T cell repertoire complexity is conserved after LLME treatment of donor lymphocyte infusions

Abstract

Slow reconstitution of the T cell repertoire following allogeneic blood or bone marrow stem cell transplantation is a major risk factor for patient mortality. The delivery of immunocompetent T cells as delayed donor lymphocyte infusions (DLI) is one potential way to counteract this problem. However, the development of graft-versus-host disease (GVHD) is a potential complication of this procedure. We previously found in P→F1 haploidentical murine models that the ex-vivo treatment of donor lymphocytes with L-leucyl-L-leucine methyl ester (LLME) can prevent the onset of GVHD after DLI, likely by inducing cell death the majority of perforin-positive CD8⁺ T cells, and a fraction of CD4⁺ T cells. Our previous preclinical studies have formed the basis of an ongoing phase I clinical trial in which patients received LLME-treated DLI from their original donor in an attempt to accelerate T cell reconstitution. In order to understand how this treatment strategy might impact upon the complexity of the DLI T cell repertoire, we used TCR Vβ spectratype analysis to evaluate the DLI product pre and post-LLME treatment. The results of the spectratype analysis indicated that the LLME-treated DLI product exhibited CDR3-size distribution complexities similar to its untreated donor sample. In addition, comparisons of the CD4⁺ and CD8⁺ T cell repertoire from the donor before LLME treatment to that of the patient post DLI demonstrated equal complexity for most of the resolvable Vβ families. Lastly, the in vitro proliferative capacity of LLME-treated DLI product in response to allo-stimulation in a one-way mixed lymphocyte reaction was comparable to the untreated product.

Introduction

Delayed and/or incomplete T cell reconstitution and the correlating lack of graft-versus-leukemia (GVL) effects and opportunistic infections following allogeneic blood or bone marrow transplantation (BMT) are major risk factors for patient morbidity and mortality. This is particularly true in the case of CD34-enriched, T cell-depleted hematopoietic stem cell transplantation (HSCT), where, in order to avoid the development of acute graft-versus-host disease (GVHD), no immediate source of mature donor T cells is provided¹. Immune reconstitution in this situation, therefore, depends solely upon the ability to generate de novo T cells via the host thymus, which may have limited functional capabilities in these patients². One approach to rectify this problem of T cell reconstitution is to administer immunocompetent T cells in the form of a delayed donor lymphocyte infusion (DLI), 1–3 months post-transplant, with the intent of providing the lymphocytes in the setting of an environment with diminished host inflammatory responses associated with the pre-transplant conditioning regimen^3,4. Nevertheless, DLI’s effective reconstitution potential is still seriously hampered by the development of GVHD^5,6,7, and these opposing effects need to be separated by some means. In this regard, several murine studies have indicated that both the CD4 and CD8 T cells that mediate GVHD immunopathology depend on the utilization of perforin cytolytic mechanisms^8,9. In addition, immunohistochemical studies of epidermal tissue from transplant recipients have correlated acute GVHD lesions with the presence of perforin-containing cytolytic T cells^10,11. In contrast, investigation of GVL mechanisms in some myeloid leukemia models have found significant CD4-mediated GVL potential remaining in the absence of perforin pathways ^12,13 and the transfer of monoclonal populations of tumor-specific CD4⁺ versus CD8⁺ T cells were more efficient at tumor rejection when tested against several different tumor lines¹⁴. Thus, elimination of perforin-positive T cells from DLI inoculum could be a means of separating GVHD and GVL responses, and may allow for rapid reconstitution of immune capabilities in BMT recipients.

L-leucyl-L-leucine methyl ester (LLME) is a lysosomotropic agent that preferentially acts upon cells containing dipeptidyl peptidase I enzymes (DPPI) in their cytolytic granules, which normally also contain perforin ¹⁵. The overall effect is to induce cell death of most NK cells, monocytes, granulocytes, the majority of CD8⁺ T cells, and a small fraction of CD4⁺ T cells. LLME has been used previously to treat hematopoietic cell grafts in an effort to reduce the occurrence of GVHD in both murine and canine models^16,17,18. However, its use as GVHD prophylaxis in human BMT was abandoned when clinical trials indicated that the effective concentration required for the ex vivo elimination of GVHD-reactive T cells from marrow inoculum resulted in engraftment failure¹⁹. We thus considered the possibility that LLME could still be used as an immunomodulating agent for BMT if used to eliminate perforin-positive alloreactive T cells from a separate DLI product administered following T cell-depleted HSCT.

