Peripheral blood late mixed chimerism in leucocyte subpopulations following allogeneic stem cell transplantation for childhood malignancies: does it matter?

The impact of persistent mixed chimerism (MC) after haematopoietic stem cell transplantation (HSCT) remains unclarified. We investigated the incidence of MC in peripheral blood beyond day +50 after HSCT and its impact on rejection, chronic graft‐versus‐host disease (c‐GvHD) and relapse in 161 children receiving allogeneic HSCT for haematological malignancies. The 1‐year incidence of late MC was 26%. Spontaneous conversion to complete donor chimerism (CC) occurred in 43% of patients as compared to 62% after donor lymphocyte infusions. No graft rejection occurred. The 1‐year incidence of c‐GvHD was 20 ± 7% for MC, and 18 ± 4% for CC patients (P = 0·734). The 3‐year cumulative incidence of relapse (CIR) according to chimerism status at days +50 and +100 was 22 ± 4% for CC patients vs. 22 ± 8% for MC patients (day +50; P = 0·935) and 21 ± 4% vs. 20 ± 7% (day +100; P = 0·907). Three‐year CIRs in patients with persistent MC and patients with CC/limited MC were comparable (8 ± 7% vs. 19 ± 4%; P = 0·960). HSCT for acute leukaemia or myelodysplastic syndrome as secondary malignancies (hazard ratio (HR) 4·7; P = 0·008), for AML (HR 3·0; P = 0·02) and from mismatched donors (HR 3·1; P = 0·03) were independent factors associated with relapse. Our data suggest that late MC neither protects from c‐GvHD nor does it reliably predict impending disease relapse.

In the last decade routine chimerism testing has been established in most transplant centres to verify haematopoietic engraftment but also to guide possible interventions, such as modification of immunosuppression or donor lymphocyte infusions (DLI). Well-established methods of chimerism testing of peripheral blood (PB) or bone marrow (BM) samples include polymerase chain reaction (PCR)-based investigation of variable number tandem repeats (Antin et al, 2001) or short tandem repeats (STR) (Lion, 2003;Schraml et al, 2003;Watzinger et al, 2006;Lion et al, 2012;Clark et al, 2015) for donors of identical gender, and fluorescence in situ hybridization (FISH) of sex-chromosome markers (Kogler et al, 1995;Seong et al, 2000) in case of gender mismatched donor/recipient pairs. Chimerism testing of unsorted leucocytes has a limited sensitivity with a detection limit in the range of one in one hundred cells (10 À2 ) when commonly available approaches, such as PCR amplification of microsatellite or STR markers, are applied (Preuner & Lion, 2014). Taking into account that many patients within the first months after HSCT display very low numbers of circulating lymphocytes, it is conceivable that the analysis of fluorescence-activated cell sorting (FACS)-sorted leucocyte subpopulation increases the sensitivity substantially and provides more precise information on engraftment dynamics than analysis of whole blood samples .
Mixed chimerism (MC), defined as the coexistence of recipient-and donor-derived cell populations, is frequently observed after HSCT following reduced intensity conditioning (RIC) regimens but also after myeloablative conditioning (MAC). Haematopoietic engraftment and chimerism represent a dynamic post-HSCT process, which may be influenced by various factors, such as underlying disease, intensity and type of conditioning, graft composition, viral reactivation and post-transplant immunosuppression.
Mixed chimerism has been associated with an increased risk of graft rejection in patients transplanted for severe aplastic and Fanconi anaemia, whereas no impact was reported in patients with thalassemia (Lisini et al, 2008;Lawler et al, 2009). Early post-transplant MC in CD3+ and CD3À/CD56+ cell subsets was linked particularly to graft rejection following RIC and MAC (Matthes-Martin et al, 2003;Breuer et al, 2012).
Early overall complete donor chimerism (CC) and T-cell CC were proposed to be predictive of acute and chronic graft-versus-host disease (a-GvHD and c-GvHD, respectively) (Balon et al, 2005;Park et al, 2011;Rupa-Matysek et al, 2011;Nikolousis et al, 2013;Elkaim et al, 2014), but data on the chimerism status at the occurrence of GvHD are not available.
The impact of MC in PB or BM samples on the relapse risk is discussed controversially (Bader et al, 2004a,b;Lamba et al, 2004;Doney et al, 2008;Rettinger et al, 2011;Pochon et al, 2014;Terwey et al, 2014), and an increase of recipientderived cells in leukaemia-lineage specific subpopulations was reported to predict impending relapse (Gardiner et al, 1998;Zetterquist et al, 2000;Mattsson et al, 2001;Miura et al, 2006). The majority of published observations are based on serial analyses of chimerism between day +14 to day +100 whereas data on late chimerism are scarce (Schaap et al, 2002;Stikvoort et al, 2013). The availability of serial chimerism analyses of FACS-sorted leucocyte subpopulations over a time course of several years post-HSCT prompted us to retrospectively investigate the incidence and the dynamics of late MC, defined as the presence of MC beyond day +50 and up to 12 years after HSCT. We analysed its impact on rejection, c-GvHD and relapse rates in 161 consecutive paediatric haemato-oncological patients treated with allogeneic HSCT at our paediatric centre between 2000 and 2013.

