Impact of ABO mismatching on the outcomes of allogeneic BMT:

IPD based meta-analysisElianna Saidenberg

November 2009

A brief introduction to HSCT

• What is bone marrow?• What are hematopoietic stem cells?• What is the purpose of transplantation?

– Solid organ transplantation• For treatment of organ dysfunction or failure

– Hematopoietic stem cell transplantation• High dose chemotherapy +/- radiation eradicate the

malignant clone or abnormal marrow but also results in bone marrow ablation. Transplanted marrow repopulates the marrow with normal cells.

Do these tissues match?• Major Histocompatibility

Complex (MHC) and Human Leukocyte Antigens (HLA)

• MHC genes (on chromosome 6) encode human leukocyte antigens and a variety of cell surface markers, antigen-presenting molecules, and other proteins which allow immune recognition of self and non-self

• MHC genes are inherited as a group (haplotype), one from each parent. Thus, a heterozygous human inherits one paternal and one maternal haplotype, each containing three class-I (B, C and A) and three class II (DP, DQ and DR) loci

• Engraftment– Resumption of function by the transplanted organ or

tissue• Rejection

– A problem in solid organ transplant (rarely in BMT)– Host’s immune system recognizes transplanted organ

as foreign and attacks it• “Host-versus-graft disease”

• Graft-versus-host disease (GVHD)– A problem in BMT, never in SOT– Donor immune system recognizes all host tissues as

foreign and attacks them• The closer the HLA match the less the risk of

rejection and GVHD• Graft-versus-tumour effect

– Donor immune system recognizes host’s tumour cells as foreign and removes them too

You mean there is more than one kind of bone marrow transplant?

• Syngeneic- From an identical twin• Allogeneic- From an HLA matched donor

– MRD- Matched related donor– MUD- Matched unrelated donor

• Autologous- Patients own stem cells are harvested before high dose chemotherapy and are later re-infused.– Not a curative therapy

• Reduced intensity transplantation– Recipient bone marrow not completely ablated before

transplantation of donor marrow– Relies on GVT effect

You mean there are even more kinds of BMT?

• Bone marrow source– HSC are harvested from a large bone of the donor, through a

large needle that reaches the center of the bone, performed under general anesthesia

• PBSCT– Donor stem cells harvested by apheresis procedure after G-CSF

therapy to boost levels of peripheral blood stem cells

• Umbilical cord blood – Cord blood has a higher concentration of HSC than is normally

found in adult blood. Stem cells are harvested at time of delivery• Dose of stem cells low, adult cord blood BMTs usually requires 2

cords• Lesser degree of HLA matching possible compared to other sources

of stem cells

Causes of mortality and morbidity in BMT

• Treatment related morbidity/mortality (TRM)– Chemotherapy/ radiation toxicities

• Includes secondary malignancies– Complications related to cytopenias (infection, bleeding,

anemia)– Veno-occlusive disease causing severe liver dysfunction– GVHD

• Acute GVHD occurs in the first 3 months after transplantation and may involve the skin, intestine, or the liver, and is often fatal.

• Chronic GVHD is the major source of late treatment-related complication. Can lead to the development of fibrosis causing functional disability

• High-dose corticosteroids are the mainstay of treatment but increase risk of infection

– Drug side effects• Drugs needed to prevent GVHD include cyclosporine, MMF• All have serious side effects (ie HTN, ARF. TTP)

• Disease relapse

ABO incompatibility in BMT

• Genes for ABO groups and for HLA are inherited independently

• ABO incompatibility may occur in 20-40% of HLA-matched allogeneic SCT

• ABO incompatibility between donor and recipient is NOT considered a contraindication to transplantation

Major ABO Incompatibility• Donor ABO antigens are not compatible with recipient’s immune system – Example: Donor is group A, recipient is group B or

group O and hence has anti-A in his or her serum

• Can have acute hemolysis due to lysis of incompatible RBCs contained in the graft– Purge grafts of RBCs– Plasmapheresis may be effective in reducing the anti-donor

hemagglutinin in the recipient plasma with a goal to reduce the titer to 1:16 or lower

