Direct evidence for new T-cell generation by patients after either T-cell–depleted or unmodified allogeneic hematopoietic stem cell transplantations: Presented in part in abstract form at the 42nd Annual Meeting of the American Society of Hematology, San Francisco, CA, December 2000.

Impaired reconstitution of T-cell–mediated immunity is a major cause of morbidity and mortality after allogeneic hematopoietic stem cell transplantation (allogeneic HSCT). T-cell regeneration after allogeneic HSCT can occur by several mechanisms, including thymic-dependent and thymic-independent pathways. Investigators have usually followed the recovery of specific T-cell subsets, such as CD4⁺, CD8⁺, and CD45RA⁺, to monitor immune reconstitution after allogeneic HSCT, as well as high-dose chemotherapy or infections such as HIV.1-7 Freedom from opportunistic infections (OIs) correlates strongly with numerical recovery of these and other phenotypically distinct lymphocyte subsets.1,2,6 7 Even when using rare event analysis for cytofluorographic assessments of circulating CD45RA⁺ T cells, however, the numbers of circulating T cells or T-cell subsets are just too low for detection during the early months after transplantation when patients are at greatest risk of complications from immune deficiency.

As progenitor T cells undergo T-cell receptor (TCR) gene rearrangement in the thymus, chromosomal sequences are excised to produce episomal DNA byproducts termed TCR rearrangement excisional circles (TRECs). TRECs in circulating T lymphocytes therefore mark these cells as recent thymic emigrants (RTEs).8 TRECs persist but decrease with age. TRECs can also increase after effective therapy for HIV infection or after thymic transplantation.8-12 Because TRECs are nonreplicating episomal DNA, TRECs are lost only by cell death, dilution through cell division, or both.13 TRECs therefore represent the composite production, proliferation, and death of RTEs.9,10,14 15 

Our study was designed to determine the mechanism and kinetics of T-cell regeneration after allogeneic HSCT, especially in a series dominated by older adult recipients of T-cell–depleted allografts. This group has long posed a quandary regarding immune reconstitution, as to whether there are host limitations to T-cell regeneration and/or graft limitations from the quantity and quality of transplanted T-cell precursors. TRECs offer a means of directly addressing these important unknowns. TREC measurements are also potentially more sensitive than immunophenotypic monitoring, especially when circulating T cells are numerically low or undetectable by this method early after transplantation. We therefore evaluated the relation of TRECs to other host and transplant factors such as donor and recipient age, type of graft, degree of HLA matching, clinically significant graft-versus-host disease (GvHD), and radiation dose. We also assessed the physiologic significance of measured TRECs with respect to the presence or absence of severe OIs.

We obtained 5-mL peripheral blood samples from consecutive, consenting patients who were in the hospital at least 1 month after transplantation and engrafted, or who were returning for regularly scheduled outpatient followup after allogeneic bone marrow or granulocyte colony-stimulating factor (G-CSF)–elicited peripheral blood stem cell transplants (Table1). Regardless of whether the regimen included total body irradiation or not, all patients received myeloablative conditioning. This series does not include any adult patients treated with nonmyeloablative preparative regimens for allogeneic HSCT. Most of the patients received transplants for lymphohematopoietic malignancies. All patients who received unmodified allografts also received standard pharmacologic prophylaxis against GvHD with cyclosporin or tacrolimus plus methotrexate. T-cell–depleted allograft recipients received no additional GvHD prophylaxis, but 106 (84%) of 127 T-cell–depleted allograft recipients received a brief course of antithymocyte globulin (ATG) and steroids in the peritransplantation period to prevent graft rejection.16-20 HLA allelic mismatching was based on DNA nonidentity at any one of the 6, codominant HLA-A, -B, or -DRβ1 loci. All patients were in hematologic and/or cytogenetic remission and were fully donor engrafted at the time of sampling. Patient sampling was conducted in accordance with Institutional Review Board sanctioned protocols.

