MRD Kinetics Can Predict the Time to Relapse after Autologous Stem Cell Transplantation (SCT) in Chronic Lymphocytic Leukemia (CLL).

Introduction: Ongoing studies suggest, that myeloablative conditioning chemotherapie with autologous SCT is not a curative treatment option in high risk CLL patients. For this reason a method to predict progression free survival (pfs) in individual patients (pts) is desirable.

Minimal residual disease (MRD) short after auto SCT in pts with CLL are known to be homogeneous low with increasing MRD level at later time points. Early MRD increase is associated with clinical risk factors and high risk for clinical relapse.

Methods: To estimate time to progression in individual patients we studied MRD kinetics between 12 and 36 months after SCT (observation period) by quantitative ASO-primer IgH PCR and/or MRD flow in 37 poor risk CLL pts (5 % with mutated IgH, advanced Binet stage, high lymphocyte count). We postulated LOG-linear growth kinetics from 12 months after SCT until clinical relapse, which allowed calculation of time to predicted hematologic relapse (pHR) after auto SCT. All patients were in clinical remission before myeloablative conditioning regimen of TBI and high-dose cyclophosphamide following autologous SCT. 16/37 pts. with relapsed disease served as a control group whereas 21 pts in continuing clinical remission defined an observation group. LOG-MRD kinetics were described by a linear standard curve defined by 2 or more samples of patients in clinical remission more than 6 months (mos) apart. Significant MRD change was defined by a change of more than 0.5 orders of magnitude within the observation time. By this standard curve time to pHR, defined as a CLL level of 0.5 (i.e. 50% of all blood cells are CLL cells), was estimated.

Results: 28 of all 37 pts. showed increasing, 4 stable and 5 decreasing MRD level during the observation period. In the control group of 16 pts. with clinical relapse, MRD level of one pt. remained stable until 36 mos after SCT, whereas 15 pts. showed significant MRD increase with a median slope of 0.093 (0.04 to 0.25) LOG-MRD level and a median time to pHR of 51 (27 – 92) mos compared to an observed median pfs of 39 (28 – 64) mos after SCT (ns). Median difference between pfs and time to pHR was 3.5 (5.4 – 60) mos. Only in 5 of 15 pts this difference exceeded 12 mos of whom all relapsed earlier than the estimated time point.

In the observation group of 21 pts 9 (4 and 5 respectively) pts showed stable MRD level or significant MRD decrease. 12/21 pts showed significant MRD increase with a median slope of 0.08 (0.03 – 0.17) and a median time to pHR of 57 (28 – 160) mos. Only in 2/12 pts the clinical relapse preceded the pHR (0.4 and 7.4 mos) within the median clinical follow up period of 45 (25 – 69) mos.

Conclusions: LOG-linear MRD models can characterize CLL increase from 12 momths after SCT: Increasing MRD kinetics predict the time to clinical relapse with acceptable accuracy in the majority of CLL pts, although this simple model tends to overestimate the time to relapse. Further improvement of the model, e.g. by calculating absolute CLL numbers and/or by more sophisticated statistical methods might minimize this error. Nevertheless, this overestimation might also be caused by biological reasons, e.g. clonal evolution or subclone selection.