Objectives: The choice of whether to use bone marrow or peripheral blood in autologous transplantation remains controversial. Posttransplant relapse and long-term survival are critical issues.

Materials and Methods: Studies that compared bone marrow transplant versus peripheral blood stem cell transplant in lymphoma patients were searched. Our search resulted in 15 studies.

Results: Pooled data showed contradictory results with no conclusive differences in overall survival (for randomized controlled trials vs nonrandomized controlled trials: hazard ratio = 0.69 vs 1.17; 95% confidence interval, 0.44-1.10 vs 0.90-1.51; and P = .12 vs P = .25), progression-free survival (for randomized controlled trials vs nonrandomized controlled trials: hazard ratio = 0.89 vs 1.14; 95% confidence interval, 0.57-1.38 vs 0.82-1.58; and P = .60 vs P = .43), and relapse rates. However, we observed an overall trend toward lower relapse rate after bone marrow transplant. Lower relapse rate was likely associated with better progression-free survival (P = .052), and lower transplant-related mortality was associated with better overall survival (P = .043).

Conclusions: Autologous bone marrow transplant with mobilization should be reconsidered for lymphoma patients to reduce recurrence and improve quality of life. More powered randomized controlled trials are warranted to update our findings.

Key words : Autologous transplantation, Bone marrow transplant, Hematopoietic stem cells

Introduction

High-dose chemotherapy followed by autologous hematopoietic stem cell transplant (auto-HSCT or ASCT) has been a standard treatment modality for refractory or relapsed non-Hodgkin lymphoma (NHL) and Hodgkin disease. Early consolidative transplant improves the prognosis of patients with high-risk aggressive lymphoma who gain little benefit from conventional regimens alone. Autologous bone marrow and peripheral blood progenitor cells permit the use of significantly intensified chemotherapy or radiotherapy irre-spective of fatal myelosuppression or hematologic toxicity. As a first-line treatment, high-dose chemotherapy plus ASCT substantially reduces the bone marrow tumor load, resulting in a high rate of complete response, but late relapses are common. Bone marrow transplant (BMT), which contains a number of different stem cell populations, including hematopoietic stem cells, mesenchymal stem cells, endothelial progenitor cells, and fibroblasts, may have a protective effect. Peripheral blood progenitor cells have become more popular due to their faster trilineage engraftment or because NHL usually involves the bone marrow. However, whether bone marrow or mobilized peripheral blood has more tumor cell contamination or greater clonogenic potential is unclear because tumor cells can also be mobilized to the peripheral blood. Although “solid tumors” like lymphoma are distinct from leukemia or other blood cancers, most systematic reviews and meta-analyses on this topic have taken all hematologic malignancies into account. To further investigate whether the stem cell source is associated with better survival outcomes, lower relapse rates, or nonrelapse mortality, we did a meta-analysis to compare BMT and peripheral blood stem cell transplant (PBSCT) among lymphoma patients.

Materials and Methods

This meta-analysis was registered with the International Prospective Register of Systematic Reviews (PROSPERO, Centre for Reviews and Dissemination, University of York, York, UK) on March 6, 2016 (Registration No. CRD42016036152). We prepared a prospective protocol a priori in adherence to PRISMA-P (Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols) 20151 and MOOSE (Meta-analysis of Observational Studies in Epidemiology),² which are centered around the PICO principle.

Literature search
PubMed, EMBASE, and the Cochrane Library were searched for electronic records to be exported into EndNote X7.5 (Thomson Reuters, New York, NY, USA), augmented by Google Scholar, the Lymphoma Research Foundation, the National Cancer Institute, the Center for International Blood and Marrow Transplant Research, OpenGrey, and ClinicalTrials.gov. The databases were searched from their inception until January 3, 2016. No language or outcome restrictions were imposed. The search strategy was uploaded to PROSPERO. In addition, we scanned the obtained full-text articles and scrutinized the references of included papers and pertinent reviews. We also manually searched abstracts from conference proceedings of annual meetings of the American Society of Hematology, American Society of Clinical Oncology, European Hematology Association, and the European Bone Marrow Transplantation Group from 2000 to 2015 for other eligible studies.

Inclusion and exclusion criteria
All randomized controlled trials (RCTs), controlled clinical trials, and cohort or case-control studies comparing PBSCT with BMT for lymphoma patients of any age, with at least one of the quantitative data sets for survival or death, progression/relapse, and engraftment (absolute neutrophil count > 0.5 × 10⁹/L; platelet count > 20 × 10⁹/L), whether in a full-text or an abstract format, were included. Those studies that compared either PBSCT or BMT with a mixed combination of the 2 or a second HSCT were ineligible. We intended to include hematologic malignancies if the data of lymphoma patients could be separately extracted, but no population in our practice fit this case.