Peripheral blood mononuclear cells (PBMCs) were isolated from the interface after centrifugation of approximately 5 mL whole venous blood on Ficoll-Paque Plus (Amersham Pharmacia Biotech, Piscataway, NJ). Cells were frozen in a pellet at −80°C until DNA extraction. Frozen cell samples were thawed on ice, and genomic DNA was extracted by using a QIAamp DNA Blood Mini Kit (no. 51106; QIAgen, Valencia, CA) according to the manufacturer's instructions. The DNA from each processed sample was then frozen at −80°C until tested for TRECs. TRECs and the corresponding number of cell equivalents were quantified by using real-time polymerase chain reaction (PCR).9 A molecular beacon, designed to recognize a region upstream from the signal joint, was included in the reaction mixture throughout PCR to serve as a real-time detector for the amplified product. This beacon was specifically designed to have a hairpin structure, with a 6-bp stem and a 26-nucleotide target-recognition loop, plus a fluorophore [FAM (6-carboxyfluorescein)] and quencher [DABCYL (4-dimethylaminophenlazo benzoic acid)] in close proximity at the 2 ends of the oligonucleotide.

The PCR reaction was conducted in a spectrofluorometric thermal cycler (ABI PRISM 7700; Applied Biosystems, Foster City, CA) that monitors changes in the fluorescence spectrum of each reaction tube during the annealing phase, while simultaneously carrying out programmed temperature cycles. The cycle number during PCR that yields a fluorescent intensity significantly above the background is designated as the threshold cycle (C_T). The C_Tis directly proportional to the log of the copy number of the target sequence in the input DNA. We have previously reported that the assay is efficient (< 4 hours), specific, sensitive (100 copies), dynamic (7-log range), and accurate (10% coefficient of variance).9 

Immunophenotyping used fluorochrome-conjugated antibodies to CD3, CD4, CD8, CD45, and CD45RA, after which cells were analyzed on a FACScan (BD Pharmingen, San Diego, CA).2Cytomegaloviremia (CMV antigenemia) was detected by positive immunofluorescent staining of the pp65 antigen (Chemicon, Temecula, CA) in circulating leukocytes. CMV disease was defined by identification of CMV in the liver, gut, lungs, and/or other tissue site, together with circulating CMV antigenemia. Table2 lists additional severe OIs and their incidences. Clinically significant acute GvHD included overall grade II to IV disease,21 and clinically significant chronic GvHD comprised extensive but not limited disease.22 

TRECs were quantified per 10⁶ input PBMCs. All values less than 100 were normalized to 100 for purposes of further calculations and analysis, because the valid detection limit of the assay was 100 TRECs/10⁶ PBMCs. Almost all patients had simultaneous lymphocyte phenotyping as above at the time of TREC sampling. Excluding those patients who had undetectable CD3⁺ T cells (TREC values relative to CD3⁺ T cells could not be calculated when CD3⁺ T-cell denominators were zero) and the rare patients whose CD3⁺ T cells were not simultaneously assayed with TREC sampling, TRECs were also calculated per 10⁶ CD3⁺ T cells [TRECs per million CD3⁺ T cells = (TRECs per 10⁶PBMCs)/(percent CD3⁺ T cells/100)]. These calculations and analyses generated similar curves and P values to those of TRECs per 10⁶ PBMCs, so these findings have not been separately illustrated.

Stratified permutation tests were performed using the Wilcoxon rank sum statistic to determine the level of association between TRECs and prognostic factors. The stratification variable was the time after transplantation at which the TREC value was measured. The test statistic was stratified into time intervals so that TREC (rank) values could be compared at comparable time periods. Although there were multiple observations for more than half the patients, the permutation procedure was performed at the patient level to maintain the correlation structure induced by multiple observations. For the purpose of this stratified analysis, the time since transplantation was divided into 7 groups: 1 to 3 months, 3 to 6 months, 6 to 9 months, 9 to 12 months, 12 to 24 months, 24 to 48 months, and more than 48 months.

To depict the graphic relationship between TRECs and the prognostic factors studied, controlling for the elapsed time since transplantation, a nonparametric smoothing procedure was performed.23 The figures depict all data points, but the connecting curves have been truncated for groups in which data became too sparse to allow a valid fit.

