Bone Marrow Transplantation, (1997) 19, 927–932  1997 Stockton Press All rights reserved 0268–3369/97 $12.00

Donor lymphocyte infusions (DLI) in patients with chronic myeloid leukemia following allogeneic bone marrow transplantation

A Bacigalupo1, M Soracco1, F Vassallo1, M Abate1, MT Van Lint1, F Gualandi1, T Lamparelli1, D Occhini1, N Mordini1, S Bregante1, O Figari1, F Benvenuto1, M Sessarego2, G Fugazza2, P Carlier3 and M Valbonesi3

1Divisione Ematologia II, Ospedale San Martino, Genova; 2Dipartimento Medicina Interna, Universita Genova; and 3Centro Trasfusionale Ospedale San Martino, Genova, Italy

Summary: and discontinuation of infusions before life-threatening GVHD has developed. Donor lymphocyte infusions (DLI) were given between Keywords: chronic myeloid leukemia; graft-versus- June 1990 and March 1996 to 18 patients with chronic leukemia; donor lymphocyte infusions myeloid leukemia (CML) for the treatment of cytog- following (12 ؍ or hematologic relapse (n (6 ؍ enetic (n an allogeneic bone marrow transplant (BMT). Patients Graft-versus-leukemia (GVL) is thought to play an (8 ؍ were divided in two groups: patients in group A (n received a large dose of donor lymphocytes important role in preventing relapse in patients undergoing allogeneic bone marrow transplants (BMT).1,2 Although (10 ؍ у1 ؋ 108/kg), whereas patients in group B (n) received escalating numbers of cells (2 ؋ 105 up to GVL was described in animals almost 20 years ago3,4 little ؋ 108/kg). The median number of DLI in group A was is known about the magnitude of this effect on effector cells 2 2 (range 1–3); the median number of infusions in group and target antigens, at least in humans.5 In the experimental B was 7 (range 3–9). Acute GVHD occurred in 12 animal, cells separately mediating GVL or graft-versus-host patients (grades I–III) and was a major cause of death disease (GVHD) have been described and can be identified in two. The risk of developing GVHD correlated with by surface markers.3,6 In humans we have indirect evidence the number of cells infused: 37%, 14%, 5% and 0% for GVL: programs involving T cell depletion expose the 8 7 6 for DLI with cells у1 ؋ 10 ,2؋10 /kg, 2 ؋ 10 /kg, and 2 5 patients to a high risk of leukemic relapse, and this has Median transamin- now been shown in very large numbers of patients.1,2 .(0.01 ؍ ؋ 10 /kg, respectively (P 2 ase levels were found to be significantly increased in If mature donor T cells infused on the day of marrow patients with, as compared to patients without, acute transplant (an average of 6 × 107/kg) have such an GVHD (GPT 412 vs 28 IU/l; P 0.03). Severe aplasia -important role in preventing relapse, the use of large num ؍ occurred in four and was a contributing cause of death bers of donor T cells which have undergone maturation in in two patients. Overall, four patients died as a conse- the donor thymus, would possibly be helpful in treating quence of DLI and all received Ͼ1 ؋ 108/kg cells: the relapse. Indeed this has been shown to be the case, actuarial risk was 38% in group A and 14% in group especially in patients with chronic myeloid leukemia There were 10 complete and three partial .(0.1؍B(P (CML), both for unfractionated7–17 and CD8-depleted cytogenetic responses: the actuarial probability at 5 18 years of being Ph negative was 69%: it was 46% for cells. The longest The exact mechanism by which the leukemic clone can .(0.1 ؍ group A and 85% for group B (P patient is now 6 years post-DLI, Ph negative, BCR-ABL be reduced is unknown. GVHD has a deleterious effect on 19,20 negative. The actuarial 3 year survival is 38% in group hematopoietic cells and it is quite possible that the so- The study confirms called GVL effect seen in humans is an allo-effect mediated .