We also questioned whether the differences between peripheral blood and bone marrow varied with donor type (ie, allogeneic or autologous donors); therefore, the initial search strategy was designed not to exclude allogeneic HSCT for lymphoma. However, only 3 documents (2 full-text articles3,4 and 1 abstract5) were returned, which were inadequate to answer this question. Thus, “allo-HSCT” studies were excluded. Although no language restriction was imposed during the search, only English and Chinese studies were eligible for comprehension. Two reviewers (YZ and YC) independently evaluated all selected articles meeting the predefined criteria, and the strength of agreement on the final inclusion was graded using the Cohen statistic.⁶

Data extraction
Data were extracted by YZ and confirmed by YC. The extracted data were filled into a standardized Excel file (Microsoft Corporation, Redmond, WA, USA). If data at both diagnosis and transplant were provided, we chose the latter one for analysis. Median (range) days to neutrophil and platelet recovery were converted to the mean and standard deviation using estimation methods recommended by Hozo and associates.⁷ The original publication was retained to clarify the study characteristics, using other updated versions only to supplement longer follow-up data. Standard error and the natural logarithm of hazard ratio (HR) were indirectly calculated from the reported number of events with the log-rank P value or by digitizing the Kaplan-Meier curves using Engauge Digitizer8 version 4.1 (Mark Mitchell) and Adobe Photoshop CS6 Extended as described by Tierney and colleagues.⁹ An attempt was made to contact all primary authors for incomplete information by sending E-mails and Research Gate messages; however, no responses were received. Any discrepancies would be addressed by joint crosschecks for validity.

Quality assessment
We applied the revised Cochrane “risk of bias” tool (described in version 5.1.0) to estimate the methodologic quality of included RCTs, whereas non-RCT comparative studies were critically appraised using the ROBINS-I tool (Risk of Bias in Non-randomized Studies of Interventions) version 7 (March 2016),¹⁰ which considers 7 domains of bias and 1 overall bias. Each domain is judged as low, moderate, serious, or critical risk or as having too little information to determine. One study mixed the randomized and nonrandomized data and was assessed using both tools. Disagreements would be resolved through discussions to reach a consensus.

Data synthesis and statistical analyses
Data from interventional and observational studies were synthesized separately. Subgroup analyses were performed according to study design. In sensitivity analyses, we tested whether omitting a single study affected the final results. Continuous outcomes were expressed as the mean differences and binary outcomes as odds ratios (ORs) with 95% confidence interval (95% CI). In time-related adverse events, an HR > 1 favored BMT over PBSCT. Interstudy heterogeneity was explored using the Cochran’s Q test with significance set at P < .1 and I2 statistic with a cut-off value of > 50%, which indicated severe heterogeneity. A random-effects model would then be chosen; otherwise, the fixed-effects model would be used. An L’Abbe plot mirrored the relapse rate in the experimental group versus in the control group and reflected the heterogeneity of binary variables. Meta-regression models helped to investigate potential covariates that might contribute to the heterogeneity of continuous variables.

Publication bias was visualized by funnel plots and quantified by using the Egger test. Meta-regression and publication bias were not conducted routinely when the study number was less than 10. Unless otherwise specified, P < .05 (2-sided) was considered statistically significant. All statistical analyses were performed using RevMan 5.3.5 (Nordic Cochrane Centre, Copenhagen, Denmark) and Stata/MP 14.1 (StataCorp LP, College Station, TX, USA).

Results

Search results
Study selection results are illustrated in Figure 1. Our initial search returned 8596 citations, which included 5113 from EMBASE, 3085 from PubMed, and 398 from Cochrane Central Register of Controlled Trials, without additional sources of publications identified. Of the 7024 nonduplicate records, 6962 were quickly discarded (reviews, case reports, biomarker and animal/cell studies, and those not relevant). The remaining 62 references were retrieved for eligibility assessment, and 46 publications were excluded for reasons shown in Figure 1. Finally, 15 studies (5 RCTs,^11-16 2 controlled clinical trials,^17,18 and 8 observational studies^19-26) comprising 16 publi-cations (15 accessed as full text11-18,20-26 and 1 as an abstract¹⁹) were included in this meta-analysis. The Cohen statistic for agreement was 0.90.

