Methods

Patients

Data from 201 consecutive adult patients with nonmyeloid malignancies and available peripheral CD34+ cell concentrations on the day of first collection were included. All patients had undergone mobilization with chemotherapy and recombinant G-CSF at LDS Hospital or the University of Utah. Most patients had received G-CSF 10 (n=161) or 6 (n=33) g/kg/day subcutaneously beginning on day 5 (n=100) or 6 (n=79) following the initiation of chemotherapy. Complete blood counts were followed daily and, when the WBC had risen to greater than 800 cells/l, daily pCD34 was determined. PBSC collections were begun at the discretion of the attending physician, usually following a recovery of the peripheral white blood count (pWBC) to >5.0 10⁹ cells/l (median: 10 10⁹ cells/l) and a pCD34 >10 10⁶ cells/l (median: 60 10⁶ cells/l).

Prior chemotherapeutic agents

Data on prior chemotherapy were obtained by chart review. Chemotherapeutic agents received by the patients were divided into seven classes: platinum analogs, alkylating agents, topoisomerase interactive agents, antimicrotubule agents, antimetabolites, corticosteroids, and bleomycin. The individual drugs were usually administered in combination and a course of the combination therapy was considered a course for the individual agent. For the small number of instances of oral continuous administration, 1 month of therapy was considered a course.

CD34+ cell determination

CD34+ cells were enumerated in the peripheral blood by flow cytometry utilizing a Beckman Coulter Epics XL-MCL according to the recommendations of the International Society of Hematotherapy and Graft Engineering¹⁴ modified for three-color fluorescence.

Statistical analysis

Statistical analyses were performed with SPSS Version 10.1 or higher. Pearson's correlation (continuous variables) or one-way analyses of variance (ANOVA) (categorical variables) were utilized to determine the significant associations with pCD34. Pairwise comparisons were performed if ANOVA or ANCOVA were significant at the 0.05 level. Univariate regression models and a parsimonious multivariable regression model were created to investigate the effect of the chemotherapy classes on pCD34. Results were considered significant if P<0.05 (two-sided).

Results

The characteristics of the 201 patients are summarized in Table 1. The patients had received a mean of 6.13.3 premobilization chemotherapy courses utilizing various agents. In all, 40 patients were mobilized following standard-dose salvage chemotherapy used in the treatment of their specific primary tumors, while the remainder received various intermediate-dose chemotherapy regimens designed for mobilization. The patients were relatively uniform in the dosage and timing of G-CSF.

Table 1 - Characteristics of the 201 patients.

Full table

The specific chemotherapeutic agents received by the 201 patients prior to mobilization are summarized in Table 2. Cyclophosphamide and doxorubicin had been used most often. Some agents such as fludarabine and venorelbine were received too infrequently for a significant evaluation of their effects on pCD34.

Table 2 - Numbers of patients exposed to individual chemotherapeutic classes and agents.

Full table

The correlation of several possible predictive factors for pCD34 in our patient population is shown in Table 3. The mobilization chemotherapy regimen, G-CSF dose, and G-CSF schedule did not correlate with pCD34. A statistical trend was noted for prior radiation. However, only the number of prior chemotherapy courses was significant (P=0.001). Regression modeling suggested that each course of chemotherapy resulted in a decrease of 9.4 cells/l for the pCD34 on the first day of harvest.

Table 3 - Significance of patient characteristics in prediction of pCD34.

Full table

The relation between pCD34 and number of prior courses of drugs from the various chemotherapeutic agent classes is shown in Table 4. On univariate analysis, prior exposure to platinum compounds and alkylating agents had the strongest negative effect on pCD34. Only the number of platinum courses was more predictive than the total number of chemotherapy courses. Within drug classes, DNA crosslinking agents appeared more damaging than methylating agents and DNA-intercalating agents (doxorubicin) than nonintercalating topoisomerase inhibitors (etoposide). On a multivariate analysis, which included the total number of chemotherapy courses, only the number of prior courses of platinum compounds (P=0.001) and alkylating agents (P=0.01) were found to be independent negative predictive factors for pCD34. Regression analysis suggested that each course of platinum decreased the pCD34 on the first day of collection by 18 cells/l, and each alkylating agent course by 6.6 cells/l.

