Mixed Chimerism Predicts Early Time to Relapse in Myeloablative Conditioning in Allogeneic Hematopoietic Stem Cell Transplant Patients

Chimerism evaluation after allogeneic hematopoietic stem cell transplant (AHCT) is recommended to help predict relapse, graft loss, or acute graft-vs-host-disease (GVHD). While most studies associate early mixed chimerism with higher relapse, its predictive power is still debated. Thus, using chimerism to guide interventions, such as adjusting immunosuppression or adding donor lymphocyte infusions, remains uncertain. To evaluate the timing of chimerism changes and their impact on relapse and GVHD prediction at the University of Michigan A retrospective, single center analysis of relapsed AHCT recipients between years 2014-2024. Chimerism was measured using semi-quantitative PCR of short tandem repeats (STR-PCR) for CD33+ from peripheral blood mononuclear cells and CD3+ T cell compartment. On average, the first chimerism was obtained on day +46.7, and then monthly until complete CD3/CD33 chimerism. Complete donor chimerism (CDC) was defined as ≥95% and mixed chimerism as <95%. Overall, 161 AHCT patients were evaluated with median age of 58 years (23–74), and most received a PBSC graft (79.5%). Donors were 52% matched unrelated, 46% matched related, and 2% cord blood. The majority were transplanted for acute leukemia or MDS (87%). Conditioning regimens included both full intensity (FIC, 77.6%) and reduced intensity (RIC, 22.4%). GVHD prophylaxis was tacrolimus/methotrexate (80.7%) or tacrolimus/MMF (17.4%). Overall, 50% achieved CDC in both CD3 and CD33, leading to a longer time to relapse than the 41% with only CD33 chimerism (344 vs 151 days, p<0.01); this was especially true for FIC (382 vs 129 days, p<0.01). FIC patients (54.4%) who failed to achieve CDC relapsed sooner than RIC patients (33.3%) without CDC (129 vs 255 days; p<0.01). Interestingly, these FIC patients had both a higher rate of aGVHD (49% vs 33%) and steroid use (48% vs 25%) than RIC patients without CDC, suggesting greater immunosuppression and diminished GVL. Finally, loss of CD33 or CD3 chimerism was rarely observed before relapse (14 patients, 8.7%), with most losing donor chimerism only at relapse. Among those who lost donor chimerism prior to detectable relapse, patients achieving both full chimerism had significantly longer time to relapse than those with only CD33 chimerism (552 vs 149 days; p<0.02); no differences were observed with FIC or RIC in this group. Overall, tracking of chimerism post AHCT is variable between patients and providers. In our data, loss of donor chimerism was rarely detected prior to relapse, suggesting higher rates of chimerism analysis may be needed. Our data also suggest that failure to achieve CDC in FIC was a predictor of early relapse, likely due to worse disease, increased GVHD, and higher levels of immunosuppression. This highlights a focus for further investigation to clarify which chimerism patterns predict relapse after AHCT.