Study characteristics
Of the 8 observational studies, four were cohort studies (2 prospective studies^20,22 and 2 retrospective studies^19,25) and four^21,23,24,26 were retrospective case-control studies (3 with matched-pair designs^21,24,26). The main study characteristics and summary outcomes are shown in Tables 1 to 3. Nine studies originated from Europe, 5 from the United States, and 1 from Asia. These studies were published between 1994 and 2005, enrolling a total of 1993 patients with lymphoma, among whom 1949 patients received high-dose chemotherapy plus ASCT (988 PBSCT and 961 BMT). Data from 1883 patients were analyzed. Sample sizes varied across studies, ranging from 26 to 710. Most patients were 30 to 50 years old and diagnosed with advanced relapsed/refractory NHL or Hodgkin disease. The median age was generally larger within the PBSCT groups.

Quality assessment
The risk of bias assessment of the included studies is presented in Tables 4 and 5.

Primary outcomes
No studies provided HR information directly on overall and progression-free survival (PFS) assessments. By indirect means, 12 studies qualified for overall survival (OS) analysis in 1543 patients (Figure 2a) and 10 for PFS analysis in 1570 patients (Figure 2b). When RCTs and non-RCTs were examined, the opposite results were shown. We found a significant heterogeneity (I2 = 51%) within the non-RCT subgroup regarding PFS. Therefore, a random-effects model was applied. No significant heterogeneity was found in both OS subgroups (I2 = 0% and 49%, respectively); however, given the subgroup differences (P = .05), we strongly estimated a publication bias for RCTs, which is mostly dotted on the left side of the OS funnel plot (Figure 2c).

We found that RCTs might favor PBSCT that improved OS (pooled HR of 0.69; 95% CI, 0.44-1.10; P = 0.12) and PFS (pooled HR of 0.89; 95% CI, 0.57-1.38; P = .60) and non-RCTs might favor BMT that improved OS (pooled of HR 1.17; 95% CI, 0.90-1.51; P = .25) and PFS (pooled HR of 1.14; 95% CI, 0.82-1.58; P = .43). However, no significant differences were found between PBSCT and BMT regarding OS and PFS.

Incidence of relapse was derived from 13 of 15 studies with 1584 patients analyzed. Great discre-pancies were observed among different study types (Figure 3a). Overall, the L’Abbe plot indicated large heterogeneity and a higher relapse rate after PBSCT. No significant publication bias regarding relapse rate was determined from the funnel plot (Figure 3b) or the Egger test (P = 0.157). In cohort studies, BMT resulted in significantly lower relapse rates than PBSCT (pooled OR of 0.30; 95% CI, 0.13-0.69; P = .005). However, no significant differences were found in RCT (pooled OR of 2.13; 95% CI, 0.74-6.13; P = .16) and in case-control studies (pooled OR of 0.63; 95% CI, 0.27-1.47; P = .29).

Secondary outcomes
Ten studies involving 1368 patients reported nonrelapse mortality or early toxic death events (Figure 4a).

Eleven studies of 845 patients and 7 studies of 453 patients reported the median days and ranges to recovery for neutrophils and platelets, respectively. A significantly faster engraftment was shown after PBSCT (Figure 4b and 4c). Further removal of any single study did not virtually change the total effects.

In non-RCTs, PBSCT had a significantly reduced transplant-related mortality (TRM) (pooled OR of 0.52; 95% CI, 0.33-0.80; P = .003). In RCTs, no statistical significance was found that BMT increased TRM (pooled OR of 0.34; 95% CI, 0.11-1.06; P = .06). Both subgroups had little heterogeneity (I2 = 0%).

Meta regression
Ten data sets from 9 studies11,15,16,19,21-26 provided both PFS linear HR and relapse rate data for a single covariant meta-regression. We found no significant positive linear relation between the relapse rate difference (PBSCT minus BMT) and PFS (P = .052).

The relapse rate difference alone could explain 52.76% of the PFS heterogeneity. However, it failed to explain the heterogeneity of OS (11 datasets^{11,14-16,18,19,22-26}; P = .426). Surprisingly, when we added TRM difference to the relapse rate difference to conduct a multicovariant meta-regression, both covariates together could explain 80.18% of the OS heterogeneity, whose residual variation was reduced to only 16.27% (8 datasets^{11,14,19,22,24-26}; P = .282 for relapse rate; P = .058 for TRM). The TRM difference alone contributed to 84.41% of the OS heterogeneity, following a significant positive linear relationship in 9 datasets^{11,13,14,19,22,24-26} (P = .043; Figure 3d). Other covariates such as year, sample size, age, sex, Hodgkin disease/NHL ratio, and follow-up difference did not significantly affect the OS or PFS (data not shown).

