Early molecular prediction of response to TKI David Marin, Imperial College London I do not believe that we have been making the best of molecular monitoring. • Molecular monitoring is normally used to assess responses like MR3 and MR4.5 or CMR that are achieved late on therapy (while cytogenetics are used to assess responses early on therapy) • The current definitions of molecular response are partially arbitrary and with very little prognostic relevance One example of this is MR3 (MMR) PFS is similar in patients with CCyR regardless of the depth of molecular response 18 months 1.0 0.9 0.8 Probability of PFS CCyR with no MMR, n=91 0.7 CCyR with MMR, n= 41 0.6 p= 0.4 0.5 0.4 0.3 0.2 0.1 0.0 0 6 12 18 24 30 36 42 48 54 60 Months from starting imatinib therapy Marin et al, Blood 2008 PFS similar in patients with CCyR regardless of depth of molecular response Druker BJ, et al. NEJM, 2006;355(25):2408-17. PFS is similar in patients with CCyR regardless of depth of molecular response Kantarjian HM, et al. Blood. 2006;108:1835-1840. Maybe we have the concept of MMR wrong? The real question is: What is the additional reduction in the transcript level that confers survival advantage in patients already in CCyR ? Patient characteristics (n=282) Characteristics Age Median-yr Range-yr 46.3 13-86.4 Sex- no.(%) Male Female 157 (55.7) 125 (44.3) Sokal risk group - no.(%) Low Intermediate High 88 (31.8) 111 (40.1) 78 (28.1) Interval since diagnosis – (months) Median Range 1.5 0-6 Chromosomal abnormalities in addition to the Ph chromosome no. (%) 16 (6.0) Splenomegaly n (%) Spleen size ≥10cm below the costal margin, n (%) 186 (66.4) 75 (26.8) Marin et al, JCO 2011 Outcome in 282 patients treated with imatinib first line in Hammersmith Hospital 94% 6% death non CML 84% 10% death because the CML 77% 7% alive but not in CCyR 29% in CCyR but not on imatinib 48% 48% in CCyR on imatinib Marin et al, JCO 2011 Using appropriate statistical methodology we can identify the cut off in the 12 month transcript level for patients in CCyR that predicts for OS with the maximal sensitivity and specify. BCR-ABL/ABL= 0.53% Marin et al, JCO 2011 166 out of 282 patients who received imatinib as first line therapy were in CCyR at 12 months Transcript level >0.1% <0.1% 125 41 OS (%) EFS (%) p=0.5 94.2 96.3 p=0.08 80.4 93.7 Marin et al, JCO 2011 166 out of 282 patients who received imatinib as first line therapy were in CCyR at 12 months Transcript level >0.1% <0.1% >0.53% <0.53% OS (%) EFS (%) 125 41 p=0.5 94.2 96.3 p=0.08 80.4 93.7 20 146 p=0.01 81.5 94.4 p<0.0001 29.5 74.3 Marin et al, JCO 2011 Can we define robust molecular levels at 3, 6 or 12 month that have prognostic value? Patients outcome according to the transcript level measured at 3 month Outcome OS PFS EFS CCyR MR3 CMR cutoff (%) ≤9.84 >9.84 ≤9.54 >9.54 ≤9.84 >9.84 ≤8.58 >8.58 ≤2.81 >2.81 ≤0.61 >0.61 Number of patients at risk Eight years probability of the outcome 211 68 p<0.0001 93.3 56.9 208 71 p<0.0001 92.8 57.0 211 66 p <0.0001 65.1 6.9 169 79 p<0.0001 99.4 21.7 141 137 p<0.0001 82.5 21.1 57 222 p<0.0001 84.7 1.5 Marin et al, JCO 2011 Survival for 282 patients treated with imatinib first line at the Hammersmith Hospital according to molecular response achieved at 3 months Probability of survival BCR-ABL/ABL<9.8% OS= 93.3% BCR-ABL/ABL>9.8% OS= 54% p<0.0001 Time from onset of imatinib therapy (years) Marin et al, JCO 2011 8-year cumulative incidence of CMR on imatinib therapy according to the BCR-ABL transcript level at 3 months. Cumulative incidence of CMR 3-month transcript ratio ≤0.61% (n=57), 8-year CI of CMR of 84.7%, p<0.0001 3-month transcript ratio >0.61% (n=222), 8-years CI of CMR of 1.5% Time from onset of therapy (years) Marin et al, JCO 2011 Patient outcome according to BCR-ABL transcript level measured at 6 and 12 month 6 month 12 month Outcome cut-off (%) Number of patients at risk OS ≤1.67 >1.67 187 87 p<0.0001 93.7 74.7 PFS EFS Eight years probability of the outcome PFS ≤1.73 >1.73 188 86 