• Known complications– Pure red cell aplasia– Delayed donor RBC engraftment

• Anti-A and anti-B remain demonstrable in the recipient’s plasma for some months

Minor ABO Incompatibility• Recipient ABO antigens are not compatible

with donor’s immune system – Hence, the donor’s B cells capable of making

antibodies directed at recipient’s RBCs

• Hemolysis may develop 1-2 weeks after transplant – Due to lysis of ABO-incompatible recipient cells as the donor

lymphocytes engraft– Reactions are most common and most severe when the donor is group

O and the recipient group A

Bidirectional Incompatibility• Recipient isohemagglutinins directed against

donor RBCs and contrariwise– Occurs in 3-5% of alloSCT– Example: Donor is group A, recipient is group B– Potential complications

• Hemolysis of RBCs in graft• Hemolysis of recipient RBCs by passively transfused antibody,• Hemolysis of RBCs produced by graft• PRCA

• Special considerations in ABO incompatible transplant:– During preparatory regimen transfused blood should

be compatible with both donor and recipient• In bidirectional incompatibility (major-minor) only group O

cells should be transfused• It may be prudent to give plasma reduced platelet

transfusions • IVIg should be avoided during the post-transplant period as it

contains variable titres of red cell antibodies (esp anti-A)

Recipient blood group

Donor TransfusedRBCs

TransfusedPlatelets

TransfusedFFP

A B O Any AB

A O O Any A

A AB O,A Any AB

B A O Any AB

B O O Any B

B AB O,B Any AB

O A O Any A

O B O Any B

O AB O Any AB

AB A O,A Any AB

AB B O,B Any AB

AB O O Any AB

Rh pos Rh neg Rh neg Rh neg N/A

Rh neg Rh pos Rh pos Rh pos N/A

The Trial• Individual patient data (IPD) meta-analysis of 6

published and one unpublished cohort of donor-recipient ABO mismatched allogeneic stem cell transplants

• Primary end-point: Overall survival– OS compared among transplants with major, minor

and bi-directional mismatch

• Other end-points:– TRM– GVHD-related mortality– Engraftment or reticulocytes, neutrophils and platelets

Study Selection

• Inclusion criteria:– Original articles published in English after

1995– Study endpoints included difference in OS

between ABO matched and mismatched transplants

• Exclusion criteria– ≤80 subjects– Median follow up <6 months

11 articles eligible

Authors contacted

6 authors agreed

Kyoto University (study centre) provided previously unanalysed SCT database

Data Collection

• IPD exclusion criteria:– Patients who did not meet minimum data

requirements– SCT for conditions other than hematologic

malignancies– Cord blood grafts, combined PB and BM grafts– Known previous SCT or no info re prior SCT history

Excluded patients enrolled in other pooled cohort studies

Data Collection -2

Defined required variables (age, sex, diagnosis, stem cell source, MRD or MUD, survival status, # days survival post-BMT, donor-recipient blood types) and additional variables (HLA matching,

conditioning regimen, GVHD prophylaxis regimen, cause of death, disease status at SCT,

#days to engraftment of all blood cells)

Asked authors to complete data collection forms

Ambiguous results discussed and resolved

Patients1424 patients

133 did not meet minimum data requirements or had SCT

for diseases other than heme malignancies

28 received cord blood grafts or both PB and BM grafts

6 enrolled in other pooled studies

49 prior SCT or SCT history not known

1208 transplants697 ABO-matched

202 major ABO mismatch228 minor ABO mismatch

81 bidirectional ABO mismatch

Patients-2

• Western centres– 709 MRD– 184 MUD

• Asian centres– 214 MRD– 101 MUD

• No significant differences between matched and mis-matched groups for any category except type of donors and centres of transplantation– More MRD than MUD– Among MUD fewer HLA-mismatched

Results- Overall Survival

• Unadjusted probability of survival at 5 years:– ABO-matched: 48%– Major mismatch: 48%– Minor mismatch: 45%– Bidirectional mismatch: 37%