Patient and transplant characteristics
Between January 2000 and December 2013, 161 consecutive patients with haemato-oncological diseases were identified who survived HSCT beyond day +50. Thirteen of these patients underwent subsequent transplantations a median 462 d after their first HSCT (range: 198-989 d)for disease relapse (n = 12) or secondary leukaemia (n = 1) following the first HSCTand therefore were censored at the time point of the second transplantation.
The median recipient age was 11Á7 years (range: 0Á7-26Á6 years). The majority of patients was transplanted for acute lymphoblastic or myeloid leukaemia (ALL or AML, respectively) in first, second or subsequent complete remission and received MAC. A 12-Gray total body irradiation (TBI)-based regimen was administered in 86 transplantations, usually in combination with etoposide 60 mg/kg. A predominantly Busulfan-based chemotherapy was used for conditioning according to disease-specific standard European protocols in 75 HSCTs. The primary stem cell source was BM in 124 recipients, while 35 received peripheral stem cells (PBSC). The donors were human leucocyte antigen (HLA)identical/matched familial donors (MFD) in 61 cases, while 87 recipients were transplanted from matched unrelated donors (MUD), defined by HLA match in at least nine of ten loci. Thirteen patients received grafts from mismatched unrelated donors with two or more HLA mismatches or from haploidentical family donors. High-resolution HLA typing was performed at the allele level for HLA-A, -B, -C, -DR and -DQ in all donor-recipient pairs. Longitudinal linage-specific chimerism data were available for all patients included. Patient and transplant characteristics are summarized in Table I. Prophylaxis of GvHD for patients transplanted from unrelated donors consisted of antithymocyte globulin (ATG) Fresenius (Fresenius Biotech, Graefelfing, Germany; 20 mg/kg) or thymoglobulin (Genzyme; Polyclonals S.A.S., Marcy L'Etoile, France; 2Á5 mg/kg) given on three consecutive days (days À3 to À1), serum-level-adjusted ciclosporin (CyA) initiated on day À1, and a short-course of methotrexate (MTX) on days +1, +3 and +6. Patients transplanted from MFD received primarily chemoprophylaxis with CyA only, except for eight children treated with CyA plus Mycophenolate mofetil (MMF), and five children in whom a short-course of MTX was added. Eleven MFD recipients received additional serotherapy with thymoglobulin (n = 6), ATG Fresenius (n = 4) or alemtuzumab (n = 1). Three patients transplanted from haploidentical donors received muromonab-CD3 (OKT-3). MMF was administered as chemoprophylaxis in 18 patients as recommended by the respective protocols or according to institutional standards.

Immunomodulatory measures
If patients with MC were still under immunosuppression, the respective agents were reduced if no signs of active GvHD were present. The administration of DLI was considered for patients displaying MC despite discontinuation of immunosuppression. However, the administration of DLI was based on the decision of the treating physicians rather than protocol-driven.

