Early Emergence of PNH-Like T Cells After Allogeneic Stem Cell Transplants Utilising CAMPATH-1H for T Cell Depletion

Early emergence of PNH-like T cells after allogeneic stem cell transplants utilising CAMPATH-1H for T cell depletion

RJ Garland1, SJ Groves1, P Diamanti1,4, SE West1, KL Winship1, PF Virgo2, SP Robinson3, A Oakhill3, JM Cornish3, DH Pamphilon4, DI Marks3, NJ Goulden1,3 and CG Steward1,3

1Department of Pathology and Microbiology, University of Bristol, University Walk, Bristol, UK; 2Department of Immunology and Immunogenetics, North Bristol NHS Trust, Bristol, UK; 3Bone Marrow Transplant Unit, Royal Hospital for Children, Bristol, UK; and 4Institute for Transfusion Sciences, National Blood Service, Bristol, UK

Summary: to T cell depletion (TCD) in the SCT setting.5–7 The CD52- protein lacks a transmembrane region and is tethered to cell CAMPATH-1H (C-1H)is widely used in vivo and/or membranes via a glycosylphosphatidyl-inositol (GPI) anchor.8 in vitro for T cell depletion in hematopoietic SCT. This GPI-anchored proteins are present on many blood cell types, humanised monoclonal antibody is specific for CD52, a including CD55 and CD59 on erythrocytes, CD16 on marker coexpressed on the majority of human lymphocytes neutrophils, and CD48, CD55 and CD59 on lymphocytes.9 with CD48 and other glycosylphosphatidyl-inositol (GPI) Several studies of patients in nontransplant settings anchored proteins. We detected CD52/CD48 dual expres- (rheumatoid arthritis (RA), B cell CLL and B cell non- sion on 499% of CD3 þ lymphocytes from normal Hodgkin’s lymphoma (NHL)) have shown that C-1H individuals and all 15 post-SCT patients whose transplants treatment can result in the emergence of T cells lacking did not utilise C-1H. By contrast, 23/26 patients with expression of CD52 and that of other GPI-anchored transplants involving C-1H (in vivo, in vitro or both) proteins.10–13 These cells resemble those found in paroxysmal exhibited populations lacking CD52 expression that nocturnal haemoglobinuria (PNH), an acquired clonal stem accounted for 49.7% (4.2–86.2%)of the CD3 þ lympho- cell disorder characterised by chronic intravascular haemo- cytes (median and range)in samples evaluated at a median lysis and bone marrow failure.14 In PNH there is a total or of 2 months post-SCT. Most CD52À cells also lacked partial deficiency of GPI-anchored membrane proteins, CD48 expression. These GPIÀ T cells were of either donor with blood cell subsets affected to varying extents.9,15 The or mixed donor/recipient origin. They were predominant in involvement of lymphoid cells is less well recognised than the early months after SCT at times of profound that of myeloid lineages, but GPI-deficient T cells have been lymphopenia and inversely correlated with the recovery shown to comprise 0.2–30% of the T-cell compartment in a of the absolute lymphocyte count (r ¼À0.663, Po0.0001). high proportion of PNH patients (84%). Similar proportions The presence of CD52À cells has been correlated of GPI-deficient NK and B cells were also observed.16–18 previously with clinical outcome after CAMPATH ther- Little is known about the role of GPI-anchored proteins apy for both malignant and nonmalignant diseases. on lymphocytes, despite CD52 being present at a level of Bone Marrow Transplantation (2005) 36, 237–244. approximately 500 000 molecules per cell,5,19 although doi:10.1038/sj.bmt.1705049; published online 13 June 2005 defects in their expression do not appear to affect activation Keywords: CAMPATH-1H; alemtuzumab; PNH; immune or proliferation in vitro.10,17 However, CD52À cell emergence reconstitution; CD52 after C-1H therapy may have clinical relevance: in patients treated for RA, the presence of CD52À T cells was correlated with prolonged remission while in CLL a role has been þ þ 10,13 CAMPATH-1H (C-1H) is a humanised IgG1 antibody, which postulated for CD8 CD52 cells in controlling relapse. has lower immunogenicity and prolonged half-life after We therefore decided to monitor the presence of these cells infusion compared to the rat antibody (CAMPATH-1G) following SCT where C-1H was used for prophylaxis of from which it is derived.1–4 Its target antigen is CD52, which is GVHD and graft rejection, in comparison to a control expressed on virtually all lymphocytes but not stem cells; a population who did not receive C-1H. Serial sampling distribution pattern that allows the application of this antibody allowed us to examine the persistence and, in some cases, the donor/recipient haemopoietic origin of these cells.

