Shorter Post-Relapse Survival (PRS) of Myeloma (MM) Patients Randomized to the Thalidomide (T) Arm of Total Therapy 2 (TT2) Related to Chromosome 1q21 Amplification (AMP1Q21, FISH) at Relapse.

Background: T2 randomized 668 patients up-front to +/− T as part of an intensive combination chemotherapy program prior to and after melphalan-based tandem autotransplants (Barlogie, ASCO 2005). Patients randomized to T had superior 2-yr CR (52% vs 43%, p<0.001) and 5-yr EFS (56% vs 45%; p=0.02). The similar 5-yr OS (67% on T vs 63% without T) was due to the shorter PRS of patients on the T arm. Given our recent reports of an adverse impact in TT2 of amp1q21 at diagnosis, we examined whether there were differences in amp1q21 between the 2 study arms at baseline and relapse.

Patients and Methods: All standard pre-treatment laboratory features including the frequency of cytogenetic abnormalities (CA) were similar in the 2 study arms. FISH data to assess amp1q21 were available for 475 patients at baseline and at relapse in 46 patients, of whom 30 also had baseline data. The amp1q21 index is a weighted average of the proportions of cells with 3 and >3 gene copies, ranging from 0 through 100. The post-relapse therapies were similar among the 2 arms and included T in all patients previously not on T and for most previously on T, often together with bortezomib + DEX (VTD regimen; Barlogie, Blood, 2004); DT-PACE and 3^rd autotransplants as well as mini-allotransplants were also applied either as first or subsequent salvage maneuver - with similar frequencies among the 2 study arms.

Results: The median PRS was 2.7 years among 95 patients (28%) in the no T group compared to 1.1 years for 67 patients (21%) on T (p=0.002). The amp1q21 index was similar in the 2 arms considering all 475 patients with FISH data available at diagnosis (p=0.28); this also applied for baseline levels among those who have relapsed (p=0.22). However, FISH analysis applied to 23 relapse samples from each arm showed that the 1q21 amp index was significantly higher in those on T than not on T (median, 69 vs 22; p<0.001). When restricted to the 30 paired baseline/relapse samples, the amp1q21 index increased at relapse in both study arms (p=0.0015), and more markedly in some on no T with very low levels at baseline. Among all 46 patients with relapse samples, randomization to T loses significance as a predictor of PRS after adjustment for amp1q21 (p=0.5), whereas amp1q21 remains a strong predictor (p=0.003, HR: 1.15 for an index difference of 10).

Conclusion: Similar OS in the 2 study arms of TT2, despite higher CR rate and superior EFS on T, reflects the shorter PRS of T-randomized patients, which could be traced to differences in tumor genetics between the 2 arms:

1. amp1q21 indices were similar at baseline but higher at relapse for patients on T;

2. those not on T relapsed earlier;

3. (3) among baseline and relapse variables examined, PRS was most strongly affected by amp1q21 index at relapse, neutralizing the notion of a potentially detrimental effect of initial treatment with T on subsequent salvage therapies.