Introduction

ASCT of PBSCs is the first option for patients with relapsed lymphoma or multiple myeloma (MM).1, 2, 3, 4, 5, 6, 7, 8 Successful engraftment correlates with the number of CD34+ hemopoietic progenitors cells infused;9, 10, 11, 12 however, a proportion of MM or lymphoma patients fail to mobilize PBSCs and cannot proceed to ASCT.12, 13, 14, 15, 16, 17, 18

The definition of ‘poor’ mobilization varies, owing to the different parameters set to evaluate the extent of mobilization: peak of CD34+ cells in PB, cumulative apheresis yield or simply percent of candidate patients undergoing ASCT. Various criteria have been proposed to define a successful CD34+ cell mobilization and the adequate apheresis yield, but these data are difficult to analyse and compare with each other.10, 12, 15, 19, 20 Therefore, a systematic review of the literature does not provide a clear definition of poor mobilization.

Standardization of criteria for diagnosis, prognosis and response is a major goal of the haematology community and may facilitate comparison among retrospective and prospective data. For these reasons, Gruppo italianoTrapianto di Midollo Osseo GITMO (Italian Group for Stem Cell Transplantation) convened a Working Group (WG) to clarify the definition of ‘poor mobilizer’ (PM). This is essential for the assessment of PBSCs’ mobilization in clinical trials and as a decision criterion for streamlining mobilization strategies in clinical practice. To develop the criteria for the definition of the PM, GITMO-WG adopted a flexible decision-making method: the analytic hierarchy process (AHP) was developed to establish priorities and make the best decision when both the quantitative and qualitative aspects of a decision needed to be considered and a poor information base was available.21 AHP is a multistep process, including four major phases: (1) defining the goal; (2) decomposing the problem and identifying critical issues; (3) categorizing/framing the main criteria; and (4) defining a hierarchy of the criteria. The participants’ subjective judgment allows them to overcome the scarcity or inconsistency of available information on the problem; finally, they are forced to quantify their subjective judgments by pairwise comparisons among the decided criteria. AHP has been successfully applied to develop the criteria of refractoriness, resistance or response in haematology.22, 23 The final definition of PM developed by the GITMO-WG is reported here, along with details with regard to the AHP method employed for defining criteria for poor PBSC mobilization.

Methods

A GITMO-WG of seven experts, with specific expertise in the field of PBSCs’ mobilization, harvest, CD34+ cell count and ASCT, was selected to develop a standard definition of patients with lymphoma or MM failing to mobilize adequate CD34+ cells to proceed to ASCT (referred to as PM).

English-language articles were extracted focusing clinical studies of PBSC mobilization in lymphoma or MM patients and particularly the studies aimed at defining the adequate dose of CD34+ cells for ASCT; AHP was used to identify the values and criteria that define poor mobilization. The participants first framed the conceptual criteria, and then the operational criteria: they are qualitative and quantitative, respectively. For instance, ‘old age’ is a conceptual criterion, whereas ‘older than 65 years’ is an operational criterion. Conceptual criteria were selected by the first questionnaire, if there was >80% agreement among the participants. Rank of the core set criteria was assigned through a number from 1 to 9 and the geometric mean of the ranks was calculated. Framing of operational criteria was obtained through the second questionnaire and by consensus through the Nominal Group Technique.24 The participants set the hierarchy of the operational criteria by pairwise comparison. Each participant analysed 55 couples of operational criteria and assigned a relative weight of one criterion with respect to the other one: if the former criterion was judged to have a higher importance than the latter, a weight from 1 to 9 was indicated; if the former criterion was less important than the latter, a weight from 1/9 to 1 was indicated. Inter-participant standardized geometric means of the weights for each criterion were calculated, and subsequently, inter-participant means were also calculated.

Finally, 36 scenarios were built by the combination of operational criteria: for each scenario, the participants were requested to check whether the definition of ‘predicted PM’ held. A representative scenario is: ‘The patient is a predicted PM if he or she is older than 65 years and shows extensive BM involvement at mobilization’. Each scenario received approval by each participant. The percent agreement of each scenario was compared with the sum of weights (obtained by pairwise comparison) of the criteria composing the scenario itself, and plotted.21

Results

The GITMO-WG agreed that establishing a clear definition of PM would help to optimize mobilization and transplant strategies (Figure 1); this might eventually reduce the need for: (1) repeated mobilization procedures; (2) SC harvest from BM; and (3) switch to allogeneic transplantation or other strategies.

Figure 1
figure 1

Goals and reasons of the GITMO-WG project.

Full size image

Decomposition of the problem

GITMO-WG deemed that the issue of ‘poor mobilization’ may pertain to three sequential phases:

  1. i)

    before the mobilizing treatment, when the mobilization procedure is planned;

  2. ii)

    during the mobilization procedure; and

  3. iii)

    at the completion of the whole PBSCs collection process.