Intervals of chimerism analysis
Routine chimerism testing on FACS-sorted PB leucocyte subpopulations before day +100 was performed at least once every 2 weeks. Thereafter, the intervals of analyses in patients with stable MC with donor levels >90% or complete donor chimerism were extended to at least every 2 months during the first year post-transplant, followed by chimerism analyses at least once a year.
Patients with increasing recipient MC were monitored more closely until stable MC or CC was documented in at least two consecutive samples.
Late MC was defined as MC after day +50, provided that the number of circulating T-cells at this time point was sufficient for cell sorting and chimerism analysis in all cell-subsets of interest.

Cell sorting and flow cytometric techniques
All chimerism analyses were performed on flow-sorted CD45+ PB leucocyte subtypes defined by their unique antigen co-expression. Myeloid cells were CD33+ monocytes and CD15+ granulocytes; T-cells were defined by their positivity for CD3+/CD4+ (helper T-cells) and CD3+/CD8+ (cytotoxic T-cells), NK-cells and B-cells were defined by their CD3À/ CD56+ and CD19+ phenotypes, respectively. Cell sorting was performed on a FACSAria instrument (BD Biosciences, San Jose, CA, USA), after eight-colour staining using a lyse-andwash cell preparation procedure as described previously (Fritsch et al, 1997). The FACSDiVa software (BD Biosciences) was used for data evaluation. All cell types exceeding 1% of nucleated cells were targets for cell sorting. The number of T-cells isolated for subsequent PCR or FISH analysis ranged between 1000 and 15 000. The purity of the flow-sorted leucocyte fractions was usually >98%.

Chimerism analysis
If patients were transplanted from gender-mismatched donors (n = 80), the samples for chimerism testing were analysed using Dual-colour FISH. Cells were dropped onto slides, air-dried and fixed with increasing concentrations of ethanol. Dual-colour FISH was performed according to standard procedures with commercially available probes specific for the centromeric and heterochromatic regions of the X and Y chromosomes, respectively. Depending on the number of cells available, up to 500 leucocytes were analysed within each sorted cell fraction.
For donor-recipient pairs with identical gender (n = 81), a quantitative PCR technique was used for chimerism testing. DNA was extracted from nucleated cells using the QIAamp Blood kit (Qiagen, Hilden, Germany). Recipient and donor DNA were tested before transplantation by a panel of highly polymorphic STR markers to select an informative primer set suitable for the monitoring of chimerism during the posttransplant course (Thiede et al, 2001;Schraml et al, 2003). The detection limit was in the range of 1%. For the analysis of patient/donor origin of cells isolated by flow sorting, the amount of DNA serving as template in individual PCR reactions was generally in the range of 1-20 ng. Upon amplification by PCR, the alleles were quantified by two different approaches: (i) gel electrophoresis and video densitometry using the Kodak Digital Science System with the 1D Image Analysis Software (Kodak, Rochester, NY, USA) and, from CD34+/kg BW 4Á15 9 10 6 /kg (0Á4 -62 9 10 6 /kg) CD3+/kg BW 41 9 10 6 /kg (0 -1920 9 10 6 /kg) ALL, acute lymphoblastic leukaemia; AML, acute myeloid leukaemia; CML, chronic myeloid leukaemia; MDS, myelodysplastic syndrome; JMML juvenile myelomonocytic leukaemia; NHL, non-Hodgkin lymphoma; GvHD, graft-versus-host disease; BW, body weight; HSCT, haematopoietic stem cell transplantation; ATG, anti-thymocyte globulin; OKT-3, muromonab-CD3; BM, bone marrow; PBSC, peripheral blood stem cells. *Including 13 patients >18 years (median 19Á6 years; range: 18Á1-26Á6 years). All but one of these patients were pre-treated in our centre and transplanted for relapsing disease. A 24-year-old patient was referred from an adult centre due to mental retardation for treatment of NHL and received HSCT at the age of 26Á6 years after relapse. †Including one sibling donor with a single HLA mismatch (9/10).

