Principles of Bone Marrow and Stem Cell Transplantation

Total-Body Irradiation

Several parameters associated with TBI can be adjusted to deliver different amounts of radiation and different degrees of end organ effect. Total dose, dose rate, fraction size, interfraction interval, and shielding are among the TBI parameters most manipulated in SCT. An ideal schedule would maximize malignant cell kill, hematopoietic ablation, and, in the case of allogeneic SCT, immunosuppression while limiting acute and chronic toxicity. In principle, higher total dose, higher dose rate, and larger fraction size are associated with greater hematopoietic ablative effect (and therefore potentially greater antileukemic efficacy) and with better immunosuppression. Fractionation, the division of the total dose of radiation over time usually in a twice-daily schedule, theoretically provides improved tolerance for nonhematopoietic tissues, leading to decreased acute toxicity and reduced late effects, but carries the risk for decreased antileukemic efficacy. However, a series of studies from the 1980s supports the use of fractionated radiation, and this remains the standard today. Studies from Seattle have demonstrated increased antileukemic efficacy with increased total dose, but lung, liver, and kidney tolerances were limiting. Parameters of total dose, fractionation, dose rate, and schedule have been largely extrapolated to pediatric HSCT from adults.

Busulfan

Developed in the mid-1970s, busulfan (BU)/cyclophosphamide (CY) rapidly became an established regimen for patients with acute myelogenous leukemia (AML) receiving either allogeneic or autologous SCT. The original regimen, called BU/CY4 (“big BU/CY,” BU 4 mg/kg/day × 4 days followed by CY 50 mg/kg/day × 4 days) was joined by BU/CY2 (“little BU/CY,” in which the CY was delivered as 60 mg/kg/day × 2 days), and this regimen has been used in patients with acute lymphocytic leukemia (ALL) or chronic myelogenous leukemia (CML). The BU/CY regimen has been used extensively in pediatric patients, particularly those with nonmalignant diseases, in an attempt to avoid radiation-associated toxicity. The current practice of using an intravenous formulation of BU and dose adjustment to achieve a target level has decreased toxicity; effects on efficacy are more difficult to demonstrate. Treosulfan, a related agent already approved in Europe, has limited nonhematologic toxicity, and pediatric trials are ongoing in national trials.

TBI versus BU

Several randomized studies and meta-analyses have compared BU/CY with TBI-based regimens. Overall, results are similar although suggestive that both relapse rates and hepatic veno-occlusive disease (VOD) may be less in the TBI-containing regimens. Although recent data support these observations, variability of the chemotherapy component of TBI-based regimens and potential differences in the antileukemic efficacy of BU/CY4 and BU/CY2 regimens render definitive generalizations difficult. Reports of decreased bioavailability, increased volume of distribution, and increased clearance rate of BU in children raise the question of how to interpret studies in which such data are unavailable. Alternative BU dosing, based on plasma levels, has been suggested, and comparison of “optimal” BU/CY with CY/TBI has not been reported. BU-containing regimens have often been reported to result in greater toxicity when compared with TBI-based regimens; thus increases in BU dose may further increase regimen-related morbidity. Monitoring of BU pharmacokinetics has generally, but not always, demonstrated an association of increased area under the curve with increased volume of distribution. Individualizing BU dosing may possibly increase the risk-benefit ratio, and the recent availability of an intravenous form of BU with more dependable pharmacokinetics now makes this achievable.

Alternative Myeloablative Regimens

The addition or substitution of a variety of chemotherapeutic agents to the backbone of CY/TBI, and less commonly BU/CY, has met with mixed success. Etoposide, cytarabine, and BU have been added to or substituted for CY in TBI-based allogeneic regimens. Etoposide (VP-16) and melphalan have also replaced CY in some BU-based protocols. Clofarabine, a purine nucleoside analogue, is a relatively new drug known to be very effective in relapsed/refractory pediatric ALL; ongoing studies have shown promise in using it in combination with BU to decrease relapse in advanced cases of pediatric leukemia. Fludarabine has become an increasingly used drug in allogeneic transplant preparative regimens. It has potent immunosuppressive properties and minimally extramedullary toxicity. Combined with BU (BU/FLU) it provides adequate ablation for full donor engraftment, and in the pediatric population it appears to have similar efficacy but less transplant-related toxicity and faster neutrophil engraftment.