Materials and methods Correspondence: Dr RJ Garland, Department of Pathology and Microbiology, School of Medical Sciences, University of Bristol, University Walk, Bristol BS8 1TD, UK; Patient cohort and sample collection E-mail: [email protected] Received 14 February 2005; accepted 22 April 2005; published online 13 A cohort of 41 patients (23 adults and 18 children) June 2005 undergoing allogeneic SCT for haematological malignan- PNH-like T cells after transplants involving CAMPATH RJ Garland et al 238 cies in Bristol between April 2002 and April 2004 met the cytes. This minimised the risk of misleading percentages following criteria: (1) consent to take part in research study derived from low cell numbers. Absolute lymphocyte E5237 approved by the Local Research Ethical Committee counts (ALC) were obtained from the Haematology of the United Bristol Healthcare Trust; (2) at least one Laboratory of the United Bristol Healthcare Trust and peripheral blood sample available between engraftment and absolute numbers of CD3 þ subsets derived by multiplying 4 months post-SCT in which CD3 þ cells exceeded 10% of the ALC by the percentage expression of CD3 on gated the lymphocyte population and (3) greater than 1000 gated lymphocytes. lymphocytes on flow cytometry. The stem cell source was BM in 21 cases and PBSC in 20. Donors and patients were matched for 10 antigens (HLA-A, B, C, DR and DQ) by Immunomagnetic sorting of cells PCR with sequence-specific primers to the allelic level. Donors were matched siblings (21 cases) or unrelated Some patient samples were subjected to immunomagnetic þ À donors (13 matched and seven mismatched). In 26 cases the cell sorting in order to separate CD3 CD52 and þ þ patients or their stem cell donations were treated with C-1H CD3 CD52 cells. This was performed using anti-human (Schering AG, Berlin, Germany). For in vivo conditioning, CD3 Dynal bead selection followed by a two-step selection this included a range of daily dosing schedules within the for CD52 using CAMPATH (unconjugated rat anti human range 0.2–0.3 mg/kg or 10–30 mg over 2–5 days, resulting in CD52, Serotec) and sheep anti-rat immunoglobulin an overall median total dose of 48.9 mg (range 11.4– Dynal beads (Dynal, Oslo, Norway), according to the 102.0 mg overall, 9.9–54.9 mg for children and 42.0–90.0 mg manufacturer’s instructions. The fractions were analysed for adults). TCD of the donor material involved addition of by flow cytometry to confirm the purity of each C-1H in vitro (‘in the bag’, five cases) or CD34 selection of subset, using antibodies to CD4 (FITC, Beckman stem cells prior to infusion in seven cases (by CliniMACS Coulter, High Wycombe, UK), CD8 (RPE-Cy5, Serotec) selection, Miltenyi Biotech). TCD by C-1H in vitro and CD52. consisted of addition of 20 mg to the final product followed by incubation at 221C for 30 min. The product was then infused within 30 min following premedication with methyl- Chimeric analysis prednisolone, paracetamol and chlorpheniramine. Where Chimeric analysis was performed as described pre- CD34 selection was performed, T cell addbacks of 5 Â 23,24 4 6 viously. Briefly, DNA was extracted using the Qiamp 10 –10 /kg were infused along with the TCD donor DNA mini kit (Qiagen, Hilden, Germany), according to the material. The policies of our unit for prophylaxis against manufacturer’s instructions and DNA quantification per- GVHD and cytomegalovirus infection have been detailed formed by spectrophotometry. Donor and recipient DNA 20–22 previously. Patients were serially monitored for were amplified in a multiplex reaction containing primer immune reconstitution parameters and compared to a pairs for 16 polymorphic STR markers (Powerplex 16 panel of normal blood donors. system, Promega, Madison, WI, USA). The allelic lengths and relative positions of donor and recipient alleles Flow cytometry determined the suitability of each marker for chimerism evaluation. Post transplant chimerism was determined Staining of whole blood was performed using CD3-RPE using the beaded samples in single locus reactions using Cy5, CD48-FITC and CD52-PE (all from Serotec, Oxford, primers for a minimum of two selected informative markers UK) or isotype controls obtained from Serotec (RPE Cy5 (Sigma, Poole, UK). Fluorescent data were collected on the and FITC) or Immunotech (PE) (Luton, UK). Antibody ABI Prism 3100 Genetic analyser using the collection incubation was performed for 25 min. Lysis solution (BD software and analysed with Genescan Software (Applied Biosciences, Oxford, UK) was then added and the samples Biosystems, Warrington, UK). The relative percentages of vortexed and incubated for 10 min. All incubations were donor and recipient DNA present in sorted cell populations performed in the dark and at room temperature. Samples were determined.24 This analysis was applied to sorted cell were then pelleted and washed once with wash buffer, populations (CD3 þ CD52À and CD3 þ CD52 þ ) obtained consisting of cold calcium/magnesium-free PBS (Gibco, from peripheral blood. Life Technologies, Paisley, Scotland, UK) supplemented with 0.1% sodium azide (BDH, Poole, UK). Cells were fixed using wash buffer supplemented with 1% parafor- Statistical analysis maldehyde (BDH). Samples were stored at 41C and flow cytometric analysis performed within 24 h using a FACS The lymphocyte subset constitution between treatment Calibur Flow Cytometer and Cell Quest Software (BD groups and normal individuals were compared using the Biosciences). A population of gated lymphocytes, as Kruskal–Wallis test. Where differences were identified, determined by forward vs side scatter properties, were these were further examined using one-way analysis of analysed for cell surface expression of CD3, CD48 and variance with post hoc test. When only two groups were CD52. However, many patients were lymphopenic in the compared, the Mann–Whitney test was used. In order to early post transplant period, so these results were restricted assess whether the proportion of GPI-deficient T cells to those patients from whom at least one sample was correlated with the ALC or CD3 þ counts the bivariate available within 4 months of transplantation in which the Pearson coefficient was calculated (SPSS 12.0 for Windows, CD3 þ population accounted for 410% of the lympho- Apache software foundation).

