Scholarly article on topic 'Longitudinal Analysis of T-Cell Receptor Variable β Chain Repertoire in Patients with Acute Graft-versus-Host Disease after Allogeneic Stem Cell Transplantation'

Longitudinal Analysis of T-Cell Receptor Variable β Chain Repertoire in Patients with Acute Graft-versus-Host Disease after Allogeneic Stem Cell Transplantation Academic research paper on "Clinical medicine"

CC BY-NC-ND
0
0
Share paper
OECD Field of science
Keywords
{"T-cell receptor variable β chain" / "Graft-versus-host disease" / "Allogeneic T cell–depleted peripheral blood stem cell transplantation" / Graft-versus-leukemia}

Abstract of research paper on Clinical medicine, author of scientific article — Congxiao Liu, Min He, Barbara Rooney, Thomas B. Kepler, Nelson J. Chao

Abstract T-cell receptor variable β chain (TCRBV) repertoire spectratyping involves the estimation of CDR3 length distributions for monitoring T-cell receptor diversity and has proven useful for analyses of immune reconstitution and T-cell clonal expansions in graft-versus-host disease (GVHD) and graft-versus-leukemia after allogeneic stem cell transplantation. We performed a longitudinal spectratype analysis of 23 TCRBV families in 28 patients who underwent allogeneic T cell–depleted peripheral blood stem cell transplantation. Sixteen patients subsequently developed acute GVHD. We recently developed statistical methods that bring increased power and flexibility to spectratype analysis and allow us to analyze TCRBV repertoire development under appropriately complex statistical models. Applying these methods, we found that patients with acute GVHD demonstrated TCRBV repertoire development statistically distinct from that repertoire development in patients without GVHD. Specifically, GVHD patients showed spectratypes indicative of lower diversity and greater deviation from the spectratypes expected in healthy individuals at intermediate times. Most individual TCRBV subfamilies had spectratypes statistically distinguishable between GVHD and non-GVHD patients at 6 months after transplantation. These results suggest that the T-cell receptor repertoire perturbations associated with acute GVHD are widely spread throughout the TCRBV families.

Academic research paper on topic "Longitudinal Analysis of T-Cell Receptor Variable β Chain Repertoire in Patients with Acute Graft-versus-Host Disease after Allogeneic Stem Cell Transplantation"

Biology of Blood and Marrow Transplantation 12:335-345 (2006) © 2006 American Society for Blood and Marrow Transplantation l083-879l/06/l203-00ll$32.00/0 doi:l0.l0l6/j.bbmt.2005.09.0l9

AS BMI

American Society for Blood and Marrow Transplantation

Longitudinal Analysis of T-Cell Receptor Variable P Chain Repertoire in Patients with Acute Graft-versus-Host Disease after Allogeneic Stem Cell Transplantation

Congxiao Liu,1 Min He,2 Barbara Rooney,1 Thomas B. Kepler,2 Nelson J. Chao1

1The Division of Cellular Therapy, Department of Medicine, and 2Department of Biostatistics and Bioinformatics, Duke University Medical Center, Durham, North Carolina

Correspondence and reprint requests: Nelson J. Chao, MD, MBA, Division of Cellular Therapy/BMT, Duke University Medical Center, 2400 Pratt St., Suite 1100, Box 3961, Durham, NC 27710 (e-mail: chao0002@mc.duke.edu).

Received January 24, 2005; accepted September 21, 2005

ABSTRACT

T-cell receptor variable P chain (TCRBV) repertoire spectratyping involves the estimation of CDR3 length distributions for monitoring T-cell receptor diversity and has proven useful for analyses of immune reconstitution and T-cell clonal expansions in graft-versus-host disease (GVHD) and graft-versus-leukemia after allogeneic stem cell transplantation. We performed a longitudinal spectratype analysis of 23 TCRBV families in 28 patients who underwent allogeneic T cell-depleted peripheral blood stem cell transplantation. Sixteen patients subsequently developed acute GVHD. We recently developed statistical methods that bring increased power and flexibility to spectratype analysis and allow us to analyze TCRBV repertoire development under appropriately complex statistical models. Applying these methods, we found that patients with acute GVHD demonstrated TCRBV repertoire development statistically distinct from that repertoire development in patients without GVHD. Specifically, GVHD patients showed spectratypes indicative of lower diversity and greater deviation from the spectratypes expected in healthy individuals at intermediate times. Most individual TCRBV subfamilies had spectratypes statistically distinguishable between GVHD and non-GVHD patients at 6 months after transplantation. These results suggest that the T-cell receptor repertoire perturbations associated with acute GVHD are widely spread throughout the TCRBV families. © 2006 American Society for Blood and Marrow Transplantation

Graft-versus-host disease • Allogeneic T cell-depleted peripheral Graft-versus-leukemia

KEY WORDS

T-cell receptor variable p chain • blood stem cell transplantation •

INTRODUCTION

Allogeneic hematopoietic stem cell transplantation (allo-SCT) has become the treatment of choice for a variety of hematopoietic disorders [1]. The effectiveness of allo-SCT is often limited by graft-versus-host disease (GVHD), which is initiated by alloreactive T cells in the graft. Various strategies have been used to separate GVHD from graft-versus-leukemia activities. An immunosuppressive regimen administered in the months after allo-SCT is the most efficient preventive treatment but remains only partially effective [2]. T-cell depletion can prevent GVHD [3-5] but increases the risk of graft rejection, leukemic relapse, and opportunistic infection [6,7]. The selec-

tive depletion of alloreactive T cells, as an adaptation of total T-cell depletion, has effectively prevented GVHD without impairing other immune functions by targeting activation markers after host-reactive T cells are activated in vitro [8-12]. Only a few of these strategies, however [13,14], have been explored in human trials so far because of the complexity of the processes involved. In many of these approaches, immune reconstitution is an important area for continued study and monitoring.