Fluorouracil, high-dose methotrexate, doxorubicin, cisplatin, carboplatin, thiotepa, melphalan, and many other agents have been used in a variety of high-dose chemotherapeutic regimens supported by autologous SCT. Common conditioning regimens referred to by acronyms are CBV (CY, carmustine, VP-16); BACT (BCNU [carmustine], cytarabine [Ara-C], CY, 6-thioguanine); BEAM (BCNU, etoposide, cytarabine [Ara-C], melphalan); and CBP (CY, BCNU, cisplatin). In the allogeneic setting, alternative regimens have generally shown equivalent benefit in terms of disease control but often have increased toxicity. No regimen has established unequivocal superiority in terms of disease-free or overall survival when compared with “standard regimens.” Given the more limited follow-up and the extraordinary diversity of diseases and regimens, establishment of relative efficacy in the realm of autologous SCT is likely to be a prolonged process.

Nonmyeloablative Regimens

One of the target tissues of allogeneic T lymphocytes is the hematopoietic precursors. As a result, the infusion of donor lymphocytes can be sufficient to cause marrow aplasia. Indeed this is the basis of the marrow aplasia seen in inadvertent cases of transfusional GVHD. Observations made initially in the 1980s by the Seattle group have led to a new approach to conditioning based on the immunologic elimination of normal and malignant hematopoiesis. Because this technique minimizes the doses of conventional conditioning drugs and radiation, it has been called “nonmyeloablative” or reduced-intensity conditioning. Although allogeneic graft-versus–bone marrow can cause the necessary marrow aplasia and antileukemic efficacy, it does not supply the immunosuppression needed to avoid graft rejection. This is usually achieved by preinfusion conditioning with purine-analogue chemotherapy and/or low-dose TBI. A variety of agents including lower-dose BU, melphalan, and cytarabine have also been used for additional effect and partial HSC depletion. A major advantage of this approach is that the toxicity of conventional myeloablative approaches can be reduced. Comparisons of conventional versus nonmyeloablative regimens have shown reduced nonrelapse mortality, especially in older adults or patients with preexisting illnesses for whom HSCT is associated with significant toxic death rates. The safety and effectiveness of this approach has led to its wider application in recent years. However, the low toxicity of nonmyeloablative regimens is balanced against the increased risk for relapse in acute leukemias and the increased risk for chronic GVHD. The indications for nonmyeloablative transplant may therefore remain principally for slower-growing tumors. Because childhood hematologic malignancies rarely fall into this category, the use of nonmyeloablative transplant in children has been more limited. Moreover, the increase rates of graft rejection after standard HSCT for nonmalignant conditions such as thalassemia and sickle cell anemia have precluded the broad use of nonmyeloablative transplants in these conditions.

Typing

Genetic disparity between host and donor can lead to graft rejection or GVHD. The genes encoding the antigens most associated with these complications lie in a group of genes known as the major histocompatibility complex (MHC). In humans, the MHC maps to a region of the short arm of chromosome 6, known as the human leukocyte antigen (HLA) system. The MHC codes for many genes are still being identified, and the exploration of the functional importance of some of these gene products is still ongoing. Among the best characterized gene products are proteins comprising the class I and class II antigens. In humans, MHC class I molecules include HLA-A, HLA-B, and HLA-C and MHC class II molecules are called HLA-D. There are at least five subregions, including DR, DQ, and DP, but the number of genes transcribed differs between different HLA class II haplotypes.

The MHC class I antigens are present on nearly all human cells (with rare exceptions such as the erythrocyte and corneal endothelium). MHC class II antigen expression is more restricted, with expression limited to “professional” antigen-presenting cells, including dendritic cells, B cells, monocytes, macrophages, and Langerhans cells. Class II antigens can also be induced on activated T cells and endothelium. MHC genes are codominantly expressed, that is, each parent contributes half of the MHC antigens expressed on each cell of their offspring. MHC class I and II antigens normally function to present partially degraded products of intracellular proteins to T cells. In this way, damaged or virally infected cells can be identified by the immune system. However, in the setting of transplantation, the recognition of MHC molecules by T cells is known as alloimmunity and is the basis for GVHD and graft rejection.

The genes of the MHC region are highly polymorphic, and less than 30% of potential recipients of HSCs have HLA-identical siblings. Moreover, even HLA-identical sibling transplants can be associated with GVHD because of differences in the sequence of the intracellular proteins associated with MHC molecules caused by polymorphisms in non-MHC genes. These polymorphisms are mostly unidentified, and those that are known to elicit alloreactive T-cell responses are termed minor histocompatibility antigens. Although these antigens may not be directly responsible for organ and bone marrow rejection, human typing studies suggest that these minor antigens contribute to the generation of GVHD.