The criteria supporting a prediction of ‘poor mobilization’ in the first phase may help one to identify patients at ‘high risk’ of ‘poor mobilization’ (predicted PM) and cannot be amended because they pertain to a patient's history, age or disease status. The criteria supporting a judgment of ‘poor mobilization’ during the second phase are markers of the biological capability of the patient to mobilize SC. They are dynamic and subject to change while mobilization is ongoing (for example, increase in the dose of G-CSF or addition of a new drug). The third phase involves both the host (history and mobilization ability) and the performance of the apheresis procedure (apheresis volumes, timing of collections, efficiency, and so on). Keeping in mind an operational definition of poor mobilization, GITMO-WG agreed that the criteria for defining a ‘proven PM’ pertain both to the second and third phase. However, even if an adequate CD34+ cell peak after mobilization predicts a successful collection, it cannot be considered a sufficient condition for a good harvest, as the latter is dependent on correct timing and performance of the apheresis procedure.

Framing of criteria

Through the first questionnaire, the GITMO-WG analysed 33 candidate conceptual criteria (Table 1); among them two (nos. 2 and 7) were selected for ‘proven PM’ and ‘proven poor mobilization’ (PPM) and 10 for ‘predicted PM’ (nos. 14–19 and 21–24); these latter were reduced to eight as ‘disease status’ was split into ‘advanced disease’ and ‘refractory disease’; ‘previous therapy with lenalidomide’ and ‘previous therapy with fludarabine’ were included in the same category, and the new criteria ‘previous exposure to other therapies potentially affecting SC mobilization’ include criteria numbers 18,19,22. The reasons for excluding the remaining criteria included characteristics of the mobilization strategy (nos. 10–12), redundancy, scarce clinical application or poor predictive value (that is, mononuclear cell count). As for the mobilization strategy, GITMO-WG deemed it necessary to include the optimal dose of G-CSF in the definition of PM: G-CSF dose ⩾10 μg/kg if used alone or ⩾5 μg/kg after mobilizing chemotherapy. Through a second questionnaire, GITMO-WG ranked the 10 conceptual criteria from the core set and chose among 2–4 operational definitions for each conceptual criterion (Table 2). Among predicting criteria, a high rank was assigned to previous extensive radiotherapy and exposure to drugs known to affect the mobilization capacity. The operational definitions of the two criteria for the proven PM (CD34+ cells peak in PB) and PPM (the final CD34+ cell harvest) were scored and discussed by Nominal Group Technique.24 As both the criteria concurred to the definition, harmonization of their operational wording was requested. Moreover, GITMO-WG established that the cutoff of ⩾2.0 × 106 CD34+ cells per kg harvested must be integrated with a pre-fixed number of apheresis procedures performed: ⩽3 apheresis days within a single mobilization attempt, after G-CSF alone or after chemo-mobilization. Finally, the peak of PB CD34+ cells was timed according to the mobilization strategy adopted, as a larger variability in PB CD34+ kinetics is expected after chemo-mobilization. Operational criteria for defining ‘predicted PM’ require some adjustments about the age cutoff and BM cellularity; there was major disagreement with regard to exposure to drugs or therapies potentially affecting SC mobilization, as a larger body of evidence supports a detrimental effect of selected alkylating drugs, that is, melphalan and fludarabine, whereas fewer reports were published on the detrimental effect of lenalidomide. Finally, GITMO-WG expressed the requirement to consider only full-course therapies damaging SC mobilization, as lower doses or shorter therapies are not ascertained detrimental factors.

Table 1 Candidate conceptual criteria evaluated in the first questionnaire; the core set criteria selected after the first questionnaire are represented in italic bold characters
Full size table
Table 2 Relative importance of the selected core set criteria, expressed both as conceptual and by operational definitions
Full size table

Hierarchy of the operational criteria

The GITMO-WG compared the 10 operational criteria by pairs and elaborated the relative importance weight of one criterion to another (Table 2). Pairwise comparison of the two criteria defining ‘proven PM’ and PPM (harvested CD34+ cells and peak CD34+ cells) showed two similar scores, thus confirming similar ranks. Indeed, GITMO-WG considered each of the two criteria itself sufficient to sustain a judgment of ‘predicted PM’, but recognized that a poor harvest may be caused by technical problems affecting the extraction efficiency and the final yield of circulating CD34+ cells. These problems may consist of delayed or anticipated timing of apheresis, small volume of processed blood and any troubles during the procedure that may prejudice the harvesting, even though the patient achieved a satisfactory peak of CD34+ cells in PB. Pairwise comparison was particularly important to rank the importance of the eight criteria for defining ‘predicted PM’. Advanced disease, refractory disease and previous extensive radiotherapy were the three criteria that decidedly had a higher importance. However, GITMO-WG observed that the criteria were biologically and clinically dependent on the others and, therefore, covariate.