Table 4 - Univariate analysis of the correlation of pCD34 with exposure to different classes of chemotherapy agents.

Full table

The potential for recovery from the effects of the different drug classes was investigated by correlating the time from the last drug administration with the pCD34 on the first day of harvest (Table 5). Although some of the patients had relatively long intervals, we could find evidence of recovery for none of the drug classes.

Table 5 - Correlation between time from last chemotherapy class and pCD34.

Full table

Discussion

In this study, we have examined a number of factors potentially related to the ability to obtain higher levels of pCD34 at the initiation of harvest. Of several possible end points for mobilization efficiency, we chose pCD34 on the first day of harvest as the most direct measurement. For practical reasons, many other studies have utilized the harvest yield. We and others have shown a high correlation between these two end points.¹⁵ However, we have also suggested previously that the end point of harvest yield does add additional confounding variables not related to mobilization such as cell separator collection efficiency.¹⁶

We also attempted to design our study in a manner that would minimize the effect of mobilization regimen variables as confounding factors. As randomized trials have demonstrated the superiority of chemotherapy+growth factors over growth factors alone,¹⁷ we included only patients receiving both chemotherapy and G-CSF. Randomized trials have also suggested that the dose of G-CSF correlates with the mobilization efficiency when G-CSF is used alone¹⁸ or following chemotherapy,¹⁹ although the latter finding has not been consistent.²⁰

In our patient population, the G-CSF dose did not correlate with pCD34. There are several possible reasons why such an effect, if it exists, may have been more difficult to detect. The first is that our patient population is relatively homogeneous in terms of the G-CSF dose received. Thus, 80% of our patients received a dose of 10 g/kg/day and all but seven patients received doses of 6 or 10 g/kg/day. A second reason may be that the range of doses of G-CSF received by our patients is not sufficient to show a dose effect in a study of our sample size. If so, our results would be consistent with those of Andre et al,²⁰ who found no significant difference in CD34+ cell harvest yields between 131 patients randomized to receive 5 or 10 g/kg/day of G-CSF following mobilization chemotherapy. That doses greater than 10 g/kg/day may be superior to lesser doses, or that wider disparity in the doses administered compared to the 5 g/kg/day difference in the study arms in the trial of Andre et al, are needed to show an effect was suggested by the small study of Demirer et al,¹⁹ which found superior collections following 16 as compared to 8 g/kg/day following mobilization chemotherapy.

It has also been suggested that patient case mix may affect the sensitivity of detecting significant differences among groups receiving varying doses of G-CSF. Thus, using G-CSF alone, Weaver et al¹⁸ noted that the differential effect of an increased dose (30 vs 10 g/kg/day) on CD34+ cell collections seemed to be greater in patients who had received lesser doses of premobilization chemotherapy. However, this result was not seen in the study of Andre et al. These findings may suggest that differences between harvest results from good and poor risk patients are minimized with chemotherapy+G-CSF mobilization regimens and/or use of G-CSF doses <10 g/kg/day such as were received by our patient population and that of Andre et al.

In agreement with previous studies,^3,4,6,7,8 the amount of prior chemotherapy is a significant predictor for mobilization efficiency in our study. Prior radiation therapy appears to have less effect, a result that conflicts with some^3,4,9 but not other^11,12 studies. However, our ability to detect an influence was likely related to the use of radiation in only 24% of our patients, and only one course of radiation was received by 88% of these. While such limited radiation appears to have little impact, our results do not rule out a negative effect from more extensive treatments.

More data are needed with regard to the effects of individual classes of drugs received prior to stem cell mobilization. Drake et al²¹ proposed a scoring system based largely on murine studies that evaluated the repopulating ability of bone marrow cells that had been exposed to various chemotherapeutic agents. In general, Drake's system classified DNA alkylating agents as most, platinum analogs and topoisomerase inhibitors as intermediate, and antitubular agents as least damaging. Both Drake et al²¹ and Clark and Brammer²² found a correlation between the scoring system and CD34+ cell collections. Some problems of the Drake's system, for current use, include a lack of initial grounding in human patient data, exclusion of antimetabolites (especially when 5-FU, a commonly used drug, had shown adverse effects in the murine studies that formed the basis of their system), and a lack of inclusion of newer and frequently used agents such as the taxanes.