Bone Marrow Transplantation PNH-like T cells after transplants involving CAMPATH RJ Garland et al 239 Results possible, 10 000 gated lymphocytes were acquired. This was dictated by patients’ ALC and fell in the range 1000–10 000 CD52 expression on T cells from normal individuals and (median 7000), while for normals the range of values was SCT recipients the same, with a median of 10 000. The proportion of CD3 þ T lymphocytes expressing CD52 Expression of CD48 and CD52 on T cells was examined in a panel of 22 normal blood donors. Only 0.05% (0–1.0%) of CD3 þ cells failed to express CD52 CD52 is normally expressed on lymphocytes with other (Figure 1; Table 1). In total, 41 SCT recipients were then GPI-anchored proteins such as CD48. In order to be examined, 26 with and 15 without C-1H utilisation in the conﬁdent that any differences in CD52 expression were due transplant regimen. Of the 26 patients with transplants to a true global defect in GPI-anchored proteins (rather involving C-1H, 23 showed greater percentages of CD3 þ than the tail of a normal distribution curve for CD52 lymphocytes lacking CD52 expression than controls expression), three colour staining was performed to (Figure 2a and b). These cells were present at differing monitor the expression of CD3, CD48 and CD52.9 In the levels between patients: at the time of the ﬁrst evaluable panel of normal individuals, 99.9% (99.3–100%) of CD3 þ sample (0.6–3.9 months post-SCT), CD52À cells accounted cells were positive for both CD48 and CD52 (Figure 3; for 47.8% (0.9–86.2%) of the CD3 þ lymphocytes. Three of Table 1). The patient group with SCTs not involving C-1H the 26 patients receiving SCT involving C-1H and all 15 (C-1H untreated GPI þ ) also exhibited this pattern (99.7% patients with transplants not involving C-1H showed a of CD3 þ cells were positive for both CD52 and CD48). pattern of CD52 expression on CD3 þ lymphocytes, which In contrast, CD48 staining of CD3 þ T cells in the was indistinguishable from that observed in normal C-1H treated GPIÀ group revealed a population individuals. Based on C-1H treatment and whether of CD52ÀCD48À cells in addition to the normal CD52À T cells were detected, patients were assigned to CD52 þ CD48 þ T cells. The latter accounted for a median one of three groups designated (i) C-1H untreated GPI þ of 39.1% of the T cell population in these patients although (n ¼ 15), (ii) C-1H treated GPIÀ (n ¼ 23) and (iii) C-1H the relative proportions varied between individuals (range treated GPI þ (n ¼ 3, Table 1). The SCT characteristics of 0.4–82.6%, Table 1, Figure 4a and b) and over time (Figure these groups of patients are displayed in Table 2. Where 4b–e). Thus, these patients exhibit a population of T cells

104 104 104 0.0 64.5 0.01 72.57 0.05 77.96

103 103 103

102 102 102

0.5 1.02 26.40 0.7 CD3 RPE-CY5 CD3 RPE-CY5 35.0 CD3 RPE-CY5 101 101 101

21.29 100 100 100 0 1 2 3 4 10 10 10 10 10 100 101 102 103 104 100 101 102 103 104 CD52 PE CD52 PE CD52 PE Figure 1 Expression of CD3 and CD52 on normal lymphocytes. Whole blood staining of lymphocytes from a panel of 22 normal individuals conﬁrmed the high expression level of CD52 on CD3 þ lymphocytes and rare occurrence of CD52À T cells. Three representative individuals are displayed.

Table 1 Characterisation of CD52 and CD48 expression in normals and SCT patients

Group (n) Time of ﬁrst CD3+a CD52Àa CD52+a CD52+/CD48+a CD52À/CD48Àa sample (months) (as % of lymphocytes) (as % of CD3+ cells)

Normals (22) N/A 67.1 (39.1–80.8) 0.05 (0–1.0) 99.9 (98.9–100) 99.9 (99.3–100) 0 (0–0.05) C-1H untreated GPI+ (15) 1.0 (0.6–1.6) 61.9 (12.7–85.4) 0.2 (0–2.7) 99.8 (97.3–100) 99.7 (98.9–99.7) 0 (0–0.02) C-1H treated GPI+ (3) 1.4 (1.2–1.6) 22.1 (13.7–75.7) 0.3 (0.2–0.4) 99.7 (98.7–99.8) ND ND C-1H treated GPIÀ (23) 2.0 (0.6–3.9) 22.0 (12.4–83.4) 47.8 (0.9–86.2) 52.3 (13.9–99.1) 39.1 (3.2–99.0) 50.0 (0.4–82.6)

Patterns of CD52 and CD48 expression on CD3 cells from normal individuals (n ¼ 22) and SCT patients (n ¼ 41). Patients were categorised as to whether their transplant regime included C-1H or not and by whether they had evidence of a population of CD3+CD52ÀCD48À T cells (GPIÀ) or not (GPI+). All data represented are median (and range) values. ND ¼ not done. aStatistical analysis (excluding the C-1H treated GPI+ group) showed signiﬁcant differences between C-1H untreated GPI+ and C-1H treated GPIÀ (Po0.0001) but not between C-1H untreated GPI+ and normal individuals (P40.1 for CD3+% and P40.9 for other cell population readouts) for each of these parameters.

