack of adequate metaphases. An additional 60 cases did not have CC ordered. Together, these account for the 21.8% of cases that had only MRP results available in our study. Of these, there were 4 cases in which RT-PCR revealed a t(15;17). In 3 of these cases, there were classic hypergranular APL morphologic features, whereas 1 was a microgranular variant. In one case, FISH confirmed the translocation, and in another, the immunophenotype was compatible with APL. In these 4 cases, it is unclear whether the decision not to order CC analysis was made by the clinical team before or following the reporting of the positive t(15;17) result obtained by MRP. It is likely, however, that the positive t(15;17) by MRP led the clinical team to make therapeutic decisions and opt not to order CC.

In our study, in cases with an optimal RNA yield, the MRP assay showed a diagnostic sensitivity of 100%, being able to detect translocations in all 50 of 50 cases in which they were identified by CC. By contrast, of the 57 cases that harbored 1 of the 7 translocations identified by MRP or CC and that could be directly compared, 7 (12%) had cryptic fusion transcripts that were detected by MRP only but missed by CC (Figure 1 and Table 3 and Table 4). This frequency of false-negative CC results is somewhat higher than in other studies in which cryptic fusion transcripts were seen in 0.5% to 10% of patient samples.[15,18,35–38] The reason for this is unclear, although 1 study demonstrated cryptic fusion transcripts in 15% of cases; however, here, RT-PCR included numerous translocations not detected by our assay, perhaps highlighting the need for more comprehensive MRP studies.[39]

A targeted review of the 7 cases with cryptic fusion transcripts identified 2 patients in whom therapy had been initiated before the submission of specimens for MRP and CC analysis (Table 3). One was a patient with AML with t(15;17)/PML-RARA who had already been treated with 2 doses of ATRA more than 2 days before sample analysis, whereas another was a patient with ALL with t(9;22)/BCR-ABL1 who had been treated with 1 dose of methotrexate at an outside institution. In both cases, it is likely that the therapy contributed to the false-negative cytogenetic results and that these translocations may not have been truly cryptic. The t(15;17)/PML-RARA case was the same case alluded to earlier in which there was a suboptimal yield of metaphases (6 instead of 20). However, these specimens were legitimately submitted for clinical diagnostic purposes, albeit not under ideal circumstances. Accordingly, although it might be argued that these are not bona fide false-negatives, they nevertheless bolster the value of MRP, given its ability to detect diagnostically relevant genetic events despite the putative negative effects of prior therapy on CC analysis.

Although this study and others highlight the usefulness of molecular (RT-PCR and/or FISH) assays that are rapidly and reliably performed in urgent (eg, APL) and nonemergency situations, the availability of molecular testing should not overshadow the critical role that CC can fulfill in the diagnostic setting. CC analysis, although time-consuming relative to molecular analysis by RT-PCR, reveals global chromosomal abnormalities that are not readily discernible by targeted molecular methods, which have the limited capacity to detect only what they are designed to detect. The importance of analyzing the comprehensive karyotype in AL is highlighted by the recent additions to the list of ALs defined by recurrent translocations or numeric abnormalities in the WHO 2008 classification,[1] all of which were initially identified by CC, as they were in the WHO 2001 classification. Furthermore, CC analysis may well lead to the continuous discovery of new, relevant genetic events in leukemia, despite our nascent molecular centricity.

In our study, without CC analysis, we would have missed 6 cases of AML with inv(3)(q21q26.2) or t(3;3)(q21q26.2)/RPN1-EVI1 translocation, which has been shown to be an extremely aggressive disease.[40] We also identified 1 case with a t(1;22)(p13q13), RBM15-MKL1, a rare translocation seen in fewer than 1% of AML cases that is associated with megakaryoblastic features.[41,42] This case was diagnosed as AML without maturation with nonspecific morphologic features (small blasts with scant basophilic cytoplasm and round nuclei); megakaryoblastic markers (CD41, CD42, CD61, and von Willebrand factor) were not evaluated by flow cytometry or immunohistochemical analysis. In another case of AML, we identified a t(6;9)(p23;q34)/DEK-NUP214, an abnormality associated with bone marrow basophilia and multilineage dysplasia and a poor prognosis.[43] We did not encounter any cases of AML t(9;11), which confers an intermediate prognosis and would not have been detectable by our RT-PCR assay. Curiously, we identified 9 AMLs with 11q23/MLL translocations, involving partners other than MLLT3, which is involved in the entity defining t(9;11).

It is also now well established that there are a number of specific gene mutations that afford prognostic significance in AL. This is especially true for AML, in which 40% to 49% of patients are CN, a designation that historically held an "intermediate" prognosis.[1,44] It is now recognized that a number of the CN patients (as well as many with recurrent translocations) have submicroscopic gene mutations that impart additional prognostic information.[1,44] Examples include mutations in the NPM1 gene, reported in 46% to 62% of CN patients with AML, which affords a relatively good prognosis, and internal tandem duplication of the FLT3 gene (28%–33% of CN patients), which predicts a poor outcome.[44] In ALL, recent studies have shown recurrent mutations in genes involved in B-cell differentiation, including PAX5 and IKZF, and these may have prognostic significance as well.[44,45] Detection of these mutations, as well as numerous others, relies on molecular methods.

There are a number of limitations in our study. First, the MRP assay is not comprehensive, and it should be updated to incorporate as many WHO 2008 genetically designated entities as possible. Future multiplex assays might also include the ability to test for prognostically relevant mutations alluded to earlier. In addition, the assay includes AML- and ALL-associated translocations in a single reaction; it would be more logical to group AML and ALL translocations separately, to preclude unnecessary testing. Second, our patient population is restricted to adults, with an expected predominance of AML (85% of our cases). However, while acknowledging that the epidemiology of AL in the pediatric population is very different from that seen in adults, we believe that the usefulness and complementarity of these 2 methods would likely hold in a similar study performed in the pediatric population.

This study has demonstrated that molecular and CC analyses for translocations are critical in determining diagnosis and prognosis and in guiding therapy in AL. This is highlighted by the occurrence of cryptic translocations that are missed by CC (false-negatives) but detected by MRP. However, although molecular technology has emerged as the more sensitive (diagnostically and analytically) and facile modality for genetic analysis, it is clear that CC continues to fill an important niche in the diagnosis of AL. Thus, it is important to appreciate the pitfalls of both strategies, and we believe that these methods should be used routinely in a complementary manner in clinical practice.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: