Background

The molecular response at 3 months (mo) after commencement of tyrosine kinase inhibitor (TKI) therapy for CML patients (pts) has prognostic significance and has been confirmed by many groups. Analyses by Marin, JCO 2012 and Neelakantan, Blood 2013 went further and suggest that measuring BCR-ABL at 3 mo is the only requirement to predict outcome, with BCR-ABL at later timepoints adding little useful information. However, Nazha (Haematologica 2013) using cytogenetic measurement suggest that assessing the 6 mo response for pts with a poor response at 3 mo may provide a better predictor of long-term outcome.

Aim

In this study, we aimed to dissect the prognostic importance of assessing both the 3 and 6 mo molecular response in our large cohort and determine whether a poor response at 3 mo identifies a group with high ongoing risk of treatment failure regardless of subsequent molecular response.

Method

We evaluated 528 pts enrolled in consecutive trials of first line imatinib from 2000-2009. Many pts were treated before alternative TKIs were available, but 89 switched. The utility of BCR-ABL as a predictor of death, progression (AP/BC/death: PFS) and treatment failure (according to ELN 2013: FFS) was assessed by ROC analysis, Kaplan-Meier survival with log rank test, and cumulative incidence with Fine & Gray regression for MMR. Pts were divided according to the 2013 ELN 3 and 6 mo molecular response definition: 3 mo, Optimal (Opt) ≤10% or Warning (Warn) >10%; 6 mo, Opt <1%, Warn 1-10% or Failure (Fail) >10%. The Warn category requires more frequent monitoring and Fail mandates a therapy change.

Results

Optimal BCR-ABL threshold values at 3 mo for outcome prediction were concordant with Marin, JCO 2012: death 16.1%; progression 9.5%; failure 8.4%; MMR 1.4%, demonstrating the commutability of the international reporting scale. Consistent with other studies, responses after 4 years were significantly superior for pts ≤10% (Opt) compared with >10% (Warn) at 3 mo: survival 97% v 90%, P = .003; PFS 96% v 84%, P < .0001; FFS 83% v 42%, P < .0001; and MMR 89% v 42%, P< .0001. 100 pts met the BCR-ABL Warn criterion at 3 mo: 14 of these progressed; 4 before 6 mo and a further 7 before 12 mo.

When modeling BCR-ABL measurements at 0, 1, 2, 3, 6 and 12 mo as a time-dependent continuous covariate in Cox regression, a BCR-ABL decrease at any time significantly reduced the risk of death (P = .0002), progression (P < .0001) and treatment failure (P < .0001) indicating that response after 3 mo also has prognostic significance. Of the 528 pts, 45 had missing molecular analysis at 3 or 6 mo, including 13 who ceased before 6 mo. The remaining 483 pts were categorized according to the ELN response definition at 3 and 6 mo and included 89 pts >10% at 3 mo. The Table and Figure detail outcomes at 4 years. Any reduction below 10% at 6 mo for pts in the 3 mo Warn category significantly reduced the subsequent risk of death, progression and failure, and increased MMR rates. Notably, there was no significant difference in outcome for pts >10% at 3 mo and <1% at 6 mo (Warn / Opt) compared to pts with an optimal response at both timepoints.

Table. Outcome beyond 6 mo of therapy according to the ELN 3 and 6 mo molecular response definition

Molecular response 3 mo / 6 mo*No. of evaluable ptsSurvival %PFS %FFS %MMR %
Opt / Opt 342 97 # 97 # 87 # 88 # 
Opt / Warn 42 100 # 97 # 79 # 71 #+ 
Opt / Fail 10 89 90 51 56 #+ 
Warn / Opt 18 100 # 100 # 76 # 88 # 
Warn / Warn 36 100 # 94 # 79 #+ 69 #+ 
Warn / Fail 35 74 + 69 + 11 + 3.3 + 
Molecular response 3 mo / 6 mo*No. of evaluable ptsSurvival %PFS %FFS %MMR %
Opt / Opt 342 97 # 97 # 87 # 88 # 
Opt / Warn 42 100 # 97 # 79 # 71 #+ 
Opt / Fail 10 89 90 51 56 #+ 
Warn / Opt 18 100 # 100 # 76 # 88 # 
Warn / Warn 36 100 # 94 # 79 #+ 69 #+ 
Warn / Fail 35 74 + 69 + 11 + 3.3 + 
*

3 mo Opt ≤10, Warn >10; 6 mo Opt <1, Warn 1-10; Fail >10

Significant difference from

#

Warn / Fail

+

Opt / Opt

Small number with Opt / Fail - results should be interpreted with caution

Conclusion

Our data confirms the importance of achieving BCR-ABL values of 10% or below at 3 mo after starting therapy. Based on the 3 mo assessment, the prognosis was significantly superior for these pts. Progression before the 6 mo timepoint occurred in 4% of pts with BCR-ABL >10% at 3 mo. However, the vast majority of pts with >10% at 3 mo continued therapy and the 6 mo BCR-ABL assessment provided important long-term prognostic information. Any reduction below 10% at 6 mo led to significantly superior outcomes, approximating those with optimal response at 3 mo. The impact and optimal timing of therapeutic intervention for pts in the poor risk category at 3 mo still needs to be determined in prospective studies and may require large patient cohorts. However, pts who are >10% at both 3 and 6 mo have inferior outcomes and are undoubtedly in need of a therapy change.

Disclosures:

Branford:Novartis: Honoraria, Research Funding; BMS: Honoraria, Research Funding; Ariad: Honoraria, Research Funding. Yeung:Novartis: Honoraria, Research Funding; BMS: Honoraria. Hughes:Novartis: Honoraria, Membership on an entity’s Board of Directors or advisory committees, Research Funding, Speakers Bureau; BMS: Honoraria, Membership on an entity’s Board of Directors or advisory committees, Research Funding; Ariad: Honoraria, Membership on an entity’s Board of Directors or advisory committees, Research Funding.

Author notes

*

Asterisk with author names denotes non-ASH members.

This icon denotes a clinically relevant abstract

Sign in via your Institution