Discussion

Relapse occurs in more than 50% of lymphoma patients after autologous transplant^27,28 and is undoubtedly the main reason for rehospitalization. Indeed, relapse-free survival is always of paramount concern to any patient with malignancy. For lymphoma, “relapse-free” is simply “disease-free” (ie, complete remission). Once relapse occurs, it means progression, because relapse is also a type of progression. Progression has a broader definition that includes partial remission and refractory disease. Because relapse rate more directly describes relapse than relapse-free survival and because relapse-free survival is closer to describing OS if there is a rare relapse over a short time, PFS was used more often than relapse-free survival in our included studies.

In this study, we found that PFS was not different between BMT and PBSCT studies; however, significant heterogeneity was observed regarding PFS in individual studies, which might be explained by different relapse rates. Relapse rate varied by study design, but subgroup analysis and meta-regression analysis failed to reveal the cause of lymphoma relapse. However, stem cell source was highly suspected, and the overall results appeared to favor BMT. Overall survival was more closely related to TRM than relapse rate. Both TRM and engraftment favored PBSCT, and their pooled results were highly consistent. However, PBSCT had no significant OS benefit.

We found no reasons why RCTs and non-RCTs showed conflicting results. Indeed, relapse itself is a disputed issue. Tumor burden originates from minimal residual disease after high-dose chemotherapy or from reinfused contaminating autografts. However, some groups have found that peripheral blood progenitor cells contain more tumors,^29,30 others have found that bone marrow contains more tumors,^31-33 and others have shown that peripheral blood progenitor cells and bone marrow are similarly contaminated.^20,34,35 One study revealed that peripheral blood progenitor cell contamination was highly variable.²⁷ However, all studies implied that peripheral blood progenitor cells were frequently contaminated. In fact, tumor cells were found in the blood before mobilization or were mobilized into the circulation regardless of marrow infiltration circulating tumor cells, although the blood was initially polymerase chain reaction negative.³⁶ Because Hodgkin disease has notoriously poor peripheral blood progenitor cell collections,³⁷ it may result in repeated mobilization. Because more cells are required for peripheral blood progenitor cell infusion than for BMT, peripheral blood progenitor cell autografts may actually be even more contaminated by lymphoma cells.²⁶

We speculated that immunosurveillance may play a role in the lower relapse rate of patients after BMT. This was based on the evidence that all subsets of natural killer T cells in malignant marrow were more activated than in peripheral blood38 and that long-lasting expansion of the CD56+and CD16+ natural killer T-cell subsets was observed only in BMT.³⁹ Natural killer cells are known to monitor cancer development and to kill tumor cells. Natural killer T cells are T cells that express killer cell immunoglobulin-like receptors; these receptors can recognize the down-regulated major histocom-patibility complex class I-like molecule CD1d on the tumor cell surface or secrete cytokines such as interferon gamma and tumor necrosis factor alpha, which participate in the antitumor immune response.⁴⁰ The invariant natural kill T cells are a type I CD1d-reactive natural killer T-cell subset that can protect host immunosurveillance against a B-cell lymphoma.41 A high-level invariant natural killer T-cell line within the graft is the only subset associated with less recurrence and improved PFS.⁴² In BMT, faster recovery predominantly involved CD8+ cytotoxic T cells and CD19+/CD20+ B cells. Conversely, PBSCT was more profound for CD4+ helper T and T regulatory cells.^39,43 Peripheral blood progenitor cells were reported to have an increased secondary cancer risk, particularly for myelodysplastic syndrome/acute leukemia occurrences.^44,45 Transplanted bone marrow cells might be capable of mounting an immunologic assault against the chemoresistant residual lymphoma cells, resembling a graft versus leukemia effect.