The amount of variation in TRECs per 10⁶ PBMCs predicted by the percentage of circulating CD4⁺ CD45RA⁺ T cells and the time since transplantation were summarized by using the goodness of fit measure r². This statistic is based on a nonparametric generalized additive model24 that relates these 3 variables and is similar to ther² statistic used to summarize the goodness of fit in linear regression models. The statistic varies between zero and 1, with high r² values indicating that an accurate prediction of an individual TREC outcome can be obtained through knowledge of an individual's percentage of circulating CD4⁺ CD45RA⁺ T cells and the time since transplantation that it was recorded.

TRECs were significantly higher in those patients who were younger than 19 years at the time of allogeneic HSCT, compared with those 19 years and older (P = .02; Figure1). Even adults, however, had detectable TRECs after myeloablative and lymphoablative conditioning for allogeneic HSCT. There were too few pediatric recipients of unrelated adult donor grafts and insufficient disparity in the overall group between recipient and donor age to detect an independent effect of donor age on TRECs.

We compared circulating TRECs in recipients of T-cell–depleted and unmodified (T-cell replete) allografts. Of the T-cell–depleted allograft recipients, 83%, who were all adults, had received ATG for the prevention of graft rejection. Recipients of unmodified allografts were treated with standard pharmacologic prophylaxis against GvHD, using methotrexate and either cyclosporine or tacrolimus. TRECs were initially more numerous in the recipients of unmodified allografts (P < .01; Figure2), but this difference abated beyond 9 months after transplantation. Furthermore, when pediatric patients were excluded and only adults aged 19 years and older at the time of transplantation were analyzed, the relationship between TRECs and type of graft (unmodified versus T-cell depleted), controlling for elapsed time from transplantation, diminished (P = .07 instead ofP < .01). Adult recipients of either type of transplant recovered TREC values during the second year after allogeneic HSCT that were as high or higher than those measured in historical healthy controls9 (median TRECs per 10⁶ PBMCs in adults aged 19-62 years were 3620 in healthy donors,9 1087 in patients between 9 and 12 months after transplantation, and 10 853 in patients between 12 and 24 months after transplantation). There were unfortunately too few data points from patients younger than 19 years to make similarly valid comparisons.

The only naive T-cell phenotype followed in this series was coexpression of the CD45RA isoform by CD4⁺ T cells. The absolute number and the percentage of CD4⁺CD45RA⁺ T cells were evaluated, and the percentage values were more predictive of the TREC values. Eighty percent of the CD4⁺ CD45RA⁺ values were 5% or less, and 1% was the median. We therefore compared TRECs between samples dichotomized by the median of 1% CD4⁺CD45RA⁺ T cells. Higher TRECs indeed predominated in those samples with more than 1% CD4⁺CD45RA⁺ T cells (P < .01; Figure 3). The variation in TRECs attributable to the time after transplantation and the observed percentage of circulating CD4⁺CD45RA⁺ T cells was only 0.45, however. Differences in the proportion of T cells expressing the naive, CD4⁺ CD45RA⁺ surface phenotype therefore could not predict more than half (55%) of the variation in measured TRECs (r² derived from the generalized additive regression model). The percentage of circulating CD4⁺ CD45RA⁺ T cells exerted a 7.5-fold greater contribution to the predicted TRECs in samples obtained more than 9 months after transplantation, compared with those obtained earlier.

We compared TREC values from patients who had received allogeneic HSCTs from genotypically HLA-identical, related donors with those who had received allogeneic HSCTs from either HLA-nonidentical, related donors or any unrelated donors (Table 1). Unrelated donors, even if HLA-identical by DNA, are by definition phenotypically but not genotypically identical to their recipients because of different parentage. There was no effect on TRECs exerted by the presence or absence of genotypic HLA identity between host-donor pairs (P = .79; Figure 4).