Historically, HLA typing was performed by using antibodies derived from postpartum sera, from persons who have undergone transplants or transfusions, or even from persons immunized for the purpose of generating specific serologic reagents. These antibodies were used in a standardized complement-dependent, microcytotoxicity assay with purified T or B lymphocytes as their target cell. The extensive cross-reactivity among products coded by the alleles of the HLA-A and HLA-B loci, known as cross-reactive groups, made precise delineation of the MHC class I antigens by this technique difficult. In contrast, HLA-D specificities were originally defined by their ability to stimulate T-lymphocyte proliferation in a mixed leukocyte culture.

Because of the difficulty in precise delineation of HLA-type by serologic methods, modern HLA typing now uses high-resolution techniques that resolve the DNA sequence of specific alleles. The first attempts to use molecular genetic techniques were based on application of restriction fragment length polymorphism analysis of genomic DNA by Southern blotting using HLA region probes. This soon gave way to current technology in which sequence-specific oligonucleotide probes and, in some cases, direct sequence analysis are performed. This has also led to a great increase in the number of identified HLA specificities. This fine analysis is increasingly being used to determine the haplotypes of patients and potential donors. The implications of more exact typing within families, where the haplotypes are “conserved,” are probably confined to the potential identification of minor histocompatibility antigen differences, allowing for better selection between possible donors. In the UD setting, high-resolution HLA matching has significantly contributed to the improved outcomes reported in UD HSCT in the past decade. Historically, an allogeneic donor must match the patient at five of the six major loci (e.g., A, B, DR) for a transplant to be feasible. Stem cells obtained from UCB are more permissive, and four of six matches are acceptable.

Sources of HSC

In the past, most HSCT has used bone marrow mononuclear cells as a source of HSCs. Bone marrow is usually obtained by repeated aspiration of the posterior iliac crests while the donor is under general anesthesia. Multiple aspirations are performed to a total volume of 10 to 15 mL/kg recipient body weight. The marrow is filtered through a fine wire mesh to rid the product of aggregates and bony spicules. This is a surprisingly well-tolerated procedure and serious complications are rare, probably because the donor is generally a healthy individual undergoing an elective procedure. Potential short- and long-term complications include risks of anesthesia, blood loss and potential transfusion, pain, neurologic deficits, and psychosocial complications. In the most truly altruistic setting, UD marrow harvest, only 5% to 8% of donors later expressed any ambivalence about their participation. This was most marked if the transplant was subsequently unsuccessful. Once harvested, the red cells must be removed from the bone marrow if there is ABO incompatibility before infusion into the recipient. The marrow may be manipulated ex vivo to remove donor T lymphocytes (see later) or tumor cells. It is estimated that only 1% of the donor pluripotent HSC population is removed in a typical marrow harvest.

HSCs continuously recirculate from the marrow into the peripheral blood; therefore HSCs can also be harvested by collection of peripheral blood mononuclear cells, usually by leukapheresis. The frequency of cells that express the surface marker CD34 is widely used as a proxy for actual HSC frequency and allows the enumeration of stem cells. HSCs in the peripheral blood are rare, and granulocyte colony-stimulating factor (G-CSF) is usually administered to increased the detachment of HSCs from bone marrow niches and promote their egress into the peripheral blood, increasing their frequency. Moreover, studies in adults suggest that additional strategies to further mobilize marrow HSCs are possible. AMD3100 is a reversible inhibitor of the CXC chemokine receptor (CXCR4), a receptor on the surface of HSCs that promotes adhesion to the bone marrow milieu. Combining AMD3100 with G-CSF increases HSC yield in adults, but this approach has not been investigated in children. Complications of PBSC harvest have included bone pain, as a side effect of G-CSF, and inability to collect via peripheral access, necessitating central line placement. There is an increased risk for splenic rupture resulting from G-CSF administration in healthy individuals, although this is an extremely rare complication. The implications of mobilizing and harvesting PBSCs from pediatric donors have not been fully explored, although cytopenias and hemostatic changes of unknown short- and long-term significance have been observed in healthy donors.

Since the first description of its use in a child with Fanconi anemia in 1988, UCB has become an increasingly used HSC source. Placental blood is recovered from the umbilical vein by drainage or catheterization, and techniques to optimize UCB HSC collection are being investigated. Cord blood is immediately frozen and stored until needed. If indicated, in the related setting, determination of the HLA type of the fetus can be performed before delivery. Otherwise, typing and analysis is performed on intrapartum or postpartum samples.