Because high-dose chemotherapy and radiation therapy damage the hematopoietic system, especially the immune system, TRM basically refers to infection as its adverse event. Because of mobilization required for PBSCT, both hematopoietic recovery and immunologic reconstitution can be quicker, resulting in reduced infectious complications and thus significantly lower TRM in PBSCT. Although OS but not relapse is dependent on TRM, PBSCT is not associated with a better OS, indicating that OS might have other confounding factors besides nonrelapse mortality. However, the use of granulocyte colony-stimulating factor and granulocyte/macrophage colony-stimulating factor could result in inconclusive outcomes. In the study from Weisdorf and associates,¹⁷ both bone marrow collection and PBSC collection were unmobilized, with the collections in the study from Damiani and associates¹² being mobilized. Interestingly, in both studies, use of BMT had better results. We attributed the severe heterogeneity of neutrophil recovery to diverse granulocyte colony-stimulating factor doses in the studies. Despite rapid blood cell recovery, mobilization impaired the long-term marrow reconstitutive ability of autografts in lymphoma patients, which caused persistent deficiency for several years after PBSCT.⁴⁶

Hemopoietic stem cells are the most relevant targets for high-dose stem cell toxic drugs. Hemopoietic stem cells residing in the bone marrow microenvironment maintain a lifelong self-renewal capacity to repopulate the blood system. It has been reported that peripheral blood progenitor cell grafts are more easily affected by previous cytotoxic chemotherapy, whereas bone marrow progenitor cells are not as sensitive as peripheral blood progenitor cells, since little or no impact was found in early and long-term recovery after hemopoietic stem cell BMT.^47,48 This was particularly advantageous in heavily pretreated lymphoma patients with poor peripheral blood progenitor cell mobilization. Epigenetic detection revealed that only bone marrow-derived mesenchymal stem cells exhibited hypomethylation and increased expression of transcriptional genes, which facilitated hematopoietic niche formation and permitted homing and maintenance of long-term hemopoietic stem cells.⁴⁹ Mesenchymal stem cells have also been shown to promote early immune reconstitution for lym-phomas after ASCT.⁵⁰

To our knowledge, this is the second systematic review but the first meta-analysis comparing bone marrow versus peripheral blood ASCT in patients with lymphoma with a focus on survival and relapse. The previous systematic review only included RCTs and focused on engraftment.⁵¹ This meta-analysis had several limitations. First, only 5 studies were included in the meta-analysis, which might be one of the most important reasons for publication bias and subgroup disparity. However, because observational studies tend to reflect more real world situations, objective phenomena can be shown versus with the more idealized RCTs.⁵² Second, there have been advances in transplant methods, with established transplant methods in all hematologic malignancies. Because solid tumors are different from blood cancers, future studies should separate leukemia and lymphoma as much as possible. Other factors, including patient age, adjuvant therapy, disease stage, and treatment protocols, may have also affected the strength of the study conclusion. Further meta-analyses should be performed to verify our study based on better-powered RCTs with comparable disease stages and treatment protocols.

Conclusions

Our meta-analysis shows contradictory results from RCTs and non-RCTs, and we failed to determine whether BMT or PBSCT showed superior OS and PFS. Nevertheless, the strong likelihood that lower TRM is associated with better OS and that less relapse is associated with better PFS suggest that physicians should choose a proper ASCT according to patient’s conditions and prognosis.

Mobilization with growth factors is encouraging especially in BMT, but caution is warranted. The protective role that bone marrow plays in relapse-free survival needs further consideration. Large-volume and well-designed RCTs with longer follow-up are warranted to update our findings. Future translational research studies are needed to confirm our ideas.

References:

1. Shamseer L, Moher D, Clarke M, et al. Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015: elaboration and explanation. BMJ. 2015;350:g7647.
   CrossRef - PubMed
   
2. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis of Observational Studies in Epidemiology (MOOSE) group. JAMA. 2000;283(15):2008-2012.
   CrossRef - PubMed
   
3. Murashige N, Kami M, Kishi Y, et al. Allogeneic haematopoietic stem cell transplantation as a promising treatment for natural killer-cell neoplasms. Br J Haematol. 2005;130(4):561-567.
   CrossRef - PubMed
   
4. Marcais A, Porcher R, Robin M, et al. Impact of disease status and stem cell source on the results of reduced intensity conditioning transplant for Hodgkin's lymphoma: a retrospective study from the French Society of Bone Marrow Transplantation and Cellular Therapy (SFGM-TC). Haematologica. 2013;98(9):1467-1475.
   CrossRef - PubMed
   
5. Mato AR, Pecora AL, Rowley SD, et al. A decade of stem cell transplantation in lymphoma: single center experience and outcome of 938 consecutive allogeneic and autologous hematopoietic stem cell transplants performed At John Theurer Cancer Center. Blood. 2011;118(21):3086-3086.
   