TREC values were compared between patients who did or did not develop clinically significant GvHD. We did not identify an association between TRECs and acute GvHD. In contrast, the presence of extensive chronic GvHD was strongly associated with low TRECs (P < .01; Figure 5). This association is despite the fact that only 10 patients in the overall series had extensive chronic GvHD, 2 of whom had inactive disease at the time of sampling. One case of extensive chronic GvHD had occurred after donor leukocyte infusions, 4 had occurred after unrelated allografts, and one had occurred after a mismatched related allograft. Nine of the 10 cases were in adults.

TRECs were also compared between patients who did or did not develop severe OIs (Table 2). Of the 58 individuals who developed a severe or life-threatening OI, 8 individuals had more than one OI; but analyses were based only on whether an individual patient did or did not have a history of any severe or life-threatening OI. Both children (younger than 19 years) and adults (19 years and older) who developed higher numbers of TRECs over time from transplantation were less likely to suffer from severe OIs. Conversely, low TRECs were associated with the presence of severe OIs (P < .01; Figure6A). The relationship between OI and TREC values remained strong (P < .01) after adjustment for other factors (eg, age, CD4⁺ T-cell count, CD4⁺CD45RA⁺ T-cell count, type of transplant, and presence or absence of chronic GvHD).

Among the various OIs seen in this series, the most common was reactivation of CMV detected in circulating leukocytes (CMV antigenemia). Actual CMV infection or disease also occurred, as did chronic or refractory CMV antigenemia, but these were included in the above analysis in Figure 6A. These events were relatively rare within the overall group with reactivated CMV, given the current clinical practice of routine surveillance for earlier detection and preemptive antiviral therapy of CMV antigenemia. The occurrence of CMV antigenemia and/or its treatment with antiviral drugs was nevertheless also associated with low TREC values (P = .05; Figure 6B).

The percentages of circulating CD4⁺CD45RA⁺ T cells were also compared between patients who did or did not develop OIs after transplantation. Lower percentages of circulating T cells expressing this naive phenotype correlated strongly with the presence of severe OIs (P < .01) and slightly less so with CMV antigenemia or disease (P = .06). Again, adjustments for other potentially contributing factors did not affect the association between OIs and the percentage of circulating CD4⁺ CD45RA⁺ T cells.

Successful reconstitution of T-cell–mediated immunity substantially reduces morbidity and mortality after allogeneic HSCT. We therefore used TRECs to investigate potential mechanisms of immune reconstitution after allogeneic HSCT. In a series dominated by adult recipients of T-cell–depleted allografts, most of whom also received total body irradiation and ATG, active T-cell neogenesis did occur; TRECs recovered at least age appropriate levels by the second year after transplantation. Measurable deficiencies in TREC values significantly correlated with clinical morbidity and mortality, especially during the early posttransplantation period of greatest risk for severe OIs. This finding was in contrast to the detection of circulating CD4⁺ CD45RA⁺ cells, often used as an indicator of new T-cell generation, but whose recovery actually heralds a decreasing risk of severe OIs.1,2 6 

Hosts have no measurable T cells after myeloablative preparative regimens for allogeneic HSCT. Only newly synthesized T cells or proliferation of T cells transferred with the allograft could account for any real expansion of T cells. Most of the patients in this series, however, received T-cell–depleted allografts comprising fewer than 10⁵ clonable T cells per kilogram at the outset.25 This limits the initial pool for T-cell expansion and dilution of TRECs, independent of T-cell neogenesis. It also contrasts with the higher numbers of residual T cells after other depletion methods or the approximate dose of 10⁷ clonable T cells per kilogram administered to unmodified transplant recipients. Given that TRECs are undetectable at the time of transplantation, alterations in T-cell proliferation alone would not explain changes in the absolute numbers of TREC⁺ cells. The observed increases in TREC levels must therefore have been associated with the generation of new donor-derived T cells, notwithstanding the influence of T-cell activation and proliferation on TREC measurements after initial T-cell repopulation.15 

Other investigators have measured TRECs to evaluate immune recovery after autologous26 or allogeneic HSCTs,27-30but most of these studies have focused on children and young adults. Recent work has demonstrated a reduction in T-cell proliferation in the steady state, contrasted with increased proliferation in response to chronic antigenic stimulation associated with infection or GvHD after allogeneic HSCT.15 This research highlights the influence that clinical status can have on measured TRECs and reinforces the paradigm that TREC levels represent new T-cell production counterbalanced by the competing factors of either T-cell death or T-cell proliferation that dilutes TRECs among the progeny, or both.