In preclinical experiments utilizing a haploidentical murine model, lethally irradiated (B6x DBA/2) mice were transplanted with T cell-depleted B6 bone marrow (ATBM) at day 0 and injected with mock- or LLME-treated B6 DLI at day 14 post-transplant and monitored for symptoms of GVHD and immune reconstitution. All mice receiving mock-treated DLI succumbed to acute GVHD, whereas those injected with LLME-treated DLI exhibited no signs of disease, with 100% survival, and demonstrated immune reconstitution comparable to that of normal B6 mice by 10 weeks post-transplant ⁸.

These preclinical studies formed the basis of an on-going phase I clinical trial in which patients received LLME-treated DLI from their original donor in an attempt to accelerate T cell reconstitution following CD34-enriched, allogeneic peripheral blood stem cell transplantation. Donors were either HLA-identical siblings, HLA-identical or single-antigen mismatched unrelated donors, or mismatched (haploidentical) related donors. The results of this clinical study are fully described elsewhere (Filicko et al, submitted for publication), but an important question was raised as to whether this LLME treatment strategy might impact upon the complexity of the DLI T cell repertoire being infused into patients, thereby affecting their reconstitution potential. This is particularly relevant to the CD8 subset which has been recently reported to contain between five and nine discrete subpopulations, each of which was observed to express different levels of perforin in addition to being represented in different proportions among individuals^20,21. Taken together with its mechanism of action, LLME treatment of the DLI could therefore potentially create major deficits in the repertoire within the different Vβ families. To address this issue, we used TCR Vβ CDR3-size spectratype analysis to evaluate the DLI product, pre- and post-LLME treatment, from peripheral blood mononuclear cells (PBMC) of six donor/patient pairs. PBMC cells obtained from donors at the time of DLI donation, prior to LLME treatment, served as the baseline spectratype for Vβ complexity. Thus, the CD4⁺ and CD8⁺ T cell repertoire complexity of each LLME-treated DLI product was determined by comparison to its untreated sample.

The results of the spectratype analysis indicated that overall the LLME-treated DLI products exhibited CDR3-size distribution complexities similar to the untreated donor sample for the resolvable Vβ families. In addition, we were also able to examine the complexity of the post-transplant CD4⁺ and CD8⁺ T cell repertoires in 4 of the 6 patients and observed that both subsets exhibited reconstitution complexities similar to the DLI donor samples for the resolvable Vβ families. Lastly, the in vitro proliferative capacities of LLME-treated PBMC in response to allostimulation in a one-way mixed lymphocyte reaction (MLR) were observed to be comparable to that of untreated PBMC. In summary, it appears that ex vivo LLME treatment of DLI does not adversely impact the complexity of the T cell repertoire or its functional capacity, and thus this transplant strategy is likely to afford advantages with regard to rapid immune reconstitution.

Materials and methods

Patients and Donors

Twenty-three patients with hematologic malignancies were enrolled on study at Thomas Jefferson University or the Medical College of Wisconsin following CD34 enriched HSCT. Ten patients had HLA identical sibling donors, eight had unrelated donors (four HLA-matched, four mismatched) and five had haplodisparate donors. Three patients expired prior to receiving DLI. One received unmanipulated DLI, and in one patient, the LLME depletion was not complete. Eighteen patients, aged 6–65 years (median 47 years) received LLME DLI as originally planned including a variety of preparative regimens. The six patients discussed herein, received CD34 selected HSCT following fludarabine, cytarabine and melphalan (FAM) + anti-thymocyte globulin (ATG) (Thymoglobulin^®, SangStat) as previously described²².