In order to address some of these issues, we performed a retrospective study correlating the exposure to seven chemotherapeutic drug classes in common use with mobilization efficiency. Our results revealed some similarities and some differences with those of Drake et al. In our patients, prior exposure to DNA crosslinking agents and platinum analogs appeared to carry the greatest risk of impairment, while antimetabolites seemed to have little, if any, effect. In addition, we could find no evidence of recovery over time for any of the drug classes. This result is consistent with some^3,4 but not other^1,7,8 prior work that has examined the relation between time since last chemotherapy and mobilization efficiency.

The mechanism(s) by which prior chemotherapy decreases the concentration of mobilized peripheral blood CD34+ cells is not entirely clear. Studies in rodents have suggested that alkylating agents and platinum analogs can produce a sustained reduction in the number of effective marrow stem cells as evidenced by decreased repopulating ability.²³ While this may result, at least in part, from a direct effect on stem cells, cyclophosphamide has also been shown to negatively affect the ability of the marrow stroma to support hematopoietic progenitors.^24,25

We conclude that prior chemotherapy is the most important potentially controllable variable for stem cell mobilization. Minimization of exposure to chemotheraputic agents, especially platinum analogs and DNA crosslinking agents, should increase CD34+ cell harvest yields.

References

1. Kotasek D, Shepherd KM, Sage RE et al. Factors affecting blood stem cell collections following high-dose cyclophosphamide mobilization in lymphoma, myeloma and solid tumors. Bone Marrow Transplant 1992; 9: 11–17. | PubMed | ISI | ChemPort |
2. Chabannon C, Le Coroller AG, Faucher C et al. Patient condition affects the collection of peripheral blood progenitors after priming with recombinant granulocyte colony-stimulating factor. J Hematother 1995; 4: 171–179. | PubMed | ChemPort |
3. Koumakis G, Vassilomanolakis M, Hatzichristou H et al. Predictive factors affecting mobilization and peripheral blood stem cell (PBSC) collection using single apheresis (SA) for rescuing patients after high-dose chemotherapy (HD, CHE) in various malignancies. Bone Marrow Transplant 1996; 18: 1065–1072. | PubMed | ISI | ChemPort |
4. Haas R, Möhle R, Frühauf S et al. Patient characteristics associated with successful mobilizing and autografting of peripheral blood progenitor cells in malignant lymphoma. Blood 1994; 83: 3787–3794. | PubMed | ISI | ChemPort |
5. Keung YK, Cobos E, Dunn D et al. Determining factors for the outcome of peripheral blood progenitor cells harvests. J Clin Apheresis 1996; 11: 23–26. | Article | PubMed | ISI | ChemPort |
6. Humpe A, Riggert J, Vehmeyer K et al. Comparison of CD34+ cell numbers and colony growth before and after cryopreservation of peripheral blood progenitor and stem cell harvests: influence of prior chemotherapy. Transfusion 1997; 37: 1050–1057. | Article | PubMed | ISI | ChemPort |
7. Olivieri A, Offidani M, Ciniero L et al. Optimization of the yield of PBSC for autotransplantation mobilized by high-dose chemotherapy and G-CSF: proposal for a mathematical model. Bone Marrow Transplant 1994; 14: 273–278. | PubMed | ISI | ChemPort |
8. Shepherd KM, Charles P, Sage RE et al. Mobilization of haemopoietic stem cells by cyclophosphamide into the peripheral blood of patients with haematological malignancies. Clin Lab Haematol 1991; 13: 25–32. | PubMed | ISI | ChemPort |
9. Bensinger W, Appelbaum F, Rowley S et al. Factors that influence collection and engraftment of autologous peripheral-blood stem cells. J Clin Oncol 1995; 13: 2547–2555. | PubMed | ISI | ChemPort |
10. Imrie K, Stewart K, Crump RM et al. Blood stem cell collection: factors influencing the recovery of granulocyte–macrophage colony forming cells. Cell Transplant 1996; 5: 379–383. | Article | PubMed | ISI | ChemPort |
