Relationship between graft-versus-host disease and graft-versus-leukaemia in partial T cell-depleted bone marrow transplantation


L. F. Verdonck, Department of Haematology, # G03·647, University Medical Centre Utrecht, Heidelberglaan 100, 3584 CX Utrecht, The Netherlands. E-mail: L.F.Verdonck@DIGD. AZU.NL


The success of allogeneic bone marrow transplantation (BMT) is limited by the major complications, graft-versus-host disease (GVHD) and relapse. The very beneficial effect of maximal T-cell depletion of the graft for prevention of GVHD has been counterbalanced by an increase in graft failure and relapse of disease. Therefore, we started an approach of partial T-cell depletion of the graft. GVHD and graft-versus-leukaemia (GVL) are strongly correlated after non-T cell-depleted BMT. Here, we report whether the correlation between GVHD and GVL also exists in partial T cell-depleted BMT from sibling donors. We retrospectively studied 117 adult patients with early haematological malignancies. Our method of partial T-cell depletion gave a relapse rate in patients with acute leukaemias similar to that observed in non-T cell-depleted BMT. However, patients with chronic myeloid leukaemia had a relapse rate that was similar to that observed in maximal T cell-depleted BMT. We found a significant correlation between the presence of chronic GVHD and an improved disease-free survival. Nevertheless, overall survival was lower in patients with chronic GVHD. There was no correlation between the occurrence of acute GVHD and disease-free or overall survival.

Graft-versus-host-disease (GVHD) is the major adverse complication of allogeneic bone marrow transplantation (BMT), but is also strongly correlated with the development of the beneficial graft-versus-leukaemia (GVL) effect (Weiden et al, 1981; Gale & Champlin, 1984). Donor T cells in the marrow graft play a pivotal role in both GVHD and GVL (Horowitz et al, 1990; Marmont et al, 1991). To date, it is still unclear whether GHVD and GVL are two different reactions or exert their effects through the same mechanism(s) (Bortin et al, 1979; Weiden et al, 1979, 1981; Moscovitch & Slavin, 1984; Butturini et al, 1987; Butturini & Gale, 1987; Tutschka et al, 1987; Papadopoulos et al, 1998).

The correlation between GVHD and GVL has evolved from non-T cell-depleted allogeneic BMT. After maximal T cell-depleted BMT, the intensity and frequency of both GVHD and GVL are markedly reduced (Gale & Reisner, 1986; Maraninchi et al, 1987; Apperley et al, 1988; Horowitz et al, 1990; Marmont et al, 1991). As GVL is not related to the severity of GVHD but merely to its occurrence (Sullivan et al, 1989), we initiated an approach of partial T-cell depletion aimed at reducing the severity of GVHD while conserving the beneficial effect of GVL (Verdonck et al, 1990). To investigate whether the correlation between GVHD and GVL also exists after partial T cell-depleted BMT, we retrospectively analysed these data in 117 patients with early haematological malignancies treated with this approach of marrow grafting from their sibling donors.

Patients and methods

Patients A cohort of 121 consecutive adult patients with early haematological malignancies [i.e. acute leukaemias in first remission and chronic myeloid leukaemia (CML) in first chronic phase] who received a partial-T cell-depleted allogeneic BMT from HLA-identical (n = 110) or one HLA-antigen mismatched siblings (n = 11) were studied. The study period was from January 1986 until December 1999. Patients were analysed as of 1 January 2000. Patients needed to survive at least 2 months post transplant to be included in the study. Four patients died within this period and were excluded from the study (117 patients were therefore included). Except for AML-M3, all adults patients with acute myeloid leukaemia (AML) and acute lymphoblastic leukaemia (ALL) in first complete remission (CR) were included. However, since 1993, patients with ‘good-risk AML’: t[8;21], t[15;17], M3, or inv[16], were not transplanted in first CR. The characteristics of the patients are shown in Table I. All patients were treated with protocols approved by the local Investigational Review Board and gave informed consent. Engraftment was achieved in all patients and no graft rejections were observed.

Table I.   Characteristics of all patients receiving an allogeneic BMT.
Sex of recipient
Age of recipient(years) median,range40 (19–62)
 AML in first remission45
 ALL in first remission34
 CML in first chronic phase38

Bone marrow collection, T-cell depletion and transplantation Donor bone marrow cells were aspirated from the iliac crest under general anaesthesia in 104 cases and 13 patients received peripheral blood stem cells. T cells were removed by soybean lectin (SBA) agglutination and sheep red blood cell (SRBC)-rosette formation. After this maximal T-cell depletion, the number of T cells in the graft were counted and the required number of T cells (set apart at the start of the manipulation procedure) were reinfused to the graft to obtain the fixed number of 1 × 105 T cells/kg bodyweight in the graft, as described before (Verdonck et al, 1990). From January 1997, SBA plus SRBCs was replaced by the immunorosette technique with CD2 and CD3 monoclonal antibodies [because of good manufacturing practice (GMP) developments]. The T cell-depletion approach, however, remained similar. Peripheral blood stem cells (PBSCs) were depleted of T cells with the CD34 column (Cellpro Ceprate System). The latter technique, however, resulted in a small increase of T cells (median 1·5, range 1·0–3·0 × 105 T cells/kg) in the graft. The partial T cell-depleted stem cell graft was infused through a central venous access.