Results- Overall SurvivalThe Subgroups

• MRD– No difference in OS between ABO-matched

and mismatched groups • Consistent across each stratified group

• MUD– Minor and bi-directional mismatch associated

with poorer OS when adjusted for age and sex

• Strongly observed in some stratified categories: acute leukemia, SCT after 1998, Asian centres

Results- Secondary Endpoints

• TRM– Cumulative incidences of overall TRM was not

significantly different between ABO matched and unmatched groups

• Engraftment– No impact of ABO matching on engraftment in any

cell line among patients who had MRD grafts– Marginally significant impact of minor and bi-

directional mismatch on MUD grafts on delayed reticulocyte engraftment as compared to matched grafts

The Authors’ Conclusion

• “Our IPD-based meta-analysis demonstrated no adverse association between any type of ABO mismatching and survival in allogeneic SCTs for hematologic malignancies, although the possible association of minor or bi-directional ABO mismatching with lower OS was observed among recipients of unrelated SCTs”

Intro to IPD

• A Systematic Review and meta-analysis based on Individual Patient Data (IPD) involves collecting original individual patient data from trials included in the systematic review, and using this data to undertake meta-analysis.

• http://www.liv.ac.uk/medstats/ipd.htm• As of 2005 IPD represented <5% of total

meta-analysis literature

Conventional vs IPD meta-analysis

• Conventional meta-analysis extracts aggregated data – Analysis requires calculating a weighted average for

effect across trials

• Limitations:• Risk of publication bias• Heterogeneity in trial results• Inability to perform intention-to-treat analyses when relevant

data are excluded/ missing• Limited methodological quality of source studies• Sub-group analysis generally not possible

– Subgroup or meta-regression analysis using study level covariates can lead to questionable conclusions

Differences between conventional and IPD meta-analysis

• Advantages:– Enables detailed statistical analysis including time-to-

event and subgroup analysis with adjustment for important baseline differences among patients

– Enables intention-to-treat analysis • A necessary prerequisite for this is that all randomized

patients, including those excluded from trialists’ analyses, must be included

– Can use common definitions, coding and cutpoints– May be able to address questions not addressed in

original publication– Can assess adequacy of randomization – Can check data, update data and check analyses

• Limitations– Biased pooling of data

• If IPD investigators do not include ALL databases or explain why some studies were not included

• IPD investigators should also indicate how missing data was handled

– Presently no standardized method for IPD data analysis

• A review of 44 IPD analyses found that clear reporting of statistical methods was rare and little reporting on why particular methods were chosen (Simmonds et al Clinical Trials 2005)

Why would someone ever want to do this?

• According to the Cochrane Collaboration, IPD analysis is desirable when:– Reporting of trials is selective, inadequate or

ambiguous– Long term or time-to-event outcomes are

considered– More detailed analyses are planned– Subgroup effects are of primary interest

Some helpful tips from Sud and Douketis (EBM August 2009)

• Analysis should occur as per an a priori statistical plan that accounts for across study and within study variation– “Prespecifying hypotheses and analytic methods is

essential to prevent data dredging” • Important details to ascertain when looking at

IPD analyses:– Any missing studies?

• Do missing studies differ from those included in the IPD?– Any missing patients? Were statistical methods used

that account for missing patient data and variability of patient characteristics

How did our trial do?

No pre-specification of sub-group analyses Missing data

Publications in languages other than English not considered

Databases other than PubMed not checked Only 6 of 11 possible studies used Inclusion of 1 previously unanalysed data set

Insufficient detail on choice of statistical methods Missing patients

Ex: Data on (i) primary cause of death available for only 85% of patients, (ii) engraftment available for 24-55% of patients

Does IPD have a role in transfusion medicine?

• According to the Cochrane Collaboration, IPD analysis is desirable when:– Reporting of trials is selective, inadequate or ambiguous

• Off label use of rFVIIa and thrombo-embolic complications

– Long term outcomes are considered• ESAs- and risk of cancer relapse/ metastasis

– Time-to-event outcomes are considered• Red cell to plasma ratios in massive transfusion

– More detailed analyses are planned• Effect of platelet transfusion dose

– Subgroup effects are of primary interest• ICU and perio-operative transfusion triggers in special groups (ie

cardiac patients)