11. McQuaker IG, Haynes AP, Stainer C et al. Stem cell mobilization in resistant or relapsed lymphoma. Superior yield of progenitor cells following a salvage regimen comprising ifosphamide, etoposide and epirubicin compared to intermediate-dose cyclophosphamide. Br J Haematol 1997; 98: 228–233. | Article | PubMed | ISI | ChemPort |
12. Goldschmidt H, Hegenbart U, Wallmeier M et al. Factors influencing collection of peripheral blood progenitor cells following high-dose cyclophosphamide and granulocyte colony-stimulating factor in patients with multiple myeloma. Br J Haematol 1997; 98: 736–744. | Article | PubMed | ISI | ChemPort |
13. Morton J, Morton A, Bird R et al. Predictors for optimal mobilization and subsequent engraftment of peripheral blood progenitor cells following intermediate dose cyclophosphamide and G-CSF. Leukemia Res 1997; 21: 21–27. | Article | ISI | ChemPort |
14. Sutherland R, Anderson L, Keeney M et al. The ISHAGE guidelines for CD34+ cell determination by flow cytometry. J Hematother 1996; 5: 213–226. | PubMed | ChemPort |
15. Ford CD, Chan K, Reilly WF, Petersen FB. An evaluation of predictive factors for CD34+ cell harvest yields from patients mobilized with chemotherapy and growth factors. Transfusion 2003; 43: 622–625. | Article | PubMed | ISI | ChemPort |
16. Ford CD, Greenwood J, Strupp A, Lehman CM. Change in CD34+ cell concentration during peripheral blood stem cell collection: effects on collection efficiency and efficacy. Transfusion 2002; 42: 904–911. | Article | PubMed | ISI |
17. Narayanasami U, Kanteti R, Morelli J et al. Randomized trial of filgrastim versus chemotherapy and filgrastim mobilization of hematopoietic progenitor cells for rescue in autologous transplantation. Blood 2001; 98: 2059–2064. | Article | PubMed | ISI | ChemPort |
18. Weaver CH, Birch R, Greco L et al. Mobilization and harvesting of peripheral blood stem cells: randomized evaluations of different doses of filgrastim. Br J Haematol 1998; 100: 338–347. | Article | PubMed | ISI | ChemPort |
19. Demirer T, Ayli M, Ozcan M et al. Mobilization of peripheral blood stem cells with chemotherapy and recombinant human granulocyte colony-stimulating factor (rhG-CSF): a randomized evaluation of different doses of rhG-CSF. Br J Haematol 2002; 116: 468–474. | Article | PubMed | ISI | ChemPort |
20. Andre M, Baudoux E, Bron D et al. Phase III randomized study comparing 5 or 10 g per kg per day of filgrastim for mobilization of peripheral blood progenitor cells with chemotherapy, followed by intensification and autologous transplantation in patients with nonmyeloid malignancies. Transfusion 2003; 43: 50–57. | Article | PubMed | ISI | ChemPort |
21. Drake M, Ranaghan L, Morris TCM et al. Analysis of the effect of prior therapy on progenitor cell yield: use of a chemotherapy scoring system. Br J Haematol 1997; 98: 745–749. | Article | PubMed | ISI | ChemPort |
22. Clark RE, Brammer CG. Previous treatment predicts the efficiency of blood progenitor cell mobilization: validation of a chemotherapy scoring system. Bone Marrow Transplant 1998; 22: 859–863. | Article | PubMed | ISI | ChemPort |
23. Neben S, Hemman S, Montgomery M et al. Hematopoietic stem cell deficit of transplanted bone marow previously exposed to cytotoxic agents. Exp Hematol 1993; 21: 156–162. | PubMed | ISI | ChemPort |
24. Molineux G, Xu C, Hendry J, Testa NG. A cellular analysis of long-term haematopoietic damage in mice after repeated treatment with cyclophosphamide. Cancer Chemother Pharmacol 1986; 18: 11–16. | Article | PubMed | ISI | ChemPort |
25. Adler SS, Kuznetshy RD. Effects of chemotherapeutic agents on hemopoietic stromal tissue: I. Effects of cyclophosphamide and bleomycin on hemopoiesis in subcutaneous femur implants. Exp Hematol 1984; 12: 153–161. | PubMed | ISI | ChemPort |