Good-prognosis acute myelogenous leukemia (AML) is associated with 3 chromosome translocations (8;21, INV(16)/16;16, and 15;17) that generate the fusion genes AML1/ETO, CBFβ/MYH11, and PMLRARα, respectively. Each fusion gene provides a disease-specific mRNA detectable by reverse transcriptase–polymerase chain reaction (RT-PCR) with more than 10-fold higher sensitivity (103-104) for detecting residual leukemia than standard methods (≤ 102). In AML1/ETO and CBFβ/MYH11 AMLs, RT-PCR positivity was variably found during continued clinical remission, complicating predictive assessment. In acute promyelocytic leukemia (APL), however, confirmed PML-RARα mRNA positivity after consolidation therapy was linked to high relapse risk. This criterion was incorporated into a multi-institutional protocol as a basis for salvage therapy, and pilot results suggested that this intervention produced superior outcome compared with a historic group treated after clinical relapse.1
Recently, real-time quantitative RT-PCR (RQ-RT-PCR) has been applied at higher sensitivity (104-105) with goals of improving predictive accuracy by quantifying RT-PCR positivity after consolidation and of disclosing possible quantitative predictive associations at earlier time points. Studies of small groups with 8;21 (eg, Viehmann et al2 ) or INV(16)/16;16 (eg, Guerrasio et al3 ) AML and of a larger APL group4 identified tentative quantitative cutoff levels for good-versus poor-risk patients after consolidation therapy but did not find an association between pretreatment or preconsolidation fusion mRNA levels and outcome.
Schnittger and colleagues (page 2746) now report a “new score” using pretreatment and immediate postconsolidation RQRT-PCR determinations in good-prognosis AML. By combining values for the ratios of fusion gene/ABL (housekeeping) gene transcripts from higher than median value postconsolidation samples with higher than 75th percentile value pretreatment samples (poor-risk) and comparing them to lower than median value postconsolidation samples (good-risk), they found zero incidence of “events” in the good-risk group, which impressively differed by Kaplan-Meier analysis from the poor-risk group for each AML subclass. They also identify other subgroups with poor-risk for event-free survival, including patients from each AML subclass with pretreatment transcript ratios greater than the 75th percentile. By deduction, all postconsolidation patients studied had pretreatment ratios lower than 75th percentile and this group was enriched for APL cases, raising some concern that analyses including postconsolidation data might partly involve a comparison of APL with low transcript ratios to the other 2 subclasses with higher transcript ratios and poorer clinical outcome. How the “new score” relates to relapse-free survival with longer follow-up seems of particular interest, since its potential clinical value will pertain only to patients who do not suffer an “event” before completing consolidation therapy.
Surprisingly, the investigators conclude that individual patient monitoring seems a more promising application of RQ-RT-PCR. This is partly based on analysis of 15 cases, in which an increase from a previous negative RQ-RT-PCR assay was demonstrated in all 8 patients who relapsed within 6 months. This limited data set does not specify individual assay sensitivities, and the reported 2000-fold ABL transcript variation could affect data subsets, especially at/near zero during the postconsolidation period, even though no overall correlation between transcript ratios and ABL transcript numbers was found. One hopes that a forthcoming report5 and future trials will address essential RQ-RT-PCR standardization requirements to derive reliable predictive values for both cohort and individual case analyses.
In answer to the title-posed question: not quite yet.