The series reported here is unique, however, in its large number of samples; a 4:1 preponderance of T-cell–depleted over unmodified allograft recipients; the exclusive use of myeloablative conditioning regimens for transplantation of mostly lymphohematopoietic malignancies; and the presence of full donor T-cell chimerism in all patients at the time of sampling. There were also several variables that we anticipated might have negatively biased the results but did not. For example, approximately half the adults were 35 years or older at transplantation. Most adults also received T-cell–depleted allografts, and 83% received prophylaxis against graft rejection with ATG that could have destroyed RTEs because of its persistence in the circulation. Despite this, adult patients recovered age-appropriate and possibly even higher levels of TREC⁺ cells than age-matched historical healthy control subjects9 during the second year after transplantation.

There was a strong association between TRECs and the percentage of circulating CD4⁺ CD45RA⁺ lymphocytes, but this association was almost entirely limited to samples obtained beyond 9 months after transplantation. Use of this phenotype as a surrogate approximation of T-cell neogenesis can therefore confound accurate estimates of thymic activity. There were in fact patients early after transplantation who had detectable TRECs but no detectable circulating T cells by flow cytometry. The subsequent recovery of CD4⁺CD45RA⁺ T cells heralds the time after transplantation when the risk of OI markedly decreases,2 6 rather than providing a discriminating correlate during the period of greatest risk.

Other variables beyond coexpression of CD4 and CD45RA, either not observed or not assessed in this study, could also have contributed to measured TRECs. Possibilities include T cells expressing additional naive phenotypes such as CD8⁺CD103⁺³¹ and/or CD8⁺CD45RA⁺; additional surface markers such as CD62L, CD11a, CD27, and CD2832; or Ki67, indicating active proliferation and dilution of TRECs among the T-cell progeny.15 The CD4⁺ CD45RA⁺phenotype could also overestimate T-cell neogenesis, as T cells expressing the CD45RA isoform may expand in the absence of specific antigenic stimulation and maintain CD45RA expression.33CD45RA⁺ T cells may also be phenotypic revertants from CD45RO⁺ memory T cells.34 

Higher TREC values were indeed associated with a decreased incidence of CMV antigenemia and a decreased incidence of severe OIs overall. Conversely, lower TREC values were associated with a significantly higher incidence of these complications. We considered a number of other factors (eg, age, CD4⁺ T-cell count, CD4⁺CD45RA⁺ T-cell count, type of transplant, and presence or absence of chronic GvHD) that might affect the relationship between severe OI and TREC values, but none of these factors independently influenced this association. On the basis of these data alone, we cannot distinguish whether the lower TREC levels result from severe OIs or alternatively reflect graft-host interactions independent of infection that decrease TRECs and increase infection risk in the posttransplantation period.

That TRECs recovered to levels as high or higher than historical healthy controls in adult patients could indicate rapid entry of TREC⁺ naive T cells into a virtually empty T-cell pool, which would markedly increase the TREC concentration without a meaningful increase in the absolute TREC number. Alternatively, a “supranormal” TREC concentration may reflect both robust T-cell neogenesis and a concomitant reduction in T-cell proliferation over time.15 Although restoration of TCR diversity was not specifically evaluated in this series, such recovery of the T-cell repertoire has been associated with the ability to respond to vaccination6 and a lower susceptibility to OIs.2 

Despite the low incidence of extensive chronic GvHD in this study, there was a strong association between this posttransplantation complication and low TRECs as previously reported.28Extensive chronic GvHD is deleterious to clinical immune competence,22 and mouse data have demonstrated GvHD-induced injury to host lymphoid organs required for reconstitution of T-cell immunity.35 In addition, the chronic T-cell stimulation and turnover associated with extensive chronic GvHD could decrease TREC numbers.15 

