Background: Early identification of predictive factors of failure to mobilise CD34+ cells could enable rational use of plerixafor during first mobilisation, avoiding the need for a second mobilisation course. However, "on demand" administration of plerixafor needs to be driven by established parameters to avoid inappropriate use.
Materials and methods: To address this issue, we studied the value of the peripheral blood CD34+ count, measured early (on days +10, +11, +12 and +13), in predicting the mobilisation outcome in the ensuing days. We retrospectively collected data from three Italian centres on 233 patients affected by multiple myeloma or lymphoma who underwent a first or second attempt at mobilisation with cyclophosphamide 4 g/m(2) and granulocyte colony-stimulating factor. To assess the diagnostic value of peripheral blood white blood cell and CD34+ cell counts with respect to "mobilisation failure", we considered failed mobilisation as "disease" and the CD34+ cell count in peripheral blood, on a specific day, as a "diagnostic test". For various thresholds, we measured sensitivity, false positive rate, specificity and positive predictive value (PPV) as well as the area under the receiver-operating characteristic curves (AUC).
Results: A CD34+ cell count <10 × 10(6)/L on day 13 had high sensitivity (1.00) and high specificity (1.00) for predicting subsequent mobilisation failure, with an AUC of 1.0. However, good prediction was also obtained using a lower threshold (CD34+ cell count: <6 × 10(6)/L) at an earlier time (day 12). The PPV of the day 13 threshold was 1.00 while that of the day 12 one was 0.87.
Discussion: We propose that patients with <6 × 10(6)/L CD34+ cells in peripheral blood on day 12 and <10 × 10(6)/L on day 13 following mobilisation with cyclophosphamide 4 g/m(2) and granulocyte colony-stimulating factor are candidates for "on demand" use of plerixafor, making the administration of this expensive agent more efficient and avoiding its inappropriate use.