1 Introduction

Engraftment is defined as the first of 3 consecutive days with an absolute neutrophil count higher than 0.5 × 109/L (sustained >20 × 109/L platelets and hemoglobin >80 g/L, free of transfusion requirements).

In the setting of RIC protocols, it is also recommended to confirm the donor origin through chimerism studies.

The incidence of GF is <3–5% in the auto- and matched allo-HSCT setting, but it increases up to 10% in the cases of haploidentical or CBT.

The prognosis of graft failure (GF) is poor, and most patients die because of causes related to infections or bleeding, with an OS at 3–5 years after the diagnosis of GF less than 20%.

2 Definitions

Primary graft failure (GF)

ANC <0.5 × 109/L by day +28

Hemoglobin <80 g/L and platelets <20 × 109/L

RIC: Confirmation of donor cell origin is required

CBT: Up to day +42

Secondary GF

ANC <0.5 × 109/L after initial engraftment not related to relapse, infection, or drug toxicity

RIC: Loss of donor hematopoiesis to <5%

Poor graft function

Two or three cytopenias >2 weeks, after day +28 in the presence of donor chimerism >95%

Graft rejection

GF caused by immune rejection of donor cells mediated by host cells

3 Causes and Risk Factors

The etiology of GF is multifactorial in most of the cases (Fig. 41.1, Table 41.1).

Fig. 41.1
figure 1

Causes associated with the development of GF

Full size image
Table 41.1 Risk factors for GF
Full size table

3.1 Donor Type, HLA Matching, and Graft Source

Classical studies showed a close relationship between the degree of HLA mismatch and the incidence of GF, but it is difficult to draw conclusions because most of them used a limited HLA matching including only low-resolution A, B, and DR locus (Anasetti et al. 1989; Petersdorf et al. 2001). More recent studies, using high-resolution techniques for HLA typing and including 10–12 loci (A, B, C, DR, DQ, and DP), did not find differences in GF rates between no HLA antigen mismatch and a single HLA mismatch in both conventional MAC (Lee et al. 2007) or RIC (Passweg et al. 2011).

URD transplant was associated with a higher risk of GF (HR 1.38, p < 0.001 compared to HLA identical sibling) that was even higher when there were two or more mismatches (HR 1.79, p < 0.001) (Olsson et al. 2015).

In the haploidentical setting, the incidence of GF is around 10% which seems higher than the 3–5% currently reported MSD or URD HSCT although there are not well-designed comparative studies.

3.2 Graft Source and Cellular Content

BM is consistently associated with delayed neutrophil and platelet engraftment across all types of transplant; the impact on GF depends on donor type. GF incidence is not different for HLA MRD (Bensinger, 2012), but it is higher in the setting of URD (9% vs 3%, for BM and PB, respectively, p < 0.001) (Anasetti et al. 2012). There are no prospective randomized data either looking at MAC or RIC, but retrospective results from EBMT and CIBMTR suggest there were no differences in GF between BM and PB (less than 5% in all cases). In contrast, in a study evaluating donor characteristics, the use of BM was the only factor associated with GF after RIC (HR 2.3; p = 0.02) (Passweg et al. 2011).

The minimum cellular content required is still a matter of debate. Table 41.2 depicts a conservative proposal based on the literature review.

Table 41.2 Minimum cell content recommended
Full size table

3.3 Anti-HLA Antibodies

The presence of donor-specific anti-HLA antibodies (DSA) is associated with higher risk of GF in the context of haploidentical CBT and URD transplants, and it may in fact translate into a reduced OS (Spellman et al. 2010; Ciurea et al. 2009; Ciurea et al. 2015). The high prevalence of anti-HLA antibodies (10–40%) (Morin-Zorman et al. 2016) and the increasing use of mismatched donors prompted the EBMT to write a set of advices and recommendations on this issue (Table 41.3) (Ciurea et al. 2018).

Table 41.3 Considerations regarding the presence of anti-HLA antibodies
Full size table

3.4 Conditioning Regimen

Increasing the intensity of MAC conditioning protocols does not reduce the incidence of GF. In contrast, RIC may be associated with a higher risk.

Although it is well accepted that TBI reduces the risk of GF, there are no comparative studies that confirm this latter point. In combination with CY, the use of full-dose TBI does not seem to reduce GF in comparison with BU. The use of ATG in the preparative regimen in combination with CY seems to reduce the incidence of GF in patients with aplastic anemia. Also, in aplastic anemia patients, the addition of two Gy TBI to FLU/CY did not reduce the incidence of this complication.