(0.06 ؍ A and 86% in group B (P that DLI post-BMT is not innocuous and that there is a by donor T cells recognizing minor differences between definite long-lasting antileukemic effect in patients with donor and recipient:7 this would lead to cytotoxicity against CML. It also suggests that: (1) the risk of developing recipient leukemic cells, and would spare donor hemo- GVHD correlates with the number of infused cells; (2) poietic progenitors. The consequence would be induction that significant elevations of serum GPT levels are asso- of remission, also maintained over time, suggesting ciated with GVHD; and (3) that the use of escalating engraftment and expansion of infused donor T cells. doses of cells may allow the identification of side-effects In the present study we report 18 patients who were given donor lymphocyte infusions (DLI) either for prophy- laxis or treatment of leukemia relapse following allogeneic BMT: we describe two groups of patients, the first given Correspondence: Dr A Bacigalupo, Divisione Ematologia 2, Ospedale San large numbers of donor cells and the second given escalat- Martino, 16132 Genova, Italy ing, programmed numbers at intervals of 21 days. Received 3 September 1996; accepted 26 December 1996 Donor lymphocyte infusions in CML after BMT A Bacigalupo et al 928 Materials and methods further processing. In a second group of 10 patients, donor cells were cryopreserved in liquid nitrogen in aliquots of Patients 20–50 × 106 and then administered in escalating doses every 21 days: the program was designed to give DLI All patients were transplanted from an HLA-identical sib- 2 × 105/kg on days 1 and 22, then 2 × 106/kg cells on days ling between 1984 and 1994 in our Unit. Clinical details 43 and 64, 2 × 107/kg on days 85 and 106, 2 × 108/kg on are outlined in Table 1: 12 had hematologic relapse days 127 and 148. If acute GVHD developed the program (chronic phase n = 5; accelerated phase n = 6; blastic trans- was stopped. formation n = 1) and six had cytogenetic relapse at the time of DLI. Statistical analysis Transplant procedures ␹2 and Student’s t-tests were used to analyze the data. Conditioning therapy consisted of cyclophosphamide Kaplan–Meier plots were used to analyze survival data. (120 mg/kg) followed by a 2-day rest and fractionated total body irradiation (TBI) (3.3 Gy/day on each of 3 consecu- tive days. Patients receiving T cell-depleted marrow were prepared with single dose TBI 7.5 Gy on day −7, thiotepa Results 5 mg/kg on day −6, Campath 1G 20 mg i.v. on days −5, − − − − − − 4, 3, 2 and 1, CY 50 mg/kg on days 3 and 2 and Donor lymphocyte infusions (DLI) marrow treated with Campath 1M. Campath 1G and Cam- path 1M were kindly provided by Prof H Waldman and Dr DLI were given between days +182 and days +2545 post- G Hale, Cambridge, UK. BMT and between 3 and 2231 days after relapse (Table 1). There were no immediate reactions, except in one patient receiving cryopreserved cells, who had a serious allergic GVHD prophylaxis reaction with respiratory problems, which resolved on high- Patients receiving unmanipulated marrow were all given dose intravenous prednisolone. The first group of patients cyclosporin A (CsA) with methotrexate. Patients receiving (n = 8, group A) received the entire leukapheresis product T cell-depleted marrow were not given in vivo immuno- (range 1–2 × 108 cells/kg): the median number of infusions suppression post-BMT. in this group was two (range 1–3). The second group of patients (n = 10) received a median number of seven infusions (range 3–9). In this group, patients were designed Donor lymphocyte infusions (DLI) to receive escalating doses of DLI starting from 2 × 105/kg Donors were the same as donated marrow. They underwent up to 2 × 108/kg: three patients received the programmed leukapheresis on a cell separator. In a first group of eight eight infusions, six patients stopped at 2 × 107 cells/kg and patients, the entire leukapheresis cells were infused without one patient at 2 × 106 cells/kg because of acute GVHD.

Table 1 Clinical data of patients

UPN Diagnosis Sex Age Disease status at BMT Relapse Interval BMT–Relapse TCD Interval relapse–DLI days days

112 CML M 27 Acc HEM Acc 314 Yes 2231 294 CML F 23 CP CYTO 730 Yes 1192 318 CML F 43 CP CYTO 731 Yes 1011 343 CML F 36 Acc HEM Acc 334 Yes 944 414 CML M 46 CP HEM Acc 121 Yes 1007 539 CML M 49 CP HEM CP 365 Yes 159 568 CML F 47 Acc HEM CP 153 Yes 55 664 CML M 51 Acc CYTO 167 Yes 15