Prophylactic antibiotics consisted of an amphotericin product, valacyclovir and trimethoprim-sulfamethoxazole in all patients. All patients received transfusion support with leukodepleted and irradiated cellular blood products. Patients who were CMV-seronegative received CMV-seronegative blood products. Patients were followed with weekly ATG levels and were eligible to receive LLME DLI when the ATG level was <2.0 mcg/ml

LLME Treatment

LLME was synthesized under GMP conditions by Bachem (Switzerland). LLME dry powder was prepared in Normosol R at a 500 uM concentration. Washed cells (blood or apheresis product) were incubated with LLME at a concentration of 10×10⁶ cells/ml for 60 min. Cells were washed in Normosol (4°C X 1, then Room Temperature X 2). Release criterion for the LLME-treated products was 80% depletion of NK cells (CD 16/56+, CD3). Samples of the cells were labeled with fluorescein isothiocyanate (FITC)-conjugated anti-CD4, anti CD8, anti-CD16 or CD56 monoclonal antibodies (mAb), or phycoerythrin (PE)-conjugated anti-CD3 mAb, and analyzed by flow cytometry with a FACScan (Becton Dickinson, Finger Lakes, NJ). Patients received LLME-treated DLI, based on viable CD3⁺ cells.

Blood Samples

PBMC were enriched from donor and patient peripheral blood samples, as previously described²³, by centrifugation over Ficoll-histopaque. CD4⁺ and CD8⁺ T cell subsets were separated by standard antibody-panning techniques and the enriched subset populations were solubilized in Ultraspec (Biotex Laboratories, Houston, TX). The single exception was the LLME-treated DLI product from patient #5, which remained unseparated. Vβ spectratype analysis of separated PBMC from untreated donor samples served as the reference point for full repertoire complexity, and was used as the standard of comparison for the LLME-treated, as well as post-transplant samples. At the time of the post-transplant spectratype analysis, all patients exhibited >90% donor chimerism, as determined by molecular analysis of short tandem repeats of marrow or PBMC samples²⁴.

CDR3-Size Spectratype Analysis

Total RNA was isolated from Ultraspec samples and cDNA was prepared, as previously described23. Semi-nested PCR was performed using a panel of human Vβ sense oligoprimers and 2 Cβ antisense oligoprimers; the second Cβ being fluorescently labeled²³. The PCR products were run on a sequencing gel and analyzed by the Genotyper Genescan software program (Applied Biosystems, Foster City, CA). This approach allowed direct comparison of the Vβ repertoire complexity in the pre- and post-LLME DLI product.

Spectratype Complexity Index

The complexity index within a Vβ family was determined as a percentage of the number of peaks found in its spectratype histogram in relation to the number of peaks in the corresponding donor Vβ family histogram, as previously described^23. Any Vβ family with a complexity index of ≥85% was considered to be fully complex. Histogram peaks were identified by the Genotyper Genescan program. For some of the donor/patient samples, several Vβ families in the untreated DLI product did not exhibit an evaluable spectratype, but the LLME-treated product did. In these cases, we compared the number of peaks to the average number of peaks for that particular Vβfamily obtained from the other samples. Any donor or patient sample exhibiting fewer than 12 evaluable Vβ family spectratypes was excluded.

Statistical Analysis

Data comparisons were analyzed by the nonparametric Wilcoxon rank-sum analysis. P values of ≤0.05 were considered statistically significant.

MLR Assay

LLME- and mock-treated PBMC from four healthy volunteers were cultured at 1×10⁵ cells/well with irradiated (30 Gy) PBMC from a fifth healthy volunteer in 96-well tissue culture plates, at a 1:2 responder:stimulator ratio. Proliferative responses were determined after six days in culture (37°C; 8% CO₂) by measuring H³-thymidine (1 μCi/well) incorporation following a 6 h pulse/label period. Quadruplicate samples were used to determine the mean incorporation counts per minute (CPM) on a 1205 Betaplate liquid scintillation counter (Perkin-Elmer, Gaithersburg, MD).

Results

Patient Characteristics and GVHD Status

The ages, gender, underlying diseases, and transplant characteristics of the six patients presented are listed in Table 1. All patients received CD34 selected HSCT following fludarabine, cytarabine and melphalan (FAM) + anti-thymocyte globulin (ATG) One patient developed Grade III GVHD following a second dose of LLME DLI. This patient had received an initial dose of 1×10⁷ LLME DLI, but had not reached the target of 200 CD4⁺ cells/ul, and thus received a second LLME treated DLI 43 days later. Following the second dose of LLME DLI, at 4.1×10⁷ CD3⁺ cells/kg (target 1×10⁸) he developed GVHD of the skin and gastrointestinal tract one week later. None of the other patients reported here developed GVHD.

