The Predictive Value of Early Molecular Response in Chronic Phase CML Patients Treated with Dasatinib First Line Therapy

Abstract 785

We assessed the correlation between molecular response at 3 and 6 months of dasatinib 100mg daily treatment and subsequent cytogenetic and molecular responses in 150 newly-diagnosed chronic phase CML patients treated with front line dasatinib in the UK SPIRIT 2 study (imatinib vs dasatinib). The median age was 54 years (range 18.4–82.1); 90 patients were male. The Sokal risk distribution was: 39 low, 65 intermediate and 46 high. At diagnosis 26 patients had splenomegaly >10cm below the costal margin; median WBC and platelet count were 65.7 (2.2-428) and 404 (101-2,433). The median hemoglobin level was 11.0 g/dl (4.17-15.8). The median percentage of blasts and basophils in peripheral blood was 0.4% (0-13.5) and 3.6% (0-19.2) respectively. The dose of dasatinib was adjusted according to tolerance. BCR-ABL1 transcripts in the peripheral blood were analyzed at 12 week intervals using RQ-PCR. Results were expressed as percentage ratios relative to an ABL1 internal control and expressed on the international scale. Complete molecular response (CMR) was defined as two consecutive samples with no detectable transcripts (RQ-PCR negative) and ABL1 control >40,000 (the median ABL1 control in the CMR samples was 96,000). In addition, we also explored a less stringent definition of CMR, namely CMR4.5 which was recently defined by the EUTOS group as BCR-ABL1 ratio of 0.0032 on the international scale, consistent with a 4.5 log reduction in the transcript level, without necessarily being RQ-PCR negative.

With a median follow up of 15 months (range 6–29) the 2 year cumulative incidences (CI) of CCyR, MMR, CMR4.5 and CMR were 84.5, 72.1, 24.1 and 5.6% respectively. The median BCR-ABL/ABL ratios at 3, 6, 12 and 24 months were 0.830%, 0.093%, 0.040% and 0.034% respectively. We investigated the predictive value of the BCR-ABL1 transcript levels at 3 (>10% vs ≤10% and >1% vs ≤1%) and 6 months (>1% vs ≤1%) of dasatinib therapy on the 2 years CI of cytogenetic and molecular responses. The 135 patients who at 3 months had a BCR-ABL1/ABL1 ratio ≤10% and the 81 patients who had a ratio ≤1% had a significantly better 2 year CI of CCyR (89.1% vs 50.2%, p=0.02 and 100% vs 84.7%, p=0.01), MMR (83.7% vs 14.2%, p=0.004 and 85.2% vs 54.3 p<0.001), CMR4.5 (25.0% vs 0%, p<0.18 and 37.6 v 3.3% p=0.001) but not CMR (6.7 vs 0%, p=0.51 and 7.1 vs 0% p=0.46). Similarly, the 109 patients who at 6 months had a transcript ratio ≤1% had a better 2 year CI of MMR (86.3 vs 13.9%, p<0.001), CMR4.5 (31.2 vs 0%, p=0.03) and CMR (14.3 vs 0%, p=0.04) than the remaining patients.

We used a receiver operating characteristic (ROC) curve to identify the optimal cut-off in the transcript level at 3 and 6 months that would predict the probability of each outcome with maximal sensitivity and specificity. Table 1 shows the results of applying the optimal cutoffs for each outcome in the 3-month analysis. Then we investigated whether the various outcomes could be better predicted using the cut-offs defined at 3 or at 6 months (including both the 1 and 10% cut-offs and the newly identified cut-offs) by using a multivariate model. For each outcome the cut-off defined at 3 months shown in Table 1 was superior. No pre-therapy patient characteristics were an independent predictor for cytogenetic or molecular response.

The key finding from this analysis is that patients who achieve a transcript level ≤10% after 3 months of dasatinib (135 of 150) have an 89.1% probability of eventually achieving CCyR, compared to 50.2% for patients with higher transcript levels (p=0.02). This preliminary observation may allow the identification of around 10% of dasatinib-treated patients for whom other forms of treatment might be considered although our conclusions require verification in further studies. The predictive power of RQ-PCR assessment can be greatly improved by identifying the optimal cut-offs for the specific outcomes, which is particularly important when predicting for the achievement of CMR. It remains uncertain whether these differences in response will translate into differences in survival and the SPIRIT 2 study continues to address this question.