580 CML M 29 BT HEM Acc 1061 No 3 787 CML M 44 Acc HEM Acc 225 No 223 675 CML M 35 CP CYTO 181 Yes 417 696 CML M 41 CP HEM CP 283 Yes 206 665 CML F 36 Acc HEM CP 229 Yes 294 618 CML F 35 CP CYTO 789 No 357 660 CML M 46 CP CYTO 828 Yes 101 739 CML F 41 Acc HEM Acc 728 Yes 10 693 CML F 51 CP HEM CP 748 Yes 214 774 CML M 34 CP HEM BT 365 No 81

BMT = bone marrow transplantation; Acc = accelerated; CP = chronic phase; BT = blastic transformation; TCD = T cells depletion; DLI = donor lympho- cytes infusion; CML = chronic myeloid leukemia; HEM = hematological; CYTO = cytogenetic. Donor lymphocyte infusions in CML after BMT A Bacigalupo et al 929 Acute GVHD Mortality due to DLI

The actuarial incidence of acute GVHD was 67%. GVHD Overall, four patients died as a consequence of DLI and all developed in 7/19 (37%) infusions of у1 × 108 cells/kg, received у1 × 108/kg cells: three deaths occurred in group in 4/28 (14%) infusions of 2 × 107 cells/kg, in 1/21 (5%) A (eight patients) and one in group B (10 patients). The infusions of 1 × 106 cells/kg and in 0/12 infusions of actuarial risk of treatment-related mortality (TRM) was 2 × 105/kg (P = 0.01) (Figure 1). GVHD was scored as 38% in group A and 14% in group B (P = 0.1). grade I, II, III respectively in four, four, four (Table 2). All patients were off CsA at the time of DLI, but six had to be started on CsA and/or prednisolone. Death due to Graft-versus-leukemia (GVL) aGVHD was seen in three patients (two in group A): all of these had received a dose of cells equal to or exceeding An antileukemic effect was seen in the 18 CML patients × 8 1 10 /kg. with 10 complete and three partial cytogenetic and/or hem- atologic responses; four patients did not respond, one patient was unevaluable because of severe aplasia which Transaminase levels prevented cytogenetic analysis. Of the 12 hematologic relapses six (50%) had a complete response, compared to GOT and GPT levels were scored and found to be elevated 4/6 (66%) complete responses in patients with cytogenetic in patients with aGVHD: the median maximum GPT level relapse (P = 0.4 in Fisher’s exact test). The median duration (IU/ml) was 412 (range 35–1860) for patients with aGVHD of response is 18 months (range 8–72). The actuarial prob- vs 28 (range 13–54) for patients without aGVHD ability of becoming Ph negative was 69% at 6 years, with (P = 0.03). Similarly, of the nine patients with elevated GPT levels, eight had acute GVHD. a median interval of 10 months between the first DLI and cytogenetic response (Figure 2).

Pancytopenia Overall survival Severe aplasia occurred in four patients and was a contribu- ting cause of death in two (all in group A). Pancytopenia The overall survival of the whole group was 52% at 60 was severe in all four patients and lasted 2, 3, 4 and 8 months; 86% in group B and 38% in group A (P = 0.06) months, respectively. The last patient was dependent on (Figure 3). The median follow-up for surviving patients is platelet and red blood cell transfusions for 8 months and 960 days (range 433–2221). The longest survivor is over 6 then slowly showed hematologic recovery: she is now 1 years post-DLI: this patient relapsed 334 days after unman- year after recovery with normal blood counts and is Ph ipulated BMT, was given one single DLI in July 1990, and negative. is now BCR-ABL negative (March 1996).

Table 2 Results of DLI

UPN Schedule aGVHD Response Side-effects Status A/D Cause of Disease status Days FU from of DLI post-DLI % Ph neg death at last FU DLI

112 High dose I No Aplasia D Infec 186 294 High dose 0 100% No A Ph− 2035 318 High dose III 90% Aplasia D Infec + GVHD 170 343 High dose 0 100% No A Ph− 2282 414 High dose 0 No No D Leuk 1181 539 High dose I 5% No A Ph+ 1985 568 High dose I No No D Leuk 559 664 High dose III 25% IP D GVHD + IP 81