p<0.0001 92.8 68.9 EFS ≤1.67 >1.67 186 87 p<0.0001 70.7 18.3 98 66 p<0.0001 92.0 24.7 CCyR ≤2.70 >2.70 MMR ≤0.73 >0.73 136 123 p<0.0001 81.6 20.4 CMR ≤0.21 >0.21 73 197 p<0.0001 42.7 6.1 Outcome OS MMR CMR cut-off (%) Number of patients at risk ≤0.53 >0.53 164 93 p<0.0001 95.4 74.7 ≤0.53 >0.53 164 92 p<0.0001 94.9 73.1 ≤ 0.57 >0.57 168 78 p<0.0001 82.1 41.4 90 114 p<0.0001 81.6 20.4 59 182 p<0.0001 52.1 4.1 ≤0.22 >0.22 ≤0.036 >0.036 Eight years probability of the outcome Marin et al, JCO 2011 These cut-offs can be used in other centres when the transcript level is expressed on the international scale Transcript ratio We have validated our findings using 95 patients treated with imatinib first line at the Royal Liverpool University Hospital 3 month ≤9.84% >9.84% 6 month ≤1.67% >1.67% 12 month ≤0.53% >0.53% Marin et al, JCO 2011 n OS 63 32 p=0.003 98.3 69.1 54 40 p=0.009 98.1 7.2 46 37 p=0.01 98.0 74.4 The predictive value of the transcript level measured at 3 months is not affected by imatinib dose reductions n 3 months BCR-ABL/ABL ratio CCyR (RR) c-CCyRS (RR) OS (RR) 186 5.8% 1 1 1 Non-haematological toxicity 23 7.1% p=0.8 0.94 p=0.6 1.12 p=0.5 0.97 p=0.6 Haematological toxicity 70 8.9% p<0.001 0.44 p<0.001 0.53 p<0.001 2.3 p<0.001 No discontinuation Marin et al, JCO 2011 What is the best moment to identify the poor risk patients? 3 months is the optimal time point to predict a patient’s outcome Off Imatinib high transcript level at 3 month BCR-ABL1/ABL1 (log) low transcript level at 3 month CMR Marin et al, JCO 2011 Results of transcript levels at 3 and 6 months are generally consistent 66% (L/L) (85%) 77% 6 month (1.67%) 11% (L/H) 3 month milestone 9.84% 2% (H/L) 23% 6 month (1.67%) 21% (H/H) (91%) 8 year CI of CCyR Low/Low (L/L) Low/ high (L/H) (L/L) Low at 3 and low at 6 months p<0.001 P<0.001 (L/H) Low at 3 and high at 6 months (H/L) High at 3 and low at 6 months (H/H) High at 3 and high at 6 months High/Low (H/L) P=0.09 High/High (H/H) 8 year probability of OS 93.5 % (L/L) 92.4 % (L/H) 83.3 % (H/L) P<0.001 55.6 % (H/H) (L/L) Low at 3 and low at 6 months (L/H) Low at 3 and high at 6 months (H/L) High at 3 and low at 6 months (H/H) High at 3 and high at 6 months The same principle applies to patients treated with 2G-TKI after imatinib failure OS and c-CCyRS in 107 patients treated with dasatinib or nilotinib after imatinib failure 3 months BCR-ABL/ABL <10%, n=77 3 months BCR-ABL/ABL >10%, n=30 Probability of OS Probability of c-CCyRS p=0.001 p=0.02 Time (years) Time (years) Milojkovic et al, Blood 2012 What about first line 2G-TKI? SPIRIT 2: Study Design Arm A Imatinib 400 Chronic phase CML within 3 months of diagnosis R Arm B Dasatinib 100 Randomised open label study Dasatinib patient characteristics N=142 Sex Male, n(%) Female, N(%) 79 (55.6) 63 (44.4) Age Median (range) 54.5 (18-82) Sokal risk group Low, n(%) Intermediate, n(%) High n(%) 35 (29.9) 51 (43.6) 31 (26.5) EUTOS risk group Low, n(%) High, n(%) 86 (83.5) 17(16.5) Marin et al, Blood 2012 24 month cumulative incidences of cytogenetic and molecular responses Cumulative incidence of response CCyR = 91.5% MR3 = 70.1% MR4.5 = 39.1% CMR = 6.5% Time from onset of therapy (months) Marin et al, Blood 2012 The BCR-ABL1 transcript levels at 3 month strongly predict for the 2 year CI of CCyR, MR3 and, MR4.5 MR4.5 BCR-ABL/ABL >10%, n=11 p<0.001 Time from onset of therapy (months) Cumulative incidence of MR4.5 BCR-ABL/ABL ≤10%, n=117 Cumulative incidence of MR4.5 Cumulative incidence of CCyR CCyR p<0.001 BCR-ABL/ABL ≤10%, n=117 BCR-ABL/ABL >10%, n=11 Time from onset of therapy (months) Marin et al, Blood 2012 ….. but the optimal cut-offs may turn out to be lower MR4.5 CCyR BCR-ABL/ABL ≤2.26%, n=88 BCR-ABL/ABL >2.26%, n=40 Cumulative incidence of MR4.5 Cumulative incidence of CCyR p<0.0001 BCR-ABL/ABL ≤0.57%, n=62 BCR-ABL/ABL >0.57%, n=66 p<0.0001 Time from onset of therapy (months) Time from onset of therapy (months) Marin et al, Blood 2012 Thanks to: Dragana Milojkovic & John Goldman