   
6. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33(1):159-174.
   CrossRef - PubMed
   
7. Hozo SP, Djulbegovic B, Hozo I. Estimating the mean and variance from the median, range, and the size of a sample. BMC Med Res Methodol. 2005;5:13.
   CrossRef - PubMed
   
8. Mitchell M, Muftakhidinov B, Winchen T, et al. Engauge Digitizer Software. http://markummitchell.github.io/engauge-digitizer.
   
   
9. Tierney JF, Stewart LA, Ghersi D, Burdett S, Sydes MR. Practical methods for incorporating summary time-to-event data into meta-analysis. Trials. 2007;8:16.
   CrossRef - PubMed
   
10. Sterne JAC, Hernán MA, Reeves BC, et al. ROBINS-I: a tool for assessing risk of bias in non-randomized studies of interventions. BMJ. 2016;355:i4919. doi: 10.1136/bmj.i4919.
    CrossRef
    
11. Vose JM, Sharp G, Chan WC, et al. Autologous transplantation for aggressive non-Hodgkin's lymphoma: results of a randomized trial evaluating graft source and minimal residual disease. J Clin Oncol. 2002;20(9):2344-2352.
    CrossRef - PubMed
    
12. Damiani D, Fanin R, Silvestri F, et al. Randomized trial of autologous filgrastim-primed bone marrow transplantation versus filgrastim-mobilized peripheral blood stem cell transplantation in lymphoma patients. Blood. 1997;90(1):36-42.
    PubMed
    
13. Vellenga E, van Agthoven M, Croockewit AJ, et al. Autologous peripheral blood stem cell transplantation in patients with relapsed lymphoma results in accelerated haematopoietic reconstitution, improved quality of life and cost reduction compared with bone marrow transplantation: the Hovon 22 study. Br J Haematol. 2001;114(2):319-326.
    CrossRef - PubMed
    
14. Kanteti R, Miller K, McCann J, et al. Randomized trial of peripheral blood progenitor cell vs bone marrow as hematopoietic support for high-dose chemotherapy in patients with non-Hodgkin's lymphoma and Hodgkin's disease: a clinical and molecular analysis. Bone Marrow Transplant. 1999;24(5):473-481.
    CrossRef - PubMed
    
15. Schmitz N, Linch DC, Dreger P, et al. Randomised trial of filgrastim-mobilised peripheral blood progenitor cell transplantation versus autologous bone-marrow transplantation in lymphoma patients. Lancet. 1996;347(8998):353-357.
    CrossRef - PubMed
    
16. Kottaridis PD, Peggs K, Schmitz N, et al. Survival and freedom from progression in autotransplant lymphoma patients is independent of stem cell source: further follow-up from the original randomised study to assess engraftment. Leuk Lymphoma. 2002;43(3):531-536.
    CrossRef - PubMed
    
17. Weisdorf DJ, Verfaille CM, Miller WJ, et al. Autologous bone marrow versus non-mobilized peripheral blood stem cell transplantation for lymphoid malignancies: a prospective, comparative trial. Am J Hematol. 1997;54(3):202-208.
    CrossRef - PubMed
    
18. Hou SL, Zhang Q, Han WE, Gui W, Wang YL. Autologous hematopoietic stem cell transplantation for lymphoma: An evaluation of grafts source and minimal residual disease. Chinese J Cancer Res. 2005;17(3):6.
    CrossRef
    
19. Goździk J, Hansz J. Comparison of the results of high-dose chemotherapy followed by transplantation of autologous bone marrow or peripheral blood stem cells in poor prognosis Hodgkin's disease patients. Acta Haematologica Polonica. 2003;34(1):11.
    
    
20. Sharp JG, Kessinger A, Mann S, et al. Outcome of high-dose therapy and autologous transplantation in non-Hodgkin's lymphoma based on the presence of tumor in the marrow or infused hematopoietic harvest. J Clin Oncol. 1996;14(1):214-219.
    CrossRef - PubMed
    
21. Liberti G, Pearce R, Taghipour G, Majolino I, Goldstone AH. Comparison of peripheral blood stem-cell and autologous bone marrow transplantation for lymphoma patients: a case-controlled analysis of the EBMT Registry data. Lymphoma Working Party of the EBMT. Ann Oncol. 1994;5 Suppl 2:151-153.
    CrossRef - PubMed
    