A third questionnaire was finally completed including 36 scenarios combining the above eight criteria (Figure 2) and identified previous extensive radiotherapy as the most powerful independent criterion. The scenarios also identified previous exposure to therapies potentially affecting SC mobilization as synergic independent factors, whereas disease status itself was not sufficient to fulfil the definition of ‘predicted PM’. Therefore, the panel decided to join two conceptual criteria into a unique exhaustive one, which included therapies definitely proven to affect mobilization and all the other therapies that have been or will be proven to negatively affect SC mobilization. Finally, GITMO-WG decided to extend the definition of predicted ‘PM’ to those patients with a history of failure, not otherwise specified, and listed an additional specific criterion. On the basis of the above information, GITMO-WG separated the criteria for defining ‘predicted PM’ into two categories: major and minor. The former category included the three most powerful criteria, which are: previous failed mobilization, not otherwise specified, previous extensive radiotherapy and previous therapies detrimental to SC mobilization. The second category included: advanced phase disease, refractory disease, extensive BM involvement, BM cellularity <30% (before mobilization) and age >65 years. Although one major criterion was sufficient to qualify a patient as ‘predicted PM’, the presence of at least two minor criteria was requested to qualify a patient as ‘predicted PM’ (Table 3).

Figure 2
figure 2

Scoring of 36 scenarios by sum of pairwise weights of direct judgment after the third questionnaire evaluation; all the scenarios have been framed by combining different operational definitions, generated from the selected conceptual criteria. The best scored scenarios concurred to the final definition of proven and predicted poor mobilizer.

Full size image
Table 3 Final definitions of proven and predicted poor mobilizer
Full size table

Discussion

In this paper, GITMO-WG applied the AHP methods to select the operational criteria useful to identify the PM in current medical practice; at least 15% of lymphoma or MM patients fail to produce the target cell dose of ⩾2 × 106/kg CD34+ cells and cannot proceed to ASCT.10, 12, 15, 16, 89 Re-infusion of high doses of CD34+ cells is associated with fast platelet and neutrophil engraftment, leading to a significant cost sparing25, 26, 51, 90, 91, 92, 93, 94 and increased survival rates.52, 95, 96, 97, 98, 99, 100 GITMO-WG selected two conceptual criteria to identify proven PM and/or PPM: (I) peak of CD34+ cells in PB and (II) harvested CD34+ cells.

The WG agreed about the definition of PM also for those patients who did not achieve at least 2.0 × 106 CD34+ during 3 apheresis days. Previous reports on the remobilization in this setting show that only a minority of patients was able to achieve the 2.0 × 106 CD34+ threshold after a second mobilization attempt;12, 15 in particular, Pusic et al.16 reports that among 269 patients remobilized, only 62 (23%) yielded ⩾2.0 × 106 CD34+.

In the past, the quality of PB harvest has been evaluated both as CFU-GM and as CD34+ cell content, whereas limited data suggest a relationship between mononuclear cell count and CD34+ cell content in mobilized PB.27, 32 Despite persistence of controversies about the most reliable technique for enumerating the CD34+ cell content both in PB and in the harvest,101, 102, 103 GITMO-WG agreed that pre-apheresis CD34+ cell count in PB is the best predictor of CD34+ cells in the apheresis products12, 16, 30, 31, 33, 34, 35, 83, 84 and pragmatically considered a peak of CD34+ cells >20 μL in PB, as a reliable indicator of a satisfactory mobilization ability; on the other hand, the strong correlation between the pre-apheresis CD34+ cell count and the final harvest also indicates that a PPM could be considered a reliable marker of a proven PM.12, 16, 30, 31, 33, 34, 35, 83, 84

The GITMO-WG agreed on 2.0 × 106 CD34+ cells per kg as the minimum safe dose for ensuring rapid neutrophil and platelet recovery both in lymphoma and in MM patients;10, 11, 12, 16, 25, 26, 93, 94 others suggested a different dose, such as 2.5 (refs. 9,11,26,28,29,104) or 1.5 × 106 CD34+ cells per kg,105 but below 1 × 106/kg CD34+ cells, a high risk of delayed platelet recovery has been reported.28