580 Escalating 0 100% Aplasia A Ph− 1122 787 Escalating 0 No No A Ph+ 402 675 Escalating III 100% Pericarditis D GVHD Ph− 469 696 Escalating III 100% Malabsorption A Ph− 917 665 Escalating II 100% Aplasia A Ph− 1047 618 Escalating I 100% No A Ph− 759 660 Escalating II 100% Oral GVHD A Ph− 669 739 Escalating II 100% No A Ph− 418 693 Escalating II 100% GVHD A Ph− 465 774 Escalating 0 No No A Ph+ 484 aGVHD = acute graft-versus-host disease; Ph = Philadelphia chromosome; A = alive; D = dead; Infec = infection; Leuk = leukemia; IP = interstitial pneu- monia; FU = follow-up. Donor lymphocyte infusions in CML after BMT A Bacigalupo et al 930 40 100 37% 90 B = 86% 80 30 70 60 P = 0.01 50 20 40 A = 38%

14% % Surviving 30 20 10 P = 0.06 10 % of patients with aGVHD 5%

0 0 8 16 24 32 40 48 0 108 107 106 105 Months (median) from DLI Number of cells infused/kg Figure 3 Actuarial probability of 4-year survival in eight patients receiv- Figure 1 Incidence of acute graft-versus-host disease (GVHD) grade II ing single high-dose DLI (group A) (38%) and in 10 patients receiving or more: it was recorded in 37% of infusions containing 1 × 108 cells/kg, escalating doses of DLI (group B) (86%) (P = 0.06). 14% of the infusions with 2 × 107/kg cells and 5% with 2 × 106/kg cells and none with 2 × 105/kg cells. The difference is statistically significant (P = 0.01). a complete response. This is an important point because of the large number of patients who could be treated with DLI: close to 700 CML patients are being allografted every year 100 in Europe:21 if 50% survive the transplant and 25% of these 90 B = 85% relapse, there will have been approximately 800 relapses in the past 10 years. A large number of these patients may 80 still be alive, and would thus be eligible for DLI, with a 70 change of success in the order of 50%. It is interesting that 60 one of our CML patients was successfully treated 1192 days after relapse and is now Ph negative. In the future 50 we can expect to treat an increasing number of cytogenetic 40 A = 46% relapses with an even higher success rate. Despite convincing evidence in animals of strong GVL 30 3,4,6 % of patients Ph neg activity in acute leukemia, clinical data, so far reported, 20 suggest a significantly higher chance of cure in patients P = 0.1 with CML as compared to patients with acute leukemia.16 10 The reasons for this difference are unknown: possible 0 explanations include a preferential expression on CML 0 4 8 12 16 20 24 cells of potential target antigens, such as minor or major Months (median) from DLI histocompatibility antigens. If GVL is an allo-effect, as suggested by the association with GVHD, then cells not Figure 2 Actuarial probability of becoming Ph negative in eight patients expressing allo-antigens would be more likely to escape receiving single high-dose DLI (group A) (46%) and in 10 patients receiv- ing escalating doses of DLI (group B) (85%) (P = 0.1). killing. It is interesting that selecting for DR-positive cells has been shown to enrich for CML precursors.22 Cytotoxic T cell clones specific to recipient alloantigens have been Discussion demonstrated in patients with GVHD:23 it is very important to prove their presence in patients responding to DLI. A In this study we have confirmed that excellent results can second hypothesis for the different effect of DLI in CML be achieved using donor lymphocytes to treat CML relapse would be a smaller number of clonogenic cells in this dis- after allogeneic BMT.7 We have also shown that: (1) the ease, following recent studies indicating that the majority risk of developing acute GVHD correlates with the number of early precursors (LTC-IC) in patients presenting with of infused cells; (2) that the use of escalating doses of cells CML, are Philadelphia chromosome negative.24 There is may allow the identification of side-effects and discontinu- one additional difference: patients with CML are often ation of infusions; and (3) that significant elevations of given DLI with or after interferon, and this may well have serum GPT levels are associated with GVHD and may be an immune-modulating