Bone Marrow Transplantation PNH-like T cells after transplants involving CAMPATH RJ Garland et al 240 Table 2 SCT characteristics of patient groups a 104 5.54 12.49 C-1H C-1H C-1H untreated treated treated GPI+ GPIÀ GPI+ 3 10 Age (o18 : 418 years) 6 : 9 11 : 12 1 : 2 Sex (male : female) 10 : 5 13 : 10 2 : 1

Disease 2 ALL 5 7 0 10 AML 4 7 0 CML 1 2 2

CD3 RPE-CY5 Multiple myeloma 3 0 0 5.2 76.77 Other neoplastic 2 7 1 1 10 Stem cell source BM : PBSC 8 : 7 12 : 11 1 : 2

T cell depletion 0 None (D or R) 10 0 0 10 C-1H involved 0 23 3 100 101 102 103 104 C-1H in vitro only 0 2 0 CD52 PE C-1H in vivo only 0 16 2 C-1H in vivo+CD34 selection 0 2 1 b 104 C-1H in vivo and in vitro 030 31.21 6.45 CD34 selection only 5 0 0 (+T cell addback/kg) 5/5 (5 Â 104–106)0 0

Conditioning 103 TBI yes/no 10/5 15/8 2/1

Donor MSIB : MMUD : MUD 14/1/0 5/6/12 2/0/1

102 CMV status R+/D+ 6 5 0 R+/DÀ 3112 CD3 RPE-CY5 0.63 61.72 RÀ/D+ 0 3 1 RÀ/DÀ 640 101 Total 15 23 3

MMUD ¼ HLA mismatched unrelated donor; MSIB ¼ HLA matched sibling donor; MUD ¼ HLA matched unrelated donor. 100 exhibiting PNH-like T cells included 18 treated with C-1H 0 1 2 103 104 10 10 10 in vivo only (two of whom received CD34 selected grafts), CD52 PE two treated with C-1H ex vivo alone (UPN 14 and 15) and a Figure 2 Populations of CD3 þ lymphocytes lacking CD52 are observed further three who received C-1H treated donations in in most C-1H treated patients. Flow cytometric proﬁles of CD3 and CD52 addition to in vivo C-1H (UPN 5, 6 and 11). This group þ À staining lymphocytes in two SCT patients show populations CD3 CD52 consisted of 11 children ( 18 years) and 12 adults, and 12 cells in addition to the CD3 þ CD52 þ cells observed in normal individuals o (Figure 1). Displayed are UPNs 2 C-1H in vivo,(a) and 14 C-1H in vitro, BM and 11 PBSC transplants and included ﬁve matched (b) at 1.7 and 2.7 months post-SCT respectively. sibling and 18 unrelated donor transplants (six of which were mismatched).

lacking GPI-anchored proteins, which can be described as Persistence of CD52À T cells in C-1H treated patients GPI-deficient or ‘PNH-like’. The C-1H treated GPIÀ group had a significantly higher proportion of CD3 þ CD52À and Of the 23 patients demonstrating GPI-deficient T cells CD3 þ CD52ÀCD48À cells than both normal individuals (UPNs 1–7, 11–12 and 14–15), 11 had more than five blood and the C-1H untreated GPI þ patients (Table 1, samples taken (median 8, range 5–15) over a period of 9.5 Po0.0001). No significant differences in any of these (3.5–19.3) months post-SCT, which allowed monitoring of parameters were observed between C-1H untreated GPI þ the persistence of these cells. Of these patients, six were patients and normal individuals (P40.1). The C-1H treated in vivo C-1H treated alone (UPN 1–4, 7 and 12), three GPIÀ group also showed a lower proportion of CD3 þ received both in vivo and ex vivo C-1H (UPN 5, 6 and 11) lymphocytes at the time of first evaluable sampling (22.0%) and two received ex vivo but not in vivo C-1H (UPN compared to 67.1% in normals and 61.9% in C-1H 14 and 15). untreated GPI þ patients (Po0.0001; Table 1). Figure 5 shows the persistence pattern of GPI-deficient The SCT characteristics of the C-1H treated GPIÀ CD3 þ CD52À cells as a percentage of the CD3 þ cell subset patients are displayed in Table 3. This group of 23 patients for these 11 patients. The majority of patients treated with

Bone Marrow Transplantation PNH-like T cells after transplants involving CAMPATH RJ Garland et al 241 104 104 104 0 100 0.00 99.99 0.04 99.88

103 103 103

102 102 102 CD48 FITC CD48 FITC CD48 FITC 101 00101 0.01 0.00 101 0.04 0.04

100 100 100 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 CD52 PE CD52 PE CD52 PE

Figure 3 CD3 þ CD52 þ lymphocytes from normal individuals also express CD48. Staining samples from a panel of normal individuals (as described in Figure 1) conﬁrmed that the CD52-expressing T cells were also CD48 þ . Three representative individuals are displayed. CD48ÀCD52À T cells are rarely observed.