The T-cell receptor (TCR) repertoire has been studied in the immune recovery processes of patients with a variety of hematopoietic disorders after allo-SCT. Analysis of the complementarity-determining

region 3 (CDR3) via T-cell receptor variable p chain (TCRBV) spectratyping [15] could elucidate the clonal expansion of T cells associated with graft-ver-sus-leukemia and GVHD. In this study, we analyzed changes in each of the subfamilies of the TCRBV repertoire from peripheral blood and compared the extent of perturbation in each TCRBV family between patients with and without GVHD. We investigated the relative changes over time in the TCRBV repertoire in these patients.

MATERIALS AND METHODS Patients and Cells

Blood samples were obtained before transplantation and at regular intervals after transplantation from 28 patients who underwent allogeneic T cell-depleted peripheral blood stem cell transplantation (allo-PBSCT) from February 2000 to March 2003 at Duke University Medical Center after informed consent for sample collection. All donors and recipients were related. Twenty-six donor-recipient pairs were matched for HLA, and 2 pairs in the GVHD group were mismatched (5/6 and 4/6). Peripheral blood mononuclear cells (PBMCs) were prepared by density gradient sedimentation by using Ficoll-Hypaque, cryopreserved in medium with 10% dimethyl sulfoxide, and stored at — 190°C until analysis. Jurkat cells were cultured in RPMI-1640 containing 10% fetal calf serum and served as templates for reverse transcriptase-polymer-ase chain reaction-positive control.

CDR3 Size Spectratyping and Statistical Analysis

CDR3 size spectratyping was used to analyze the TCR repertoire. RNA was extracted from 0.5 to 5 X 106 PBMCs and Jurkat cells by using an RNeasy Mini Kit (Qiagen, Hilden, Germany). First-strand complementary DNA (cDNA) was generated from 1.5 ^g of total RNA by using random hexamer primer (Pro-mega, Madison, WI) and Moloney murine leukemia virus reverse transcriptase (Gibco BRL Carlsbad, CA) in 30 ^L of reaction. The quality of the cDNA was determined by careful examination of the product of the housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (G3PDH). PCR was performed using a panel of 23 different TCRBV primers and 1 Cp primer, which have been described previously [15]. The TCRBV8 and Cp primers were used in the PCR amplification for Jurkat cells. Each TCRBV segment was amplified from 1 ^L of the cDNA in a 20-^L reaction. The final mixture contained 0.5 ^mol/L of each TCRBV primer, 0.5 ^mol/L Cp primer, 0.25 mmol/L deoxynucleoside triphosphate, and 0.025 U of Taq polymerase (Applied Biosystems, Foster City, CA). Amplification was performed in a DNA thermal cycler (Gene Amp PCR system 9700, PE Applied

Biosystems, Foster City, CA). After 30 seconds of denaturation at 94°C, 40 cycles consisting of denatur-ation at 94°C for 1 minute, annealing at 60°C for 45 seconds, and extension at 72°C for 45 seconds were performed. After the 40 cycles of PCR, an additional extension at 72°C lasted 15 minutes.

Two microliters of the beta variable (BV)-Cp PCR products was subjected to 5 cycles of elongation (runoff) reaction with a nested carboxyfluorescein-labeled Cp primer. The runoff products were then subjected to electrophoresis on an ABI sequencer (PE Applied Biosystems) in the presence of fluorescent size markers. Initial analysis was performed with the Genescan software provided by the manufacturer (Applied Biosystems). Products differing in size were thus separated and quantitated as discrete peaks. Because the positions of the BV and Cp primers were fixed, the length of distribution observed in the BV-Cp products directly reflected the CDR3 lengths as determined by the rearrangement of the V-D and D-J gene segments and the randomly inserted nucleotides. The TCR spectratype in healthy adult subjects displays an average of 8 to 10 peaks per subfamily, corresponding to in-frame transcripts, contained within a symmetric, bell-shaped, or quasi-gaussian distribution. Alterations of the normal repertoire result in measurable deviations from this ideal distribution.