effect, although the European com- used to monitor alloreactivity. bined data did not suggest a significant impact of IFN on As to the antileukemic effect, this was very clear in response.16 patients with cytogenetic relapse and, most interestingly, The anti-leukemia effect of DLI is often associated with also in patients with hematologic relapse, in keeping with GVHD, and the latter can be life-threatening. In the first the European data.16 One patient had hematologic relapse group of eight patients receiving the entire leukapheresis with an additional chromosomal abnormality, and still had product, three died, two with cytopenia, one with over- Donor lymphocyte infusions in CML after BMT A Bacigalupo et al 931 whelming acute GVHD: similar side-effects have been mentation of graft-versus-host reactivity. Nature 1979; 281: described.7 Due to this experience we started a program of 490–491. DLI with escalating numbers of cells. 4 Slavin S, Weiss L, Morecki S et al. Eradication of murine Patients were designed to receive eight DLI starting at a leukemia with istoincompatible marrow grafts in mice con- low dose (2 × 105/kg) escalating up to a high dose of cells ditioned with total lymphoid irradiation (TLI). Cancer Immu- nol Immunother 1981; 11: 155–159. × 8 (2 10 /kg). Only three patients completed the scheduled 5 Barret AJ, Malkovska V. Graft-versus-leukemia: understand- infusions, whereas seven had the program discontinued at ing and using the alloimmune response to treat haematological 7 6 2 × 10 /kg or 2 × 10 /kg cells due to acute GVHD: the latter malignancies. Br J Haematol 1996; 93: 754–761. patients still achieved complete and durable Philadelphia 6 Truitt RL, Atasoylu AA. Contribution of CD4+ and CD8+ negativity. In this group, one patient died as a consequence T cells to graft-versus-host disease and graft-versus-leukemia of DLI, and he had received 1 × 108 cells/kg. The overall reactivity after transplantation of MHC compatible bone mar- mortality for patients receiving cells у1 × 108/kg is 6/11 row. Bone Marrow Transplant 1991; 8: 51–58. vs 0/7 for patients receiving lower doses. We take these 7 Kolb HJ, Mittermuller J, Clemm Ch et al. Donor leukocyte results to indicate that the infusion of cells equal or greater transfusions for treatment of recurrent chronic myelogenous than 1 × 108/kg exposes recipients to a significant risk of leukemia in marrow transplant patients. Blood 1990; 12: 2462–2465. death, which may be unnecessary, since complete responses 8 Porter DL, Roth MS, McGarigle C et al. Induction of graft- were seen in CML patients with lower doses. The actuarial versus-host disease as immunotherapy for relapsed chronic probability of becoming Ph negative was 85% in patients myeloid leukemia. New Engl J Med 1994; 330: 100–106. given escalating doses of DLI, which is superior, though 9 Bar BMAM, Schattemberg A, Mensink EJBM et al. Donor not significantly so, to the 46% probability of group A. lymphocyte infusions for chronic myeloid leukemia relapsed An elevation of transaminases often preceded GVHD: after allogeneic bone marrow transplantation. J Clin Oncol this was proven by significantly higher GPT levels in 1993; 11: 513–519. patients with, as compared to patients without, GVHD 10 Helg C, Roux E, Beris P et al. Adoptive immunotherapy for (P = 0.03). Although increased transaminase levels have recurrent CML after BMT. Bone Marrow Transplant 1993; been described in acute GVHD,25 they are not considered 12: 125–129. 11 Drobyski WR, Keever CA, Roth MS et al. Salvage immuno- a hallmark of liver GVHD, which is, indeed, graded by 25 therapy using donor leukocyte infusions as treatment for bilirubin levels. In our transplant practice looking at 541 relapsed chronic myelogenous leukemia after bone marrow patients, median levels of GPT on day +100 after BMT transplantation. Efficacy and toxicity of a defined T-cell dose. were 26, 34, 40, 45, 131 IU/l, when stratified for acute Blood 1993; 82: 2310–2318. GVHD grade O, I, II, III, IV, respectively (unpublished 12 Jiang YZ, Cullis JO, Kanfer EJ et al. T cell and NK cell data), and median maximum GPT levels for