Preparative regimen and GVHD prophylaxis plus treatment The conditioning regimen consisted of cyclophosphamide (60 mg/kg) infused on each of two successive days followed by total body radiation (TBI; 10 or 12 Gy) in two fractions. The irradiation was delivered by a 10-MV linear accelerator. BM infusion occurred on the second day of TBI.

All patients were hospitalized in conventional single rooms with reversed isolation. They received infection prophylaxis with oral ciprofloxacin and amphotericin B, together with oral acyclovir and cotrimoxazole (Verdonck et al, 1990). During the first 10 d after transplant, intravenous cephalotin was given. Furthermore, parenteral alimentation and semisterile food was administered. Infection prophylaxis, parenteral alimentation, semisterile food and the reversed isolation were stopped when granulocyte counts were > 0·5 × 109/l. Prophylaxis with cotrimoxazole and acyclovir was continued for 12 months after BMT. For GVHD prophylaxis, patients received a short course of cyclosporine, starting with 3 mg/kg/d intravenously by continuous infusion the day before BM infusion and continued for 28 d, after which it was given orally. Generally, cyclosporine was stopped within 3 months after transplantation. Patients with grade II and higher acute GVHD were treated with prednisone, 1 mg/kg orally twice a day for at least 14 d; thereafter, the prednisone dose was lowered and stopped, if possible, after 21 d. Chronic GVHD was treated with systemic corticosteroids, sometimes combined with cyclosporine (Sullivan et al, 1988). Since June 1989, patients with latent cytomegalovirus (CMV) infection (CMV-seropositive, 22 patients) received ganciclovir (2·5 mg/kg i.v. twice a day for 14 d) when the CMV infection reactivated or when they were treated with high-dose corticosteroids during the first 4 months after BMT (Verdonck et al, 1997). Patients receiving a one antigen HLA-mismatched graft received rabbit anti-thymocyte globulin (4 mg/kg/d for 5 d) preceding the cyclophosphamide treatment.

Diagnosis and evaluation of GVHD Acute GVHD (grades 0–IV) and chronic GVHD (limited or extensive) were diagnosed clinically, confirmed pathologically by skin or mucosal biopsy, and classified according to standard criteria (Thomas et al, 1975). Chronic GVHD was defined if GVHD was present on d 90.

Response of disease CR was defined as the absence of signs and symptoms of leukaemia, the normalization of blood counts and bone marrow cellularity, and the absence of cytogenetic abnormalities. For CML, CR included absence of the Philadelphia chromosome on all metaphases and interphases analysed using fluorescence in situ hybridization (FISH) (at least 30 metaphases and 400 interphases were examined) and/or a negative result for polymerase chain reaction assays for BCR-ABL mRNA (sensitivity of 10−4 in our laboratory). Haematological response (HR) for CML was defined as normalization of leucocyte counts, with no immature forms, and platelet counts below 450 × 109/l, and disappearance of splenomegaly.

Relapse of leukaemia was defined as the recurrence of leukaemia cells including cytogenetic recurrence. In CML, relapse was defined as the cytogenetic recurrence of the Philadelphia chromosome with or without haematological relapse.

Statistical analysis The probabilities of survival and disease-free survival from acute and chronic GVHD were estimated from the day of transplant (d 0) using the method of Kaplan and Meier (Kaplan & Meier, 1958). The log-rank test was used to compare survival outcomes.



Of 117 patients, 86 (74%) patients developed acute GVHD at a median 18 d (range 11–63 d) after transplant; 35 patients had grade I, 48 patients had grade II, one patient had grade III and two patients had grade IV GVHD.

Of the 100 patients surviving more than 90 d, 37 (37%) patients developed chronic GVHD. Chronic GVHD was graded as limited in 19 (19%) patients and as extensive in 18 (18%) patients. Chronic GVHD occurred after preceding acute GVHD in 31 patients and developed de novo in six patients. Twenty-five patients (21%) did not develop acute or chronic GVHD.

Relapse, disease-free and overall survival

Median follow-up of all patients was 26 months (range, 2–163 months). Thirty-four out of 117 (29%) patients relapsed, nine out of 45 (20%) patients with AML, eight out of 34 (24%) patients with ALL, and 17 out of 38 (45%) patients with CML. The probability of relapse for patients with AML was 30% (CI 95: 16–51%), for ALL 29% (CI 95: 15–50%) and for CML 70% (CI95: 47–89%). Of all 117 patients, 73 (62%) were alive and well at a median follow-up of 45 months (range 2–163 months), including 25 out of 45 (56%) patients with AML, 23 out of 34 (68%) patients with ALL, and 25 out of 38 (66%) patients with CML.