Table 1.

Patient Characteristics

Patient	Age/Gender	Disease	Graft Type	DLI Dose/kg	GVHD	Post DLI sample day	Days Post CD34HSCT of DLI
1	61 M	ALL - 1st CR	Mat Sib	1 × 107	None	64	43
2	38 M	ALL - 1st relapse	Haplo-M	NA	NA	NA	NA
3	22 M	CML – (LBP-1stCR)	Mat Sib	1 × 106	None	49	101
4	53 M	CML - Acc Phase	Mat Sib	(1st)1 × 107 (2nd)4×107	Grade III	42	(1st)27 (2nd)70
5	19 M	ALL - 2nd CR	URD	1 × 105	None	329	118
6	52 F	AML - 1st CR	Mat Sib	1 × 107	None	68	70

Median age was 45 (range, 19–61 years). Four patients received transplants from HLA-identical sibling donors (Mat Sib), one from completely matched (10/10) unrelated donor (URD), and one from a haplodisparate maternal donor (Haplo-M). ALL, acute lymphoblastic leukemia; LBP, lymphoid blast phase; Acc, accelerated, CR, complete remission; AML acute myelogenous leukemia;

One patient expired following relapse of her AML 175 days after LLME DLI. One patient required high dose steroid therapy for interstitial pneumonitis and expired subsequently due to multi-system organ failure related to sepsis. The one patient who did develop GVHD was treated with immune suppressive therapy including high dose steroids and subsequently succumbed secondary to infections complications. The other three patients presented here are alive, without evidence of GVHD, opportunistic infection or relapse of their leukemia at 1399, 1029 and 620 days post-LLME DLI.

Determination of Complexity Index Following LLME Treatment

LLME is known to cause rapid apoptosis of perforin- and granzyme-containing cells, thereby affecting a majority of NK cells, macrophages, CD8⁺ T cells, and a small subset of CD4⁺ T cells. However, due to the differential expression of perforin, particularly within the CD8 subpopulations, we evaluated whether LLME treatment of the DLI product might affect the complexity of the overall T cell repertoire. CDR3-size spectratype analysis was used to compare the complexity of 24 Vβ families in the CD4 and CD8 T cell subsets from the DLI product, before and after LLME treatment. Spectratype histograms of the pre- and post-LLME-treated donor CD4⁺ T cells for three representative Vβ families from each of five patients can be seen in Figure 1. The complexity index for the Vβ families was determined, as described in “Materials and Methods,” by comparing the donor pre and post-LLME treatment spectratype histograms. The results summarized in Tables 2 and 3 for the evaluable patients and Vβ families, indicated overall complexity indices for the CD4 and CD8 compartments of 82.7% (SEM 6.2) and 79.6% (SEM 6.2), respectively. There was no statistical difference (P≥0.68) between the overall complexity of the CD4⁺ and CD8⁺ T cell subsets suggesting that LLME treatment did not cause biased apoptosis of any particular Vβ family in either subset, even though there was a much greater depletion (P=0.001) of CD3⁺/CD8⁺ cells (mean 60.9%, range 35.1–81.9%) than CD3⁺/CD4⁺ cells (mean 27.5%, range 0–53%) by LLME.

Figure 1.

CDR3-size spectratype analysis of PBMC enriched from the DLI products of five donors before and after LLME treatment. The spectratype analysis was performed as described in “Materials and Methods.” Three representative Vβ spectratypes are shown for each donor.

Table 2.