In contrast, we did not ascertain an independent effect of radiation dose used in the preparative regimen (1350-1500 cGy range), which we had hypothesized might affect host lymphoid niches for regenerating T cells. In the absence of lymphoid sites where new T cells can encounter antigen and undergo proliferation and dilution of TRECs, values may remain misleadingly elevated. We did at least confirm in vitro, however, that engrafted T cells have the capacity to divide and decrease their TREC content in response to stimulation with a polyclonal mitogen (data not shown).

Comparisons between T-cell–depleted and unmodified allograft recipients demonstrated a more significant increase in TREC levels after unmodified allografts, but only in the first 9 months after transplantation. When adults only were examined, however, the relationship between TRECs and type of transplant diminished (P = .07 versus P = .02). Age may therefore have contributed only to the early differences between type of transplant and TRECs, because of the disproportionate representation of children in the unmodified transplant group and the preponderance of adults in the T-cell–depleted transplant group. This finding is consistent with prior studies from our institution in which children have always reconstituted normal immunity earlier than adults regardless of type of donor or transplant.1 2 There were too few pediatric recipients of unrelated adult donor grafts and insufficient disparity in the overall group between recipient and donor age to detect an independent effect of donor age on TRECs.

Because pediatric patients reconstitute normal cellular immunity and achieve a lower risk status for complications earlier than adults, most return to their referring physicians for followup sooner than adults. Accordingly, there were very few pediatric samples after 24 months from transplantation, given the manner in which we sampled consecutive, consenting patients returning for regular followup at this center. When we analyzed each parameter over the entire sampling period but included only adults, almost identical curves and similar P values were obtained as for the whole population (not shown). There was one expected exception, which was the relationship between TRECs and type of transplant that was influenced by the inclusion of children.

Donor leukocyte infusions (DLIs) administered in this series were predominantly for molecular or cytogenetic relapse of CML and were dosed according to the content of bulk CD3⁺ T cells. Only 12% of the patients in this series received DLI, however, and these patients only accounted for 10% of the total samples. We detected no influence on TRECs, either because the numbers were too small or because DLI really exerted no effect.

Finally, we considered whether late sampling after transplantation could have selected for patients surviving long enough to generate higher TRECs. This is unlikely, however, because in all situations in which substantive differences in TRECs were noted, the differences occurred within the first 9 to 12 months rather than later.

In conclusion, these results demonstrate that adult recipients of T-cell–depleted allografts, most of whom also received total body irradiation and ATG, can actively generate new T cells based on increasing TREC levels. Recipient hosts of all ages did in fact increase their TREC values after transplantation of either unmodified or T-cell–depleted allografts. Low TRECs and low percentages of circulating CD4⁺CD45RA⁺ T cells were strongly associated with an increased incidence of severe OIs. TRECs were measurable, however, even before flow cytometry could detect circulating CD4⁺CD45RA⁺ T cells. TREC values are therefore an important correlate for severe OIs during the period of greatest risk, and their predictive value in this setting warrants further prospective evaluation. Extensive chronic GvHD, although limited in incidence in this study, also had a strong negative association with TRECs, consistent with GvHD-mediated destruction of lymphoid niches for the regeneration of T-cell immunity35or with chronic T-cell stimulation and turnover,15 or a combination of both processes.

Select individuals with low TREC values may therefore benefit from targeted interventions to hasten new T-cell production or from preemptive antimicrobial therapy against severe OIs when they are at greatest risk. The active generation of new donor-derived T cells also justifies measures to enhance acquired cellular immunity against residual disease or opportunistic microbes. These hypotheses warrant prospective evaluation in clinical trials.

We thank Frieda Toomasi, RN, the nurses, and attending physicians of the Allogeneic Bone Marrow Transplantation Service and Drs Jan Baggers and Gudrun Ratzinger, all at MSKCC, for assistance with sample procurement and processing.