a 104 b 104 c 104 0.10 70.60 0.63 14.12 0.61 31.50

103 103 103

102 102 102 28.47 82.60 65.62 CD48 FITC CD48 FITC CD48 FITC 1 1 2.27 10 0.87 10 2.65 101

100 100 100 100 101 102 103 104 100 101 102 103 104 100 101 102 103 104 CD52 PE CD52 PE CD52 PE

d 104 e 104 0.98 68.37 0.65 92.10

103 103

102 102 24.89 6.99 CD48 FITC CD48 FITC 0.65 101 5.76 101

100 100 100 101 102 103 104 100 101 102 103 104 CD52 PE CD52 PE

Figure 4 Association between CD52 and CD48 expression on CD3 þ cells in C-1H treated SCT patients. Flow cytometric proﬁles of CD52 and CD48 expression patterns on CD3 þ lymphocytes in two SCT patients. Displayed are representative UPNs 2 (a) and 14 (b) at 1.7 and 2.7 months post-SCT respectively. Further evaluation of UPN 14 at 5.7 (c), 8.7 (d) and 12.5 (e) months post-SCT revealed the proﬁle of relative proportions of GPI þ and GPIÀ CD3 þ cell populations over time.

C-1H in vivo possessed CD52À populations (expressed as a Po0.0001) or CD3 þ counts (r ¼À0.518, n ¼ 61, Po0.0001). proportion of the CD3 þ population), which declined over Only one patient showed stable absolute counts of CD3 þ time. The rate of decline varied between patients, ranging CD48ÀCD52À cells (UPN 4, data not shown). from those in whom CD52À T cells were only observed in the ﬁrst sample to others with levels of 410% at 410 þ À months post-SCT (UPN 4, 5 and 14). The proportion Origin of CD3 CD52 cells of CD3 þ CD52À cells after in vivo C-1H treatment was T cell chimerism was assessed as part of the clinical inversely correlated with ALC (r ¼À0.663, n ¼ 56, management of a subset of the patients on one or two

Bone Marrow Transplantation PNH-like T cells after transplants involving CAMPATH RJ Garland et al 242 Table 3 SCT characteristics of C-1H treated GPIÀ patients

UPN Age at SCT (years) Donor match Conditioning Stem cell source Diagnosis T-cell depletion R/D

1 8.5 MUD CYC/TBI BM Ph+ ALL C-1H/none 2 19.2 UD MEL/CARM/VP16/ARA-C BM HD C-1H/none 3 16 MMUD CYC/TBI PBSC CML C-1H/none 4 52.2 MSIB FLUD/MEL PBSC CMML C-1H/none 5 44.3 MMUD CYC/TBI BM AML C-1H/ITB 6 28 MUD ETOP/TBI PBSC T ALL C-1H/ITB 7 43 MSIB MEL/CARM/VP16/ARA-C PBSC NHL C-1H/none 8 6.1 MMUD CYC/TBI BM AML C-1H/none 9 3.9 MUD CYC/TBI BM ALL C-1H/none 10 6.7 MUD CYC/TBI BM LL C-1H/none 11 24.4 UD CYC/TBI BM CML C-1H/ITB 12 52.7 MSIB FLUD/MEL PBSC NHL C-1H/none 13 6.8 MUD CYC/TBI PBSC ALL C-1H/CD34 14 43.6 MSIB CYC/TBI PBSC AML None/ITB 15 58.9 MSIB CYC/TBI PBSC AML None/ITB 16 12 MMUD CYC/TBI PBSC Ph+ ALL C-1H/CD34 17 24.9 MUD MEL/CARM/VP16/ARA-C PBSC HD C-1H/none 18 10.9 MUD CYC/TBI BM ALL C-1H/none 19 5.3 MMUD CYC/TBI BM ALL C-1H/none 20 11.9 MUD CYC/TBI BM AML C-1H/none 21 4.9 MUD CYC/TBI BM AML C-1H/none 22 46.4 MUD CYC/MEL/CARM/VP16/ARA-C PBSC TCL C-1H/none 23 40.2 MUD FLUD/MEL BM AML C-1H/none

Bu ¼ Busulphan; Carm ¼ carmustine (BCNU); C-1H ¼ CAMPATH-1H; CD34 ¼ CD34 selected SCT; CMML ¼ chronic myelomonocytic leukaemia; FLUD ¼ Fludarabine; HD ¼ Hodgkin’s disease; ITB ¼ In the bag (ex vivo); LL ¼ lymphoblastic lymphoma; MEL ¼ Melphalan; MMUD ¼ HLA mismatched unrelated donor; MSIB ¼ HLA matched sibling donor; MUD ¼ HLA matched unrelated donor; NHL ¼ non-Hodgkin’s lymphoma; Ph+ ALL ¼ Philadelphia positive ALL; TCL ¼ T-cell lymphoma; UD ¼ unrelated donor; VP16 ¼ Etoposide.

occasions. The results are shown in Table 4. On occasions, 100 patients 12, 14 and 15 showed majority donor CD3 C-1H in vivo

cells) 90 C-1H in vitro À + C-1H in vivo and in vitro chimerism (86–100%) at times when CD52 cells con- 80 stituted 61.8–84.6% of their CD3 population, implying that 70 À CD52 cells were largely or completely of donor origin. 60 UPN 12 had in vivo C-1H depletion and UPN 14 and 15 50 had ex vivo depletion. 40