Spectratypes were analyzed by using the methods developed by 2 of the co-investigators [16] and implemented online for public use as SpA: Spectratype Analysis [17]. An abbreviated description is as follows: Genescan provides tables of peak heights, central locations, and areas. These tables are processed in SpA to assign each reported output peak to an integer-valued CDR3 length by using a maximum-likelihood method that accounts for random single-nucleotide additions (as is known to occur with spectratyping), as well as among-run stochastic variability in electrophoretic mobility and within-run random error variation. More than 1 reported peak may be assigned to each CDR3 length; the relative frequency assigned to any 1 CDR3 is the sum of the areas under the associated reported peaks. For each TCRBV family, a reference distribution was estimated from spectratypes performed independently on peripheral blood from each of 5 individual healthy volunteers who did not undergo transplantation. The estimation of these distributions was performed by maximum likelihood. For each observed spectratype distribution, we computed the Kullback-Leibler divergence (DKL) between it and the corresponding reference distribution [16]. The Dkl, which we will from here on refer to as simply the divergence, quantifies the departure of each observed CDR3 length distribution from the corresponding reference distribution. The divergences for each TCRBV, each patient, and each available time after transplantation were subjected to logarithmic transformation to regularize the variance and were collected for

analysis by linear modeling by using the S-plus software package (version 6.1; Insightful Corp., Seattle, WA). We fit a series of linear models with diagnosis, time, TCRBV family, and GVHD as predictors (independent variables). Because of the small number of distinct sampling times, time was treated as a categorical variable.

Immunologic Recovery

Immune recovery was analyzed before transplantation and at 6 weeks and 3, 6, and 12 months after transplantation. PBMCs were stained with fluorescent-labeled monoclonal antibodies and analyzed by flow cytometry. The data were analyzed by using Flow Jo software (Becton Dickison, Franklin Lakes, NJ). Monoclonal antibodies used in this study included anti-CD3, -CD4, and -CD8. The Student t test was used to analyze the difference in cell populations between patients with acute GVHD and without GVHD.

RESULTS

Patient Characteristics

Twenty-eight patients' characteristics are described in Table 1. The median age was 46 years (range, 24-61

years). Five patients received allo-PBSCT for myelodys-plastic syndromes, 4 for breast cancer, 3 for acute myeloid leukemia, 3 for acute lymphocytic leukemia, 3 for renal cell cancer, 2 for chronic myelogenous leukemia, 2 for mantle cell lymphoma, and 1 each with thalassemia, aplastic anemia, myelofibrosis, myeloma, chronic lymphocytic leukemia, and non-Hodgkin's lymphoma. Sixteen patients had acute GVHD grade II or greater affecting the skin, gut, or both. Seven of the 16 patients developed chronic GVHD. Nineteen patients had a positive cytomegalovirus (CMV) DNA test.

CDR3 Size Analysis of Normal Human PBMCs

Figure 1 shows a representative CDR3 size distribution from normal individual human PBMCs. A total of 23 of 24 TCRBV-Cp subfamilies displayed mostly bell-shaped, quasi-gaussian size distributions of 8 to 10 identifiable peaks spaced by 3 nucleotides. TCRBV 13B gave a very weak signal in most PBMCs. TCRBV 24 sometimes failed to yield interpretable profiles. Panel TCRBV 13A is poorly expressed in most individuals. Instead of this panel, a PCR-positive control, Jurkat cells' cDNA, were run with TCRBV8 and Cp primers, showing 2 discrete peaks. Overall, the patterns observed

Table 1. Patient Characteristics

Patient No. Age (y)/Sex Diagnosis Preparative Regimen GVHD Prophylaxis Acute GVHD Chronic GVHD CMV Status

