© 1997 AlphaMed Press
"Less is More": The Role of Purging in Hematopoietic Stem Cell TransplantationDepartment of Haematology, Royal Free Hospital School of Medicine, London, United Kingdom Correspondence: Mark W. Lowdell, Ph.D., Department of Haematology; Royal Free Hospital School of Medicine, London, United Kingdom. Telephone: 44-171-830-2183; Fax: 44-171-794-0645; e-mail: heg{at}rfhsm.ac.uk
The epithet "less is more" is usually applied to the essentials of good design, but it might be equally true of autologous blood or marrow transplantation. Ever since autologous marrow transplantation was first used to reconstitute recipients of high-dose chemotherapy or radiotherapy, there has been much discussion about the relative contribution of residual tumor cells in the graft to the occurrence of subsequent relapse. It was not until the early 1990s that this risk was finally confirmed by the use of gene marking [1]. A retroviral vector was used to mark a proportion of the autologous remission bone marrow from patients with acute myeloid leukemia (AML) before marrow infusion after high-dose therapy. Two recipients relapsed and both had leukemic blasts with the marker, the neomycin-resistance gene. As the safety of autologous hematopoietic stem cell transplantation has increased, the use of high-dose therapy followed by stem cell "rescue" is becoming more widespread. The malignancies treated in this way include leukemia, lymphoma, myeloma, neuroblastoma, breast cancer, and ovarian tumors. In each of these conditions a number of important questions should be addressed: Can we identify and quantitate tumor cells in the grafts and establish their oncogenic potential? If so, how best can we remove them? Can they be removed without compromising the graft, and will such purging produce a clinically significant reduction in relapse risk? Finally, will the procedure be cost-effective?
Contaminating tumor cells are inadvertently harvested along with hematopoietic stem cells and subsequently infused into patients undergoing high-dose therapy and transplantation. Such cells may be responsible for a proportion of relapses, but the extent of this clinical problem is still unknown and is likely to be determined by the frequency or number and biological properties of tumor cells in different types of hematopoietic harvests (marrow versus non-mobilized blood versus mobilized blood).
Several methods for detection of low-level tumor contamination have been reported. Generally, hematopoietic and immune cell tumors are detected using molecular techniques, and solid tumor micrometastases are detected by immunocytochemical assays (Table 1
Peripheral blood stem cell collections and bone marrow from patients with breast cancer have been evaluated for the presence of tumor cells using sensitive immunocytochemical techniques. It is believed that peripheral blood progenitor cell collections are less likely than bone marrow harvests to be contaminated with tumor cells [7]. Tumor cells and progenitor cells both are mobilized after chemotherapy and administration of hematopoietic growth factors (granulocyte colony-stimulating factor) [8]. Several investigations have demonstrated tumor cell contamination of 10% to 30% of leukapheresis products used for transplants [7], and, in patients with neuroblastoma, stem cell harvests have been shown to contain contaminating tumor cells irrespective of marrow involvement [9]. In other studies in patients with breast cancer or lymphoma, tumor-contaminated peripheral blood stem cell collections resulted in a poor clinical outcome [10, 11]. The sensitivity of immunocytochemical methods for epithelial tumor cell detection is limited by the number of cells examined [12]. The 95% confidence limit of sensitivity is approximately one tumor cell in 105 nucleated cells for 3 x 105 cells evaluated. The time-consuming aspects of this assay might be solved by automated image analysis. The application of polymerase chain reaction (PCR) in breast cancer tumor detection is limited because of the lack of relevant molecular markers. Only recently has PCR been applied to the detection of occult tumor micrometastases in solid tumor malignancies [13], but the sensitivity and specificity in large studies involving patients with solid malignancies have not yet been evaluated. Reverse-transcription-PCR produced a significant number of false-positive results in detection of hematogenous lung cancer dissemination [14]. Although the false-positive rate can be decreased by decreasing the number of PCR cycles and by using fewer cells, this limits the sensitivity of the method. The likelihood of detecting legitimate epithelial tumor cells can be increased by including other cytokeratins or other genes, increasing the workload and cost, as well as the risk of inappropriate results because of tumor cell heterogeneity. This approach may not become routinely feasible until technical development of automated gene detection equipment, similar to that being applied in mutation analysis, becomes available.