Definition of chimerism groups
Complete chimerism. In line with other investigators, complete donor chimerism (CC) was defined by the detection of >95% donor-derived cells by FISH or >95% donor-specific signals by STR-PCR on at least two consecutive analyses for every leucocyte subset (Michallet et al, 2005;Baron & Sandmaier, 2006;Stikvoort et al, 2013). All chimerism analyses were performed in sorted cell-subsets. The percentage of overall chimerism was calculated on the basis of absolute cell numbers and the respective percentages of donor-cells at time points displaying the lowest level of donor chimerism in the most prevalent cell subset.
Mixed chimerism. Mixed chimerism was defined as the presence of ≥5% recipient-derived cells in at least one cell type of the sorted leucocyte subpopulations at two consecutive time points. Patients with MC were subdivided according to 1 the time point of onset: a) patients with early MC persisting after day + 50 and b) patients with MC arising de novo after day +50; 2 the outcome of MC: a) recipients with persisting MC (defined as MC present on days +50, +100, +150 and +200) and b) recipients who lost their MC during the course of subsequent analyses; 3 the dynamics of MC: a) stable MC, b) MC with decreasing autologous cell subsets at least within two consecutive assessments, and c) MC with increasing recipient-derived cell subsets.

Definition of outcomes and GvHD
Graft rejection, chronic GvHD and disease relapse were considered the main clinical outcomes of interest for our retrospective analysis. Patients surviving without event were censored at last follow-up, with a median follow-up time of 5Á0 years (range: 0Á2-13 years). GvHD was scored and classified according to the published NIH consensus criteria (Filipovich et al, 2005).

Statistical analysis
We designed a Cox proportional hazard model with timedependent covariates to investigate the impact of the appearance of MC (modelled as time-dependent variable) on the occurrence of relapse and c-GvHD with consideration of additional risk factors (Klein et al, 2001a,b).
Cumulative incidences of relapse (CIR) were estimated according to the chimerism status on day +50, +100, +150 and +200 (excluding patients who relapsed before the respective time points or had a shorter follow-up), considering the competing risk of death. Groups were compared by the method of Gray (Gray, 1988;Fine & Gray, 1999). The date of HSCT was taken as starting point for the statistical analysis.
The evaluation of c-GvHD included 157 patients who did not relapse before day +100 and who had a follow-up period of at least 100 d after HSCT. Cumulative incidence of c-GVHD was estimated according to chimerism status at day +100 and considering the competing risks of death, relapse and second HSCT. Groups were compared using the test of Gray (Gray, 1988;Fine & Gray, 1999).
For non time-to-event variables, the Chi-Square test was used to compare groups for categorical variables. The statistical analysis was performed on the Statistical Analysis System (SAS Institute, Cary, NC, USA).

Engraftment and chimerism
All patients achieved a stable neutrophil engraftment >0Á5 9 10 9 /l between days +11 and +29 (median day +20). Eight patients had stable MC and in five patients recipient MC was increasing. In 29 patients, recipient MC decreased over time, and 24 of them converted to CC between days +84 and +322. Conversion to CC occurred spontaneously in 18 patients, after DLI in five patients (administered for increasing MC, predominantly in the T-cell compartment), and after a G-CSF-stimulated PB stem cell boost (administered for increasing MC observed in all leucocyte subsets) in one patient. In four patients, chimerism was undulating with reappearance of MC after conversion to CC. Twenty-one patients displayed on-going MC at last follow-up. The percentage of maximum overall recipient chimerism was calculated based on absolute cell numbers and the respective percentage of recipient-cells, and varied between 88% and 1% (median 6%). In 24/161 patients, overall recipient chimerism was >5%, and ≥1% in 34/161 patients. As shown in Fig 1, MC would not have been detected in eight patients on day +50, in five on day +100 and in three patients on day +150 by chimerism analysis performed in unsorted PB samples. No rejections were observed amongst these cases, but three patients relapsed (in one case despite having received pre-emptive DLI).