001 24/M Thalassemia Mini, Cam/Flu/Cyc CSA/MMF — — +

013 37/F Aplastic anemia Mini, Cam/Flu/Cyc — — +

020 60/M MCl Mini, Cam/Flu/Cyc — — +

024 43/M ALL Full, TBI/Cyc — — +

033 40/F AML Mini, Cam/Flu/Cyc — — +

061 41/M CML Full, busulfan/Cyc — — -

090 61/M Renal cell cancer Mini, Cam/Flu/Cyc — — +

0103 25/F ALL Full, TBI/Cyc/ARA-C CSA — — +

0164 47/F MDS Mini, Cam/Flu/Cyc — — +

0165 57/M MCl Mini, Cam/Flu/Cyc — — +

0121 50/F Myelofibrosis Mini, Cam/Flu/Cyc — — -

0212 39/F Breast cancer II Mini, Cam/Flu/Cyc — — +

010 47/F Breast cancer Met. IV Mini, Cam/Flu/Cyc III — +

012 59/F AML Mini, Cam/Flu/Cyc II — +

014 54/F Breast cancer Met. IV Full, busulfan/Cyc CSA IV — +

021 66/F MDS Mini, Cam/Flu/Cyc II + skin +

023 45/F ALL Full, TBI/VPI6 CSA/prednisone III + skin -

031 44/M MDS Mini, Cam/Flu/Cyc II — -

070 45/M Renal cell cancer Mini, Cam/Flu/Cyc CSA III — +

072 51/F MDS Mini, Cam/Flu/Cyc II — +

0101 61/M MDS Mini, Cam/Flu/Cyc IV + -

0102 44/F Myeloma Mini, Cam/Flu/Cyc II — +

0116 52/M CLL Mini, Cam/Flu/Cyc II + skin -

0118 45/M Renal cell cancer Mini, Cam/Flu/Cyc III + skin & gut -

0120 45/M CML Full, busulfan/Cyc CSA/methotrexate IV — +

0162 50/F Breast cancer Met. IV Mini, Cam/Flu/Cyc II — -

0166 41/F NHL Mini, Cam/Flu/Cyc III + skin +

0210 45/F AML Mini, Cam/Flu/Cyc III + skin -

Cam indicates campath; Flu, fludarabine; Cyc, cyclophosphamide; CSA, cyclosporin A; MMF, mycophenolate mofetil; TBI, total body irradiation; Mini, nonmyeloablative; Full, myeloablative; VP16, etoposide; ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; CML, chronic myeloid leukemia; MDS, myelodysplastic syndrome; CLL, chronic lymphoblastic leukemia; NHL, non-Hodgkin's lymphoma; MCL, mantle cell lymphoma; CMV status, patient's DNA by hybrid capture; ARA-C, cytosine arabinoside; breast cancer stage II, breast cancer metastatic, stage IV.

Figure 1. CDR3 size distribution patterns in PBMCs from individual healthy humans. Total RNA from PBMCs was reverse-transcribed and amplified by PCR with 23 pairs of TCRBV- and Cß-specific primers. The PCR product was further copied in runoff reactions by a fluorescent Cß primer and loaded to electrophoresis on an automated DNA sequencer. The size and intensity of each peak were determined with the help of the Immunoscope software package (PE Applied Biosystems).

in PBMCs from healthy humans were reproducible from one individual to another, but deviations from the usual gaussian-like profiles could be found in runoff profiles.

CDR3 Size Analysis in Acute GVHD after Allo-PBSCT

The changes in the TCRBV gene repertoire were analyzed at different time points (days 42, 90, 180, and

360 after allo-PBSCT). The CDR3 size distribution in most TCRBV transcripts analyzed from non-GVHD patients demonstrated the expected bell-shaped patterns of more than 5 peaks from 6 weeks after transplantation. Moreover, the peak numbers in most families remained stable. In contrast, the major alternation of CDR3 size distribution in patients with acute GVHD was the modification of the regular

gaussian distribution by 1 or several peaks. This alteration in some BV families did not improve by 6 months or even by 1 year. This phenomenon may be due to 1 or several clonal T-cell expansions. Figure 2 shows the repertoire diversity as a histogram in representative patients with and without GVHD. The peak number in most TCRBV subfamilies of acute GVHD patients was 1 to 4, and some of them did not demonstrate restoration until 2 years (Figure 2A). In contrast, TCRBV subfamily peak numbers in non-GVHD patients ranged from 5 to 10 peaks even at early time points after transplantation. Furthermore, most expressed TCRBV gene subfamilies displayed a relatively stable picture at the following time point (Figure 2B). Meanwhile, some BV families showed a slight decrease in the repertoire complexity by 6 months compared with 6 weeks, such as TCRBV3, 4, 5, 12, 16, and 23. This may be relative to the complex clinical situation that occurs during the transplantation course. Infections are a common complication after transplantation.

To compare the extent of perturbation that is attributable to the difference in GVHD status among patients, we first performed a stepwise regression, stepping in both directions, by using diagnosis, age, sex, GVHD status, TCRBV family, and days after transplantation (as a categorical variable), as well as the estimable interactions, as the possible predictor variables for the log of the divergence as the response. The best model found excluded age and sex altogether and used time, TCRBV family, diagnosis, GVHD status, and the interactions between GVHD and time and between GVHD and TCRBV family as predictors. We confirmed these results with direct likelihood ratio tests and, in particular, found that the omission of GVHD was rejected quite soundly (P < 10~6) and that omission of the TCRBV-GVHD interaction was also rejected (P < 10~3).

Figure 3A shows the effects of GVHD status over time by plotting the components of the predicted response attributable to GVHD and time only. This plot allows for the appropriate correction for diagnosis while averaging over TCRBV families. Note that the GVHD patients and the non-GVHD patients are indistinguishable before transplantation and then again at 6 months after transplantation. At intermediate times, the divergence in the GVHD group was significantly higher than that in the non-GVHD group, corresponding to greater perturbation from the healthy profile and decreased diversity in the GVHD group. We similarly examined the contributions to the GVHD effect from each of the TCRBV families (Figure 3B). This figure shows the 95% confidence intervals for the TCRBV-GVHD interaction terms. Although these terms differ from each other statistically, 14 of 23 families have nonzero coefficients, and all of these are positive, thus indicating

smaller diversity in the GVHD patients for all of these families. Altogether, 21 of 23 families had a positive GVHD-TCRBV coefficient.

These results are suggestive of multiple antigen-driven T-cell clonal expansions, with greater magnitudes of such expansions in GVHD patients compared with non-GVHD patients. Figure 4 further demonstrates the dynamic changes of perturbed TCRBV families 7, 11, 15, 16, 17, and 23 in representative patients after transplantation.