CD4+ Repertoire Complexity Analysis

	Patient
Vβ	1	2	3	4	5	6
1	+	+	+	+	(+)	+
2	+	+	+	+	(+)	+*
3	+	+	+	+	(+)	+
4	+	ND	ND	+	(+)	ND
5	+	+	+	+	(+)	+*
6	+	ND	+	ND	ND	ND
7	+	−	+	+	(+)	ND
8	+	ND	+	+	(+)	−
9	+	+	+	+	(+)	+
10	+	ND	ND	−	ND	ND
11	+	+	+	ND	(+)	+*
12	−	ND	ND	+	(+)	ND
13	+	−	+	−	(+)	−*
14	+	+	+	−	(+)	−*
15	+	+	+	+	(+)	+
16	+	−	+	+	(−)	−*
17	+	+	+	+	(+)	+
18	+	−	+	+	(+)	ND
19	−	−	+	+	(+)	ND
20	ND	+	ND	+	ND	−*
21	+	−	+	+	(+)	ND
22	+	ND	ND	−	ND	ND
23	+	+	+	+	(+)	ND
24	−	ND	ND	ND	ND	ND
Overall % Complexity	87	65	100	85	95	64

− indicates not full complexity; +, full complexity;

^*, complexity as compared to averaged control donors; ( ), CD4 and CD8 subsets not separated;

ND, not determined

Table 3.

CD8+ Repertoire Complexity Analysis

	Patient
Vβ	1	3	4	5	6
1	+	−	−	(+)	+
2	+	+	−	(+)	+
3	+	+	+	(+)	+
4	+	+*	−	(−)	+
5	+	+	+	(+)	+
6	+	+	−	ND	ND
7	+	+	+	(+)	+
8	+	+	+	(+)	+
9	+	+	+	(+)	+
10	+	ND	+	ND	ND
11	+	+	+	(+)	+
12	−	ND	ND	ND	ND
13	−	+	−	(+)	+
14	+	+	+	(−)	+
15	+	+	ND	(+)	+
16	+	−	ND	(−)	+
17	+	+*	−	(+)	+
18	+	−	ND	(+)	−
19	+	ND	+	(+)	+
20	ND	ND	ND	ND	−
21	+	−	+	(+)	+
22	+	ND	−	ND	ND
23	−	+	ND	(+)	+
24	ND	ND	ND	ND	ND
Overall % Complexity	86	78	56	89	89

− indicates not full complexity; +, full complexity;

^*, complexity as compared to averaged control donors; ( ), CD4 and CD8 subsets not separated;

ND, not determined

Donor Chimerism and Flow Cytometric Analysis

During the peritransplant period after infusion of the CD34-enriched stem cell graft, all patients received ATG as part of GVHD prophylaxis. Patients were subsequently eligible to receive LLME-treated DLI when the serum ATG level was <2mcg/ml. Molecular analysis performed on bone marrow or PBMC indicated near complete T cell donor chimerism (>90%) for the five evaluable patients at the time of the sample collection for spectratype analysis. Furthermore, PBMC was collected from these five patients on a regular basis post-DLI, and flow cytometric analysis was performed to evaluate the overall T cell reconstitution. One patient presented here never recovered a CD3/CD4⁺ cell count >25/ul. The other five patients achieved a CD3⁺/CD4⁺ count >100 (median 163.9/ul, range 115.31–471.0/ul) at a median of 34 days post DLI. This was associated with a CD3⁺/CD8⁺ recovery that was also robust (median 608.8/ul, range 225.81–2304.96/ul).

Determination of Complexity Index Following DLI Infusion

To determine if there were any significant differences in the complexity of the reconstituting CD4⁺ and CD8⁺ T cell subsets, CDR3-size spectratype analysis was used to analyze PBMC samples from the evaluable patients after infusion of the LLME-treated DLI product. Spectratype histograms of the donor LLME-treated CD4⁺ T cell DLI product and the post-transplant sample for a representative Vβ family from each of three patients can be seen in Figure 2. In all three examples, there were no significant differences in the overall level of Vβ family complexity between the two samples. In regard to all of the data sets, although there were some individual Vβ family differences between a donor LLME-DLI sample and the post-transplant patient sample, the results for the four evaluable patients indicated mean overall complexity indices for the CD4 and CD8 compartments of 75% (SEM 5.5; Table 4) and 80% (SEM 3.5; Table 5), respectively. Additionally, for both T cell subsets there were no significant differences in the overall complexity indices (P≥ 0.29, CD4; P>0.71, CD8) between the LLME-DLI products at the time of infusion and the patient post-transplant. Furthermore, there was also no statistical difference between the overall complexity indices of the CD4 and CD8 compartments in the reconstituting repertoires (P>0.68). Taken together, these results suggested that LLME-treated DLI did not appear to undergo any biased reconstitution with regard to the complexity of the recovering subsets.