Patient and transplant characteristics associated with late MC
As expected, we found higher rates of late MC in patients who received RIC as compared to MAC (36% vs. 24%), although this difference was not statistically significant (P = 0Á188). Likewise, a non-significant trend towards an increased incidence of late MC was observed after TBI conditioning as compared to chemotherapy-based regimens (30% vs. 20%; P = 0Á137). Other previously reported risk factors for MC, such as recipient gender, graft source (BM vs. PBSC), number of CD34+ or CD3+ cells transplanted per kg body weight, donor type (MFD vs. MUD vs. MMD) or the application of serotherapy did not impact significantly on the incidence of late MC in our cohort (Table II).

Chimerism in cell subsets
Mixed chimerism was most frequently observed in the CD3+ T-cell compartment, which was documented in 38/42 patients (90%). Fourteen patients (33%) showed MC in the NK-cell or B-cell subsets, respectively, and in 16 recipients (38%), the myeloid compartment was affected. Approximately half of the patients (52%) displayed MC confined to the T-cell subset, whereas the remaining patients had patterns of MC involving also other cell populations, as detailed in Table III.

Donor lymphocyte infusions
A total of 10 patients received a median of one DLI (range: 1-4) between day +50 and +1095 after HSCT. Indications for DLI were: the detection of persistent MC despite modification of immunosuppression in patients at high risk of disease relapse and/or delayed engraftment of donor-derived CD3+ T-cells (n = 8), molecular relapse of chronic myeloid leukaemia (CML) 3 years after HSCT without evidence of MC (n = 1, treated successfully with DLI) and ALL relapse 160 d after HSCT in one patient, who received two DLI on days +199 and +211 in attempt to boost graft-versus-leukaemia reaction without success. Of eight patients who received between one and four DLIs for MC between days +50 and +111 (median +70), three displayed persistent MC until last follow-up despite DLI treatment, and 5/8 converted to CC between 34 and 58 d after DLI. One patient relapsed 19 d after DLI. De novo occurrence of c-GvHD was observed in 4/5 patients following conversion to CC between 50 and 168 d after the DLI administration.

Rejection
No graft rejections have been observed. Four patients (3%) experienced secondary graft dysfunction with severe neutropenia (neutrophil granulocytes <0Á5 9 10 9 /l) and hypocellular bone marrow. Chimerism analysis, however, revealed CC in 4/4 patients. All patients were treated successfully with stem cell boosts from the same donors between day +90 and +190.

Graft-versus-host disease
The evaluation of c-GvHD was restricted to 157 patients with a follow-up of at least 100 d after HSCT and absence of relapse within this time period. Forty patients (25%) developed GvHD between days +75 and +349 after transplantation. Four patients were classified as having late onset (n = 2) or recurrent late (n = 2) a-GvHD, and seven patients had persistent late a-GvHD. Grading according to the NIH consensus criteria (Filipovich et al, 2005) included mild GvHD in six patients (4%), moderate in 16 patients (10%) and severe in 18 patients (11%). Twenty-nine of 119 patients (24%) who never displayed any level of MC developed GvHD, compared to 11 of 42 patients (26%) who displayed MC in one or more cell subsets at any time point after day +50. In seven patients with MC, chronic GvHD was diagnosed between six and 155 d after conversion to CC. Five patients developed GvHD with on-going MC, and three of them had persistent MC until the last follow-up. The 1-year incidences of c-GvHD were calculated according to the chimerism status on day +100, and revealed c-GvHD incidences of 20 AE 7% in patients with MC and 18 AE 4% in patients with CC on day +100 (P = 0Á734) as shown in Fig 2. A multivariate Cox regression model, taking into account the chimerism status on day +100 and additional variables including the recipient age groups, donor types, conditioning regimen and the administration of serotherapy, revealed no significant association with the occurrence of c-GvHD in our cohort (Table IV). *Two out of the 37 patients received peripheral blood stem cells and bone marrow. †ATG Fesenius (n = 53), ATG Thymoglobulin (n = 51), Alemtuzumab (n = 1), OKT-3 (muromonab-CD3) (n = 3). ‡Data not available in five patients. §One patient with MC onset at day +438 excluded. (P = 0Á907) on day +100 (Fig 4A and B). Results obtained for days +150 and +200 were similar (data not shown).
To assess the impact of chimerism dynamics on disease relapse, we compared patients with persistent MC to patients with persistent CC or with CC following transient MC. Again, the results did not differ significantly between the subgroups, with 3-year CIRs of 8 AE 7% in patients with persistent MC and 19 AE 4% in patients with CC or with CC following transient MC (P = 0Á960) ( Fig 4C). However, due to low patient numbers, no further statistical evaluation was possible for the groups with increasing, stable or decreasing MC.
For exploratory purposes, the 3-year CIRs were estimated for 42 patients with MC on day +50 or later and for 126 patients with CC on day +50 (patients with subsequent switch to MC were censored). For both groups, the 3-year CIRs were identical (21 AE 7% for patients with MC vs. 22 AE 4% for patients with CC).