Lymphocyte Subset Reconstitution after Transplantation

To better understand the changes in the TCRBV repertoire between patients with and without acute GVHD after transplantation, we evaluated the reconstitution of T lymphocytes and their subsets, including CD3+, CD4+, and CD8+, by immunophenotypic analysis of PBMCs at the same serial time points after transplantation. Table 2 lists the results of the lymphocyte recovery after transplantation. Both the lymphocyte count and the CD8+ cell count were statistically lower in the GVHD group than in the non-GVHD group.

DISCUSSION

Mature a/p T cells in the graft play a major role in the induction of GVHD. Through their receptors, T cells recognize a subset of host peptides, called minor histocompatibility antigens (miHAs), which are derived from the expression of polymorphic genes that distinguish host from donor [18]. Each TCR molecule includes a variable and a constant region. CDR3 of the BV chain of the TCR forms the contact site for binding peptides in HLA complexes and plays a key role in defining the specificity of antigen recognition. The human genome contains 52 functional BV segments belonging to 24 subfamilies, 2Dp and 13 Jp segments that undergo variable, diversity, joining regions (VDJ) rearrangements to express a Vp region [19,20]. This combinatorial diversity is greatly enhanced by random insertion and deletion of junctional flexibility and N-region nucleotide additions. The generated size heterogeneity of the CDR3 region during rearrangement contributes to the shaping of the TCR repertoire.

According to PCR amplification of the BV CDR3 region, healthy individuals demonstrate highly diverse and polyclonal TCR repertoires with a typical quasi-gaussian distribution of approximately 8 to 10 different sizes for each BV region separated by 3 nucleotides. Strong immune responses, such as GVHD, organ transplant rejection, and infection, are associated with oligo-clonal or monoclonal CDR3 patterns in the peripheral blood and in the affected tissues [21-23]. It is difficult, however, to simply compare the extent of variability of

Number of peaks per familg

rn rn rn rn

il n 1-1 m 1-1 - n il

VB1 VB2 VB3 VB4 VB5 VB6A VB6B VB7 VB8 VB9 VB11 VB12 VB13B VB14 VB15 VB16 VB17 VB18 VB20 VB21 VB22 VB23 VB24 BMT-0116 3-month

Number of peaks per familg

rn - - rn rn - rn rn rn

- 1 0 □ 0 1 -

VB1 VB2 VB3 VB4 VB5 VB6A VB6B VB7 VB8 VB9 VB11 VB12 VB13B VB14 VB15 VB16 VB17 VB18 VB20 VB21 VB22 VB23 VB24 BMT-0116 6-month

VB1 VB2 VB3 VB4 VB5 VB6A VB6B VB7 VB8

VB9 VB11 VB12 VB13B VB14 VB15 VB16 VB17 VB18 VB20 VB21 VB22 VB23 VB24

BMT-0116 1-year

L. №

VB1 VB2 VB3 VB4 VB5 VB6A VB6B VB7 VB8

VB9 VB11 VB12 VB13B VB14 VB15 VB16 VB17 VB18 VB20 VB21 VB22 VB23 VB24

BMT-0116 2-year

Figure 2. A representative composite spectratyping of the TCRBV gene at serial time points after allo-PBSCT. TCRBV subfamilies are shown on the x-axis, and peak numbers per family are shown on the y-axis. A, TCRBV repertoire in a patient with acute GVHD. The peak numbers in most of the TCRBV subfamilies were less than 4. B, TCRBV repertoire in a patient without acute GVHD. The peak numbers in most of the TCRBV subfamilies were more than 5.

W M If à m - - pi - - pi - - ri - - - ri ri - - - ri - - - -

0 fl » « ú £

Z VB1 VB2 VB3 VB4 VB5 VB6 VB6B VB7 VB8 VB9 VB11 VB12 VB13B VB14 VB15 VB16 VB17 VB18 VB20 VB21 VB22 VB23 VB24 BMT-024 day-42

a 4 j - pi pi - -

u. E 5" o m •a E

Z VB1 VB2 VB3 VB4 VB5 VB6 VB6B VB7 VB8 VB9 VB11 VB12 VB13B VB14 VB15 VB16 VB17 VB18 VB20 VB21 VB22 VB23 VB24 BMT-024 3-month

'H1 0 H

VB1 VB2 VB3 VB4 VB5 VB6

VB7 VB8 VB9 VB11 VB12 VB13B VB14 VB15 VB16 VB17 VB18 VB20 VB21 VB22 VB23 VB24

BMT-024 6-month

M Ji 1? 4 № 'H1 6 0 fl ï n E pi - - ri - - - ri - - - - pi - - - pi

Z VB1 VB2 VB3 VB4 VB5 VB6 VB6B VB7 VB8 VB9 VB11 VB12 VB13B VB14 VB15 VB16 VB17 VB18 VB20 VB21 VB22 VB23 VB24 BMT-024 1-year

Figure 2. (continued)

Figure 3. Comparison of the extent of perturbation in the TCRBV repertoire between groups with and without GVHD from 6 weeks to 6 months after transplantation. A, Divergence attributable to differences in GVHD status as a function of time since transplantation, estimated as described in the text (triangles, GVHD; circles, no GVHD). Corrected refers to the subtraction of components attributable to predictors other than GVHD and time. Increasing divergence is equivalent to greater perturbation from the healthy profile and decreased diversity. Note that the 2 populations are indistinguishable before transplantation and again at 6 months after transplantation. Error bars represent 95% confidence intervals. B,The95% confidence intervals for the corrected mean difference between GVHD and non-GVHD patient divergences, as described in the text. Corrected again refers to the subtraction of components of predictors other than TCRBV and GVHD. Those confidence intervals that do not include the origin correspond to hypothesis tests in which the null hypothesis, that the true mean difference is 0, was rejected.