The disease-free survival and overall survival after BMT for AML, ALL and CML are shown in Fig 1A and B. Patients with AML and ALL had a similar disease-free survival, but patients having CML had far more relapses than those with acute leukaemias, giving a statistically significant worse disease-free survival rate (P = 0·017). However, 17 patients with relapsed CML were treated with donor leucocyte infusion (DLI) and 14 out of 17 (82%) achieved a second complete remission. The overall survival rates were not different for CML, ALL or AML. The disease-free and overall survival after BMT for all patients with regard to different grades of acute GVHD are shown in Fig 2A and B. No significant differences were seen between the grade of acute GVHD and the disease-free or overall survival. The effect of chronic GVHD on the disease-free and overall survival is shown in Fig 3A and B. Chronic GVHD significantly improved the disease-free survival (P = 0·03), however, overall survival was lower (P = 0·02) in those with chronic GVHD because of the toxicity of extensive chronic GVHD.

Figure 1.

 Disease-free (A) and overall survival (B) after partial T cell-depleted BMT for AML, ALL and CML.

Figure 2.

 Disease-free (A) and overall survival (B) after partial T cell-depleted BMT for acute GVHD (aGVHD).

Figure 3.

 Disease-free (A) and overall survival (B) after partial T cell-depleted BMT for chronic GVHD (cGVHD).

Causes of death

Infection was the major cause of death. Seventeen patients died of infection. Furthermore, nine patients died of relapse, nine patients of GVHD (with secondary complications) and nine of other causes.


Allogeneic BMT in humans is hampered by immune-associated complications, such as GVHD and graft rejection. It has been clearly demonstrated that mature donor T cells in the graft are the primary mediators of GVHD (Apperley et al, 1988; Horowitz et al, 1990) but, on the other hand, relapse of the disease is diminished by the occurrence of GVHD. However, only patients without severe GVHD have a survival advantage (Weiden et al, 1981). It has been shown that maximal T-cell depletion of the marrow graft results in a relapse rate that is about three times higher than observed in non-T cell-depleted marrow grafting from HLA-identical sibling donors (Marmont et al, 1991). Also, non-engraftment or graft rejection, a highly unusual complication after non-T cell-depleted BMT from HLA-identical siblings (incidence < 1%) has markedly increased, up to 15%, following maximal T cell-depleted HLA-matched BMT (Gale & Reisner, 1986; Maraninchi et al, 1987; Marmont et al, 1991).

Our approach of partial T-cell depletion instead of maximal T-cell depletion was initiated because of the major drawbacks of the latter: graft failure and relapse. We hypothesized that the addback of unmanipulated donor T cells to the graft might be critical to avoid these complications. To date, our approach of partial T-cell depletion in HLA-identical siblings using a fixed number of 1 × 105 T cells/kg in the BM graft has proved very effective in the prevention of severe acute or chronic GVHD and, furthermore, no graft rejections have occurred in any of these patients.

To investigate whether the correlation between GVHD and GVL also exists after partial T cell-depleted BMT, we retrospectively analysed the effect of partial T-cell depletion in 117 consecutive adult patients with early leukaemias. Acute GVHD was observed in 74% of the patients. The incidence but not the severity of acute GVHD was similar to that observed in patients receiving non-T cell-depleted BMT (Atkinson et al, 1990). Chronic GVHD was observed in 37% of the patients. The incidence of chronic GVHD may be lower then generally observed in non-T cell-depleted BMT (about 50%; Atkinson et al, 1990). We found a significant effect between the presence of chronic GVHD and an improved disease-free survival. This effect was not found for the presence of acute GVHD. However, the overall survival for all patients was decreased by the occurrence of chronic GVHD, because of the fact that the patients with extensive chronic GVHD had a higher mortality rate.

The relapse rates of 20% and 24%, respectively, for patients being transplanted for AML and ALL in this study, were similar to those observed in non-T cell-depleted BMT (Marmont et al, 1991). Furthermore, it is important to note that all the surviving patients had a Karnofsky index of 90–100%, which will not be the case after non-T cell-depleted BMT. However, almost half (45%) the patients with CML relapsed. All CML patients who relapsed received DLI and 14 out of 17 patients (82%) achieved a second CR. So the survival of CML patients, eventually, might also be similar to non-T cell-depleted BMT.

We conclude that chronic GVHD has a beneficial effect on the disease-free survival in patients treated with partial T cell-depleted BMT for early leukaemias. This effect, however, is not present for acute GVHD. The beneficial effect of chronic GVHD on disease-free survival was not translated into a better overall survival; in contrast, it was even worse for those with chronic GVHD. So, the very effective GVL effect occurring after non-T cell-depleted BMT is only present for chronic GVHD after partial T cell-depleted BMT. Furthermore, partial T-cell depletion has a similar relapse rate to maximal T-cell depletion in patients with CML. For patients with acute leukaemias in CR, the approach of partial T-cell depletion is an appropriate one with regard to disease-free survival, overall survival and quality of life.