Discussion
Chimerism analyses after allogeneic stem cell transplantations have been performed since the 1970s to confirm allogeneic engraftment (Bortin et al, 1971). At present, chimerism testing is performed routinely in most centres in order to examine PB and BM at different time points post-transplant. It was shown that chimerism is not a steady state but rather a dynamic process with increasing or decreasing proportions of donor cells over time (Schaap et al, 2002). Chimerism is not only determined in nucleated cells from whole blood but is increasingly assessed in leucocyte subsets, such as lymphocyte subpopulations, monocytes or granulocytes (Antin et al, 2001;Lion, 2001;Miura et al, 2006;Breuer et al, 2012). This approach provides a higher sensitivity, especially for the T-cell compartment, which frequently has very low cells numbers in the first year post-HSCT. In our patient cohort, MC would have been missed in more than 10% of cases (e.g. eight of 35 patients at day +50) if the analyses had only been performed on unsorted material, especially within the first 200 d after HSCT (Fig 1). Consequently, one patient would not have received DLI if chimerism had been analysed in unsorted PB samples. However, DLI was not able to prevent relapse in this case. As shown earlier, results of chimerism analysis in PB are equivalent to those in BM (Rauwerdink et al, 2012). For this reason, we restricted our retrospective analysis to the chimerism results obtained from flow-sorted PB cell subsets.
Except for a trend for irradiation-containing conditioning regimens, no other risk factors for MC could be identified in our cohort. A possible explanation for the fact that RIC did not correlate with the occurrence of late MC might be because the analysis was restricted to patients with malignant diseases. The patients receiving RIC had been pre-treated with chemotherapy, and were therefore heavily immunosuppressed at the time of HSCT. Our finding, that late MC is observed more frequently following TBI, is in line with earlier reports (Minculescu et al, 2014).
For obvious immunological reasons, it was hypothesized that MC might be associated with (i) an increased risk of graft rejection (i.e. residual host T-cells rejecting the allogeneic stem cells), (ii) a decreased incidence of c-GvHD (based on the assumption that MC represents a state of tolerance) and (iii) an increased relapse incidence (due to the lack of graft-versus-leukaemia reaction). The verification of these assumptions might have a substantial impact on immunomodulatory measures, such as modification of immunosuppression or DLI.
Based on the concept that MC can be converted to CC by DLI, several prospective studies reported the inclusion of pre-emptive DLI in cases of persistent MC in order to prevent relapse (Mohamedbhai et al, 2012;Rujkijyanont et al, 2013;Horn et al, 2015). Rujkijyanont et al (2013) reported on 38 paediatric patients with malignant diseases, including seven minimal residual disease (MRD)-positive children who had received DLI for MC, mainly in the context of haploidentical transplantation. DLI was followed by conversion to CC in almost 80% of instances, but only 3/7 patients with concomitant MRD-positivity remained in remission without a second HSCT (Rujkijyanont et al, 2013). Of note, eight patients of the abovementioned cohort cleared a concomitant viraemia and converted to CC after administration of DLI, suggesting that the occurrence of MC in these patients might be attributable to the expansion of autologous virus-specific T-cells rather than impending disease relapse (Borchers et al, 2013). Conversion rates from MC to CC in response to DLI were observed in only 26% of patients with non-malignant diseases, and no randomized study has demonstrated the efficacy of pre-emptive DLI therapy in this context (Haines et al, 2015). In our patient cohort, eight patients received DLI preemptively (upon the physicians' decision) in attempts to convert MC to CC. Five of eight (63%) patients developed CC post-DLI, compared to 18/35 (51%) who converted to CC spontaneously. The high rate of spontaneous conversions to CC raises the question of whether the conversion in patients receiving DLI was attributable to this treatment.