Selection of a UD, whether bone marrow or UCB, is facilitated by the availability of epidemiologic information (e.g., age, sex) as well as HLA typing stored in a database accessible via the Internet. By comparing a patient's HLA type with donors listed and categorized in the registries, one can estimate the likelihood of a successful search, devise search strategies for patients with uncommon HLA types, and sort among potential HSC sources. Although stored UCB can be readily supplied after confirmatory typing, the completion of a UD bone marrow donor search is more involved. Potential donors must be contacted and consent to proceed, and the donor center must arrange for appropriate confirmatory studies and medical and other evaluations and schedule the harvest. The time from formal search to BMT averages 4 months for UD, whereas compatible UCB, where typing occurs at time of cryopreservation, can be found in days to weeks.

Engraftment

After allogeneic HSCT, donor stem cells must expand to repopulate hematopoietic niches and differentiate to form mature blood cells. HSCT recipients show accelerated telomere shortening in peripheral blood cells, which is presumably due to increased replicative demand for donor stem cells. Donor hematopoietic engraftment after allogeneic HSCT can be documented both by analysis of chimerism as well as by the recovery of peripheral blood cell counts. However, sensitive DNA amplification technologies have demonstrated residual host hematopoiesis for variable periods of time, certainly up to 1 year. Mixed lymphoid chimerism is also well documented. The implications of mixed chimerism with respect to subsequent relapse remain controversial, probably because the current extremely sensitive assays generate data that are difficult to interpret uniformly and because outcome may well differ by disease type. For example, chimerism after BMT for CML is quite reliably related to subsequent relapse while this may not be true for other diseases.

The rate and probability of engraftment after allogeneic BMT varies with the source of HSCs, the number of HSCs given, the extent of preinfusion cell manipulation, and the GVHD prophylactic regimen used. In general, neutrophil recovery is achieved within 2 to 3 weeks and platelet recovery occurs 1 to 2 weeks thereafter. PBSC transplants are associated with a more rapid rate of engraftment than with bone marrow, most likely a result of the higher CD34+ cell dose obtained. Neutrophil engraftment occurs between 1 and 6 days earlier with PBSC transplant than with BMT and platelet engraftment between 4 and 7 days earlier. Neutrophil and platelet engraftment after UCB transplantation is significantly delayed in comparison with that seen in other stem cell sources. It is a consequence of the lower nucleated cell and CD34+ cell dose in UCB products. This has limited the application of this modality in adults, and currently using two unrelated UCB units for a given patient is being studied.

The use of methotrexate as GVHD prophylaxis is associated with delay in count recovery; conversely, recovery after T-cell–depleted BMT is more rapid. Use of colony-stimulating factors after BMT is associated with more rapid neutrophil recovery after related BMT, a finding confirmed in small pediatric series. No change in platelet recovery has been noted. Interestingly, preliminary data do not show any colony-stimulating factor–mediated acceleration of engraftment in patients undergoing UD BMT.

The number and quality of T cells that are present in grafts from peripheral blood, bone marrow, and UCB confer different rates of GVHD. GVHD, particularly chronic GVHD is more frequent in PBSC transplant than in bone marrow–derived HSCT, although this risk may be offset by an increased graft-versus-leukemia effect. In adults this increased rate of GVHD has not diminished the efficacy of PBSC transplant. However, in children, PBSC transplant has been associated with a worse outcome than when bone marrow is used as a cell source. The precise indications of PBSCs versus bone marrow as a source in pediatric HSCT therefore remains to be defined. Conversely, the rate of GVHD in UCB is lower than with either PBSC or bone marrow sources and allows for less stringent HLA-typing requirements.

Contamination by Tumor Cells

A major limitation of autologous HSCT remains contamination of the stem cell product with residual malignant cells. Although the relative contributions of either residual tumor within the host or tumor reinfused with stem cells to subsequent relapse remain unknown, increasing evidence from gene-marking studies suggests that reinfused cells contribute to relapse. In addition, in vitro assays of tumor cell colony formation and minimal residual disease (MRD) detection by polymerase chain reaction suggest that contamination of the stem cell source by tumor cells is associated with an increased relapse rate after HSCT. No large randomized clinical trial of PBSC versus bone marrow with respect to disease-free survival (DFS) has been reported. However, paired samples of PBSCs and bone marrow have been examined in a variety of settings (e.g., breast cancer, non-Hodgkin lymphoma) in which PBSCs have revealed lower levels of tumor cell contamination in the steady-state setting. More recent data suggest that the advantage of PBSCs may be abrogated after mobilization, although impact on DFS remains unclear.