GITMO-WG reviewed the current mobilization strategies as they can influence the SC mobilization, but cannot be considered the criteria for the definition of PM. When G-CSF alone is used, it is administered at doses ranging from 10 to 16 μg/kg daily;10 higher doses resulted in better harvest,106, 107, 108 not confirmed in subsequent experiences,109 as well the combination of G-CSF plus GM-CSF.110, 41, 42 After the demonstration of the synergistic effect of CY and GM-CSF,111 G-CSF followed by chemotherapy has been reported to mobilize similar number of CD34+ cells;42, 43, 44 therefore, as this cytokine has a better safety profile, at present the mobilization with CY and G-CSF at 5 μg/kg is widespread.44 A randomized study comparing different doses of G-CSF after chemotherapy did not show substantial benefit with higher doses.45 The use of disease-oriented chemo-mobilization in the context of a well-designed treatment98, 112, 113, 114 resulted in better harvest compared with G-CSF alone,10, 16 but this did not translate into a better clinical outcome.46, 47 As regards these two mobilization strategies, GITMO-WG deemed that the same criteria for identifying the PM should be employed, even if the different timing of CD34+ cell peaks and the different doses of G-CSF must be taken in account. GITMO-WG excluded from definition of PMs those patients who cannot reach a CD34+ cell peak ⩾20/μL, but who were equally able to achieve the target CD34+ cell dose by means of ⩽3 large-volume aphereses.37

As the identification of the ‘predicted PM’ GITMO-WG defined three major and five minor criteria, the most important criteria were as follows: previous cytotoxic chemotherapy and irradiation. The average decrease estimated for each cycle of chemotherapy is of 0.2 × 106 CD34+ cells per kg per apheresis and 1.8 × 106/kg CD34+ cells after large-field radiotherapy, whereas a local irradiation is not associated with impaired mobilization.36, 48, 68, 69, 71 Although underlying disease has been reported as a factor influencing mobilization,49, 53, 54, 55, 56 GITMO-WG deemed that this is not an independent factor, being influenced by the previous treatment and by the disease status. Advanced disease is often associated with extensive BM involvement. In addition, in lymphoma patients BM involvement and platelet count before mobilization are associated with mobilization failure,9, 10, 36, 57, 58, 59, 60, 61, 62 and although it was difficult to identify a clear age cutoff,63, 64 GITMO-WG deemed that older age is an important factor associated with poor mobilization.65, 50, 66

GITMO-WG agreed that the higher percentage of PM is characterized by either morphological or functional injury to BM, caused by extensive radiotherapy or/and chemotherapy, but also by the so-called stem cell poisons.36 This heterogeneous group of drugs includes purine analogues, melphalan,67, 72 and the list has recently included thalidomide115 and lenalidomide.73, 74, 75, 76 Fludarabine negatively affects mobilization in patients with CLL;77, 78, 79, 80 however, some authors reported satisfactory collections,81 whereas others recommend paying attention to the interval from last dose of fludarabine.82

Among the other factors associated with unsuccessful mobilization, including PB white cell count, platelet count, apheresis techniques, interval between last chemotherapy and mobilization attempt,9, 34, 38, 39, 40, 61, 69, 70, 85, 86, 87, 88, 116, 117 the platelet counts before mobilization seem a reliable indicator of adequate marrow function, but this factor has not been considered sufficiently powerful like the ones selected as conceptual criteria.

In conclusion, poor mobilization of SC is a major limitation to ASCT in lymphoma and MM, and the availability of new drugs, able to increase the SC mobilization, requires a stringent definition of the PM. Several patient- and disease-related factors have been retrospectively identified, but never prospectively validated. GITMO-WG recommend that patients previously failing at least one mobilization attempt should be candidates for new mobilizing strategies.19, 20, 118 Excluding this selected group, the preventive identification of PM, by using the criteria established here, should be validated in a prospective trial. In the meantime, the GITMO-WG recommend using standard criteria for identifying both the ‘proven and the predicted PM’ before planning the use of new mobilizing agents. Recently, two large series of patients with lymphoma or MM have been retrospectively evaluated define PM, according to the standardized criteria: extensive previous chemotherapy, previous melphalan exposure and a previous failed mobilization attempt have been identified as predictive factors of poor mobilization. Pusic et al.16 reports that 19% of patients yielding <2 × 106 CD34+ cells per kg during five apheresis after mobilization with G-CSF alone,16 whereas Wuchter et al.12 identified 15% of PMs by using the criterion of a peak concentration 20/μL of CD34+ cells, after chemotherapy followed by G-CSF.12 The International Myeloma Working Group also proposed some qualitative predictive risk factors for poor mobilization, without any ranking.119 The GITMO-WG worked to define simple, but stringent operational criteria for the identification/prediction of the PM in the setting of MM and lymphoma patients. The decision to separate the criteria for defining ‘predicted PM’ into two categories (major and minor) could be questionable, but it keeps the advantage to be easy to use and to update. Recently, new mobilizing agents, such as plerixafor, proved to be effective in PM.118, 120, 121 However, the scientific community has not still provided a standard definition of PM, who may potentially benefit from this drug; therefore, these criteria could help the selection of those patients who could benefit from new mobilizing drugs, waiting for a definitive validation by a prospective trial.