TCR repertoires among samples. Therefore, in this study we introduced a novel method of analysis by using the SpA software package. This package allows for analysis of the similarity to an expected quasi-gaussian distribution in each subfamily of TCR repertoires. The changes of DKL values reflect the extent of perturbation in the TCR repertoire (ie, how far away it is from the expected distribution). When the observed CDR3 pattern is identical to the reference distribution, the DKL value vanishes; otherwise, the DKL value is positive. A high DKL value indicates a greater perturbation from the predicted reference distribution.

The data presented here demonstrate that some

BV subfamilies had clearly less complexity or even no repertoire detected in most recipients examined early after transplantation. These data were consistent with the results of previous studies [24-26]. This low level of complexity and oligoclonality may reflect an imbalance of the TCR repertoire secondary to the alterations of environmental conditions, a simple delay in the reappearance of some BV subfamily specificities, or both [27]. In this study, the data suggest that the lower complexity might be caused by limited T-cell recovery from the donor peripheral blood stem cell source (Table 2), because almost all of the patients in this study had more than 85% donor cells in their chimerism studies (data not shown), although we cannot exclude the possible contributions of residual host cells. We also did not separate the T cells into CD4+ and CD8 + subsets, so the relative perturbation of each of these subfamilies cannot be determined from this study. Most patients in both groups reactivated human CMV, but none of them developed CMV infection (Table 1). Therefore, the influence of CMV antigen-emia on the recovery of the TCR repertoire was balanced in the groups with and without GVHD.

This study further showed that patients with acute GVHD demonstrated TCRBV repertoire development statistically distinct from the repertoire development in patients without GVHD. The GVHD patients and the non-GVHD patients were indistinguishable before transplantation and then again at 6 months after transplantation. At intermediate times, the divergence in the GVHD group was significantly higher than that in the non-GVHD group, corresponding to greater perturbation from the healthy profile and decreased diversity in the GVHD group. Most individual TCRBV subfamilies had spectratypes that were statistically distinguishable between GVHD and non-GVHD patients at 6 months after transplantation. These results are suggestive of multiple antigen-driven T-cell clonal expansions, with greater magnitudes of such expansions in GVHD patients compared with non-GVHD patients. None of the patients was in relapse at the time of sample collection, but some of the patients did develop chronic GVHD in the acute GVHD group. To allow a better understanding of whether chronic GVHD enhances the perturbation of the repertoire, we performed statistical analyses between patients with only acute GVHD and those who developed chronic GVHD. The results showed distinguishable spectratypes (P = .02). These data suggest that chronic GVHD does enhance the perturbation of the repertoire. These findings further confirm the observation that oligoclonal expansions of TCRBV families occur frequently and often correlate with clinical events of GVHD [21,28]. A recent study by Tsutsumi et al. [25] also demonstrated lower TCRBV complexity in patients with chronic GVHD compared with those without chronic GVHD. These

Figure 4. Spectratyping profile of TCRBV7, 11, 15, 16, 17, and 23 in recipients with or without acute GVHD at serial time points after allo-PBSCT. Results are shown for each TCRBV family as a density peak histogram. CDR3 sizes are shown on the x-axis, and the peak fluorescence intensity is shown on the y-axis.

data suggest that the expansion of these T-cell subfamilies may have been driven in response against the patient's miHAs [29,30].

Human miHAs are recognized in the context of major histocompatibility complex molecules [30], and cytotoxic T lymphocytes against miHAs have been implicated in GVHD [28]. Skin is frequently involved in GVHD, and T cells infiltrating a GVHD target organ are likely to be implicated in a graft-versus-host reaction. Many perturbed TCRBVs from PBMCs in this study (BV 2, 6, 16, 17, and 23) have been reported in other studies to be preferentially observed in the

skin with GVHD [25,27,31,32]. Although BV 7, 8, 13B, and 21 have been not reported to be observed in the skin with GVHD, they have been observed in peripheral blood in other studies [25,27,31,32]. This difference may reflect the possibility that T cells initially expanded in blood may selectively home to the skin because of the high level of adhesion molecules, chemokines, or major histocompatibility complex molecule expression [33,34]. To further understand the T-cell differentiation pathway, a TCR excision circle assay on PBMCs was performed. All TCR excision circle numbers were below the limit detection

3 £ u

1 ) <u <J M

1 s £ £

.2 o ■o U <u Z

3 £ u

■o o Z

! « £ £ ir

■o U

■o o o

_ II _ II s.