It has been shown in HSCT for non-malignant diseases and in HSCT after RIC that overall MC is a risk factor for graft rejection (Lawler et al, 2009). We and others could show that early MC in sorted CD3+ cells, together with CD3À/CD56+ cells following RIC or MAC is associated with graft rejection (Matthes-Martin et al, 2003;Breuer et al, 2012). Moreover, in patients transplanted for non-malignant diseases, MC at later time points seems to impact on late rejection (Stikvoort et al, 2013). Late rejection is very rare following HSCT in children with malignancies, and the fact that none of our patients experienced rejection after day +50 despite MC suggests that late overall or lineage-specific MC in these patients does not represent a significant risk factor for graft rejection.
Although no data have been published regarding the incidence of c-GvHD with concomitant MC, the association of early CC with an increased risk of c-GvHD was described (Balon et al, 2005;Park et al, 2011;Rupa-Matysek et al, 2011;Nikolousis et al, 2013;Elkaim et al, 2014). This is in contrast to our findings indicating that neither the history of MC nor the presence of on-going MC seems to protect from c-GvHD. The majority of earlier studies included elderly patients with malignant diseases, in whom RIC regimens are frequently used (Rupa-Matysek et al, 2011;Nikolousis et al, 2013). It is therefore tempting to speculate that the reduced toxicity, which is associated with early MC in many cases or other hitherto unknown factors, such as chimerism of dendritic cells, have contributed to the decreased risk of c-GvHD reported in these studies, rather than MC in PB.
The question whether or not MC at any time point posttransplant indicates an increased risk of relapse has been discussed controversially during the last years (Bader et al, 2004a;Lamba et al, 2004;Doney et al, 2008;Rettinger et al, 2011;Terwey et al, 2014). Taking into account the retrospective nature of these studies and the fact that the interval between the detection of recipient cells and relapse was short in many cases, one might speculate that the time points of chimerism testing were driven by the suspicion of impending problems by individual physicians and that, in some cases, the detected autologous cells represented leukaemic blasts, which were below the detection limit of microscopy. In the patient cohort analysed in the present study, detection of MC did not correlate with increased relapse rates. This observation pertained to patients with MC at any time point, patients with a history of MC, and patients with persistent MC, indicating that chimerism status in children transplanted for malignant diseases, as assessed in the current study, did not have a significant impact on relapse incidences, in line with the findings of recently published studies (Doney et al, 2008;Nikolousis et al, 2013;Bernal et al, 2014;Pochon et al, 2014). In addition, taking into account the high incidence of c-GvHD in patients pre-emptively treated with DLI in our cohort and in other published studies, the value of chimerism-guided DLI for the prevention of relapse seems to be questionable (Horn et al, 2015). Limitations of our study include its retrospective design, the limited size of the cohort, the lack of well-defined time points for chimerism analysis and absence of clearly defined criteria for DLI. We are aware of the fact, that our approach did not include the evaluation of BM and thus CD34+ leucocyte subtypes, as the analysis was restricted to PB samples in which very little if any CD34+ cells are detectable. However, as indicated above, a bias in favour of a positive correlation between MC and relapse would be expected based on the study design. Our data suggest that testing late chimerism in paediatric patients transplanted for malignant diseases does not seem to provide a reliable tool to predict relapse, if performed within the cell subsets described in our analysis. The fact that the detection of MRD post-HSCT is cost effective and indicative of impending leukaemia relapse (Bader et al, 2015), prompted us to restrict late chimerism testing in our centre to children with malignant diseases not amenable to molecular MRD detection (e.g. AML or juvenile myelomonocytic leukaemia) and to specific, well defined prospective research tasks.
In conclusion, the clinical value of late chimerism testing in children undergoing HSCT for treatment of malignant diseases seems to be questionable with regard to the extremely low risk of graft rejection in this setting.