In several patients, at times of substantial mixed chimerism, cells (% of CD3 – 30 CD52À and CD52 þ fractions of peripheral blood CD3 cells 20

were isolated by magnetic cell sorting and subjected to CD52 10 chimeric analysis. Sorted populations of UPN 12 were both + þ À 0 50% donor in origin. In UPN 4 CD3 CD52 cells were 71% CD3 010155 þ þ donor and CD3 CD52 T cells were 44% donor. Thus, Time post SCT (months) both donor and recipient haemopoiesis were contributing to the development of CD52À T cells in these patients. Figure 5 CD3 þ CD52À cells tend to decline over time as a proportion of the CD3 þ lymphocyte subset. Serial samples demonstrated the persistence pattern of CD3 þ CD52À cells as a proportion of the CD3 þ subset after CD52 expression on cells other than T cells SCT involving C-1H in vivo, in vitro or both. Displayed are 11 patients with Expression of CD52 on B cells and NK cells was analysed in at least 3.5 months follow-up. a pooled fashion by investigating CD3À lymphocytes. À CD52 lymphocytes comprised 12.8% (1.9–51.5%) of 5,19 À blood lymphocytes, the density being higher on T cells CD3 lymphocytes in normals but only 6% (0.5–44.2%) 7 in C-1H treated GPIÀ patients. Unlike CD3 þ lymphocytes, than B or NK cells. In PNH the presence of GPI-deficient clones is well documented within erythroid and myeloid it was not possible to distinguish any effect of C-1H in the 9,16 generation of GPI-deficient CD3À lymphocytes. Further- cells. T cells are, however, also affected: GPI-deficient more, staining of GPI-anchored proteins on red cells (CD55 cells (negative for CD48 and CD59) comprised 0.2–30% of the T-cell compartment in 84 and 100% of two series of and CD59), monocytes (CD14) and neutrophils (CD16) did 16,18 not show any GPIÀ cell populations (data not shown). patients with PNH. These abnormal T cells have been shown to persist in two patients with PNH (for more than 10 and 25 years, respectively).25 Discussion GPI-deficient T cells appear to be rare in normal individuals. They were not observed by conventional flow CD52 is expressed at high density (approximately 5 Â 105 cytometry in two studies,16,25 but have been reported in antibody binding sites per cell) on over 95% of peripheral another study where a selection step (use of aerolysin which

Bone Marrow Transplantation PNH-like T cells after transplants involving CAMPATH RJ Garland et al 243 Table 4 Chimerism status in patients post-SCT at times of CD52À presence of cells bearing PIG-A mutation at very low T-cell occurrence frequency and thought to have predated C-1H therapy. By UPN CD3+CD52À Donor CD3 Time post-SCT contrast, in patients treated for RA using C-1H, PIG-A (%) chimerism (%) (months) mutations were excluded in clones derived from those who developed CD52À T cells.30 We have not tested whether the 3 25 100 3.6 CD52À cells arising post-C-1H therapy in SCT harbour 4 30 38 3.5 4 30.4 58 14.7 mutations of the PIG-A gene, but consider this to be one 5 6.1 100 7.6 important avenue of future investigation. 11 6.4 97 5.2 The potential clinical relevance of CD52À cells is 12 84.6 100 1.2 suggested by previous papers from different clinical 12 44.2 50 4.9 scenarios.13 In C-1H-treated RA patients continued disease 14 61.8 91 5.7 14 73.7 86 4.3 remission was not correlated with CD4 depletion, as 15 67.1 100 0.9 many patients relapsed despite persisting CD4 lymphopenia.10,31 However, the three patients (from 25) who had CD52À T cells showed the longest disease remission in the series. Osterborg et al also noted that two CLL lyses GPI þ cells) preceded flow cytometry: Ware et al26 patients who demonstrated clonal expansion within the found the frequency of phenotypically GPIÀ T cells in CD8 þ CD52À population had sustained remission and normals to be 17.8 (5–59.6) per million PBMC. We found hypothesised a regulatory role in controlling B cell tumour GPI-deficient T cells in 6/15 normals in whom triple growth. staining was performed on 10 000 gated lymphocytes at the It could also be argued that CD52À host T cells could levels of 0.01–0.05% of the CD3 þ population, but no such drive rejection processes by evading C-1H therapy. For cells in the other 9/15. example, one previous case report (after use of CAM- The presence of high proportions of CD52À T cells has, PATH-1G during conditioning therapy) found an expan- however, been reported following treatment for RA,10 sion of recipient CD3 þ CD8 þ CD52À T cells at the time of CLL/NHL13 and CLL.11 We now report the finding of graft rejection.32 These cells were thought to be ‘ADCC- GPI-deficient cells (accounting for 0.4–82.6% (median of resistant’ due to their lack of CD52 expression. This was 50%) of the T cells) in 85% of patients receiving C-1H not the mechanism by which CD52À T cells arose in the either as part of conditioning therapy and/or as in vitro present study as no cases of rejection were observed, these TCD, but not in other SCT patients tested. The proportion cells generally declined over time and, where studied, they of these cells declined over time in inverse proportion to the were of donor or mixed donor/recipient rather than absolute lymphocyte and CD3 counts, but CD52À cells recipient origin. were present in some patients to beyond 10 months post- It seems likeliest that the presence of CD52À T cells acts SCT. This skewing towards CD52 negativity in comparison as an inverse surrogate for immune reconstitution. We to normals was not observed in CD3À lymphocytes and we hypothesised that those showing high percentages of these could not find GPIÀ erythroid or myeloid cells. cells would have proliferation of the few T cell clones, Just one patient showed a significant rise in the which have survived the presence of C-1H. This would proportion of CD52À cells following first detection: from certainly be consistent with the inverse correlation observed 16.6% of CD3 þ cells at 1.4 months to 65.7% at 3.5 between the proportion of CD52À cells and ALC. If so, months. Relapse of T-ALL occurred 6 weeks after the last these patients might be expected to show high rates of observation and it is entirely possible that a CD52À clone(s) infection. Our data did agree with recent observations of a was responsible although this could not be tested. This high incidence of CMV reactivation following in vivo use would have been in keeping with other case reports where of C-1H:33 this occurred in four of six at risk patients four patients relapsed with CD52À leukaemia having (ie patient, donor or both CMV seropositive). However, no presented with CD52 þ T-cell prolymphocytic leukaemia correlation was apparent between high proportions of (n ¼ 3) and B cell ALL (n ¼ 1), respectively.27–29 CD52À cells and risk of CMV reactivation or other The high proportion of C-1H-treated patients exhibiting infections in this study. The power of these observations CD52À cells agrees well with findings after therapy for CLL is greatly limited by the size of the study and the small (10/15 patients had CD52À cells) but stands in contrast to number of patients who lacked CD52À cells in the C-1H- the findings of Brett et al in RA patients. This latter group treated group. documented the emergence of CD52À T cells in 3/25 In conclusion, this study showed that GPIÀ T cells patients being treated for RA despite generally higher appeared after in vitro or in vivo use of C-1H in SCT. These dosing (range 60–400 mg).10 In these patients, CD52À cells cells were predominant at times of profound lymphopenia, were present at peak levels in the range 80–100% of CD4 þ declining with post transplant follow-up as the normal T cells and 30–90% of CD8 þ T cells at 1 month into GPI þ T cell population reasserted itself. While the treatment and persisted throughout the 20 months of study. functional significance of these cells is not known, they The mechanism underlying development of CD52À T could have relevance for the incidence of infection, cells after C-1H treatment is not known. The only patient autoimmune phenomena or rates of relapse. We propose studied in detail, from those with CD52À T cells in the CLL that larger studies are now required, looking also at series mentioned above, was shown to have a mutation of correlations with serum C1-H levels and stratification the PIG-A gene.11 Highly sensitive PCR demonstrated the according to absolute CD52À T cell counts.

