Fourteen patients with high-risk B-lineage acute lymphoblastic leukemia (ALL) in complete remission underwent autologous bone marrow transplantation (BMT) using a combined immunochemopurging protocol. A monoclonal antibody (MoAb) cocktail of BA-1, BA-2, and BA-3 plus rabbit complement (C') plus 4-hydroperoxycyclophosphamide (4-HC) was used to eliminate residual occult leukemia cells from autografts. All patients were conditioned with single-dose total body irradiation (TBI) followed by high-dose Ara-C. All 14 patients engrafted at a median of 24 days (range, 12 to 36 days). Three patients are alive and disease free at 3.5 years, 3.9 years, and 4.1 years post-BMT. The Kaplan-Meiser estimate and standard error of the probability of sustained remission was 23% +/- 12% at 3.5 years post-BMT with a mean relapse-free interval of 1.4 +/- 0.4 years. The disease-free survival (DFS) at 3.5 years was 21% +/- 11%, with a mean DFS time of 1.3 +/- 0.4 years. A novel and quantitative minimal residual disease (MRD) detection assay, which combines fluorescence-activated multiparameter flow cytometry and cell sorting with leukemic progenitor cell (LPC) colony assays, was used to analyze remission BM samples from B-lineage ALL patients for residual LPC, and to evaluate the efficacy of ex vivo BM purging. Notably, the minimal residual leukemia burden before BMT, as measured by the percentage of B-lineage LPC in the pre-BMT remission BM samples, indicated the outcome of the BMT. The median value for the minimal residual leukemia burden before BMT was 0.0035% (35 LPC/10(6) mononuclear cells). The Kaplan-Meier estimates and standard errors of the probability of remaining in remission after BMT were 43% +/- 19% for patients whose BM samples contained less than or equal to 0.0035% LPC and 0% +/- 0% for patients whose BM samples contained greater than 0.0035% B-lineage LPC (P less than .05). In contrast to the minimal residual leukemia burden measured by the described MRD assay system, the percentage of blasts or TdT+ cells in the remission BM samples did not correlate with the probability of relapse. The applied purging protocol showed variable success in destroying target B-lineage LPC populations contaminating the autografts. While in some cases purging was highly effective, eliminating up to greater than or equal to 4 logs of residual B-lineage LPC, in other cases only 0.1 to 0.2 logs of B-lineage LPC were purged.(ABSTRACT TRUNCATED AT 400 WORDS)