Time to Leukaemia (TTL) Assessed in NOD/SCID Mice Transplanted with Primary ALL Leukaemia Cells Determines Early Relapse in Patients and Is Identified by a Specific Gene Signature.

Poor response to induction therapy is the major risk factor identified in ALL and is used within the ALL BFM study to identify high risk groups (e.g. prednisone poor response, PPR: more than 1000 blasts/μl peripheral blood after treatment with prednisone systemically for 8 days and MTX intrathecally once on day 1). Despite the efforts achieved by the stratification strategies the majority of relapses are recruited from the group of initially good responding patients (in ALL BFM the standard and medium risk, SR and MR groups) emphasising the need for additional independent stratification factors. In our study we transplanted primary leukaemia cells from 50 children with newly diagnosed B cell precursor ALL (BCP-ALL) into NOD/SCID mice. Time to leukaemia was determined for each patient sample transplanted as weeks from date of transplant to date of clinical manifestation of the disease. Leukaemia was verified in spleen and bone marrow by flow cytometry staining for human CD19 and CD45. Time to leukaemia of less than 10 weeks (short TLL) was observed in 6 patient samples whereas 44 leukaemia samples took more than 10 weeks until appearance of leukaemia (long TTL). A clear cut in relapse free survival (Kaplan Meier analysis, N=50, log rank: P= .0000) was found for patients whose leukaemia cell samples showed short TTL (N=6, mean survival: 12.1 months, SE: 3.9, CI: 4.6–19.6) in contrast to patients with long TTL (N=44, mean survival: 54.3 months, SE: 2.9, CI: 48.6–60.1). Of note, the same distinct difference in relapse free survival was observed considering the SR and MR groups only (N=40, log rank P= .0000, long TTL: 44.9 months; short TTL: 12.5 months). By multivariate analysis (N=50, 3-years relapse free survival) patients exhibiting short TTL in the xenotransplant model exhibited a strongly increased risk for relapse with a risk ratio of 18.31 (CI: 5.03–66.72, P= .000). Interestingly, patients in our cohort showing prednisone poor response known as important clinical risk factor revealed only a risk ratio of 6.59 (CI: 1.32–32.75, P= .021). None of the patients with long TTL encountered early relapse. These findings in 50 directly transplanted samples were confirmed transplanting different cryopreserved BCP-ALL samples. In order to further characterise the biological properties of the leukaemia cell in the two groups, gene expression profiles of samples with short or long TTL in the xenograft model were investigated using a human whole genome array (Affymetrix U133 Plus 2.0). We identified a signature of differentially expressed genes distinguishing both groups. The differential expression was confirmed for selected genes by semi-quantitative PCR. Taken together, estimation of time to leukaemia (TTL) of leukaemia samples transplanted onto NOD/SCID mice is a new promising factor in paediatric ALL. Using an expression array approach patient samples displaying short TTL can be discriminated from those with long TTL by a unique gene expression signature. This allows direct identification of patients with increased risk for relapse by this new independent risk factor avoiding transplant in the mouse model.