Figure 2.

CDR3-size spectratype analysis of CD4⁺ T cells enriched from PBMC of three patients after receiving LLME treated DLI. The spectratype analysis was performed as described in “Materials and Methods.” A representative Vβ spectratype is shown for each patient comparing the LLME-treated DLI to the reconstituting patient repertoire, post-transplant.

Table 4.

CD4+ Post-transplant Complexity Analysis

	Patient
Vβ	1	3	4	5
1	−	+	+	+
2	+	+	−	+
3	+	+	+	+
4	+	+*	+*	+
5	−	+	−	+
6	−	+	+*	+
7	−	+	−	+
8	+	+	−	+
9	+	+	−	+
10	ND	+*	+*	+
11	+	+	+*	+
12	ND	−*	ND	ND
13	−	+	+	−
14	+	+	−	−
15	−	+	−	+
16	+	+	+	+
17	+	+	−	+
18	+	+	+*	+
19	−	+	+	−
20	ND	ND	+	ND
21	+	−	+	+
22	−	+*	+	ND
23	−	+	+	+
24	+	+*	+*	ND
Overall % Complexity	57	92	65	85

− indicates not full complexity; +, full complexity;

^*, complexity as compared to averaged control donors;

ND, not determined

Table 5.

CD8+ Post-transplant Complexity Analysis

	Patient
Vβ	1	3	4	5
1	−	ND	+	+
2	+	ND	−	+
3	+	ND	−	+
4	+	+	−	+
5	−	ND	−	+
6	+	+	+	+*
7	−	+	+	+
8	+	+	+*	+
9	−	+	+	+
10	ND	+*	+	−
11	+	+	+	+
12	+	+*	ND	+*
13	+	+	+	−
14	−	−	+	−
15	−	+	+*	+
16	+	+	+*	+
17	+	ND	ND	+
18	−	+	+*	+
19	+	+*	+	−
20	ND	ND	ND	−*
21	+	+	+	+
22	ND	+*	+	+*
23	+	−	+*	+
24	ND	ND	ND	ND
Overall % Complexity	65	88	80	87

− indicates not full complexity; +, full complexity;

^*, complexity as compared to averaged control donors;

ND, not determined

Proliferative capacity of LLME-treated lymphocytes

To evaluate the proliferative capacity of LLME-treated PBMC in response to allostimulation, we utilized an in vitro one-way MLR assay. Responder PBMC from four healthy donors were mock- or LLME-treated according to the same protocol described above for patient sets, and then plated in a 96-well plate along with irradiated (30 Gy) PBMC stimulators from a fifth HLA-disparate healthy donor. Proliferative responses were determined after six days by measuring H³-thymidine incorporation following a 6-h pulse/label (1 μCi/well) period using unstimulated responder cells as the baseline. The mock-treated and the LLME-treated samples exhibited a 50-fold (range 31–84; median 41) and 56-fold (range 49–62; median 56) increase in proliferation, respectively, with no significant difference between the two groups (P>0.67; Figure 3). These results suggest that the in vitro proliferative capacity of PBMC in response to allo-stimulation is maintained after LLME treatment.

Figure 3.

Proliferative responses of mock- or LLME-treated PBMC. The immunocompetency of mock- and LLME-treated PBMC from four individuals was assessed in vitro by measuring [³H]-TdR incorporation in response to allo-stimulation.

Discussion

Rapid recovery of a functional immune system following allogeneic BMT is essential for its long term success as an enduring treatment for hematopoietic malignancies. Delayed immune reconstitution is largely responsible for the elevated levels of morbidity and mortality associated with the peritransplant period^25,26. T cell reconstitution is a vital part of this process. One potential mechanism utilized for the regeneration of the T cell repertoire consists of de novo expansion of thymic derived T cells generated from precursors found in the stem cell inoculum²⁷. This route of repertoire regeneration is confounded by both the delayed time frame in which T cell maturation and selection takes place, the diminished functional capacity of an oftentimes involuted thymus of an older patient, and as a consequence of chemotherapy and radiation induced damage to the organ²⁸.