It is likely that technological developments in automated detection of genes, image analysis, and hematopoietic stem cell harvest manipulations will have a major impact on this field. There may be a lack of coincidence between molecular detection of tumor cells and their clonal growth in culture, which may have implications for their clinical significance. These discrepancies may be associated with specific gene defects that confer aggressiveness and treatment resistance on the tumor cells. Molecular characterization would facilitate the further biological properties of contaminating tumor cells and may provide a powerful approach that can be used to detect over-amplification of various prognostic markers, as well as new insights into the course of an individual tumor. This may be as important as a determinant of time to treatment failure than the number of tumor cells inadvertently infused. This is an area of active research and we have recently developed methods for simultaneous immunocytochemistry and interphase FISH for the detection of chromosomal abnormalities and/or oncogene over-amplifications in cytokeratin-positive epithelial tumor cells in hematopoietic stem cell products (Fig. 1
The treatment of residual disease in patients after chemotherapy or transplantation is often referred to as tumor purging, but here we will concentrate on the strategies used to remove contaminating tumor cells from hematopoietic stem cell grafts. In chemotherapy-responsive tumors, research has focused on designing additional treatments to purge autologous stem cell products. Such approaches have met with variable success, a predictable outcome given the highly heterogeneous nature of the malignancies studied. The intensity of chemotherapeutic purging strategies is limited in all cases by the toxicity of the drug on the hematopoietic stem cell in the graft. Analysis of registry data from the European Blood and Marrow Transplant group (EBMT) of 270 recipients of purged versus 224 recipients of non-purged grafts [15] showed no detrimental effect on engraftment. This analysis included a broad cross-section of chemotherapeutic regimens and suggested that hematopoietic stem cells are relatively resistant to these agents. The study did not show any benefit of tumor purging with respect to progression-free or overall survival. This may have been due to residual chemoresistant tumor cells within the graft, within the patient, or both. More intensive purging chemotherapy of hematopoietic stem cell grafts may be beneficial in some cases, but eventually this will reach the threshold of toxicity for the stem cell. Efforts have been made to increase this threshold by introducing the multidrug resistance gene into human hematopoietic stem cells. Immunological approaches to ex vivo purging have been used for a number of years and have involved cytokines [16], cellular strategies [17], and specific monoclonal antibodies [1820]. These approaches are effective in leukemia resistant to chemotherapy [21]. The first clinical use of monoclonal antibodies was for the ex vivo purging of T cells from allogeneic bone marrow grafts [22] and was followed by the use of monoclonal antibody and complement to deplete residual ALL cells from autologous bone marrow in four patients [23]. The treatment was safe and allowed normal engraftment in all patients. Leukemias and lymphomas have proved good targets for ex vivo purging, since lineage and differentiation stage-specific antigens are known and high-dose therapy followed by bone marrow transplantation is an established treatment. Toxin-conjugated monoclonal antibodies, or non-conjugated monoclonal antibodies with additional complement, have been used in AML [24], ALL [18, 25], lymphoma [26], and myeloma [27, 28]. These monoclonal antibodies are targeted at differentiation antigens or, in the case of some B-cell malignancies, against surface immunoglobulin molecules, both of which are absent on normal hematopoietic stem cells [29, 30]. As autologous hematopoietic stem cell "rescue" after high-dose therapy becomes more commonly used in the management of solid tumors, there is a need for suitable purging strategies. Immunotoxin-conjugated monoclonal antibodies have been shown to selectively kill breast cancer cells in autologous graft material [31], although this awaits clinical application. Apart from targeting toxin-mediated or complement-mediated lysis of tumor cells within grafts, the flexibility of monoclonal antibody technology can also enable physical separation of the normal and residual tumor cells. Monoclonal antibodies can be bound to solid phases such as plastic tissue-culture flasks or magnetic/paramagnetic beads. Positive selection describes retention and use of the selected cells of interest, also called "enrichment." Negative selection describes selection and disposal of the sorted cells, also called "depletion." A major advance in the field of tumor purging was made with the demonstration that hematopoietic engraftment potential resides within the CD34+ cell subset [32]. Thus, instead of tumor-specific depletion strategies, in the majority of cases it has become possible to positively select normal cells. Since tumor cells rarely express CD34, the end product should, at least in theory, be free from contaminating tumor cells.