Bone Marrow Transplantation PNH-like T cells after transplants involving CAMPATH RJ Garland et al 244 Acknowledgements type in patients with paroxysmal nocturnal hemoglobinuria. Blood 1998; 92: 1799–1806. This work was funded by a grant from the COGENT trust. We 17 Tseng JE, Hall SE, Howard TA, Ware RE. Phenotypic and acknowledge the assistance of the nursing staff of the BMT Unit functional analysis of lymphocytes in paroxysmal nocturnal and Oncology Day Beds in obtaining blood samples for use in hemoglobinuria. Am J Hematol 1995; 50: 244–253. this study. This work would not have been possible without the 18 Richards SJ, Morgan GJ, Hillmen P. Analysis of T cells in agreement of the patients to participate, to whom we are deeply paroxysmal nocturnal hemoglobinuria provides direct evidence indebted. that thymic T-cell production declines with age. Blood 1999; 94: 2790–2799. 19 Bindon CI, Hale G, Waldmann H. Importance of antigen References specificity for complement-mediated lysis by monoclonal antibodies. Eur J Immunol 1988; 18: 1507–1514. 1 Riechmann L, Clark M, Waldmann H, Winter G. Reshaping 20 Cornish JM, Pamphilon DH, Potter MN et al. Unrelated donor human antibodies for therapy. Nature 1988; 332: 323–327. bone marrow transplant in childhood ALL. The role of T-cell 2 Cull GM, Haynes AP, Byrne JL et al. Preliminary experience depletion. Bone Marrow Transplant 1996; 18 (Suppl 2): 31–35. of allogeneic stem cell transplantation for lymphoproliferative 21 Oakhill A, Pamphilon DH, Potter MN et al. Unrelated donor disorders using BEAM-CAMPATH conditioning: an effective bone marrow transplantation for children with relapsed acute regimen with low procedure-related toxicity. Br J Haematol lymphoblastic leukaemia in second complete remission. Br J 2000; 108: 754–760. Haematol 1996; 94: 574–578. 3 Morris EC, Rebello P, Thomson KJ et al. Pharmacokinetics of 22 Chakrabarti S, Milligan DW, Brown J et al. Influence of alemtuzumabused for in vivo and in vitro T-cell depletion in cytomegalovirus (CMV) sero-positivity on CMV infection, allogeneic transplantations: relevance for early adoptive lymphocyte recovery and non-CMV infections following T- immunotherapy and infectious complications. Blood 2003; cell-depleted allogeneic stem cell transplantation: a comparison 102: 404–406. between two T-cell depletion regimens. Bone Marrow Trans- 4 Rebello P, Cwynarski K, Varughese M et al. Pharmacokinetics plant 2004; 33: 197–204. of CAMPATH-1H in BMT patients. Cytotherapy 2001; 3: 23 Hancock JP, Burgess MF, Goulden NJ et al. Same-day 261–267. determination of chimaeric status in the immediate period 5 Hale G, Bright S, Chumbley G et al. Removal of T cells from following allogeneic bone marrow transplantation. Br J bone marrow for transplantation: a monoclonal antilympho- Haematol 1997; 99: 403–409. cyte antibody that fixes human complement. Blood 1983; 24 Hancock JP, Goulden NJ, Oakhill A, Steward CG. Quanti- 62: 873–882. tative analysis of chimerism after allogeneic bone marrow 6 Gilleece MH, Dexter TM. Effect of Campath-1H antibody transplantation using immunomagnetic selection and fluores- on human hematopoietic progenitors in vitro. Blood 1993; cent microsatellite PCR. Leukemia 2003; 17: 247–251. 82: 807–812. 