A variety of methodologies have been developed to eliminate, or purge, tumor cells from stem cell collections. These are based on either positive or negative selection. The major clinically applied positive selection strategy is that of stem cell enrichment, most often selection of CD34+ cells. Negative selection is based on technologies that deplete tumor cells to a greater degree than normal stem cells. Strategies include the use of pharmacologic agents (e.g., mafosfamide), immunologic reagents targeting tissue specific antigens (e.g., neural crest antigens in neuroblastoma), and manipulation of physical or culture properties of cells. Combinations of both negative and positive selection strategies may improve both the yield and purity of stem cell collections. Increasing evidence in multiple clinical settings demonstrates an association between minimal residual tumor contamination and relapse and, conversely, suggests that more effective purging strategies may be associated with improved DFS. However, randomized studies comparing purged and unpurged stem cells have yet to be reported.

Engraftment

Trilineage hematologic reconstitution has been achieved after the infusion of autologous bone marrow, after bone marrow and PBSCs, and after PBSC infusion alone. The most mature studies are of patients reconstituting hematopoiesis with autologous marrow where engraftment after bone marrow infusion is both reliable and durable; durability of PBSC-reconstituted hematopoiesis, undertaken more recently, is less fully evaluated. The best predictor of engraftment and rate of hematologic recovery has not been established. Although colony-forming units, burst-forming units-erythroid (BFU-E), nucleated cell count, and other measures have correlated with time to engraftment in some studies, correlations have not been universal and may additionally depend on the patient population, method of cell procurement, and subsequent cell manipulation. CD34+ counts have emerged as the most reliable and practical method for predicting engraftment after PBSC infusion, with retrospective analyses demonstrating that the risk for graft failure is substantially higher in patients who received less than 2 × 10 ⁶ /kg CD34 + cells per kg.

Reconstitution of hematopoiesis after autologous bone marrow infusion has shown some tendency to be delayed when compared with the allogeneic setting. This delay is presumed due to the extent, nature, and duration of prior therapy. In addition, cell manipulation ex vivo (purging) may also contribute to delayed engraftment. Reconstitution after HSCT with autologous bone marrow can be hastened somewhat by the addition of hematopoietic growth factors to the post-SCT supportive care regimen, although the extent and nature of prior therapy may limit response to growth factors. Preliminary information in both the preclinical and clinical arenas suggests that prior aggressive chemotherapy may limit potential for hematologic recovery. Prior intensive chemotherapy supported by use of growth factors may also contribute to poor engraftment after autologous transplantation. In fact, despite recovery of peripheral blood counts, hematologic marrow reserves as assessed by in vitro measures of hematopoiesis may be blunted for many years after autologous HSCT.

Addition of mobilized PBSCs to bone marrow has resulted in a dramatic decrease in days of neutropenia and thrombocytopenia in most series with or without subsequent growth factor support. Further, use of mobilized PBSCs as the sole stem cell source has been almost universally associated with more rapid hematologic recovery. It is still debatable whether the use of hematopoietic growth factors in this setting further enhances the rate of engraftment, but available data suggest that it may. Whether the process of PBSC transplantation will result in durable hematopoiesis over decades is not yet evaluable, but reports of late graft failure are rare. Most studies suggest that the increased rate of engraftment observed with PBSCs results in decreased days in hospital, use of antibiotics, and associated supportive services. The overall impact of PBSCs versus bone marrow as a stem cell source on the outcome or cost of transplantation has yet to be determined.