22. Geisler CH, Hansen MM, Andersen NS, et al. BEAM+autologous stem cell transplantation in malignant lymphoma: 100 consecutive transplants in a single centre. Efficacy, toxicity and engraftment in relation to stem-cell source and previous treatment. Eur J Haematol. 1998;61(3):173-182.
    CrossRef - PubMed
    
23. Bierman PJ, Vose JM, Anderson JR, Bishop MR, Kessinger A, Armitage JO. High-dose therapy with autologous hematopoietic rescue for follicular low-grade non-Hodgkin's lymphoma. J Clin Oncol. 1997;15(2):445-450.
    CrossRef - PubMed
    
24. Majolino I, Pearce R, Taghipour G, Goldstone AH. Peripheral-blood stem-cell transplantation versus autologous bone marrow transplantation in Hodgkin's and non-Hodgkin's lymphomas: a new matched-pair analysis of the European Group for Blood and Marrow Transplantation Registry Data. Lymphoma Working Party of the European Group for Blood and Marrow Transplantation. J Clin Oncol. 1997;15(2):509-517.
    CrossRef - PubMed
    
25. Brice P, Marolleau JP, Pautier P, et al. Hematologic recovery and survival of lymphoma patients after autologous stem-cell transplantation: comparison of bone marrow and peripheral blood progenitor cells. Leuk Lymphoma. 1996;22(5-6):449-456.
    CrossRef - PubMed
    
26. Perry AR, Peniket AJ, Watts MJ, Leverett D, Goldstone AH, Linch DC. Peripheral blood stem cell versus autologous bone marrow transplantation for Hodgkin's disease: equivalent survival outcome in a single-centre matched-pair analysis. Br J Haematol. 1999;105(1):280-287.
    CrossRef - PubMed
    
27. Jacquy C, Soree A, Lambert F, et al. A quantitative study of peripheral blood stem cell contamination in diffuse large-cell non-Hodgkin's lymphoma: one-half of patients significantly mobilize malignant cells. Br J Haematol. 2000;110(3):631-637.
    CrossRef - PubMed
    
28. Rapoport AP, Meisenberg B, Sarkodee-Adoo C, et al. Autotransplantation for advanced lymphoma and Hodgkin's disease followed by post-transplant rituxan/GM-CSF or radiotherapy and consolidation chemotherapy. Bone Marrow Transplant. 2002;29(4):303-312.
    CrossRef - PubMed
    
29. Negrin RS, Pesando J. Detection of tumor cells in purged bone marrow and peripheral-blood mononuclear cells by polymerase chain reaction amplification of bcl-2 translocations. J Clin Oncol. 1994;12(5):1021-1027.
    CrossRef - PubMed
    
30. Blystad AK, Torlakovic E, Holte H, Kvaloy S, Lenschow E, Kvalheim G. CD34(+) cell enrichment depletes atypical CD30(+) cells from PBPC grafts in patients with HD. Cytotherapy. 2001;3(4):295-305.
    CrossRef - PubMed
    
31. Gribben JG, Neuberg D, Barber M, et al. Detection of residual lymphoma cells by polymerase chain reaction in peripheral blood is significantly less predictive for relapse than detection in bone marrow. Blood. 1994;83(12):3800-3807.
    PubMed
    
32. Craig JI, Langlands K, Parker AC, Anthony RS. Molecular detection of tumor contamination in peripheral blood stem cell harvests. Exp Hematol. 1994;22(9):898-902.
    PubMed
    
33. Leonard BM, Hetu F, Busque L, et al. Lymphoma cell burden in progenitor cell grafts measured by competitive polymerase chain reaction: less than one log difference between bone marrow and peripheral blood sources. Blood. 1998;91(1):331-339.
    PubMed
    
34. Yuan R, Dowling P, Zucca E, Diggelmann H, Cavalli F. Detection of bcl-2/JH rearrangement in follicular and diffuse lymphoma: concordant results of peripheral blood and bone marrow analysis at diagnosis. Br J Cancer. 1993;67(5):922-925.
    CrossRef - PubMed
    
35. Berinstein NL, Reis MD, Ngan BY, Sawka CA, Jamal HH, Kuzniar B. Detection of occult lymphoma in the peripheral blood and bone marrow of patients with untreated early-stage and advanced-stage follicular lymphoma. J Clin Oncol. 1993;11(7):1344-1352.
    CrossRef - PubMed
    
36. Brugger W, Bross KJ, Glatt M, Weber F, Mertelsmann R, Kanz L. Mobilization of tumor cells and hematopoietic progenitor cells into peripheral blood of patients with solid tumors. Blood. 1994;83(3):636-640.
    PubMed
    