25 Nakakuma H, Nagakura S, Kawaguchi T et al. Persistence of 7 Ginaldi L, De Martinis M, Matutes E et al. Levels of affected T lymphocytes in long-term clinical remission in expression of CD52 in normal and leukemic B and T cells: paroxysmal nocturnal hemoglobinuria. Blood 1994; 84: 3925–3928. correlation with in vivo therapeutic responses to Campath-1H. 26 Ware RE, Pickens CV, DeCastro CM, Howard TA. Circulat- Leuk Res 1998; 22: 185–191. ing PIG-A mutant T lymphocytes in healthy adults and 8 Hale G, Xia MQ, Tighe HP et al. The CAMPATH-1 antigen patients with bone marrow failure syndromes. Exp Hematol (CDw52). Tissue Antigens 1990; 35: 118–127. 2001; 29: 1403–1409. 9 Richards SJ, Rawstron AC, Hillmen P. Application of flow 27 Tuset E, Matutes E, Brito-Babapulle V et al. Immunopheno- cytometry to the diagnosis of paroxysmal nocturnal hemoglo- type changes and loss of CD52 expression in two patients with binuria. Cytometry 2000; 42: 223–233. relapsed T-cell prolymphocytic leukaemia. Leuk Lymphoma 10 Brett SJ, Baxter G, Cooper H et al. Emergence of CD52-, 2001; 42: 1379–1383. glycosylphosphatidylinositol-anchor-deficient lymphocytes in 28 Poynton CH, Shaw PJ, Barrett AJ. Leukaemic relapse rheumatoid arthritis patients following Campath-1H treat- associated with phenotypic change after CAMPATH-1 purged ment. Int Immunol 1996; 8: 325–334. autologous bone marrow transplantation. Lancet 1986; 1: 162. 11 Rawstron AC, Rollinson SJ, Richards S et al. The PNH 29 Birhiray RE, Shaw G, Guldan S et al. Phenotypic transforma- phenotype cells that emerge in most patients after CAM- tion of CD52(pos) to CD52(neg) leukemic T cells as a PATH-1H therapy are present prior to treatment. Br J mechanism for resistance to CAMPATH-1H. Leukemia 2002; Haematol 1999; 107: 148–153. 16: 861–864. 12 Hertenstein B, Wagner B, Bunjes D et al. Emergence of CD52-, 30 Taylor VC, Sims M, Brett S, Field MC. Antibody selection phosphatidylinositolglycan-anchor-deficient T lymphocytes against CD52 produces a paroxysmal nocturnal haemoglo- after in vivo application of Campath-1H for refractory B-cell binuria phenotype in human lymphocytes by a novel mechan- non-Hodgkin lymphoma. Blood 1995; 86: 1487–1492. ism. Biochem J 1997; 322: 919–925. 13 Osterborg A, Werner A, Halapi E et al. Clonal CD8+ and 31 Brett S, Baxter G, Cooper H et al. Repopulation of blood CD52- T cells are induced in responding B cell lymphoma lymphocyte sub-populations in rheumatoid arthritis patients patients treated with Campath-1H (anti-CD52). Eur J Hae- treated with the depleting humanized monoclonal antibody, matol 1997; 58: 5–13. CAMPATH-1H. Immunology 1996; 88: 13–19. 14 Johnson RJ, Hillmen P. Paroxysmal nocturnal haemoglo- 32 Bunjes D, Theobald M, Wiesneth M et al. Graft rejection by a binuria: nature’s gene therapy? Mol Pathol 2002; 55: 145–152. population of primed CDw52-host T cells after in vivo/ex vivo 15 Piedras J, Lopez-Karpovitch X. Flow cytometric analysis of T-depleted bone marrow transplantation. Bone Marrow glycosylphosphatidyl-inositol-anchored proteins to assess Transplant 1993; 12: 209–215. paroxysmal nocturnal hemoglobinuria clone size. Cytometry 33 Chakrabarti S, Mackinnon S, Chopra R et al. High incidence 2000; 42: 234–238. of cytomegalovirus infection after nonmyeloablative stem cell 16 Richards SJ, Norfolk DR, Swirsky DM, Hillmen P. Lympho- transplantation: potential role of Campath-1H in delaying cyte subset analysis and glycosylphosphatidylinositol pheno- immune reconstitution. Blood 2002; 99: 4357–4363.

Bone Marrow Transplantation