patients with mediated graft-versus-leukemia reactivity following donor or without GVHD were 133 and 194 (P = 0.2). Therefore, buffy coat transfusion to treat relapse after marrow transplan- it would appear that increased GPT levels after DLI largely tation for chronic myeloid leukemia. Bone Marrow Transplant exceed variations seen early after transplant: whether this 1993; 11: 133–138. is due to the large number of infused donor lymphocytes 13 Hertenstein B, Wiesneth M, Novotny J et al. Interferon alpha and donor buffy coat transfusions for treatment of relapsed can not be ascertained, but it does appear that GPT may chronic myeloid leukemia after allogeneic marrow transplan- be helpful to mark acute GVHD. tation. Transplantation 1993; 56: 1114 (Abstr.). In conclusion, donor lymphocyte infusion is an effective 14 Szer J, Grigg GL, Philips et al. Donor leukocyte infusion after and durable treatment of post-BMT relapse for CML: it chemotherapy for patients relapsing with acute leukemia fol- is not innocuous, especially if high doses of lymphocytes lowing allogeneic BMT. Bone Marrow Transplant 1993; 11: (у1 × 108/kg) are used. One may wish to use escalating 109–111. numbers of cells and monitor GPT levels as an early sign 15 Slavin S, Naparstek N, Nagler A et al. Allogeneic cell therapy of alloreactivity. with donor peripheral blood cells and recombinant human interleukin-2 to treat leukemia relapse after allogeneic bone marrow transplantation. Blood 1996; 87: 2195–2204. 16 Kolb HJ, Schattenberg A, Goldman JM et al. Graft-versus- Acknowledgements leukemia effect of donor lymphocyte transfusions in marrow grafted patients. Blood 1995; 86: 2041–2050. This work was supported by the Associazione Ricerca Trapianto 17 van Rhee F, Lin F, Cullis JO et al. Relapse of chronic myeloid Midollo Osseo (ARITMO), Genova and Associazione Italiana leukemia after allogeneic bone marrow transplant: the case for Ricerca contro il Cancro (AIRC), Milano. giving donor leucocyte transfusions before the onset of hema- tologic relapse. Blood 1994; 83: 3377–3383. 18 Nimer SD, Giorgi J, Gajewski JL et al. Selective depletion of References CD8+ cells for prevention of graft-versus-host disease after bone marrow transplantation: a randomized controlled trial. 1 Horowitz MM, Gale PB, Sondel PM et al. Graft-versus-leuke- Transplantation 1994; 57: 82–88. mia reactions after bone marrow transplantation. Blood 1990; 19 Anderson KC, Weinstein HJ. Transfusion-associated graft- 75: 555–562. versus-host disease. New Engl J Med 1990; 323: 315–321; 2 Marmont AM, Gale RP, Butturini A et al. T-cell depletion in published erratum appears in New Engl J Med 1990; 323: allogeneic bone marrow transplantation: progress and prob- 1360. lems. Haematologica 1989; 74: 235–248. 20 van Dijken PJ, Wimperis J, Crawford JM et al. Effect of graft- 3 Bortin MM, Truitt RL, Rimm AA et al. Graft-versus-leu- versus-host disease on hematopoiesis after bone marrow trans- kaemia reactivity induced by alloimmunization without aug- plantation in mice. Blood 1991; 78: 2773–2779. Donor lymphocyte infusions in CML after BMT A Bacigalupo et al 932 21 Gratwohl A, Hermans J, Baldomero H. Hematopoietic precur- poietic cells (LTC-IC) are present in Philadelphia negative sor cell transplants in Europe: activity in 1994. Report from cytaphereses collected during early recovery after chemo- the European Group for Blood and Marrow Transplantation therapy for chronic myeloid leukemia (CML). Bone Marrow (EBMT). Bone Marrow Transplant 1995; 17: 137–148. Transplant 1996; 16: 549–555. 22 Verfaille CM, Miller WJ, Boylan K et al. Selection of benign 25 Deeg HJ, Cottler-Fox M. Clinical spectrum and pathophysiol- leukemia on the basis of HLA-DR antigen expression. Blood ogy of acute graft-vs-host disease. In: Burakoff SJ, Deeg HJ, 1992; 79: 1003–1010. Ferrara J, Atkinson K (eds). Graft-vs-Host Disease. Immu- 23 Goulmy E. Minor histocompatibility antigens and their role in nology, Pathophysiology, and Treatment. Marcel Dekker: transplantation. Transplant Rev 1988; 2: 29–53. New York, Basel, 1990. 24 Podesta M, Piaggio G, Frassoni F et al. Very primitive hemo-