An alternative route for the regeneration of immune competency would then consist of transplanting exogenous T cells as part of the donor inoculum or as a DLI. Homeostatic mechanisms may allow these mature T cells, and particularly CD8⁺ T cells, to expand in vivo and repopulate the lymphoid organs²⁹. The major caveat of this approach is of course, the complication of GVHD, which cannot be easily controlled without diminishing the needed T cell response capacity for GVL and protection from opportunistic infections⁵.

Experimental mouse transplantation models have suggested that to a large extent GVHD, in contrast to GVL effects, may be preferentially mediated by perforin-positive T cells^5,6,30. Several studies have examined the outcome of pan CD8⁺ T-cell depletion of the allograft prior to its administration as DLI, and while decreases in the incidence of GVHD have been reported, there is less information available regarding retention of GVL activity and the ability to mount immune responses to opportunistic infections^31,32. LLME-treatment of the DLI differs from pan CD8-depletion in that it utilizes a mechanistic pathway that is also likely to deplete any perforin-positive CD4⁺ T cells that could be involved in potentiating GVHD³³. Conversely, any remaining perforin-low or negative CD8⁺ T cells³² would still be available to mount GVL responses via other cytolytic mechanisms, such as fas-ligand or TRAIL interactions^{34,35,36,3738}, and would still be able to respond to antigenic stimulation with cytokine production, particularly IFN-γ, which can enhance inflammatory responses against the tumor cells. Syngeneic and HLA-matched sibling transplants performed by the CIBMTR have demonstrated convincing evidence of GVL effects for leukemia patients who had no clinical evidence of GVHD³⁹. This demonstrates clearly in man that while these two phenomena may overlap, it is possible to preserve/demonstrate GVL responses in the absence of GVHD. The clinical goal is to convert the other patients (those who do have GVHD) into patients with no (or at least less) GVHD while retaining some GVL effect for them. The goal of numerous preclinical and clinical investigations has been to try to reduce GVHD potential to a greater degree than GVL capability through pharmacoprophylaxis or depletion of one or another T cell subpopulation. Consistent with this intent, selective depletion of perforin-positive T cells with LLME might allow for reduced GVHD risk with retention of GVL responses and protection from opportunistic infection, while also allowing for the increased reconstitution of the CD4⁺ T cell compartment. Whether depletion of perforin-positive T cells will accomplish this goal can only be ascertained after completion of this trial.

The reestablishment of immunocompetency would likely be dependent on regenerating the diversity and complexity of the T cell repertoire. To this end the demonstration that LLME treatment of the DLI product did not cause or induce any biased depletion of any TCR specificities is of major importance in the successful use of this treatment in the clinic. Beyond the maintenance of a complex repertoire, the functional capability of the DLI product in response to antigen stimulation is an important consideration. The LLME-treated DLI was shown to be capable of eliciting proliferative responses in vitro, with MHC alloreactivity as an example. In this regard, we would hypothesize that in addition to response capability to any potential antigen, LLME-treated DLI can also still respond to minor HA or MHC alloantigens presented in the patient. Despite this retention of proliferative capacity, GVHD pathological development is diminished because of the lack of perforin-containing effector cells. In addition, we also observed that following LLME, there was a rapid reconstitution of the CD4⁺ and CD8⁺ T cell subsets. This reconstitution compares favorably with that observed in patients receiving unmanipulated DLI⁵ in addition to being associated with decreased levels of GVHD (Filicko et al manuscript submitted).

In conclusion, our results demonstrate that LLME treatment of DLI can be associated with differing degrees of T cell repertoire complexity but overall, does not impact the complexity of either the CD8 or CD4 T cell repertoires; neither does it significantly affect the overall post-transplant repertoire reconstitution. In addition, after LLME treatment the remaining T cells retain their ability to proliferate in response to allo-antigen stimulation. Taken together these results suggest that LLME might provide a more targeted manipulation of the DLI which is likely to afford advantages with regard to post-transplant immune reconstitution, mounting GVL effects and diminishing GVHD.