CN ^ IO Ä r-*

r- >o "O

CN ^ ^

— CN

00 II X» II f r-» eo — : ^ c ^ c ^

^ m es £

S io J3 >o as m cn

2 II T II T II 1

. CN r-»

W C ^ C ^ C —

r-» — — CN

M _ r-*

0> M _ TT

ii ^ ii 9s ii

3 = ü o o

CN -©

+ CO Q I-» ¡5 * ■fCO t CN — «o fS

U « M O r- CN eo —

c c rt ■4 II üi 11 ^ 11 C»

cc c ^ = c

« _ Ifl M o

Z ■o UÏ fS

^ c ^ c o o

»O IA

2 «O S3 r- - 2 *?

■v .. — .. M IS

r-» eo

M CN — M

CN_ — o >0_ O

II .0 11

>o —

V V Si Bh

2 § & &

is "ü

i + H g

18 c .5

up to 6 months after transplantation except for 2 patients without GVHD (data not shown). These results indicate that the circulating T cells were derived from donor mature T cells, a thymic-independent pathway.

In conclusion, this study demonstrates that the extent of perturbed T-cell repertoires is significantly higher in PBMCs of patients with acute GVHD compared with those without GVHD during the first 6 months after allo-PBSCT. Moreover, the TCR repertoire perturbations associated with acute GVHD were widely spread throughout the TCRBV families. Although we cannot rule out the importance of the steroids used to treat GVHD on the basis of the observed difference between the TCR repertoires in GVHD-affected and -nonaffected subjects, it seems likely that such steroids would suppress the outgrowth of rogue T-cell clones and, thus, contribute to a more normal-looking TCR spectratype, which is the opposite of what is observed. Although the spectratype technique allows a precise measurement of the CDR3 region and the ability to distinguish different TCR clonotypes, further studies are needed to determine whether the perturbed repertoires detected in vivo correspond to T-cell clones with specific cytotoxicity against miHAs. Elimination of such clones from the graft inoculum could prevent GVHD while conserving graft-versus leukemia responses.

ACKNOWLEDGMENTS

The authors would like to thank Gregory D. Sem-powski for the TCR excision circle assay technique support, Marcella Sarzotti for discussion of the TCRBV data, and Gwynn Long for critical reading of the manuscript. Supported by National Institutes of Health grant nos. 2P0-1CA47741 (N.J.C.), 5P-01HL67314 (N.J.C.), 5 P30 AI051445-03 (T.B.K.), and U54 AI057157-02 (T.B.K.).

REFERENCES

1. Ferrara JL, Deeg HJ. Graft versus host disease. N Engl J Med. 1991;324:667-674.

2. Storb R, Deeg HJ, Pepe M, et al. Methotrexate and cyclospor-ine versus cyclosporine alone for prophylaxis of graft-versus-host-disease in patients given HLA-identical marrow grafts for leukemia: long-term follow-up of a controlled trial. Blood. 1989; 73:1729-1734.

3. Brenner MK, Wimperis JZ, Reittie JE, et al. Recovery of immunoglobulin isotypes following T-cell depleted allogeneic bone marrow transplantation. Br J Haematol. 1986;64:125-132.

4. O'Reilly RJ, Collins N, Dinsmore R, et al. Transplantation of HLA-mismatched marrow depleted of T-cells by lectin agglutination and E-rosette depletion. Tokai J Exp Clin Med. 1985; 10:99-107.

5. Champlin R. T-cell depletion to prevent graft-versus-host dis-

ease after bone marrow transplantation. Hematol Oncol Clin North Am. 1990;4:687-698.

6. Keever CA, Small TN, Flomenberg N, et al. Immune reconstitution following bone marrow transplantation: comparison of recipients ofT-cell depleted marrow with recipients of conventional marrow graft. Blood. 1989;73:1340-1350.

7. Kerman NA. T-cell depletion for prevention of graft-versus-host disease. In: Forman SJ, Korman NA, Blume KG, Thomas ED, eds. Bone Marrow Transplantation. Cambridge, MA: Blackwell Scientific Publications; 1994:124.

8. Montagna D, Yvon E, Calcaterra V, et al. Depletion of allo-reactive T cells by a specific anti-interleukin-2 receptor p55 chain immunotoxin does not impair in vitro antileukemia and antiviral activity. Blood. 1999;93:3550-3557.

9. Mavroudis DA, Dermime S, Molldrem J, et al. Specific depletion of alloreactive T cells in HLA-identical siblings: a method for separating graft-versus-host and graft-versus-leukaemia reactions. Br J Haematol. 1998;101:565-570.

10. Gribben JG, Guinan EC, Boussiotis VA, et al. Complete blockade of B7 family-mediated costimulation is necessary to induce human alloantigen-specific anergy: a method to ameliorate graft-versus-host-disease and extend the door pool. Blood. 1996; 87:4887-4893.

11. Rencher SD, Houston JA, LockeyTD, HurwitzJL. Eliminating graft-versus-host potential from T cell immunotherapeutic populations. Bone Marrow Transplant. 1996;18:415-420.

12. Mavroudis DA, Jiang YZ, Hensel N, et al. Specific depletion of alloreactivity against haplotype mismatched related individuals by a recombinant immunotoxin: a new approach of graft-ver-sus-host-disease prophylaxis in haploidentical bone marrow transplantation. Bone Marrow Transplant. 1996;17:793-799.