In vitro sterile cell sorting has been a common research tool for decades, but the difficulties in scaling-up to clinical ex vivo stem cell selection cannot be over-emphasized. Clinical-grade monoclonal antibodies and closed system devices are required and evaluation protocols have had to be devised. Currently, five methodologies have completed at least phase I clinical trials, and all are based upon one or more anti-CD34 monoclonal antibodies (Table 2
In the clinical setting, the choice of device for CD34 selection for indirect tumor purging is a matter of personal opinion and circumstance, since comparative trials have shown them to be broadly comparable in terms of yield and purity. Direct comparison between trials is difficult, as the enumeration of CD34+ cells within hematopoietic stem cell products is not standardized and there is extreme interassay variability [44]. Despite limitations, many useful studies have been and are being performed, and it is likely that the only true test of purging efficacy can come from studies of clinical outcome. It is improbable that we can ever produce an entirely tumor-free autologous graft from patients with malignant diseases of the bone marrow or with tumors that metastasize to the marrow. In addition to positive selection strategies, immunomagnetic separation methods have been used to deplete tumor cells from grafts [4548], and interest in this approach is increasing as tumor-specific determinants are identified in solid tumors.
Many groups believe that tumor purging can be improved with multistep strategies (Fig. 2
Double-sorting strategies are time consuming and expensive and will always reduce the yield of stem cells. An alternative approach is CD34+ cell selection with subsequent ex vivo expansion. This has a number of attractions, especially that a small harvest volume produces a consequent reduction in the absolute tumor burden. The culture conditions in which hematopoietic stem and progenitor cells can be expanded in vitro do not support the growth of epithelial tumor cells, thus producing specific purging. This may not be the case, however, with many hematological tumors. It has been shown that residual myeloma cells can survive ex vivo expansion [50], although more encouraging results have been seen in lymphoma [51]. Initial concerns about the potential loss of repopulating pluripotent hematopoietic stem cells in ex vivo culture and the risk of subsequent graft failure are unfounded as the high-dose therapy given to these patients is not truly myeloablative and such sorted and expanded cells have safely reconstituted patients with breast cancer [52]. Large randomized clinical trials will be needed to determine the benefit of using such grafts.
There is no doubt that our abilities to both accurately enumerate tumor cells in autologous hematopoietic stem cell products and to purge them to below detectable levels are improving and will continue to do so. However, the value of tumor purging is still debated. While one study [1] showed that tumor cells in relapsed patients can be shown to be derived from the infused graft, it did not demonstrate that this was the only source of tumor cells and could not prove that efficient purging would have prevented relapse in either of the two cases. It is likely that the benefit of purging will be dependent upon the tumor type and stage of disease, as well as the degree of purging possible in the individual case. To date, most trials in which progression-free and overall survival have been evaluable have been in hematological malignancies. A six-year single center study showed no benefit of purging in patients with non-Hodgkins lymphoma [53], nor did a large study from EBMT [15]. Registry data from the American Bone Marrow Transplant Registry (ABMTR) [54] showed an outcome which was worse in patients with AML who received syngeneic grafts from identical siblings than in those who received autologous non-purged grafts. Although one must be cautious in comparing recipients of autologous transplants with those receiving allogeneic grafts, these transplants were from genetically identical siblings and might be considered analogous to the ultimately "purged" autologous graft. Given these results, we might conclude that there is no role for purging in autologous transplantation, but this ignores the heterogeneity of different tumors and the fact that techniques for treatment of in vivo residual disease are improving. Irrespective of the contribution of residual tumor cells in vivo to disease relapse, the infusion of tumor cells into an immunocompromised host is undesirable and purging strategies should be optimized. Whether such approaches are cost-effective will depend upon their concomitant morbidity, together with effects on progression-free and overall survival. The technology of stem cell graft engineering is expanding rapidly; the challenge is to design and conduct clinical trials to properly evaluate these new techniques.
M.W.L. is supported by the Foundation for Children with Leukemia and P.T. is supported by a grant from Amgen (U.K.) Ltd.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||