Abstract The approval of three tyrosine kinase inhibitors (TKIs) for the first line treatment of Chronic Myeloid Leukemia (CML) has generated an urgent need for molecular parameters predictive of unfavorable therapeutic outcomes. Recent evidence suggests that failure to achieve early molecular responses (i.e. BCR-ABL/ABLIS levels <10% after 3 months or <1% after 6 months of TKI treatment) results in inferior rates of both overall and progression-free survival. With the current study, we wanted to establish if high BCR-ABL transcripts at diagnosis would be associated with unfavorable responses to Imatinib Mesylate (IM). Thus, we correlated quantitative determinations of BCR-ABL levels at diagnosis with the outcome of 230 newly diagnosed CML patients that were assigned to receive IM 400 mg/die. BCR-ABL transcripts were measured from peripheral blood samples drawn before exposure to any form of treatment. Real-Time Quantitative PCR (RQ-PCR) determinations were subsequently performed in triplicates using glucuronidase-beta (GUS) as the reference gene, since previous evidence has demonstrated that ABL is not a reliable control gene in samples collected at diagnosis. Values were then reported on the International Scale employing a conversion factor obtained from the laboratory of the University of Heidelberg in Mannheim, Germany. Median follow-up of the study population was 50 months. Estimated 5-year cumulative incidences of complete hematologic response, complete cytogenetic response (CCyR) and major molecular response were 97.9%, 89.5% and 64.7%. Five-year probabilities of overall survival (OS), transformation-free survival (TFS: survival without disease transformation to the accelerated phase or blast crisis) and failure-free survival (FFS: survival without IM failure as defined by the 2009 European Leukemia Net recommendations) were 93.8%, 97.8% and 76%. Elevated BCR-ABL/GUSIS correlated with inferior probabilities of optimal response (p<0.001), and lower rates of CCyR after 12 months of IM (p<0.001). Moreover, high BCR-ABL/GUSIS transcripts were associated with lower probabilities of FFS (p<0.001) and TFS (p=0.01). When we employed the 2009 European Leukemia Net criteria to subdivide our patient cohort in optimal responders, suboptimal responders and individuals failing IM, we found that increasingly elevated BCR-ABL/GUSIS transcripts accurately distinguished the three patient groups (optimal vs suboptimal p<0.001; optimal vs resistant p<0.001; suboptimal vs resistant p<0.001). Furthermore, using receiver operating characteristic curves we found that progressively higher BCR-ABL/GUSIS levels at diagnosis defined quantitative transcript thresholds (15.96% for Optimal Response, 16.01% for EFS, 16.09% for FFS, 20.36% for TFS and 22.04% for OS) that separated low risk from high risk patients. Finally, we wanted to determine the concordance rates between BCR-ABL/GUSIS levels at diagnosis and early molecular responses (eMRs) at 3 and 6 months. We therefore employed the 15.96% BCR-ABL/GUSIS threshold to identify subjects with high (<15.96%) or low (>15.96%) probabilities of obtaining an Optimal Response and found that 69% of patients displaying <15.96% BCR-ABL/GUSIS achieved <10% BCR-ABL/ABLIS levels after 3 months of IM (p<0.001). Likewise, 78% of subjects presenting <15.96% BCR-ABL/GUSIS at diagnosis attained <1% BCR-ABL/ABLIS after 6 months of therapy (p<0.001). We conclude that high BCR-ABL transcripts at diagnosis measured by RQ-PCR employing GUS as a reference gene allow the identification of CML patients unlikely to benefit from IM that should receive alternative forms of treatment. Disclosures: Muller: Novartis, BMS, ARIAD: Consultancy; Novartis, BMS: Research Funding; Novartis, BMS, ARIAD: Honoraria.