37. Haas R, Mohle R, Fruhauf S, et al. Patient characteristics associated with successful mobilizing and autografting of peripheral blood progenitor cells in malignant lymphoma. Blood. 1994;83(12):3787-3794.
    PubMed
    
38. Dean J, McCarthy D, Lawler M, Doherty DG, O'Farrelly C, Golden-Mason L. Characterization of NKR+ T-cell subsets in human bone marrow: implications for immunosurveillance of neoplasia. Clin Immunol. 2005;114(1):42-51.
    CrossRef - PubMed
    
39. Parrado A, Casares S, Prieto J, Carmona M, Vaquero A, Rodriguez-Fernandez JM. Repopulation of circulating T, B and NK lymphocytes following bone marrow and blood stem cell transplantation. Hematol Cell Ther. 1997;39(6):301-306.
    CrossRef - PubMed
    
40. Vivier E, Ugolini S, Blaise D, Chabannon C, Brossay L. Targeting natural killer cells and natural killer T cells in cancer. Nat Rev Immunol. 2012;12(4):239-252.
    CrossRef - PubMed
    
41. Renukaradhya GJ, Khan MA, Vieira M, Du W, Gervay-Hague J, Brutkiewicz RR. Type I NKT cells protect (and type II NKT cells suppress) the host's innate antitumor immune response to a B-cell lymphoma. Blood. 2008;111(12):5637-5645.
    CrossRef - PubMed
    
42. Malard F, Labopin M, Chevallier P, et al. Larger number of invariant natural killer T cells in PBSC allografts correlates with improved GVHD-free and progression-free survival. Blood. 2016;127(14):1828-1835.
    CrossRef - PubMed
    
43. Talmadge JE, Reed E, Ino K, et al. Rapid immunologic reconstitution following transplantation with mobilized peripheral blood stem cells as compared to bone marrow. Bone Marrow Transplant. 1997;19(2):161-172.
    CrossRef - PubMed
    
44. Tarella C, Passera R, Magni M, et al. Factors influencing the occurrence of secondary myelodysplastic syndrome/acute leukemia following high-dose therapy and autograft: A GITIL (Gruppo Italiano Terapie innovative neilinfomi) survey in 1,347 lymphoma patients. Haematologica. 2009;94(suppl 4):4.
45. Andre M, Henry-Amar M, Blaise D, et al. Treatment-related deaths and second cancer risk after autologous stem-cell transplantation for Hodgkin's disease. Blood. 1998;92(6):1933-1940.
    PubMed
    
46. Benboubker L, Cartron G, Roingeard F, et al. Long-term marrow reconstitutive ability of autologous grafts in lymphoma patients using peripheral blood mobilized with granulocyte colony-stimulating factor or granulocyte-macrophage colony-stimulating factor compared to bone marrow. Exp Hematol. 2003;31(1):89-97.
    CrossRef - PubMed
    
47. Dreger P, Kloss M, Petersen B, et al. Autologous progenitor cell transplantation: prior exposure to stem cell-toxic drugs determines yield and engraftment of peripheral blood progenitor cell but not of bone marrow grafts. Blood. 1995;86(10):3970-3978.
    PubMed
    
48. Martin A, Perez-Simon JA, Caballero MD, et al. Effect of pre-transplant cumulative doses of chemotherapeutic drugs on early and long-term hematological recovery after autologous bone-marrow transplantation for lymphoma. Haematologica. 2005;90(1):78-85.
    PubMed
    
49. Reinisch A, Etchart N, Thomas D, et al. Epigenetic and in vivo comparison of diverse MSC sources reveals an endochondral signature for human hematopoietic niche formation. Blood. 2015;125(2):249-260.
    CrossRef - PubMed
    
50. Batorov EV, Shevela EY, Tikhonova MA, et al. Mesenchymal stromal cells improve early lymphocyte recovery and T cell reconstitution after autologous hematopoietic stem cell transplantation in patients with malignant lymphomas. Cell Immunol. 2015;297(2):80-86.
    CrossRef - PubMed
    
51. Lewis A. Autologous stem cells derived from the peripheral blood compared to standard bone marrow transplant; time to engraftment: a systematic review. Int J Nurs Stud. 2005;42(5):589-596.
    CrossRef - PubMed
    
52. Cartwright N. Are RCTs the gold standard? Biosocieties. 2007;1(9):11-20.
    CrossRef