13. Andre-Schmutz I, Le Deist F, Hacein-Bey-Abina S, et al. Immune reconstitution without graft-versus-host disease after haemopoietic stem-cell transplantation: a phase 1/2 study. Lancet. 2002;360:130-137.

14. Guinan EC, Boussiotis VA, Neuberg D, et al. Transplantation of anergic histoincompatible bone marrow allografts. N Engl J Med. 1999;340:1704-1714.

15. Pannetier C, Levraud JP, Lim A, et al. The Immunoscope approach for the analysis of T cell repertoires. In: Oksenberg JR, ed. The Antigen T Cell Receptor: Selected Protocols and Applications. London: Chapman & Hall; 1998:287-325.

16. Kepler TB, He M, Tomfohr JK, Devlin BH, Sarzotti M, Markert ML. Statistical analysis of antigen receptor spectratype data. Bioinformatics. 2005;21:3394-3400.

17. He M, Tomfohr JK, Devlin BH, Markert ML, Sarzotti M, Kepler TB. SpA: web-accessible spectratype analysis—data management, statistical analysis and visualization. Bioinformatics. 2005;21:3697-3699.

18. Den Haan JM, Sherman NE, Blokland E, et al. Identification of a graft versus host disease-associated human minor histocom-patibility antigen. Science. 1995;268:1476-1480.

19. Wei S, Charmley P, Robison MA, Concannon P. The extent of the human germline T cell receptor V beta gene segment repertoire. Immunogenetics. 1994;40:27-36.

20. Zhao TM, Whitaker SE, Robinson MA. A genetically deter-

mined insertion/deletion related polymorphism in human T cell receptor B chain (TCRB) includes functional variable gene segments. J Exp Med. 1994;180:1405-1414.

21. Liu X, Chesnokova V, Forman SJ, et al. Molecular analysis of T cell receptor repertoire in bone marrow transplant recipients: evidence for oligoclonal T-cell expansion in graft-versus-host disease lesions. Blood. 1996;87:3032-3044.

22. Gorski J, Yassai M, Zhu X, et al. Circulating T cell repertoire complexity in normal individuals and bone marrow recipients analyzed by CDR3 size spectratyping. Correlation with immune status. J Immunol. 1994;152:5109-5119.

23. Silins SL, Cross SM, Krauer KG, et al. A functional link for major TCR expansions in healthy adults caused by persistent Epstein-Barr virus infection. J Clin Invest. 1998;102:1551-1558.

24. Hirokawa M, Matsutani T, Horiuchi T, et al. Extensive clonal expansion of T lymphocytes causes contracted diversity of complementarity-determining region 3 and skewed T cell receptor repertoires after allogenic hematopoietic cell transplantation. Bone Marrow Transplant. 2001;27:607-614.

25. Tsutsumi Y, Tanaka J, Miura Y, et al. Molecular analysis of T-cell repertoire in patients with graft-versus-host disease after allogeneic stem cell transplantation. Leuk Lymphoma. 2004;45: 481-488.

26. Matsutani T, Yoshioka T, Tsuruta Y, et al. Restricted usage of T-cell receptor a-chain variable region (TCRAV) and T-cell receptor ß-chain variable region (TCRBV) repertoires after human allogeneic haematopoietic transplantation. Br J Haema-tol. 2000;109:759-769.

27. Dietrich PY, Caignard A, Lim A, et al. Analysis of T-cell receptor variability in transplanted patients with acute graft-versus-host disease. Blood. 1992;80:2419-2424.

28. Dietrich PY, Caignard A, Lim A, et al. In vivo T-cell clonal amplification at time of acute graft-versus-host disease. Blood. 1994;84:2815-2820.

29. Parreault C, Decary F, Brochu S, GygerM, Belanger R, Roy D. Minor histocompatibility antigens. Blood. 1990;76:1269-1280.

30. Goulmy E, Gratama JW, Blokland E, Zwaan FE, Van Rood JJ. A minor transplantation antigen detected by MHC-restricted cy-totoxic T lymphocytes during graft-versus-host disease. Nature. 1983;302:159-161.

31. Martin PJ. Increased disparity for minor histocompatibility antigens as a potential cause ofincreased GvHD risk in marrow transplantation from unrelated donors compared with related donors. Bone Marrow Transplant. 1991;8:217-223.

32. Even A, Lim I, Puisieux L, et al. T-cell repertoires in healthy and diseased human tissue analysis by T-cell receptor B-chain CDR3 size determination: evidence for oligoclonal expansions in tumors and inflammatory diseases. Res Immunol. 1995;146:65-80.

33. Dustin ML, Singer KH, Tuck DT, Springer TA. Adhesion of T lymphoblasts to epidermal keratinocytes is regulated interferon R and is mediated by intercellular adhesion molecule 1 (ICAM-1). J Exp Med. 1988;167:1323-1340.

34. Picker LJ, Terstappen MM, Rott LS, Streeter PR, Stein H, Butcher EC. Differential expression of homing-associated adhesion molecules by T cell subsets in man. J Immunol. 1990; 145:3247-3255.