3.5 Other Factors Associated with the Development of GF

ABO mismatch: Major incompatibility was associated with primary GF (HR 1.24; p = 0.012).

Cryopreservation: Associated with primary GF (HR 1.43; p = 0.013).

Female donor to male recipient: Associated with primary GF (HR 1.28; p = 0.001).

Splenomegaly: Associated with primary GF in MPN (HR 3.92; p = 0.001) and MDS (HR 2.24; p = 0.002).

Use of G-CSF: Associated with reduced risk of primary GF (HR 0.36; p < 0.001) vs no growth factors.

Underlying disease: Nonmalignant diseases are associated with higher incidence.

Previous treatments: Impairment engraftment through the damage of marrow microenvironment. The absence of treatments may induce graft rejection.

Graft manipulation: Ex vivo TCD is associated with a higher risk of primary GF in most studies.

4 Management of GF

OS after GF is consistently low, even in those patients who receive a salvage transplant; thus, the most important measures should be directed to avoid graft failure GF and to identify it as soon as possible in order to adopt the measures to revert it.

4.1 Prevention and Early Diagnosis of GF

The identification of DSA is of utmost importance in the mismatch setting. Desensitization treatment in patients at higher risk seems reasonable. Although barely supported by well-designed studies, we would probably recommend the following measures to be adopted in patients at high risk of GF: the use of PB as stem cell source, include low dose TBI and/or ATG in the conditioning regimen, consider the use of G-CSF post transplant, and a close evaluation of engraftment including marrow chimerism studies shortly after transplant (day +14). In a single-CBT study, a level of donor chimerism in BM lower than 65% was associated with a higher risk of GF (Moscardó et al. 2009); these results cannot be directly extrapolated to other types of transplant.

Olson and colleagues developed a score to predict GF in patients at risk at day +21 post-HSCT (Olsson et al. 2015): age (<30, 1 point), Karnofsky status (<90%, 1 point), disease (MDS, 1; CLL or CML, 2; and MPN, 3 points), status (advanced, 1 point), HLA matching (mismatched, 2 points), graft (BM <2.4 × 108/kg, 1 point; PB, 2 points), conditioning (no TBI, 2 points), and GVHD prophylaxis (no CNI + MTX, 2 points; TCD, 3 points). A score >6 at day +21 had a positive predictive value of 28–36%, while the negative predictive value of a score <7 was 81% for GF.

4.2 Initial Measures

It is important to apply them as soon as GF is suspected.

  • Stop as many toxic drugs as possible; treat infections; although of limited utility, it would be reasonable a trial of G-CSF.

  • Adjust post transplant IS. Maintain correct IS levels in the early post transplant period. Later on, after the third/sixth month and if mixed chimerism is present, especially after a RIC transplant, a faster tapering of IST could overcome mixed chimera (in patients with SAA, it is commonly recommended to increase IST).

  • Data regarding the use of TPO analogues after transplant are scarce, but the results of eltrombopag in aplastic anemia and its favorable toxicity profile would support, in our view, a trial with this drug before considering more complex and risky options as DLI or second transplant.

4.3 DLI and CD34 Boost

DLI could be recommended if decreasing levels of donor chimerism are observed. A careful risk/benefit evaluation is warranted as this is not a risk-free approach and a high risk of development of GVHD is anticipated.

In patients with poor graft function, the use of CD34 boost can be offered. Unfortunately, it is not clear when to perform it, but probably 2–3 months without improvement after the initial measures would be a reasonable cutoff.

4.4 Second Transplant

The limited utility and low success of cryopreserved autologous stem cells do not allow to formally recommend to perform auto-HSC harvest in any type of transplant procedure.

Results and recommendations for second allogeneic transplantation are detailed in Tables 41.4 and 41.5.

Table 41.4 Second allogeneic stem cell transplant in patients with GF
Full size table
Table 41.5 Recommendations to perform a second allogeneic HSCT as treatment for GF
Full size table

Key Points

  • Graft failure is an infrequent but often fatal complication of HSCT.

  • Etiology is complex and very frequently multifactorial.

  • Preventive measures and early identification of potential causes in order to try to revert them are the key aspects to treat it.