Myelodysplasia Syndrome as a Late Complication

Chemotherapy, particularly with alkylating agents, may result in an increased risk for subsequent myelodysplasia syndrome (MDS). Many patients who come to autologous transplantation have received treatment with such agents. Over the past several years, increasing numbers of patients developing MDS after autologous transplantation have been reported, usually occurring from 4 to 7 years after HSCT. The actuarial incidence ranges from 5% to 10% in most series, although it has been reported to be as high as 18% at 5 to 6 years after HSCT in single center experiences. Identified risk factors have included etoposide use for stem cell mobilization, conditioning regimens utilizing TBI, and the amount of chemotherapy received before HSCT and/or the interval between diagnosis and HSCT. Whether the MDS results from previously damaged reinfused hematopoietic cells or from residual hematopoietic cells sustaining further injury during conditioning is unresolved. Cytogenetic analysis of bone marrow before autologous HSCT can reveal preexisting abnormalities and should be a standard component of pre-SCT evaluation. However, most patients developing MDS have been reported to demonstrate normal cytogenetics at the time of transplant. With more sensitive techniques currently available, such as fluorescence in situ hybridization, the majority of patients developing MDS post-SCT have been shown to harbor the same cytogenetic abnormality in pre-SCT specimens. Similarly, using an X-inactivation–based clonality assay in female patients, clonal hematopoiesis was demonstrated in a small percent of patients before HSCT; those patients were at significant risk for developing MDS after HSCT. This suggests that many cases of post-SCT MDS evolve from an abnormal clone already present before HSCT. The appropriate evaluation of hematopoietic status before embarking on autologous HSC collection and transplantation requires further study.

Immune Reconstitution

The period after HCST is characterized by profound immunodeficiency, and death from infectious complications is a major obstacle to the success of transplantation. Myeloablative conditioning results in the wholesale loss of immune cells that are necessary to confer protective immunity, including T, B, and natural killer (NK) cells. Although hematopoiesis can return within weeks of transplant, recovery of a functional immune system can take much longer and the risk for opportunistic infection remains high for up to a year even in uncomplicated transplants.

The degree of immunodeficiency varies with time, reflecting different contributions of donor and recipient immune function at different stages after transplant. Mature T, B, and NK cells are present in the donor inoculum of bone marrow, peripheral blood, or UCB and can contribute to some degree of host immunity in the immediate posttransplant period. The ability of these transferred cells to provide immune protection is dependent on how immunologically “experienced” the donor is. For instance, recipients of grafts from donors who are immune to cytomegalovirus (CMV), that is who have previously been infected with the virus and have mounted a protective immune response, have a lower rate of CMV disease, presumably because of transferred immunologic memory to that virus. Similarly, recipients whose donors have been vaccinated with a pneumococcal vaccine respond to pneumococcal vaccines more rapidly and with higher antibody titers after transplant, again suggesting that donor immunity can be transferred to the recipient. However, full immune reconstitution, with the ability to mount a protective immune response to new pathogenic challenges, is dependent on the differentiation of new immune cells from precursors in the donor graft inoculum.

In most patients, NK cells are the first lymphoid population to recover. In the first month after transplant, NK cells can represent the main lymphoid cell in the peripheral blood. T cells recover much more slowly, and normal numbers of CD4 and CD8 T cells are not achieved for 6 to 12 months after HSCT. The delay in T-cell neogenesis is because the development of new T cells is a complex process that takes months to complete. During the early period after transplant, T cells with a naïve phenotype predominate in the peripheral blood, suggesting repopulation of the T-cell compartment with recent thymic emigrants. Increased thymic output contributes to the expanding peripheral T-cell pool after transplant for 12 to 24 months after HSCT. This thymic output occurs even in adult patients whose thymuses have presumably involuted. The rate of increase in T cells is usually much more rapid in pediatric HSCT patients, consistent with the presence of more robust thymic function. The advantage of a functional thymus in immune reconstitution has been demonstrated by the marked decrease in the fraction of naïve peripheral blood T cells in a 15-year-old HSCT patient who had had a previous thymectomy compared with thymus-bearing HSCT patients. By using quantitative molecular measures of T-cell repertoire, several studies have shown that it can take as long as 12 to 24 months after HSCT for a full range of T-cell receptor diversity to be present in the T-cell compartment. This underscores the critical role of naïve thymic emigrants in restoring the full complement of T-cell receptor specificities.

Several factors influence the length of time that immunologic recovery takes. Age of BMT recipients profoundly affects the rate of immune reconstitution, with an inverse correlation between recipient age and the absolute number of T cells 1 year after HSCT. Using mobilized PBSCs as a source of stem cells appears to be associated with a faster rate of recovery than using HSCs from bone marrow, possibly as a result of a larger cell dose. Increasingly intensive conditioning regimens are delaying immune recovery in autologous recipients to a degree comparable with allogeneic HSCT. Although hematopoietic engraftment appears to be delayed in UCB recipients, immune reconstitution in children after UCB HSCT is comparable to those receiving unrelated BMT. The presence of GVHD can significantly delay immune reconstitution both as a result of disruption of normal T-cell development and from the added immunosuppression that is required for management of the disease. The consequence of factors that delay immunologic recovery is significant: slow recovery of immune function is associated with an increase in the cumulative incidence of infections and nonrelapse mortality.